article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
dilute low dimensional systems have been the focus of a great deal of recent attention due to their interesting physical properties at low temperature@xcite .
a number of fascinating phenomena have been reported for low - density electron systems in high mobility silicon inversion layers as a function of magnetic field . a dramatic increase of the resistivity in response to in - plane magnetic field
has been shown to be associated at high electron density with complete spin polarization of the carriers @xcite .
a number of experiments have recently shown that the spin susceptibility increases substantially as the density is decreased @xcite , indicating a possible divergence and ferromagnetic instability at finite electron density . through a detailed study of the magnetoconductivity as a function of temperature and electron density ,
we have identified an energy scale @xmath3 associated with the response of the electrons to a magnetic field applied parallel to the plane of the electrons @xcite .
the density - dependent energy @xmath3 was found to go to zero at a finite density @xmath4 , signaling critical behavior and the occurence of a zero - temperature quantum phase transition of magnetic origin .
microwave radiation may be an interesting tool for probing the behavior of these systems . at low frequencies @xmath5 the system
is expected to behave as a correlated spin liquid , while at high frequencies @xmath6 the behavior will be similar to that of an noninteracting gas . in response to microwave radiation of frequency @xmath7 , one may find particularly interesting behavior at the boundary between two metallic regions which have characteristic energies @xmath8 and @xmath9 if @xmath10 , corresponding to transport between a correlated spin liquid and a noninteracting gas . in this paper
we report unusual high frequency behavior of dilute 2d electron systems in multigated high mobility si - mosfet s .
we investigated the nonlinear microwave response of the boundary between two regions of the silicon inversion layers with different electron densities .
a rectified signal is observed whenever the electron densities @xmath1 of the two metals are different .
the polarity of the rectified signal can be changed easily by varying the electron density on the two sides of the boundary .
we find that the nonlinear response of the boundary between the two metals is unusually strong when one of the metals is close to the quantum phase transition reported earlier @xcite while the other is kept at high electron density - a condition which corresponds to @xmath11 .
we used multigated si - mosfet s with six different contacts , shown schematically in fig .
the 2d electron densities @xmath12 corresponding to different contacts were separately controlled by independent gates ( contact gates ) .
each 2d metal system was adjacent to a common 2d conducting region of density @xmath13 , also controlled independently by a main gate .
the region between the main gate and each contact gate was in the form of a thin - line split .
the narrow split between gate metallizations was obtained by reactive ion etching@xcite .
the typical width of a split ( @xmath14 nm ) was less than the thickness of the si oxide insulating layer ( @xmath15 nm ) , providing a smooth , probably monotonic , profile of electron density between two 2d layers with different gate voltages , as shown in fig . 1 ( c ) .
the orientation of two splits , shown as vertical lines on the surface of the structure in fig .
1(a ) , was perpendicular to the orientation of the remaining four horizontal splits .
the two contacts corresponding to the vertical splits were permanently connected to ground .
0.5 cm [ 1 ] by varying the frequency of the microwave radiation , which changes the configuration of the electromagnetic field near the sample , it was possible to vary the relative contributions of vertical and horizontal splits to the measured nonlinear signal .
measurements were taken at a frequency near @xmath16 ghz , where the nonlinear contribution from the vertical splits was negligibly small ( less than @xmath17% ) .
thus , the individual nonlinear properties of each horizontal split were studied in the experiment .
fig.1 ( b ) shows a simplified diagram of the experimental setup near the horizontal split ; the two two - dimensional metals with electron density controlled independently by gates # @xmath18 and # @xmath19 are connected through the conducting region under the split .
the rectified signal , a dc voltage @xmath0 , is observed in response to a microwave electric field @xmath20 applied to the structure as shown in fig.1 .
the voltage @xmath0 was measured with respect to ground using a high input impedance ( 10 g@xmath21 ) voltmeter .
-3.5 cm [ 2 ] dc currents flowing through the system were thus exceedingly small , and the odd - order nonlinear signals ( @xmath22 ... ) of the different contacts provided negligible contributions to the measured rectified signal .
microwave radiation of frequency @xmath23 ghz was provided by a loop antenna maintained at the end of a microwave coaxial line .
the sample was placed near the radiation loop at a distance of about @xmath18 cm .
the axis of the microwave loop was approximately parallel to the vertical splits , shifted from them by approximately @xmath24 cm along the sample plane .
the sample sizes are @xmath25m@xmath26 . a microwave yig generator with electronically controlled frequency supplied a regulated microwave power of about @xmath18 mw ( @xmath27 db ) at the end of the coaxial line .
the sample was mounted on a rotating platform at the end of a low temperature he-@xmath28 probe .
measurements were taken in an oxford he-@xmath28 system in the temperature range @xmath29 k for different microwave power in magnetic fields up to @xmath30 t applied parallel to the plane .
the multigated si mosfet samples used in these experiments had electron mobilities of about @xmath31 m@xmath26/vs at @xmath32 k. three different contacts were investigated ; all demonstrated similar non - linear properties . 0 cm [ 3 ]
2 shows the dependence of the rectified signal on the electron density of two adjacent 2d metals .
the density of electrons in the main 2d area is fixed by the main gate at some value , as labeled , while the electron density in the contact area was varied .
the rectified signal depends strongly on the electron density , changing polarity when the electron densities of the two metals are nearly the same .
the insert shows the relation between the two densities @xmath33 and @xmath13 for which the signal changes sign .
the correlation observed between the electron densities on different sides of the boundary demonstrates that the microwave rectification is generated by a region near to the boundary between the two 2d metals ; there is almost no rectification when the two metals have the same electron density .
this implies that the microwave rectification by the 2d metals @xcite is much smaller than the rectification associated with the boundary between them . the symmetry shown in fig .
3 provides additional support for this conclusion .
3 demonstrates an inversion of the density dependence of the rectification when the gates are interchanged .
curve ( a ) corresponds to measurements when the main gate voltage was fixed and the contact gate voltage was varied while curve ( b ) shows the results obtained for a fixed voltage on the contact gate when the main gate voltage was varied .
the almost perfect symmetry observed when the gates are interchanged demonstrates directly that the observed rectification originates near the boundary between the two two - dimensional metals . at fixed difference of electron densities of the two metals the magnitude of the rectified signal decreases considerably with increasing density ( see fig .
this decrease indicates that the mismatch in physical properties responsible for the nonlinearity at the boundary between the two metals becomes relatively small at high electron densities .
the rectified signal exhibits different behavior as a function of microwave power at low @xmath34 k ) and high ( @xmath35 k ) temperatures . the nonlinear response is weaker at high temperature than at low temperature . at high temperature ( t=@xmath35 k )
the rectified voltage is proportional to the square of the microwave electric field @xmath36 at small microwave power ( @xmath37 - 30 db ) , indicating that this is the weak ( perturbative ) nonlinear regime ( see fig.4 ( a ) ) .
however , at higher microwave excitation ( @xmath38 db ) strongly nonlinear behavior is observed : @xmath39 . at low temperature
the rectified signal is roughly proportional to the square root of the microwave power @xmath40 at all available powers if one of the two metals is kept at a low electron density ( @xmath41 @xmath42 ) , while the other is held at high electron density ( see fig . 4 ( b))@xcite .
such strong nonlinearity of the conducting boundary between two metals is unusual .
estimates based on the boltzmann equation ( see below ) show that a weak , perturbative nonlinear response is expected over the entire range of microwave powers used in our experiments .
the strong nonlinearity @xmath43 of the boundary is similar to the strong nonlinear ( high power ) behavior of an ordinary diode @xcite . using the analogy with a diode
, we can estimate the amplitude of the microwave voltage @xmath44 across the boundary .
the current voltage characteristic of a diode is @xcite @xmath45 \eqno{(1)},\ ] ] where @xmath46 is the reverse current and @xmath47 is the temperature .
below we will assume that the @xmath48 characteristic of the boundary between two metals is similar to that of a diode .
the high power ( or strongly nonlinear ) regime corresponds to the condition @xmath49 . [ 4 ] in the presence of the microwave field , the voltage drop across the boundary is @xmath50 , where @xmath51 ( @xmath52 is the thickness of the boundary ) is the voltage due to the microwave radiation and @xmath0 is the measured dc rectified voltage . in order to yield zero dc current
, the dc voltage @xmath0 must be comparable with @xmath53 at @xmath49 .
this follows directly from an estimate of the average ( dc ) current @xmath54 in the strongly nonlinear regime , ( @xmath49 ) , using the diode @xmath48 curve ( see eq .
this conclusion should be valid qualitatively for any other strongly nonlinear @xmath48 curves .
thus , the magnitude of the microwave voltage drop across the boundary between two metals should be of the order of the rectified signal in the high power regime : @xmath55 . according to eq .
( 1 ) the crossover to the perturbative regime ( @xmath56 ) should appear at @xmath57 , which gives @xmath58 at the crossover .
the rectified voltage @xmath59 at the crossover ( see fig .
4 ( a ) ) .
therefore the crossover temperature is @xmath60 k , which is of the same order as the temperature of the experiment - @xmath61 k ( see fig .
4 ( a ) ) . as shown in fig .
4 , the rectified signal @xmath0 is a weak function of temperature in the strongly nonlinear regime . in the perturbative nonlinear regime ( @xmath62 )
the microwave rectification increases substantially with decreasing temperature @xmath47 .
5 shows the temperature dependence of the rectified signal at different electron densities , as labeled . -3
cm [ 5 ] 0.5 cm [ 6 ] the data were obtained at finite magnetic field to avoid a change ( about 20%- 30% ) of the microwave field @xmath20 near the sample due to the superconducting transition of the al gate at @xmath63 k. the increase of the nonlinearity with decreasing electron temperature indicates that the conductivity itself is not the relevant parameter responsible for the rectification .
the electrical charge accumulation near the boundary due to the microwave field , one of the possible mechanisms responsible for the rectification , should be less important at lower temperature due to the higher conductivity of the dilute 2d system at low @xmath47 .
the saturation of the temperature dependence at low temperature may be related partially to overheating of the electrons by the microwave radiation as well as to the transition from the regime where @xmath64 to the regime where @xmath65 ( see fig .
4 ) . the insert to fig . 5 demonstrates the high temperature behavior of the signal . at high temperature
the rectification is proportional to @xmath66 at different electron densities .
the @xmath67 dependence of the rectification is different from the @xmath68 dependence expected for a regular diode . using the parameter @xmath69 as a small perturbation in eq .
( 1 ) , the rectified signal was found to be : @xmath70 .
the dependence of the rectified signal on in - plane magnetic field is shown in fig . 6 for different electron densities at two different temperatures .
the contact gate area was kept at high electron density @xmath71 @xmath42 .
different curves correspond to different electron density under the main gate , as labeled .
the direction of the magnetic field is parallel to the boundary between the two metals .
the rectification is a nonmonotonic function of in - plane magnetic field .
the arrows shown in fig . 6 indicate the magnetic fields corresponding to complete spin polarization of the electrons under the main gate@xcite . the magnetic field of about @xmath30 t was not sufficient to fully polarize the high density of electrons under the contact gate @xcite .
the mechanism that gives rise to the observed rectification is unknown and is the subject of current investigation .
although they are expected to be important , it is currently unclear how to include the effect of electron - electron interactions . in this section
we first consider a simplified model of non - interacting electrons and show that these simple considerations do not account for our observations .
we then speculate about other possible explanations .
the observed rectified signal depends on the difference between electron densities in the two 2d layers ( see fig.2 ) .
almost perfect symmetry of the effect with respect to exchange of the gate voltages ( see fig.3 ) indicates that the rectification is generated near the boundary separating the two 2d electron systems .
it is well known @xcite , that there is an internal electric field near the boundary separating two metals .
this electric field creates a contact difference of potentials @xmath72 .
the difference in contact potentials is equal to the difference of the work functions or chemical potentials @xmath73 of the metals considered independently ( in other words , when the metals are not connected to each other ) . at thermodynamic equilibrium
the chemical potential has to be the same when the system is connected .
thus , the potential difference @xmath74 should be generated near the boundary .
let us assume that the contact difference of potentials @xmath74 is the main reason for the rectification near the boundary between the two metals .
this suggestion is completely consistent with the symmetry of the effect with respect to exchange of the gate voltages ( see fig.3 ) .
below we estimate the rectification of the microwave radiation near the conducting boundary between two normal 2d metals using the boltzmann equation . to simplify the following calculations we introduce a model where a pair of 2d metals contiguous through the conducting boundary is replaced by a single metal of strongly inhomogeneous electron density ( see fig . 1 ( c ) )
the inhomogeneity along the direction perpendicular to the boundary between the original 2d metals is created by the voltages applied to the gates .
we choose a coordinate system where the boundary lies along `` y '' axis .
then both the electron density and the chemical potential depend on `` x '' @xmath75 where @xmath76 is the electron density of states at the fermi surface ) .
the boltzman transport equation for the electron distribution function @xmath77 is taken within the relaxation time approximation : @xmath78 where the function @xmath79 describes the equilibrium state of the electron system in the absence of the microwave radiation @xmath80 and equals the fermi distribution function : @xmath81 following [ 9 ] we expand @xmath82 in harmonic polynomials , keeping the first three terms of the expansion .
such an approximation is justified for a weak microwave field because the next terms of the expansion give smaller corrections in terms of the microwave field magnitude . substituting the expansion into the boltzman equation we get a set of equations : @xmath83 @xmath84 @xmath85 here @xmath86 are the coefficients of the expansion of the electron distribution function in harmonic polynomials
; @xmath87 is the electron effective mass , @xmath88 the term @xmath89 in ( 4 ) describes the internal `` electrochemical '' field arising from the inhomogeneity of the electron density . in the absence of the time
dependent field @xmath90 , this term balances the external field produced by the gate voltages and provides equilibrium of the electron system . solving eqs .
( 4 ) , ( 5 ) , we can calculate the dc contribution to the electron current density .
this rectified current is generated perpendicularly to the boundary ( along the `` x '' direction ) . in the present experiments , where the microwave frequency @xmath7 is smaller than the effective collision frequency @xmath91 we arrive at the simple expression for the rectified current : @xmath92 which enables us to estimate @xmath93 as follows : @xmath94 here @xmath95
is the drude conductivity corresponding to electron density @xmath96 and relaxation time @xmath97 it was observed in the present experiments that when the minimum @xmath98 is larger than a certain value @xmath99 @xmath42 ) the relaxation time @xmath100 is nearly independent on the electron concentration and can be treated as a constant in carrying out the integration over `` x''in ( 7 ) . using this approximation we get : @xmath101 these results based on the boltzman transport equation are in agreement with the experiments in several important respects .
first , it follows from ( 6)(8 ) that rectification can take place only when there exists a difference in the electron densities of the 2d metallic regions separated by a conducting boundary .
secondly , the above considerations give a correct relation between @xmath93 and @xmath102 for small power levels of the microwave field @xmath103 .
there are , however some discrepancies between the theoretical model and the experiments . using the thickness of the boundary @xmath104 , the electric field near the boundary
is estimated to be @xmath105 at microwave attenuation -30db and the temperature @xmath106 ( see fig .
4(a ) ) . using the estimates
@xmath107 is the mass of a free electron ) , @xmath108 and eq.(8 ) , we obtain @xmath109 which is less than the rectification @xmath110 observed at -30db at @xmath111 experimentally ( see fig.4(a ) ) .
moreover the perturbative parameter used in the boltzman approach is proportional to the ratio of microwave electric potential @xmath112 v at @xmath113 db ( see fig.4b ) to the contact difference of potentials @xmath114 mv formed at the boundary between two metals .
this ratio is about @xmath115 in our experiment at @xmath116=-40 db , which according to the above approach indicates we re in the perturbative regime of rectification : @xmath117 ( see eq.(8 ) ) .
however this contradicts our experimental observations at these microwave power levels ( see fig.4b ) : @xmath39 at @xmath118 @xmath119 .
the magnetic field dependence of the rectification exhibits nonmonotonic behavior .
this behavior can be qualitatively understood in terms of the competition between two effects .
the dependence of the chemical potential of the electrons with magnetic field is given by @xmath120 , where we have assumed that @xmath121 is independent of magnetic field .
the derivative of the magnetic susceptibility with respect to electron density , @xmath121 , is negative due to the decrease of the magnetic susceptibility @xmath122 with increasing electron density @xcite .
the increase of the chemical potential appears to be stronger at low electron density in si - mosfet s .
this implies that the difference of the chemical potentials @xmath123 is reduced in a magnetic field , decreasing the rectification .
on the other hand , the conductivity @xmath124 is also reduced , with a consequent increase in the rectification ( see eq.(7 ) ) .
careful analysis of these competing effects is needed in order to understand the dependence of the rectification on magnetic field .
the strong nonlinearity observed in these experiments appears to be similar to the response of an ordinary diode .
it is well known that the diode rectification is proportional to the amplitude of the electric field in the high power regime . in the case of an ordinary diode ,
the rectification is due to the potential barrier formed between the @xmath125 and @xmath126 regions , which induces an exponential dependence of the current on applied voltage @xcite .
another important property of the diode is the presence of two kinds of carriers : electron and holes .
below we will consider these possibilities .
one of the reasons for the potential barrier is simple electrostatics .
the barrier could be the result of the finite width @xmath52 of the split between two gates .
when the voltages of the gates are equal , then a drop in density will occur under the split
. however the width of the split , @xmath127 nm , is smaller than the distance between the 2d electrons and the gates @xmath128 nm .
therefore , the decrease in density is expected to be quite small . moreover at large differences between the gate voltages
( an order of magnitude in our experiment ) there is a very sharp drop in the electron density . in this case one
expects a monotonic change in electron density between the two metals @xcite . we must also note that even in the presence of such an electrostatic potential barrier the rectification should be absent at low microwave voltages @xcite .
an additional potential barrier between two dilute metals could also arise due to the presence of different charged excitations in a 2d dilute system on a si surface , as proposed in several papers @xcite .
for example , if the majority carriers in one metal are fermi particles while the other 2d metal contains mostly bose particles ( for example paired electrons ) with an energy of dissociation @xmath3 , then an additional potential barrier can , in principle , be formed between the two metals .
the origin of such an electrochemical barrier is the reaction of formation of the bose particles @xmath129 ( or other correlated structure ) .
a similar electrochemical barrier will be formed also between two metals with different strength of the electron - electron interaction .
the @xmath66 temperature tail of the rectification ( see fig .
5 ) indicates that the energy scale responsible for the rectification effect is proportional to the @xmath130 .
this energy scale has the same temperature dependence as the rate of electron - electron scattering in a metal .
finally , the observed rectification could be the result of variation of the density of electron states with energy .
however , the very strong nonlinear response observed in our experiments would require that the density of electron states be a very strong function of energy .
in summary , we report the observation of the rectification of microwave radiation by the boundary between two two - dimensional metals on the surface of si . the effect was investigated as a function of electron density , microwave power , temperature , and in - plane magnetic field .
the rectified voltage depends on the densities of the two metals , and goes to zero when the electron densities are equal . at the lowest measured temperature @xmath131
k the rectified signal is directly proportional to the amplitude of the microwave field if one of the metals is close to the quantum phase transition @xcite , while the other is kept at high electron density .
this signals a strongly nonlinear regime even when the contact potential difference between the metals is much larger than the potential associated with the microwave field .
e. abrahams , s. v. kravchenko , and m. p. sarachik , rev .
mod . phys.*73 * , 251 ( 2001 ) .
t. okamoto , k. hosoya , s. kawaji , and a. yagi , phys .
* 82 * , 3875 ( 1999 ) .
s. a. vitkalov , h. zheng , k. m. mertes , m. p. sarachik and t. m. klapwijk , phys .
lett . * 85 * , 2164 ( 2000 ) .
s. a. vitkalov , m. p. sarachik , and t. m. klapwijk , phys . rev .
* b * * 64 * , 073101 , ( 2001 ) .
a. a. shashkin , s. v. kravchenko , v. t. dolgopolov , and t. m. klapwijk , phys .
* 87 * , 086801 , ( 2001 ) .
v.m.pudalov , m. gershenson , h.kojima , preprint cond - mat/0110160 s. a. vitkalov , h. zheng , k. m. mertes , m. p. sarachik and t. m. klapwijk , phys .
87 * , 086401 , ( 2001 ) .
r. heemskerk , ph .
d. thesis , university of groningen , ( 1998 ) . v.
i. falko sov .
solid state * 31 * , 561 ( 1989 ) the small ( @xmath132 rectification signal ( not shown in the fig.4a ) at high density is comparable with an offset voltage at @xmath133 .
it needs an additional study .
s. n. levine `` principles of solid - state microelectronics '' , new york , 1963 .
d. belitz and t. r. kirkpatrick , phys .
b * 46 * , 8393 , ( 1992 ) ; _ ibid . _ * 58 * , 8214 ( 1998 ) .
p. phillips , y. wan , i. martin , s. knysh , and d. dalidovich , nature ( london ) * 395 * , 253 ( 1998 ) . s. chakravarty , s. kivelson , c. nayak , and k. voelker , philos . mag .
b * 79 * , 859 ( 1999 ) .
claudio chamon , eduardo r. mucciolo , antonio h. castro neto preprint cond - mat/0107271 , ( 2001 ) . l. d. landau and e. m. lifshitz `` electrodynamics of continuous media '' , pergamon press , ( 1984 ) .
j. g. simmons journal of appl
. phys . * 34 * , 2581 ( 1963 ) .
|
rectification of microwave radiation ( 20 - 40 ghz ) by a line boundary between two two - dimensional metals on a silicon surface was observed and investigated at different temperatures , in - plane magnetic fields and microwave powers .
the rectified voltage @xmath0 is generated whenever the electron densities @xmath1 of the two metals are different , changing polarity at @xmath2 .
very strong nonlinear response is found when one of the two 2d metals is close to the electron density corresponding to the reported magnetic instability in this system .
-0.5 cm 2
|
the soft ( nonperturbative ) odderon @xmath3 is introduced in elastic hadron - hadron scattering as the @xmath4 partner of the pomeron @xmath5 @xcite . in perturbation theory
the existence of @xmath4 contributions is clear ( 3-gluon exchange ) .
it is even believed @xcite , that 3-gluon exchange dominates the @xmath6-scattering amplitude @xmath7 for momentum transfers @xmath8 . on the other hand , for @xmath9
it seems that odderon contributions play no role asymptotically or even are absent .
an example is the measurement of @xmath10 at @xmath11 gev @xcite .
together with @xmath6 scattering data at smaller energies one deduces @xmath12 . in other words
the soft odderon couples very weakly to the nucleon . since soft elastic high energy scattering is totally dominated by the pomeron , it is useful to look at a reaction where pomeron exchange does not contribute . in the following
we discuss some results of @xcite considering exclusive @xmath0 production in high energy @xmath13-scattering .
the @xmath0 is produced ( see fig . [ b1 ] ) by @xmath14-fusion and by @xmath15-fusion , but can not be produced by @xmath16-fusion since the @xmath0 has positive @xmath17-parity .
1.0 cm ( 15.,4.8 ) ( -0.0,0.2 ) = 4.7 cm ( 4.5,2.0)@xmath18 in fig .
[ b1 ] x stands for a proton or resonances or a sum over resonances . here
we treat the very small @xmath19 range . in the h1 experiment at hera the kinematical cuts for this so called photoproduction region
are @xmath20 where , in the proton rest frame @xmath21 is the fractional energy loss of the incoming lepton .
this allowes us to use the equivalent photon approximation @xmath22 where @xmath23 is the nucleon mass , @xmath24 is the total photoproduction cross section for the reaction , and @xmath25 is the equivalent photon number .
we look at @xmath26-scattering in the c.m .
system and choose @xmath27 as 3-direction . then the photon and the proton have very large momenta in @xmath28-direction . in the following
we discuss the cases , ( i ) that the proton , which is assumed to be a quark - diquark system with a scalar diquark stays intact or ( ii ) gets diffractively exited into the resonances @xmath29 with @xmath30 and @xmath31 with @xmath32 , described as exited quark - diquark systems .
when considering unpolarised cross sections , summed over both resonances in case ( ii ) , the quark spin degree of freedom becomes irrelevant and the calculation reduces to one where a spinless state stays intact or is exited to a 2p resonance . for ( ii )
the helicity amplitudes are @xcite @xmath33 here @xmath34 is the helicity of the photon ( 2p state ) and @xmath35 is the momentum fraction of the photon , carried by the quark .
the physical picture arising from ( [ psamp ] ) is the following ( fig .
[ gampbild ] ) .
the photon fluctuates into a @xmath36 pair , described by @xmath37 . by soft colour interaction ( odderon exchange ) ,
calculated from the functional integral of two lightlike wegner - wilson loops ( @xmath38 ) the @xmath39 pair turns into a @xmath0 and the proton either stays intact or gets exited ( described by @xmath40 ) .
= 3.2 cm now , when the proton stays intact there occurs in ( [ psamp ] ) the quark - diquark density ( @xmath41 ) .
but , since this density is symmetric under a parity transformation whereas the odderon coupling changes sign , there is a cancelation when we integrate over all angles . on the other hand
when the proton gets exited to a negative parity state like the resonances @xmath29 and @xmath31 there is no cancellation ( the overlap is odd under parity ) and the odderon couples to the nucleon without any restriction @xcite .
thus in our model the odderon couples only if breakup of the proton occurs . in fig .
[ msvbild ] we show our result for the differential cross section in @xmath26-scattering , @xmath42 .
1.0 cm ( 15.,4.8 ) ( 1.0,0.0 ) = 6.5 cm the slope of @xmath43 at @xmath44 is around @xmath45 .
the integrated cross section is @xcite : @xmath46 as this photoproduction cross section is constant , i.e. independent of @xmath47 , the epa conversion to electroproduction can be achieved by simply multiplying it with a constant @xmath48 corresponding to the @xmath49 integration in ( [ epa ] ) .
this gives @xmath50 in the following we consider only @xmath13-scattering .
an experimentally prefered observable is the @xmath51 spectrum , the transverse momentum distribution of the @xmath0 with respect to the beam direction . at hera @xmath51 distributions
can be measured for values of @xmath51 greater than @xmath52(0.1 gev ) .
the photoproduction cuts of ( [ cuts ] ) restrict the transverse momentum of the incident photon to be smaller than @xmath52(0.1 gev ) , so in our case there is practically no distinction between the beam axis and the photon axis .
the @xmath51 spectrum is displayed in fig .
[ k_tmsv ] 1.0 cm ( 15.,4.4 ) ( 0.5,-0.2 ) = 6.5 cm
in this section we consider ps production mediated by photon rather than odderon exchange ( fig .
[ b1 ] ) . again the proton , now hit by the photon , is allowed to go into some hadron final state x. the process has been calculated in @xcite for x being a proton . in @xcite
we have calculated the cross section , when x is a hadron final state in the invariant mass ( @xmath53 ) range @xmath54 gev . in fig .
[ k_tmsv ] we show the distribution of the pion s @xmath51 summed over the elastic and all inelastic channels for the electromagnetic exchange .
as we can see there the photon exchange is larger than the odderon exchange only for very small @xmath51 . for @xmath55
gev the odderon exchange dominates by orders of magnitude .
the integrated cross section is @xcite : @xmath56 which has to be compared with the odderon cross section ( [ oddcross ] ) .
in addition we have estimated in @xcite contributions from reggeon ( @xmath57 ) exchange and from a possible background process , namely an additional final state photon emitted from the @xmath0 vertex ( fig .
[ b1 ] ) , where we have used realistc cuts .
both contributions are of the same order of magnitude as the photon exchange contribution .
the conclusion is : with the large odderon cross section as given in our model the process should be observable at hera .
this would establish the soft odderon as an exchange - object in high energy scattering on an equal footing with the soft pomeron .
|
we consider exclusive @xmath0 production in ep - scattering . at high energies odderon and photon
exchange contribute .
the photon exchange contribution is evaluated exactly using data for the total virtual photon - proton absorption cross section .
the odderon exchange contribution is calculated in nonperturbative qcd , using functional integral techniques and the model of the stochastic vacuum . for the proton we assume a quark - diquark structure as suggested by the small odderon amplitude in @xmath1 and @xmath2 forward scattering .
we show that odderon exchange leads to a much larger inelastic than elastic @xmath0 production cross section . observing our process at hera
would establish the soft odderon .
|
cells in connective tissue can glean mechanical information about their surroundings by pulling on the individual biopolymers of the ecm .
however , on the short length scale of a typical cell , the measured mechanical response is sensitive to the intrinsic structural disorder of the ecm . to investigate the role of local mechanical disorder in a physiologically relevant system , we considered collagen networks , which form the primary structural component of the ecm ( 14 ) .
we prepared a sample network by reconstituting fluorescently - labeled collagen type - i monomers and imaged its three - dimensional structure ( 31 , _ supporting information _ ) . the network is loosely connected ( with an average coordination number @xmath0 ) and highly heterogeneous at the cellular scale ( with an average mesh size @xmath1 m ) .
this reconstituted collagen architecture was used as an input to construct a mechanical network model where the fibers are treated as elastic beams that can bend and stretch ( fig .
1__a _ _ ) . for simplicity , we modeled the stretching and bending of the beams , respectively , as springs and torsional springs connecting point - like vertices with stretching modulus @xmath2 and bending modulus @xmath3 .
since biopolymers are expected to be much more pliable to bending than to stretching , we chose the stretching modulus such that @xmath4 .
the bending modulus @xmath3 of the torsional springs was fitted to the experimental network using data from macroscopic rheology ( _ supporting information _ )
. force - dipole stiffness distribution .
( _ a _ and _ b _ ) examples of local stiffness sensing by force dipoles
. modeled deformation under stress from a local force dipole of length @xmath5 m ( green arrows ) of ( _ a _ ) experimental collagen network and ( _ b _ ) rgg network .
magnitude of fiber deformations indicated by color ( small deformations blue , large deformations red ) .
( _ c _ and _ d _ ) distribution of local stiffnesses @xmath6 , defined as the linear response of local deformation to a force dipole of length @xmath5 m for ( _ c _ ) collagen network and ( _ d _ ) rgg network . geometric standard deviation of local stiffness @xmath7 indicated by bars ( actual distribution black , estimated distribution assuming strong locality red , estimated distribution assuming weak locality orange ;
see _ supporting information _ for details ) .
_ insets _ : stiffness loss @xmath8 , defined as the relative change in local stiffness @xmath6 upon perturbing a network by removing a single fiber , versus distance @xmath9 of center of removed fiber from probe center . for collagen , probe length @xmath10 m and removed - fiber length @xmath11 m , and for rgg ,
probe length @xmath12 m and removed - fiber length @xmath13 m .
error bars are smaller than the size of data points .
dashed lines show asymptotic scaling from continuum theory , which predicts @xmath14 for @xmath15 ( see methods for details ) . ] to study the mechanical response of the network to local forces that might be applied by cells , we defined a `` local stiffness '' @xmath16 as the linear displacement of two vertices in response to a dipole contractile force along the direction between the vertices ( 32 ) .
we then calculated this local stiffness by numerically solving the equations of force balance for the network .
strikingly , we found that local stiffness measurements yield a very broad range of values , spanning up to roughly two decades ( fig .
1__c _ _ ) .
because of the large range of stiffnesses , in what follows we characterize this variability in terms of the geometric standard deviation ( eq .
s15 ) , which we find for this collagen network to be @xmath17 . to explore the generality of this broad local stiffness distribution , we next considered a reconstituted fibrin network ( fig .
_ ) , as fibrin constitutes the main structural component of blood clots ( 14 ) .
we imaged a fibrin network prepared from a solution of fibrinogen and observed an average coordination number @xmath18 and an average mesh size @xmath19 m
( _ supporting information _ ) .
we found that the fibrin network also has a broad distribution of local stiffnesses , with @xmath20 very similar to that of the collagen network ( fig .
s1__f _ _ ) .
the broad distribution of local stiffnesses for both the collagen and fibrin architectures suggests that this feature may be a general property of biological fiber networks . collective effect of network structure on local stiffness .
( _ a _ ) distribution of local stiffnesses @xmath6 for rgg network versus average coordination number of vertices , @xmath21 ( force dipole length @xmath22 m , @xmath23m@xmath24 ) .
( _ b _ and _ c _ ) joint distribution of local stiffness and number of local fibers @xmath25 , defined as a weighted fraction of local bonds , with bond weight @xmath26 below a short - range cutoff @xmath27 and decaying as @xmath28 beyond , where ( _ b _ ) @xmath29 for rgg network ( @xmath30 m ) and ( _ c _ ) @xmath31 for collagen network ( @xmath32 m ) .
dashed lines show the macroscopic shear moduli of the networks as a function of average number of local fibers @xmath33 . ] to investigate the physical origins of the broad local stiffness distributions described above , we turned to idealized model networks .
one way to generate model disordered networks consists of arranging vertices to lie on a regular lattice , such as a simple - cubic lattice ( sc ) or face - centered - cubic lattice ( fcc ) in three - dimensions ( fig .
s2__a _ _ , 22 ) . connecting these vertices randomly with a fixed probability results in a disordered lattice network .
such lattice networks have provided a useful starting point for characterizing the mechanical response of fiber networks in two - dimensions ( 27 ) .
interestingly , although three - dimensional sc and fcc networks also yield a broad range of local stiffnesses ( fig .
s4__f _ _ , fig .
_ ) , the geometric standard deviations of these distributions ( @xmath34 and @xmath35 , respectively ) are considerably smaller than those of the experimental networks ( fig .
1__c _ _ and fig .
s1__f _ _ ) .
a possible reason for this discrepancy is that the disordered lattice networks fail to capture important structural features of real networks at the scale of the cell , such as the random positions of the vertices . to generate a model network that better matches the structural features of the experimental networks , we begin by distributing vertices randomly throughout a volume .
the density of these vertices is set roughly equal to those of the experimental networks .
pairs of vertices are then connected according to a probability function that depends on the inter - vertex distance .
this probability function was chosen to produce an average coordination number and distribution of fiber lengths that match those of the experimental architectures ( _ supporting information _ ) .
the resulting modeled networks are referred to as `` random geometric graphs '' ( rgg , fig .
1__b _ _ ) .
we computed the local stiffness distribution for the rgg network and found that the geometric standard deviation is @xmath36 , which is approximately equal to those of the experimental networks ( fig .
1__d _ _ ) .
thus , in contrast to the lattice networks , the rgg network appears to quantitatively capture this important local mechanical property of the experimental networks .
what features of the network impact the very large range of local stiffnesses we found for all types of networks ?
intuitively , more fiber - dense regions should be stiffer . for a fixed density of vertices ,
the fiber density is roughly proportional to the network connectivity , defined as the average coordination number @xmath21 .
we briefly review how the overall connectivity of a network affects its macroscopic mechanical properties ( 20 , 22 ) . for high values of @xmath21 , the bulk elastic moduli of fiber networks
are dominated by the stretching of fibers .
as @xmath21 is lowered , the macroscopic response eventually undergoes a crossover from a stretching - dominated to a bending - dominated response , as the network can be deformed without stretching fibers but not without bending them .
as @xmath21 is further lowered , another elastic transition occurs when the network ultimately loses macroscopic rigidity . over the whole range ,
the macroscopic response depends strongly and nonlinearly on the network connectivity , with the strongest dependence near the two elastic transitions .
we expect this sensitivity of the macroscopic stiffness to overall network connectivity to be reflected in a corresponding sensitive dependence of local stiffness on local network connectivity . to determine how the rapid scaling of the macroscopic response with connectivity manifests locally , we varied the average coordination number @xmath21 of modeled networks for a fixed ratio of the bending modulus @xmath3 to the stretching modulus @xmath2 ( with @xmath37 , fig .
2__a _ _ , fig .
s2__c _ _ ) . at high connectivities ,
the entire local stiffness response is dominated by stretching interactions and scales with the stretching modulus @xmath2 . as the connectivity is lowered , the entire local stiffness distribution shifts to lower values .
specifically , the median stiffness decreases rapidly and nonlinearly with the average coordination @xmath21 . near the stretching - bending crossover
, the local stiffness distribution becomes bimodal with the emergence of a subset of probes for which the measured stiffness scales with the bending modulus @xmath3 .
below the stretching - bending crossover , the number of stretch - dominated probes becomes negligible , and the local stiffness response enters a bending - dominated regime . within this regime ,
the median stiffness resumes rapid decay .
finally , as the connectivity is brought below the rigidity transition , an increasing fraction of measurements yields zero stiffness as portions of the network become floppy .
we thus find that the elastic transitions of the macroscopic response manifest locally as crossovers between stretch - dominated , bending - dominated , and zero - stiffness measurements . at these crossovers , the median
stiffness decreases most rapidly , and the local stiffness distribution becomes bimodal , yielding a maximum in the geometric standard deviation .
the low connectivity of the experimental collagen and fibrin networks suggests that they are situated in the bending - dominated regime . at these particular connectivities ,
the rgg network is poised approximately halfway between the elastic transitions . here , the median stiffness scales least rapidly , and we therefore expect the proximity to the transitions to play a minimal role in governing the width of the rgg stiffness distribution . to verify this , we systematically varied the ratio of the bending modulus to the stretching modulus for these rgg networks ( fig .
s1__b__,_d _ ) .
we found that the local stiffness measured by each individual probe is nonzero and scales with the bending modulus .
this implies that the experimental networks are far from both elastic crossovers , implicating a different source for the broad local stiffness distributions .
a very broad distribution of local stiffnesses appears to be a generic feature of disordered fiber networks .
what is the origin of this large variance in local stiffness measurements ? to investigate how a local stiffness measurement depends on the surrounding network structure , we calculated the stiffness loss @xmath38 upon removing a single fiber ( _ supporting information _ ) .
intuitively , removal of fibers that are more proximal to the stiffness probe should have a greater effect on @xmath6 , which suggests that on average , the stiffness loss should decay as a function of the distance @xmath9 from the probe center to the center of the removed fiber .
indeed , for all types of networks we studied , we found that the average stiffness loss is consistent with @xmath39 decay ( fig .
1__c__,_d _ , fig . s1__f _ _ , and fig
. s2__b _ _ , _ insets _ ) .
this apparent universality of the scaling of the stiffness loss suggests that the @xmath39 power - law decay should be calculable within continuum elasticity theory .
we therefore calculated the effect of removing a single fiber in the vicinity of a local stiffness probe for the case of a uniform lattice network ( _ supporting information _ ) .
the defect created by removing a fiber at a distance @xmath9 from the probe perturbs the dipole strain field induced by the probe .
we can treat this perturbation as an additional dipole strain field originating from the defect , with a magnitude proportional to the initial strain in the removed fiber .
since both this initial strain , and the consequent additional strain `` reflected '' back to the probe , decay as the strain field of a force - dipole , i.e. as @xmath40 ( where @xmath41 is the dimension ) , the combined effect is an increase in the strain at the location of the probe @xmath42 .
the rapid @xmath39 decay of the stiffness loss due to fiber removal in three dimensions suggests that the local stiffness is largely determined by the network structure in the immediate vicinity of the probe . within this small local region ,
all quantities are subject to large fluctuations ; for example , since a small region typically contains a small number of fibers , the variance of fiber density will be large .
the universal , rapid @xmath28 decay implies that the mechanically relevant local structure will always be _ very _ local , with consequently large fluctuations . to quantify the dependence of local stiffness on the surrounding network structure , we considered the number of local fibers @xmath43 , defined as the sum of fibers each weighted by a @xmath39-decaying function of its distance @xmath9 from the probe center ( see above and _ supporting information _ ) .
we find that @xmath6 and @xmath25 are well correlated for all types of networks studied , as shown in fig .
2__b__,_c_. importantly , @xmath6 has a strong , nonlinear dependence on @xmath25 .
specifically , the center of the marginal distribution of local stiffnesses at fixed @xmath25 increases more rapidly than linearly with @xmath25 .
this indicates that local fibers influence local stiffness in a highly cooperative manner , i.e. combining multiple fibers typically results in much larger than additive changes to local stiffness . to estimate the geometric standard deviation of local stiffness ,
we must account for these cooperative effects among fibers .
we first note that the scaling of the _ median _ local stiffness with @xmath25 is consistent with that of the macroscopic shear modulus for all types of networks we studied ( fig .
, s5__d__-_f _ , _ insets _ ) .
this suggests that much of the broad width of the stiffness distribution can be accounted for by the large variations in the local fiber density taken together with the strong , nonlinear dependence of the macroscopic shear modulus on overall fiber density . to test this notion ,
we estimated the distribution of local stiffnesses by taking @xmath25 transformed by the functional dependence of the macroscopic shear modulus , @xmath44 , where the modulus @xmath45 is that of a macroscopic network with an average number of local fibers given by @xmath46 . upon accounting for the strong collective effects of fiber density in this manner , we found that the geometric standard deviation of @xmath44 provides a very good estimate for the actual , observed geometric standard deviation of local stiffness ( fig .
1__c__,_d _ , fig .
s1__f _ _ , fig .
s2__b _ _ , fig .
s4__d__-_f _ , and fig .
s5__d__-_f _ ) .
consequently , the estimator correctly predicts the relative differences in the geometric standard deviation for the different types of networks , including the observation that the stiffness distributions for the experimental and rgg networks are much broader than for the disordered lattice networks .
our prediction of the width of stiffness distributions is not sensitive to the particular form of the weighting function , provided the decay is rapid enough .
for instance , a hard cutoff at twice the mesh size captures a majority of the broad width for the fcc network ( fig .
s4__e _ _ ) .
however , a less rapidly decaying estimator includes a larger number of fibers in the local structure , and consequently the variance in fiber density is smaller .
thus , by comparison , an estimator with weights that decay as @xmath47 yields a smaller , much poorer estimate of the width of the stiffness distribution ( fig .
1__c__,_d _ , fig .
s1__f _ _ , fig .
s4__d__-_f _ , and fig .
s5__d__-_f _ ) .
above , we demonstrated how local structure impacts local stiffness for a fixed distance between probed vertices .
biologically , this distance corresponds to the scale over which cells measure stiffness .
since cells can adopt various sizes and shapes , this means that a cell might measure on different scales depending on the length of its `` arms , '' i.e. the distance between focal adhesions .
we therefore consider the local stiffness as a function of the length of the force - dipole probe in our model . for long probes , the geometric mean and geometric standard deviation of stiffness approach constant values ( fig .
3__a__,_b _ ) .
this asymptotic saturation occurs because for very long probes , the measured deformation is simply the sum of the deformations of two independent _ monopole _ probes , namely the deformation at a vertex in response to a force applied at that vertex . as the probe length
is decreased , however , the measured stiffness increases by more than a decade before the probe length reaches the average mesh size , which is a much larger increase than predicted by continuum theory ( _ supporting information _ ) .
surprisingly , the measured stiffening deviates noticeably from the continuum prediction even for probe lengths that are several times larger than the mesh size . for the experimental and rgg networks , this anomalously large fold - stiffening
is present for all values of @xmath21 and larger near the elastic transitions ( fig .
s8 ) . in contrast , for the fcc network , the stiffness distribution has almost no dependence on probe length away from the elastic transitions ( fig .
s7__d _ _ ) .
this implies that the dramatic fold - stiffening of the experimental and rgg networks largely results from intrinsic features of these networks that are not present for the fcc network .
one relevant feature that distinguishes the different types of networks is the distribution of fiber lengths .
that is , the experimental and rgg networks both contain long fibers connecting vertices several times more distant than the mesh size , whereas for the fcc networks , the typical fiber length ( given by the number of consecutive , coaxial bonds uninterrupted by non - coaxial crosslinking bonds ) is equal to the mesh size , i.e. the lattice spacing . indeed , a simple extension of continuum elasticity that includes structural heterogeneity on the scale of the fiber length can quantitatively capture the majority of the large fold - stiffening observed for the experimental and rgg networks ( _ supporting information _ ) .
thus , our results indicate that the polydispersity of fiber length in biopolymer networks can increase the fold - stiffening as a function of probe length .
( _ a _ and _ b _ ) fold - stiffening @xmath48 , defined as the geometric mean stiffness @xmath49 divided by the geometric mean stiffness for asymptotically long probes @xmath50 ( calculated using local stiffnesses of monopole probes ) , versus probe length @xmath51 for ( _ a _ ) collagen network and ( _ b _ ) rgg network .
error bars show geometric standard deviation of local stiffness ; solid curves ( blue ) from continuum theory ( _ supporting information _ ) . ]
the intrinsic heterogeneity of fiber networks presents a challenge for cells attempting to glean information from stiffness cues .
that is , tissues with different _ global _ stiffness properties may have significant overlap of their _ local _ stiffness distributions . in this case , a single local stiffness measurement would provide only a poor estimate of the global stiffness and identity of the tissue . in view of our conclusion that local stiffness distributions are generically broad ,
how accurately can cells infer the global stiffness of their surroundings from samples of local stiffness ?
one possible strategy for cells to increase their inference accuracy is by averaging multiple stiffness measurements . with enough samples of local stiffness
, this method would allow cells to reliably distinguish between global environments with different mechanical properties .
we can visualize the local stiffness that a cell might infer at each point of the network by plotting the geometric mean stiffness of measurements obtained within a cell - sized sphere centered on that point ( fig .
4__b _ _ , _ inset _ ) .
the patterns in the stiffness landscape reflect the correlations of nearby stiffness measurements and extend over regions larger than typical cell sizes .
more precisely , we find that the correlation function @xmath52 between the log - local stiffnesses measured by two probes decays exponentially as a function of the distance @xmath53 between their centers , with decay lengths around 3 @xmath2 m ( fig .
4__b _ _ and fig .
s10__a _ _ , _ insets _ ) .
these spatial correlations of local stiffness arise because nearby stiffness measurements depend on shared local structure . since correlations reduce the effective number of independent samples , we expect less accurate global inference if stiffness measurements are made closer together in space . to quantitatively study the reduction in accuracy of cellular mechanosensing due to spatial correlations , we modeled cell inference as an idealized sampling and averaging . to be concrete , we considered the sampled stiffness to be the geometric mean of a random sample of three probes whose centers are contained within prolate spheroids of varying volume and aspect ratio ( fig .
4__a _ _ , _ inset _ ) . for all types of networks we studied ,
the shape of the sampling region directly impacts the uncertainty of stiffness inference ( defined as the geometric standard deviation @xmath54 of the inferred stiffness , fig . 4 and fig .
s10__b _ _ ) .
specifically , the uncertainty is always reduced by increasing the volume of the sampling region as well as by increasing its aspect ratio . as the volume and aspect ratio of the sampling region are increased , typical samples of stiffness become increasingly uncorrelated , and , in the asymptotic limit , the uncertainty of the sampled stiffnesses approaches the geometric standard deviation for independent samples .
minimum uncertainty of stiffness inference .
( _ a _ and _ b _ ) geometric standard deviation @xmath55 of the geometric mean of a random sample of @xmath56 local stiffnesses measured by force dipoles whose centers lie within spheres ( _ a _ , _ inset _ : blue ) and prolate spheroids of equivalent volume ( _ a _ , _ inset _ : green , aspect ratio 2:1 , and red , aspect ratio 3:1 ) versus radius of spheres for ( _ a _ ) experimental collagen network and ( _ b _ ) experimental fibrin network .
dashed black lines shows the geometric standard deviation @xmath57 of the geometric mean of @xmath56 independent local stiffness measurements .
inset _ of _ b _ shows a two - dimensional slice of the fibrin network stiffness landscape , defined at each point as the geometric mean of local stiffnesses measured by all force dipoles whose centers lie within a sphere of radius @xmath58 m centered on the point .
lower _ inset _ of _ b _ shows the two - point geometric correlation @xmath59 for local stiffness , defined as the covariance between log - local stiffnesses @xmath60 and @xmath61 divided by the logarithm of the geometric standard deviation of local stiffness squared @xmath62 , versus distance @xmath53 for collagen ( orange ) and fibrin ( cyan ) networks . ]
mechanical cues can guide cell motility and differentiation ( 3 - 6 ) .
indeed , cells are observed to reliably distinguish among two - dimensional homogeneous substrates with bulk stiffnesses similar to brain ( @xmath63 kpa ) , muscle ( @xmath64 kpa ) , and bone ( @xmath65 kpa ) ( 5 ) .
however , the three - dimensional _ in vivo _ mechanical environment at the cellular scale can be extremely heterogeneous ( 14 , 25 ) . here
, we asked how this intrinsic heterogeneity impacts mechanosensing .
interestingly , we found that within macroscopically homogeneous but locally disordered collagen and fibrin networks , local probes yield a very broad range of stiffnesses , spanning roughly two decades ( similar to the relative difference in the bulk stiffness of brain and bone ) .
moreover , the average measured stiffness is anomalously sensitive to probe length .
we quantitatively captured these striking features of the experimental networks with modeled networks , which enabled us to elucidate their physical origins using a combination of simulations and scaling arguments .
we first established that the very broad range of local stiffnesses is universal across network types and spans a wide range of connectivities , including multiple elastic regimes .
we then traced the origin of this pervasive broad distribution to variations in local structure .
specifically , stiffness probes are primarily sensitive to a very small region of local fibers , and these privileged fibers contribute to the measured response in a highly cooperative fashion .
finally , we found that the distribution of stiffnesses can be further broadened by tuning specific network features , including proximity to elastic transitions and geometrical disorder .
while our results indicate that the collagen and fibrin networks are poised squarely in a regime where the response is dominated by fiber bending , these networks include strong geometrical disorder , including randomly - oriented fiber junctions and a polydisperse fiber length distribution . these structural features
explain why the range of stiffnesses is larger for the experimental and random geometric graph ( rgg ) networks than for lattice - based networks .
the geometrical disorder of the experimental and rgg networks also introduces an anomalous fold - stiffening , i.e. a dependence of stiffness on probe length that exceeds that predicted by continuum elasticity .
our `` ideal - mechanosensor '' model does not address the internal mechanics of cells .
any internal noise in sensing or downstream signaling can only increase overall measurement uncertainty .
there is evidence that cells can fix and thus regulate the relative uncertainty in measured stiffness by linearly modulating their applied stress to maintain a constant deformation ( 8) . notwithstanding , we have shown that even if cells sense nearly optimally , any single stiffness measurement is poorly informative of global tissue properties .
this suggests that cells can benefit more from integrating the results of multiple stiffness measurements than from optimizing individual measurements , which may explain why some cells display more than a hundred focal adhesions ( 33 ) . yet
even this strategy has diminishing returns , because nearby stiffness measurements probe the same underlying local structure and are therefore correlated . to extract useful information
, cells must spread their measurements over extended regions of space , either by moving or by extending their shape .
the benefits of an elongated shape are twofold , since measurements on larger scales are both more accurate and less spatially correlated .
the biological relevance of this strategy is supported by the observation of highly polarized cells over five times longer than wide , including fibroblasts , mesenchymal stem cells , and cancer cells ( 5 , 13 , 33 , 34 ) .
since cells and the ecm evolved together , the latter may also be tuned to facilitate mechanical inference .
we uncovered two design principles for tuning the mechanical response : proximity to an elastic transition and the presence of long fibers .
both features result in an enhanced fold - stiffening for shorter probes , which may be harnessed to provide specialized signals to different cell types that measure over different scales , albeit the fold - stiffening comes at the cost of increased variance of the local stiffness . with regard to this variance , while we have only considered isotropic networks , actual tissues can have aligned fibers ( 12 ) , which may yield a narrower stiffness distribution when probed along a fixed direction .
this may explain why the alignment of the ecm has been observed to coordinate with cell polarization to promote migration ( 35 ) .
+ finally , probing stiffness beyond linear response could provide cells with additional information .
how hard would a cell need to pull to access the nonlinear regime ?
nonlinear effects will certainly become significant when fibers begin to buckle ( 36 ) .
for a single fiber equal in length to the mesh size , the euler buckling thresholds are @xmath66 and @xmath67 nn using the bending moduli we inferred for the collagen and fibrin networks , respectively .
forces of several newtons are achievable by stronger cells , as well as by colonies that migrate collectively ( 37 ) .
this suggests that the nonlinear regime is an interesting direction for future study .
+ in summary , the disorder inherent in biological fiber networks places severe physical limits on the accuracy of cellular mechanosensing , suggesting that organisms must have evolved cellular - scale strategies to cope with this uncertainty _ in vivo_. going forward , high - throughput gene deletion and mutation studies in conjunction with realistic patterned substrates ( 38 ) can help reveal the full array of internal components and pathways required for accurate mechanosensation .
fibers are modeled as simple elastic elements with a stretching modulus @xmath2 ( units of energy / length ) and bending modulus @xmath3 ( units of energy@xmath68 length ) . adding the energy contributions from the stretching and bending of all fibers
gives the hamiltonian for the total mechanical energy of the network to leading order for small deformations : @xmath69 where @xmath70 is the length of the fiber @xmath71 connecting vertices @xmath72 and @xmath73 in the unperturbed reference state , @xmath74 is the average unstretched length of fibers @xmath71 and @xmath75 , @xmath76 is the difference in vertex displacements @xmath77 , and @xmath78 is the unit vector in the direction of the fiber in the undeformed reference state . here , the first sum corresponds to stretching interactions and is taken over all fibers @xmath71 .
the second sum corresponds to bending interactions and is taken over all connected pairs of fibers @xmath71 and @xmath75 .
the cell is modeled as an idealized stiffness - measuring device that exerts a contractile , equal and opposite force of magnitude @xmath79 on vertices @xmath80 and @xmath67 , which are separated by a distance @xmath51 in the undeformed network .
the vertices are chosen such that they are not directly connected to each other , in order to avoid trivial contributions to the measured stiffness due to the stretching characteristics of a single fiber . provided the vertices @xmath80 and
@xmath67 belong to a rigid structure , in linear response the applied force deforms the network according to the force - balance equations @xmath81 , where @xmath82 is the force - constant matrix of the unperturbed hamiltonian .
the measured local stiffness @xmath6 is defined as the magnitude of the applied force divided by the relative displacement along the direction of the force @xmath83 ) .
we obtain the local stiffnesses numerically by computing the deformations of the vertices @xmath77 using the generalized inverse of the force - constant matrix ( _ supporting information _ ) .
we thank yigal meir for insightful discussions .
this work was supported in part by the national science foundation grants dmr-1310266 ( to l.m.j .
, s.m . , and d.w ) , the harvard materials research and engineering center dmr-1420570 ( to l.m.j . , s.m . , and d.w . ) , a lewis - sigler fellowship ( to c.p.b . ) , the german excellence initiative via the program `` nanosystems initiative mnich '' ( nim ) and the deutsche forschungsgemeinschaft ( dfg ) via project b12 within the sfb-1032 ( to c.p.b . and f.b . ) , national science foundation grants phy-1305525 and phy-1066293 ( to c.p.b . ,
f.b . , and n.s.w .
) , and the hospitality of the aspen center for physics ( to c.p.b . and
. 10 discher de , janmey pa , wang yl ( 2005 ) tissue cells feel and respond to the stiffness of their substrate . _
310(5751 ) : 1139 - 1143 .
vogel v , sheetz m ( 2006 ) local force and geometry sensing regulate cell functions .
nat rev mol cell biol _
7(4 ) : 265 - 275 .
lo cm , et al ( 2000 ) cell movement is guided by the rigidity of the substrate . _ biophys j _
79(1 ) : 144 - 152 .
isenberg bc , dimilla pa , walker m , kim s , wong jy ( 2009 ) vascular smooth muscle cell durotaxis depends on substrate stiffness gradient strength .
_ biophys j _
97(5 ) : 1313 - 1322 . engler aj , sen s , sweeney hl , discher de ( 2006 ) matrix elasticity directs stem cell lineage specification .
_ cell _ 126(4 ) : 677 - 89 .
guilak f , et al ( 2009 ) control of stem cell fate by physical interactions with the extracellular matrix .
_ cell stem cell _ 5(1 ) : 17 - 26 .
fletcher da , mullins rd ( 2010 ) cell mechanics and the cytoskeleton .
463(7280 ) : 485 - 492 .
trichet l , et al ( 2012 ) evidence of a large - scale mechanosensing mechanism for cellular adaptation to substrate stiffness .
_ proc natl acad sci usa _ 109(18 ) : 6933?6938 .
fabry b , lange jr ( 2013 ) cell and tissue mechanics in cell migration .
_ exp cell res _
319(16 ) : 2418 - 2423 . doyle ad , yamada km ( 2016 ) mechanosensing via cell - matrix adhesions in 3d microenvironments .
_ exp cell res _
343(1 ) : 60 - 66 .
zaman mh , et al ( 2006 ) migration of tumor cells in 3d matrices is governed by matrix stiffness along with cell - matrix adhesion and proteolysis .
_ proc natl acad sci usa _ 103(29 ) : 10889 - 10894 .
guo q , et al ( 2013 ) modulation of keratocyte phenotype by collagen fibril nanoarchitecture in membranes for corneal repair .
34(37 ) : 9365 - 9372 .
thievessen i , et al ( 2015 ) vinculin is required for cell polarization , migration , and extracellular matrix remodeling in 3d collagen .
_ faseb j _ 29(11 ) 4555 - 4567 .
frantz c , stewart km , weaver , vm ( 2010 ) the extracellular matrix at a glance .
_ j cell sci _ 23 : 4195 - 4200 .
janmey pa , amis ej , ferry jd ( 1983 ) rheology of fibrin clots .
stress - relaxation , creep , and differential dynamic modulus of fine clots in large shearing deformations .
_ j rheol _ 27(2 ) : 135 - 153 .
roeder ba , kokini k , sturgis je , robinson jp , voytik - harbin sl ( 2002 ) tensile mechanical properties of three - dimensional type i collagen extracellular matrices with varied microstructure .
_ j biomech eng _
124(2 ) : 214 - 222 .
roberts ww , lorand ll , mockros lf ( 1973 ) viscoelastic properties of fibrin clots .
_ biorheology _ 10(1 ) : 29 - 42 .
knapp dm , et al .
( 1997 ) rheology of reconstituted type i collagen gel in confined compression .
_ j rheol _
41(5 ) : 971 - 993 .
baker b , et al ( 2015 ) cell - mediated fiber recruitment drives extracellular matrix mechanosensing in engineered fibrillar microenvironments . _ nat mat _
14 : 1262 - 1268 .
broedersz cp , mackintosh fc ( 2014 ) modeling semiflexible polymer networks .
_ rev mod phys _ 86 : 995 .
heussinger c , frey e ( 2006 ) floppy modes and nonaffine deformations in random fiber networks .
_ phys rev lett _
97(10 ) : 105501 .
broedersz cp , et al ( 2011 ) criticality and isostaticity in fibre networks .
_ nat phys _ 7 : 983 - 988 .
das m , quint d , schwarz jm ( 2012 ) redundancy and cooperativity in the mechanics of compositely crosslinked filamentous networks _
plos one _ 7 : 35939 .
cioroianu ar , spiesz em , storm c ( 2013 ) an improved non - affine arruda - boyce type constitutive model for collagen networks .
_ biophys j _
104(2 ) : 511a .
lang nr , et al ( 2013 ) estimating the 3d pore size distribution of biopolymer networks from directionally biased data .
_ biophys j _ 105(9):1967?1975 .
head da , levine aj , mackintosh fc ( 2005 ) mechanical response of semi flexible networks to localized perturbations .
_ phys rev e _ 72 : 061914 .
jones car , et al ( 2015 ) micromechanics of cellularized biopolymer networks .
_ proc natl acad sci usa _ 112(37 ) : e5117?e5122 berg hc , purcell em ( 1977 ) physics of chemoreception .
_ biophys j _ 20 , 93 - 219 .
wingreen ns , endres , rg ( 2008 ) accuracy of direct gradient sensing by single cells .
_ proc natl acad sci usa _ 105(41 ) : 15749 - 15754 .
bialek w , setayeshgar , s ( 2005 ) physical limits to biochemical signaling . _ proc natl acad sci usa _ 102(29 ) : 10040 - 10045 .
jawerth lm , mnster s , vader da , fabry b , weitz da ( 2010 ) a blind spot in confocal reflection microscopy : the dependence of fiber brightness on fiber orientation in imaging biopolymer networks .
_ biophys j _
98(3 ) : l1-l3 . schwarz us , safran sa ( 2013 ) physics of adherent cells . _
rev mod phys _
85(1327 ) : 1327 - 1372 .
prager - khoutorsky m , et al ( 2011 ) fibroblast polarization is a matrix - rigidity - dependent process controlled by focal adhesion mechanosensing .
_ nat cell biol _ 13 : 1457 - 1465 .
friedl p , wolf k , lammerding j ( 2011 ) nuclear mechanics during cell migration .
_ curr opin cell biol _
23(1 ) : 55 - 64 .
fraley si , et al ( 2015 ) three - dimensional matrix fiber alignment modulates cell migration and mt1-mmp utility by spatially and temporally directing protrusions . _
sci rep _ 5 : 1458 .
ronceray p , broedersz cp , lenz m ( 2016 ) fiber networks amplify active stress .
_ proc natl acad sci usa _ 113(11 ) : 2827 - 2832 .
rourke o , et al ( 2005 ) force mapping in epithelial cell migration .
_ proc natl acad sci usa _ 102(7 ) : 2390?2395 .
kim dh , wong pk , park j , levchenko a , sun y ( 2009 ) microengineered platforms for cell mechanobiology .
_ annu rev biomed eng _ 11 : 203 - 233 .
mnster s , et al ( 2013 ) strain history dependence of the nonlinear stress response of fibrin and collagen networks .
_ proc natl acad sci usa _ 110(30 ) : 12197?12202 .
cserti j , david g , piroth a ( 2002 ) perturbation of infinite networks of resistors .
_ am j phys _
70(2 ) : 153 .
landau ld , lifshitz em , kosevich am , pitaevskii lp ( 1986 ) theory of elasticity ( pergamon press , london ) .
gao xl , ma hm ( 2009 ) green s function and eshely s tensor based on a simplified strain gradient elasticity theory .
_ acta mech _ 207 : 163 - 181 .
[ runin ] * * 1em
we study the mechanical properties of crosslinked biopolymer networks , such as those of the extracellular matrix ( ecm ) , using a fiber network model , which consists of a collection of elastic elements that are connected at point - like vertices .
the elastic elements model the stretching and bending interactions of the constituent fibers .
the mechanical energy in the fiber network model is given by ( 29 ) : @xmath84 where * @xmath2 is the stretching modulus , * @xmath3 is the bending modulus , * @xmath85 is the difference in positions of vertices @xmath72 and @xmath73 , * @xmath70 is the length of the fiber @xmath71 connecting vertices @xmath72 and @xmath73 in the unperturbed reference state ( i.e. the unstretched fiber length ) , * @xmath74 is the average unstretched length of fibers @xmath71 and @xmath75 , * @xmath86 is the angle between fibers @xmath71 and @xmath75 that meet at vertex @xmath73 , * @xmath87 is the angle between unit vectors @xmath88 and @xmath89 that point along the direction of the fibers connecting vertices in the unperturbed reference state .
the first sum corresponds to stretching interactions and is taken over all fibers @xmath71 .
the second sum corresponds to bending interactions and is taken over all connected pairs of fibers @xmath71 and @xmath75 .
small forces will generate deformations that are small compared to the lengths of fibers . in this case
, it is convenient to express the positions of the vertices in terms of their deformations @xmath77 about their positions in the undeformed reference state @xmath90 as follows : @xmath91 upon expanding both sums in the mechanical energy and discarding higher order terms , we obtain the harmonic energy in the fiber network model , which is given by : @xmath92 where @xmath93 is the relative deformation of nodes @xmath72 and @xmath73 .
the cell is modeled as an idealized stiffness - measuring device that exerts force on vertices of the network .
the effect of the applied force @xmath94 perturbs the mechanical energy as follows : @xmath95 where the index is summed over all vertices .
we assume that the ecm behaves as a viscoelastic solid , and that the forces applied by cells change slowly . assuming forces applied by cells change slower than the timescale set by viscous damping , the network will be deformed in a quasistatic manner , i.e. the network reaches mechanical equilibrium for a given force . in this case , the deformation is completely determined by minimizing the full mechanical energy : @xmath96 expressing this equation in terms of the deformations and the forces leads to the equations of force - balance in static equilibrium : @xmath97 where @xmath82 is the force - constant matrix of the unperturbed hamiltonian : @xmath98 we model a two - point measurement performed by a cell as a force dipole , which is defined as a contractile force exerted on vertices @xmath80 and @xmath67 separated by a vector @xmath99 in the undeformed reference state .
the vector of equal and opposite forces may be represented using kronecker delta notation as follows : @xmath100 where @xmath101 is the magnitude of the force applied to each vertex .
the local stiffness @xmath6 is defined as the linear response of the displacement of vertices @xmath80 and @xmath67 to the contractile force : @xmath102 to calculate the local stiffness @xmath6 defined above , we compute the deformations of the vertices @xmath77 numerically by solving the equations of force - balance in static equilibrium .
these equations do not have a well - defined solution for all possible configurations of the applied force because the force - constant matrix is singular , i.e. it contains eigenvectors with vanishing eigenvalues , or `` zero modes . ''
if the force of a local stiffness probe couples to a zero mode , the resulting deformations will diverge and the local stiffness is undefined in linear response . intuitively , this corresponds to probes that act on unconstrained , dangling portions of the network . to solve the equations of force - balance while dealing with the technical challenge provided by zero modes , we compute the generalized inverse of the force - constant matrix .
the generalized inverse allows us to first check whether a solution exists for a given force perturbation , and then to solve for the deformations by multiplying the force perturbation by the generalized inverse .
such an approach is efficient because it allows us to measure local stiffness over an entire network using only a single matrix multiplication operation per probe , which is computationally inexpensive compared with solving the system of linear equations anew for each local stiffness probe .
we obtain the generalized inverse by first computing the singular value decomposition ( svd ) of the force - constant matrix : @xmath103 where @xmath104 and @xmath105 are orthogonal matrices and @xmath106 is a diagonal matrix consisting of singular values @xmath107 equal in number to the rank @xmath108 of the dynamical matrix . from this , the generalized inverse may be computed : @xmath109 where @xmath110 is a diagonal matrix consisting of the reciprocals of the singular values @xmath111 .
the generalized inverse provides a simple way to check whether a solution exists .
that is , if a given force vector satisfies the following equation : @xmath112 then the force - balance equation has a solution .
this guarantees that a given force configuration has no projection onto any zero mode . for well - defined force probes , which satisfy ( s12 ) , the resulting deformation of each vertex is finite and given by : @xmath113 from which the local stiffness may be computed .
the network architectures for the fiber network models , i.e. the positions of the vertices and their pairwise connectivities , were obtained from experiment by analyzing images of reconstituted fiber networks .
collagen networks were prepared and imaged as described previously in ( 36 ) .
briefly , a sample of 0.2 mg / ml protein in 1x dmem with 25 mm hepes was prepared from a solution of collagen type - i monomers with a small fraction of fluorescently - labelled monomers at 4@xmath114 c. network formation was induced by neutralizing the sample s ph with 1 m naoh and incubating it at room temperature for four hours .
the resulting network was imaged using a confocal microscope ( leica sp5 , wetzlar , germany ) .
we acquired a set of fluorescent images covering a three - dimensional volume that was representative of the network structure .
to determine the fiber positions and connectivity , the image stacks were thresholded and subsequently skeletonized , which resulted in a one - voxel thick line representation of all fibers .
we define a branch - point as the junction between three or more fibers .
the number of fibers that join at a branch - point defines the branch - point s connectivity @xmath21 . in our mechanical model ,
the vertices are positioned at the branch - points and end - points of fibers . by counting the number of vertices within the network volume
, we found a vertex density of @xmath115 vertices/@xmath116 .
fibrin networks were prepared and imaged as described previously ( 44 ) .
briefly , a solution of human fibrinogen ( enzyme research labs , south bend , in ) containing a small fraction of fluorescently labelled fibrinogen with a protein concentration of 0.2 mg / ml was prepared in buffer ( 150 mm nacl , 20 mm cacl , 20 mm hepes , ph 7.4 ) .
network formation was induced by the addition of activated human alpha - thrombin ( enzyme research labs , south bend , in ) to the fibrinogen solution ( final thrombin concentration : 0.1 u / ml ) .
after the sample was allowed to polymerize for 12 hours , the resulting fibrin network was subsequently imaged and the data was processed analogously to the collagen sample .
the vertex density was around @xmath117 vertices/@xmath116 . for the collagen and fibrin networks
, we found average coordination numbers of @xmath0 and @xmath18 , respectively .
although these average coordination numbers are low , the networks are both macroscopically rigid , which is defined by a finite response to shear forces .
rheological measurements were performed on networks prepared under experimental conditions analogous to those of the imaged samples . for the collagen network ,
the measurements were done using an ar - g2 rheometer ( ta instruments , new castle de ) equipped with a custom - made plastic - plate of 25 mm and sealed with mineral oil ( 44 ) .
the macroscopic mechanical response was measured by applying a small oscillatory strain and measuring the resulting stress , which resulted in a shear modulus of @xmath118 pa .
the fibrin network was measured with a 40 mm / @xmath119 cone - plate geometry .
samples were sealed with mineral oil and allowed to polymerize for at least 12 hours .
the network was then perturbed in the same manner as for the collagen network and the shear modulus was found to be @xmath120 pa .
macroscopic rigidity requires macroscopic connectedness , i.e. the presence of a spanning cluster of vertices . in our analysis , we identified all vertices that belong to the largest cluster using a union - find algorithm and removed all other vertices from the network .
for both the collagen and the fibrin networks , we considered a spherical sample of radius @xmath121 m . within this sample volume ,
the @xmath122 vertices are distributed approximately homogeneously on scales large compared to the mesh size @xmath123 , defined as the radius of a sphere whose volume is equal to the average volume per vertex : @xmath124 for the experimental collagen and fibrin networks , we found @xmath32 m and @xmath125 m , respectively . on scales comparable to @xmath123 , the networks are intrinsically disordered due to heterogeneity in both the spatial positions of the vertices and the lengths of fibers .
the lengths of the fibers are highly polydisperse , spanning over an order of magnitude in length with a substantial fraction of fibers that are very long compared to the average fiber length ( fig .
s9__a__,_b _ ) .
the fiber length distribution peaks at small lengths ( i.e. below the average fiber length ) and decays roughly monotonically beyond the peak . for each experimental network ,
the corresponding mechanical model has two undetermined parameters : the stretching modulus @xmath2 and the bending modulus @xmath3 .
since both parameters may be rescaled together resulting only in a trivial change of units , the only remaining non - trivial parameter is their ratio @xmath126 . throughout the main text , we chose @xmath127 to capture the fact that individual fibers are much stiffer to stretching than to bending .
we compute the local stiffness distribution numerically using the procedure described above in section 1.3 ; however , rather than considering an ensemble of individual probes over many network realizations , we sample many probes throughout the volume of a single network .
these procedures are equivalent when the sample volumes are large compared to the scale over which local stiffness measurements are correlated , i.e. networks which are large enough to represent the ensemble via self - averaging .
we confirmed this equivalence by varying the sample volume and observing that the distribution converged below the network radius used in our analysis .
to determine the units of the modeled response , we first noted that both the experimental collagen and fibrin networks display a bending - dominated response .
that is , for these networks , the macroscopic shear modulus and all local stiffness measurements are directly proportional to @xmath3 ( fig .
s1__a__-_d _ ) .
we therefore fit the bending modulus of our mechanical model using the experimentally - measured macroscopic response . specifically , we first calculated the shear modulus numerically using the mechanical model for an arbitrary value of @xmath3 , which we then scaled by the ratio of the experimental shear modulus to the calculated shear modulus .
this resulted in fitted bending moduli of @xmath128 nn@xmath129m@xmath24 and @xmath130 nn@xmath129m@xmath24 for collagen and fibrin , respectively . to understand how the mechanical response emerges from the features of the network architecture , we considered idealized model architectures . for simplicity
, the resulting structures were generated randomly such that the initial positions of the vertices and fibers were not spatially correlated . throughout our analysis
, we obtained our results for these networks by averaging over a large number of network realizations to reduce sampling noise .
random geometric graphs ( rgg ) provide a way to generate model architectures with independent control over average coordination number , vertex density , and average fiber length . in particular , these parameters can be chosen to match those observed in experiment . in the rgg model , architectures are generated by first distributing vertices throughout a volume and then connecting them randomly according to a specified probability distribution .
the position of each vertex is drawn from a uniform distribution over the network volume , and the probability distribution for connecting two vertices , or the `` connection function , '' depends only on the inter - vertex distance . for simplicity , we model the connection function as a power law with an exponential cutoff : @xmath131 where @xmath132 is a length scale that describes the typical length of fibers .
this results in networks that are homogeneous and isotropic over large scales . for such networks in three dimensions ,
the expected density of vertices that are a distance @xmath133 from a particular vertex scales as @xmath134 , so the expected distribution of fiber lengths in the three - dimensional rgg model is @xmath135 . throughout our analysis , we considered networks with approximately @xmath136 vertices/@xmath116 and we chose the length scale @xmath132 equal to @xmath137 m .
we chose the average coordination number @xmath0 to match that of the experimental networks ( except in fig .
2__a _ _ where we varied the average coordination number by randomly removing fibers from a network with a high initial average coordination number ) .
this resulted in an average mesh size @xmath19 m . due to the larger resulting density of these modeled networks in comparison to the experimental networks , we considered samples of radius @xmath138 m .
we determined the bending modulus @xmath3 by fitting the geometric mean stiffness measured for the rgg network at @xmath139 m to that of the collagen network .
this resulted in a bending modulus @xmath140 nn@xmath141m@xmath24 and shear modulus @xmath142 .
for the above choices of parameters , the rgg networks quantitatively capture the local mechanical response of the experimental networks , including the broad width of the local stiffness distribution , the large fold - stiffening for short probes , and the two - point stiffness correlation length ( see _ results _ ) . because of the broad range of stiffnesses , throughout our work
, we characterize the variability of the stiffness distribution using the geometric standard deviation , defined as the exponential of the standard deviation of log - local stiffnesses : @xmath143 disordered lattice networks are an established class of modeled architectures that have successfully described many of the mechanical properties of biopolymer networks ( 20 , 22 , 23 ) .
in particular , lattice models are able to capture the macroscopic response and its striking dependence on network density .
because of their numerical convenience , two - dimensional lattice - based networks have provided a starting point for modeling the microscopic response of fiber networks ( 27 ) .
here , we provide the first numerical analysis of the microscopic response of three - dimensional disordered lattice networks and demonstrate that such models fail to quantitatively capture the broad width we observed for the experimental collagen and fibrin networks .
disordered lattice networks are generated by placing vertices on an ordered lattice , for example face - centered - cubic ( fcc ) or simple - cubic ( sc ) lattices in three dimensions , and connecting them randomly with a fixed probability @xmath144 .
for a given choice of lattice , the connectivity @xmath144 is the sole parameter that impacts the overall network structure .
in particular , the density of vertices can not be varied independently from the average fiber length .
furthermore , since the vertices are initially positioned to lie on a regular lattice , fibers only meet at discrete , fixed angles in the unperturbed reference state .
disordered lattice networks are therefore significantly simpler than the experimental networks . the local stiffness distribution for the fcc lattice network is shown in fig .
s2__b__. here , we chose a bond - probability of @xmath145 to study the bending regime relevant to biological networks . throughout our analysis
, we set the stretching modulus @xmath146 and bond length @xmath147 .
we considered local stiffness measurements for pairs of vertices separated by primitive lattice vectors of the form @xmath148 to avoid pathological effects associated with applying forces in the symmetry directions of the lattice . for both the sc and fcc disordered lattice networks
, we find the presence of a broad range of local stiffnesses spanning up to roughly two decades ( fig .
s4__f _ _ , fig .
s2__b _ _ ) .
this broad range of stiffnesses is generic over a wide range of network connectivities in the stretching and bending regimes and peaked near the crossovers ( see fig .
_ for fcc lattice results ) .
interestingly , although the range is qualitatively very large , the geometric standard deviation is substantially smaller for both the sc ( @xmath34 , fig .
s4__f _ _ ) and fcc ( @xmath149 , fig .
s2__b _ _ ) lattice networks than for the experimental ( collagen : @xmath17 , fig .
1__c _ _ ; fibrin : @xmath20 , fig .
s1__f _ _ ) and random geometric graph ( rgg , @xmath36 , fig .
1__d _ _ ) networks . in the following section ,
we study how this discrepancy arises from differences in network structure . what is the physical origin of the qualitatively broad width we observed for the disordered lattice networks ? as for the rgg network ( fig .
s5__d__-_f _ ) , the disordered lattice networks also yield macroscopic shear moduli and median stiffnesses that depend sensitively on average coordination number ( fig . s4__d__-_f _ ) .
this suggests that the broad width we observed for all types of networks we studied arises from intrinsic features common to all types of disordered networks .
indeed , as we observed for the rgg network ( see _ results _ ) , we found that for the disordered lattice networks : ( 1 ) the stiffness loss also obeys universal scaling ( see fig .
s2__b _ _ , _ inset _ for fcc lattice results ) , and ( 2 ) the scaling of the marginal distribution of local stiffnesses with the number of local fibers @xmath25 is captured by the rapid , nonlinear scaling of the macroscopic shear modulus ( fig .
s4__d__-_f _ , _ insets _ ) .
we then estimated the geometric standard deviation for the sc and fcc networks in the same way as for the experimental and rgg networks ( see next section , _ dependence of local stiffness on local structure _ ) .
as before , we estimate the local stiffness distribution as the distribution of the number of local fibers @xmath25 transformed by the macroscopic shear modulus . for both sc and fcc lattice networks , we found that counting the number of local fibers using rapidly - decaying weighting functions results in a good match between the estimated and actual geometric standard deviations ( table s1 , fig .
s4__d__-_f _ ) .
furthermore , we found that the estimate does not depend significantly on the particular form of the weighting function used to calculate the number of local fibers .
specifically , the estimate of the geometric standard deviation for the fcc network does not depend strongly on whether we chose a rapidly decaying power - law or a hard cutoff at two mesh sizes ( fig .
s4__d__,_e _ ) .
this suggests that obtaining a good estimate of the geometric standard deviation based on local structure does not require finely - tuning the details of the weighting function , provided the chosen weights are given by a function that decays sufficiently rapidly with distance .
to gain insight into the variation of local stiffness , we determined how the presence of a single fiber in the vicinity of the stiffness probe impacts the local stiffness .
we quantified the influence of a fiber by the stiffness loss @xmath38 , defined as the relative change in local stiffness upon removing the fiber from the network ( fig .
intuitively , fibers which are more proximal should have a greater effect .
we therefore expect the stiffness loss induced by removing a fiber to decay as a function of the distance @xmath9 from the probe center to the center of the fiber . in this section ,
we calculate how the stiffness loss scales with distance @xmath9 for an ordered lattice .
we calculate the stiffness loss scaling by first deriving an exact formula for the change in local stiffness when a single fiber is removed from an ordered , stretching - only network in @xmath41 dimensions , and then studying how the magnitude of the effect depends on the distance to the missing fiber from the center of the probe .
the equation of force - balance in static equilibrium ( eq .
s6 ) becomes : @xmath150 where the unperturbed force - constant matrix @xmath151 , vertex deformations , and applied forces are represented using dirac notation as follows : @xmath152 @xmath153 @xmath154 where @xmath155 if vertices @xmath72 and @xmath73 are nearest neighbors and @xmath156 otherwise , and the vectors @xmath157 form a complete orthonormal set over vertices such that @xmath158 and @xmath159 .
these equations can be inverted by fourier transformation to solve for the displacements : @xmath160 where @xmath161 is the lattice green s tensor defined by @xmath162 and given by @xmath163 in terms of the @xmath164 dynamical matrix : @xmath165 here , @xmath166 are unit vectors summed over the directions of the nearest neighbor vertices .
the lattice green s tensor relates the force applied to vertex @xmath72 to the deformation at vertex @xmath73 , which provides a simple way to express the relative deformation of the vertices induced by the dipole : @xmath167\cdot\hat{\boldsymbol{d}}.\ ] ] this relative deformation determines the local stiffness defined in ( eq .
s9 ) as the ratio of the applied force to the relative deformation @xmath168 . we will now calculate the perturbed stiffness @xmath169 between vertices @xmath72 and @xmath73 when the bond @xmath170 connecting neighboring vertices @xmath171 and @xmath172 is removed , generalizing the analogous calculation for resistor networks performed by cserti et al .
the equation of force - balance for the perturbed network is @xmath173 where @xmath174 is the force - constant matrix for the perturbed network , i.e. the unperturbed force - constant matrix @xmath151 minus the contributions associated with the missing bond @xmath170 : @xmath175 @xmath176 where @xmath177 and @xmath178 .
we solve for the deformations as before : @xmath179 in terms of @xmath180 , the lattice green s tensor for the perturbed network defined by @xmath181 . by premultiplying this equation with @xmath161
, we obtain a recursive formula for @xmath180 : @xmath182 we substitute the matrix representation of @xmath183 given above and iterate to obtain an explicit formula for @xmath180 in terms of an infinite series : @xmath184 where we have defined the constant @xmath185 .
we then sum the geometric series to obtain : @xmath186 this is an explicit formula for the lattice green s tensor of the perturbed network in terms of the lattice green s tensor of the unperturbed network .
the local stiffness measured between vertices @xmath72 and @xmath73 is now @xmath187 with the perturbation @xmath188 to the measured deformation given by : @xmath189^{2}}{1 - 2\hat{\boldsymbol{r}}_{i_{0}j_{0}}\cdot\left[\mathcal{g}_{0}(i_{0},i_{0})-\mathcal{g}_{0}(i_{0},j_{0})\right]\cdot\hat{\boldsymbol{r}}_{i_{0}j_{0}}}.\ ] ] using this exact formula for the local stiffness in the vicinity of a single removed fiber , we now approximate the stiffness loss , defined as the relative change in local stiffness upon removing the bond @xmath170 at a distance @xmath9 ( see fig .
s3 ) : @xmath190 when the location of the missing fiber is far away from the probe , i.e. @xmath191 , the perturbation is small and scales as : @xmath192 at these long distances , the lattice green s tensor approaches the continuum elastic green s tensor , which is given by the following ( 41 ) : @xmath193\frac{1}{r},\ ] ] where @xmath194 is the young s modulus , @xmath195 is the poisson s ratio , @xmath196 is the kronecker delta function , @xmath197 is a unit vector parallel to the position vector @xmath198 , and @xmath108 is the magnitude of the position vector . upon substituting the continuum elastic green s tensor in place of the lattice green s tensor and expanding the perturbation to third order in @xmath199 , we find the stiffness loss scales as @xmath200 in @xmath41 dimensions .
comparing this universal result to the average stiffness loss measured for disordered networks ( fig .
1__c__,_d _ , fig .
s1__f _ _ , and fig .
s2__b _ _ , _ insets _ ) , we find that the observed stiffness loss is consistent with the @xmath28 scaling prediction for all types of networks considered .
the rapid decay of stiffness loss suggests that the local stiffness largely depends on a very small number of proximal fibers . to quantify this dependence
, we estimate the local stiffness distribution by considering the number of local fibers @xmath25 , given by : @xmath201 where @xmath8 is a rapidly decaying function that measures the influence of a fiber based on the distance @xmath53 from the center of the probe to the center of the removed fiber : @xmath202 here , for simplicity we characterize locality by the strength @xmath203 of the power - law decay beyond a short - range cutoff @xmath27 , which we have introduced to eliminate unphysical divergences .
our introduction of a variable exponent @xmath203 will allow us to later check how the variance of @xmath25 ( and consequently our estimate for the local stiffness distribution ) depends on the strength of locality . for strong locality , i.e. equal to the rapid @xmath204 decay of stiffness loss @xmath8 in three dimensions , @xmath6
is strongly correlated with @xmath25 for all types of networks studied ( fig .
s4__a__-_c _ and fig .
s5__a__-_c _ , _ insets _ ) .
this correlation suggests that @xmath25 explains much of the variance of local stiffness .
however , the geometric mean of the marginal distribution of local stiffnesses @xmath205 scales rapidly and nonlinearly with @xmath25 .
consequently , the geometric standard deviation of @xmath25 is less than half of that of @xmath6 .
this underestimation occurs because fibers have inherently cooperative contributions to the local stiffness that are not captured by the weighted sum in eq .
( s38 ) , which implicitly assumes that fibers contribute to @xmath6 independently . to understand how the strength of the collective effects scales with fiber density , we considered the macroscopic shear modulus as a function of average coordination number @xmath21 , or equivalently @xmath33 for a fixed density of vertices .
we found that the scaling of the shear modulus with @xmath21 was consistent with the scaling of @xmath205 with @xmath25 ( fig .
s4__d__-_f _ , fig
. s5__d__-_f _ ) .
we therefore estimated the local stiffness distribution for all types of networks by transforming the numerically - obtained distribution of @xmath25 by the functional dependence of the corresponding shear modulus @xmath206 for a network with average number of fibers equal to @xmath33 , which resulted in a geometric standard deviation given by : @xmath207 ^ 2 p(n_f ) dn_f } , \ ] ] where @xmath208 for strong locality , which for concreteness we consider throughout our analysis to be @xmath209 ( equal to the rapid @xmath28 scaling of the stiffness loss ) , the estimator captures the majority of the geometric standard deviation for all types of networks we studied ( table 1 , fig .
1__c__,_d _ , fig .
s1__f _ _ , fig .
s2__b _ _ , and fig .
s4__f _ _ ) .
furthermore , the estimate does not depend sensitively on the details of the form of the weighting function , as long as the decay is rapid enough ( fig .
s4__d__,_e _ ) .
importantly , this estimator captures the large quantitative discrepancy between the width of the experimental and rgg networks versus that of the lattice networks ( table 1 ) .
[ cols="^,^,^,^,^,^",options="header " , ] + * table 1 : * numerically - simulated and theoretically - estimated geometric standard deviations for all types of networks studied .
however , for weak locality , which we consider throughout our analysis as @xmath210 , the estimated geometric standard deviation is significantly smaller than the observed value ( fig .
1__c__,_d _ , fig .
s1__f _ _ , fig .
s2__b _ _ , and fig .
s4__f _ _ ) .
in this section , we present a simple model for the mean stiffness measured by a local stiffness probe interacting with an isotropic continuum elastic material in three dimensions . to gain insight into the average mechanical response of a disordered fiber network to a local force - dipole , we approximate the network as an isotropic continuum elastic material ( 41 ) .
the response of such an isotropic continuum elastic material can be completely characterized by two macroscopic moduli that appear in the elastic free energy : @xmath211 where @xmath212 is the strain tensor for the deformation field @xmath213 , @xmath214 is the first lame coefficient , and @xmath45 is the shear modulus . to determine these elastic parameters of our model , we calculate the shear modulus @xmath45 and the bulk modulus @xmath215 of the network . to compute the shear modulus of a network
, we simulate the deformation of a cubical region of network in response to a shear stress .
we apply a shear stress via force monopoles applied to all nodes within buffer zones at the top and bottom of the network ( see fig .
s6__a _ _ ) .
the deformations of the vertices are obtained by solving the force - balance equation for static equilibrium using svd ( see _ numerical procedure _ ) .
the shear modulus is defined as @xmath216 where @xmath217 is the sum of the magnitudes of all the forces applied to the network , @xmath218 is the area of the buffer region , and @xmath219 is the average shear strain obtained from the least - squares linear fit of the deformation profile in the direction of the applied force . to compute the bulk modulus , we simulate the deformation of a spherical region of network in response to a uniform compression .
we apply a uniform compression via force - monopoles applied to nodes at the edge of the network ( see fig .
s6__b _ _ ) .
the deformations of the vertices are obtained as before by solving the force - balance equation for static equilibrium using svd .
the bulk modulus is defined as @xmath220 where @xmath217 is the sum of the magnitudes of all the forces applied to the network , @xmath9 is the distance from the center of the network to the center of the buffer zone , and @xmath221 is the average radial strain of the nodes within the bulk of the network .
here , the geometric prefactor @xmath222 arises from the spherical geometry .
the local stiffness probe consists of two force - monopoles separated by a distance @xmath223 which exert a contractile force of magnitude @xmath101 ( see figure s6__c _ _ ) : @xmath224\hat{\boldsymbol{d}}.\ ] ] the deformation produced by each force - monopole is given by the continuum elastic green s tensor ( eq .
the continuum elastic green s tensor relates applied forces to the resulting deformations of the medium as follows : @xmath225 thus , the combination of both force - monopoles acting on the medium results in a deformation @xmath226 at each point @xmath198 within the medium : @xmath227\cdot\hat{\boldsymbol{d}}.\ ] ] we would like to define a local stiffness probe that measures the sum of the deformation @xmath228 at each monopole along the direction of the probe : @xmath229 however , in contrast to the network model , here in the continuum model , the force - monopoles induce diverging deformations at the points where the forces are applied . to account for the unphysical divergences ,
we define the measured deformation at each point of force application as the average deformation within a spherical cutoff region of radius @xmath123 centered around the point , where @xmath123 is set equal to the mesh size .
intuitively , this choice of cutoff results in a modeled deformation that corresponds to a coarse - grained description of the deformation of a single vertex in the network .
the total measured deformation is then given by : @xmath230 we evaluate the integral assuming @xmath231 and find the local stiffness measured by the `` continuum dipole '' is : @xmath232 where we have defined two constants that depend on the elastic moduli of the network : @xmath233 @xmath234 the continuum theory results in monotonically decreasing stiffness as a function of probe length , with stiffness asymptotically approaching a constant value for @xmath235 .
this occurs because the measured deformations are produced by two force - monopoles , which add in a linear fashion .
since the self - deformation @xmath236 of each monopole remains constant , and the contribution of the far monopole at @xmath51 decays to zero for @xmath235 , the local stiffness saturates for large probe lengths , i.e. it approaches a constant value determined by the combined deformations of two non - interacting monopoles . to study the dynamic range of local stiffness as a function of probe length
, we consider the fold - stiffening , defined as the ratio @xmath237 of the geometric mean stiffness to the geometric mean stiffness for asymptotically long probes ( i.e. @xmath238 ) : @xmath239 this fold - stiffening predicted from continuum theory is roughly consistent with the fold - stiffening for all types of discrete networks studied when the probe length is at least five times the mesh size ( fig .
s7 ) . despite this match in the fold - stiffening at long probe lengths ,
the observed local stiffness for short probe lengths is always much stiffer than the continuum prediction for the experimental and rgg networks ( fig .
s7__a__-_c _ ) .
this discrepancy occurs because the continuum theory does not adequately describe the deformation produced by a monopole at short distances for these networks .
this results in a fixed difference between the force - monopole self - deformation predicted by continuum theory and the average deformation produced by a force - monopole applied to the discrete networks .
below around five times the mesh size , the fold - stiffening measured for the experimental and rgg networks begins to deviate rapidly from the continuum prediction ( fig .
s7__a__-_c _ ) .
in contrast , the continuum theory seems to capture the fold - stiffening of the fcc network , which is fairly constant , even down to probe lengths close to the lattice spacing ( fig .
what determines the length scale of the enhanced fold - stiffening we observed for the experimental and rgg networks ? in the previous sections , we saw that for fixed probe length , the range of stiffnesses is determined in part by the strong , nonlinear dependence of the marginal distribution of local stiffnesses on network connectivity , with the strongest dependence occurring close to the stretching and bending transitions .
does the scaling of the marginal distribution of local stiffnesses also depend on probe length ?
if so , the large fold - stiffening we observed for the experimental and rgg networks could occur due to the proximity to an elastic transition . to determine how network connectivity impacts the fold - stiffening
, we varied the average coordination number @xmath21 for a fixed ratio of the bending modulus @xmath3 to the stretching modulus @xmath2 .
we found that for the rgg network , the dependence of fold - stiffening on probe length becomes much stronger as the average coordination number is tuned to bring the networks closer to an elastic transition ( fig .
s8 ) . however , even far away from the transitions , there is still a large fold - stiffening at short probe lengths .
since only the experimental and rgg networks have this large fold - stiffening away from the transitions , this property must arise from intrinsic features of the networks that are not present for the fcc network .
one feature of the network architecture that explicitly introduces a length scale is the distribution of fiber lengths ( fig .
s9 ) .
to predict the probe length for which we expect the crossover to short - range stiffening to occur due to the presence of polydisperse fibers , we consider the distribution of fiber lengths for each network .
for the disordered lattice networks , we define fiber length as the number of bonds in a filament that are consecutive coaxial segments uninterrupted by a non - coaxial connected segment .
we approximate the distribution of fiber lengths @xmath240 as proportional to the probability of starting from a fiber endpoint and observing @xmath133 consecutive bonds followed by either a cross - linking ( i.e. non - coaxial ) bond or the absence of a bond , yielding the following distribution : @xmath241^{\ell},\ ] ] where @xmath21 is the coordination number ( 6 for the sc lattice and 12 for an fcc lattice ) .
thus , we find an exponentially decaying distribution with a decay length @xmath132 given by : @xmath242 we see that for the lattice networks , the distribution of fibers decays exponentially . for the fcc network at connectivity @xmath145 ,
the decay length is around @xmath243 .
this is approximately an order of magnitude shorter than the minimum probe length we considered ( @xmath244 ) , which explains why we did not observe the fold - stiffening to deviate from the continuum prediction away from the elastic transitions .
in contrast to the relatively monodisperse fiber - length distribution for the fcc network , the experimental and rgg networks contain long fibers that connect vertices separated by distances that are well beyond the mesh size . beyond the mesh size @xmath123 ,
the fiber - length distribution decays roughly exponentially with a length scale approximately equal to the mesh size @xmath245 .
thus , we expect the long fibers to modify the fold - stiffening out to intermediate probe lengths .
could this effect account for the larger fold - stiffening we observed for the experimental and rgg networks ? the long - ranged structure in the network , e.g. fibers that are several times longer than the probe lengths we considered , precludes us from considering the free energy in the hydrodynamic limit .
thus , to properly describe these networks at these intermediate length scales , we should also consider terms in the free energy that consist , for instance , of gradients of the strain tensor .
although many higher - order terms could contribute at such scales , for simplicity we consider an extension that contains a single additional term ( 42 ) .
specifically , for each term in the original elastic energy which contains the squared - sum of the diagonal components of the strain tensor @xmath246 or the squared - sum of the individual components @xmath247 , we also add the vector product of the gradient of each term with itself : @xmath248 where @xmath249 is the partial derivative of @xmath250 with respect to the coordinate represented by index @xmath251 , and @xmath132 is an additional material parameter that corresponds to the extent of the microscopic non - locality . by setting @xmath132 equal to the decay length obtained from the fiber length distributions of the networks ,
we show that this gradient elasticity theory captures the essential features shown in our simulations .
the green s tensor for this non - local continuum theory is given by : @xmath252,\ ] ] where , as in ( 42 ) , we have defined the two functions @xmath253 \right \},\ ] ] @xmath254.\ ] ] incorporating the additional effect of the strain gradient into the continuum prediction for the experimental and rgg networks results in a fold - stiffening that is approximately a decade at the mesh size @xmath123 , which is significantly larger than the amount predicted by conventional continuum elasticity and captures a majority of the anomalously large fold - stiffening observed for the experimental and rgg networks ( fig .
s7 ) . in contrast , for the fcc network , the gradient prediction is nearly identical to the conventional continuum prediction , due to the much smaller value of @xmath132 relative to @xmath123 .
these results suggest that accounting for the inherent non - locality from highly polydisperse fibers is crucial to describe the response of biopolymer networks at the cellular scale .
how accurately can cells infer the _ global _ stiffness properties of their surroundings based solely on _ local _ measurements ? as we showed above , the global properties are closely related to the overall behavior of the local stiffness distribution .
specifically , the median stiffness tracks the scaling with network connectivity of the macroscopic response .
this suggests that a suitable average of several samples of local stiffness could provide useful information about the bulk properties of the network . to quantify the accuracy of inference
, we considered an idealized sampling process that consists of averaging @xmath122 local stiffness measurements obtained within prolate spheroids . for a given spheroid ,
the inferred stiffness is given by the geometric mean of @xmath122 measurements randomly chosen among all the probes whose centers are contained within its volume .
we define the relative uncertainty of this estimate as the geometric standard deviation @xmath255 of the resulting distribution of inferred stiffnesses .
this uncertainty is minimized for independent samples , for which @xmath255 is related to the logarithm of the geometric standard deviation @xmath256 of the distribution of local stiffnesses as follows : in general , however , stiffness measurements are correlated in space because nearby stiffness measurements depend on shared local structure . to quantify these correlations , we consider the two - point geometric correlation function @xmath258 for local stiffness measurements @xmath259 and @xmath260 , defined as follows : we calculated the two - point geometric correlation function for the experimental and rgg networks and found @xmath258 that were consistent with a decaying exponential form ( fig .
4__b _ _ , lower _ inset _ , fig .
s10a , _ inset _ ) : where @xmath263 is the separation between the centers @xmath264 of the stiffness probes , @xmath265 is the two - point stiffness correlation length , and @xmath266 is the zero - distance geometric covariance of local stiffnesses whose centers lie at the same point ( @xmath266 is less than @xmath80 because the centers of distinct probes may coincide ) .
the correlation length @xmath265 and the zero - distance geometric covariance @xmath266 are obtained by fitting to the modeled local stiffnesses . due to the presence of these spatial correlations ,
the uncertainty of inference @xmath255 will be larger than the uncertainty @xmath267 predicted by assuming samples of local stiffness are independent . provided that we can approximate the uncertainty using only the two - point geometric correlation function , the geometric standard deviation @xmath268 ( where @xmath269 is the position of sample @xmath72 for each of the @xmath122 samples )
is given by : for our idealized sampling process , the distances between the samples vary because the probes are positioned randomly throughout the network . in this case , each configuration provides an additive contribution to the variance with a weight proportional to its probability .
since the probes are distributed homogeneously and isotropically throughout space , the overall variance of the inference is simply given by integrating the positions of the @xmath122 samples over the sampling volume and normalizing by @xmath271 .
upon substituting the form of @xmath258 from above , we find an uncertainty @xmath255 given by : we measured the uncertainty @xmath275 of the two - sample stiffness inference process for the rgg network and confirmed that it matches the above direct calculation based on the two - point geometric correlation of local stiffnesses of the network ( fig .
this consistency demonstrates that the overall uncertainty of inference for sampling two stiffnesses within a volume can be characterized in terms of the underlying two - point correlations among stiffness measurements .
|
cells actively probe and respond to the stiffness of their surroundings . since mechanosensory cells in connective tissue are surrounded by a disordered network of biopolymers , their _ in vivo _ mechanical environment can be extremely heterogeneous . here , we investigate how this heterogeneity impacts mechanosensing by modeling the cell as an idealized local stiffness sensor inside a disordered fiber network . for all types of networks
we study , including experimentally - imaged collagen and fibrin architectures , we find that measurements applied at different points throughout a given network yield a strikingly broad range of local stiffnesses , spanning roughly two decades .
we verify via simulations and scaling arguments that this broad range of local stiffnesses is a generic property of disordered fiber networks , and show that the range can be further increased by tuning specific network features , including the presence of long fibers and the proximity to elastic transitions .
these features additionally allow for a highly tunable dependence of stiffness on probe length . finally , we show that to obtain optimal , reliable estimates of global tissue stiffness , a cell must adjust its size , shape , and position to integrate multiple stiffness measurements over extended regions of space .
mechanical cues can govern cellular behavior in decisive ways ( 1 - 2 ) .
the elastic properties of a cell s substrate have been shown to guide cell migration ( 3 , 4 ) and determine cell fate ( 5 , 6 ) .
eukaryotic cells , including fibroblasts , mesenchymal stem cells , and cancer cells , attach to substrates via transmembrane protein complexes called focal adhesions , allowing the cell to sense stiffness ( 2 , 7 - 10 ) .
knockdown studies have established that this mechanosensing contributes both to motility and to the regulation of cell shape in three - dimensional _ in vitro
_ systems that closely resemble _ in vivo _ cellular environments ( 10 - 13 ) , where cells are surrounded by a loosely connected , disordered network of protein fibers such as collagen or fibrin ( 14 ) .
these biopolymers form a major structural component of the extracellular matrix ( ecm ) , which serves as the physical scaffolding within which cells live and move .
while it is clear that cells actively probe these extracellular networks , it remains unclear how ecm micromechanical properties impact mechanosensing .
+ both _ in vitro _ experiments ( 15 - 19 ) and theory ( 20 - 24 ) have demonstrated that biopolymer networks exhibit rich _ macroscopic _ mechanical behavior , depending sensitively on network connectivity .
however , because the size of a typical cell is comparable to the pore size of the ecm ( 25 ) , any mechanical information must be inferred by locally probing an extremely heterogeneous material .
although a few studies have begun to characterize this _ microscopic _ response ( 26 , 27 ) , a theoretical understanding of how local mechanics are determined by the surrounding heterogeneous structure is still lacking .
+ how does the intrinsic heterogeneity of the ecm limit a cell s ability to learn about its _
global _ environment from purely _ local _ mechanical measurements ? for the case of chemical sensing , consideration of the fundamental physical limits dates back to berg and purcell s consideration of noise due to the random arrival of diffusing particles ( 28 - 30 ) . here
, we take a similar approach to explore the fundamental limits of mechanosensing , where in contrast to chemical signals , the cues are static in time but distributed nonuniformly in space . + to quantify the physical limits of mechanosensing imposed by a cell s disordered environment , we investigate a simple model consisting of two components : the ecm as an elastic network that deforms in response to external forces , and the cell as an idealized measurement device that probes the stiffness of its surroundings .
we found that experimentally - imaged collagen and fibrin networks and randomly - generated networks all yield a _ very _ broad range of modeled local stiffness responses , spanning roughly two decades .
we trace the origin of this broad range of stiffnesses to two intrinsic features of disordered networks : ( _ i _ ) the local stiffness depends primarily on a small number of local fibers with consequently large variations , and ( _ ii _ ) these proximal fibers contribute to stiffness in a highly cooperative manner .
the combined effect is a large uncertainty in global stiffness inference , which we demonstrate is only further increased by the presence of long fibers , proximity to elastic transitions , and correlations among nearby measurements . finally , we argue that to obtain accurate estimates of global ecm stiffness
, cells integrate multiple stiffness measurements over extended regions of space .
|
very high redshift ( @xmath3 ) gamma - ray bursts ( grbs ) ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) provide an excellent probe of both the interstellar ( ism ) and intergalactic medium ( igm ) at the epoch of reionization ( eor ) using absorption spectrum techniques thanks to their simple power - law afterglow spectra and high luminosity @xcite , complimentary to quasar absorption spectrum observations @xcite . here
we present a first , detailed analysis of the physical properties of ism surrounding grbs , utilizing state - of - the - art radiation - hydrodynamic simulations , with the hope that they may aid in proper interpretations of observations of grb afterglows at eor with respect to both ism and igm .
the simulations are performed using the eulerian adaptive mesh refinement code , ramses ( * ? ? ?
* ver . 3.07 ) , with concurrent multi - group radiative transfer ( rt ) calculation @xcite .
the reader is referred to @xcite for details .
notably , a new treatment of supernova feedback is implemented , which is shown to capture the sedov solution for all phases ( from early free expansion to late snowplow ) . the initial condition for the cosmological simulations
is generated using the music software @xcite , with the wmap7 parameters @xcite : @xmath4 .
the total simulated volume of @xmath5 ( comoving ) is covered with @xmath6 root grids , and 3 more levels are added to a rectangular region of @xmath7 mpc to achieve a high dark matter mass resolution of @xmath8 . in the zoomed - in region , cells are further refined ( 12 more levels ) based on the density and mass enclosed within a cell .
the corresponding maximum spatial resolution of the simulation is 4.2 pc ( physical ) .
the simulation is found to be consistent with a variety of observations , including the luminosity function at @xmath9 .
normal and runaway star particles are created in convergent flows with a local hydrogen number density @xmath10 ( fru run , * ? ? ?
* ) , based on the schmidt law @xcite . note that the threshold is motivated by the density of a larson - penston profile @xcite at @xmath11 , @xmath12 , where @xmath13 is the sound speed at the typical temperature of the ism ( @xmath14 ) and @xmath15 is the finest cell resolution .
additionally , we ensure that the gas is jeans unstable , and that the cooling time is shorter than the dynamical time ( e.g. * ? ? ? * ) .
we assume that 2% of the star - forming gas is converted into stars per its free - fall time ( @xmath16 ) @xcite .
the mass of each star particle is determined as @xmath17 , where @xmath18 is the threshold density , and @xmath19 is a parameter that controls the minimum mass of a star particle ( @xmath20 ) .
@xmath21 is an integer multiple of @xmath20 to be formed in a cell , which is drawn from a poisson random distribution , @xmath22 with the poissonian mean @xmath23 , where @xmath24 is the simulation time step .
the resulting minimum mass of a normal ( runaway ) star particle is @xmath25 .
we adopt the chabrier initial mass function to compute the mean frequency of type ii supernova explosions per solar mass ( @xmath26 ) .
dark matter halos are identified using the amiga halo finder @xcite .
this yields 731 halos of mass @xmath27 at @xmath28 .
we adopt the assumption that long - duration grb rate is proportional to type ii supernova rate ; short - duration grbs are not addressed here since they appear not to be associated with massive stars and are hosted by elliptical galaxies @xcite . for our analysis
we use all snapshots between @xmath29 and @xmath28 .
we present results from our simulations and make comparisons to available observations . as will be clear later
, we generally find broad agreement between our model and observations , although larger observational samples of grb afterglows would be needed to fully test the model .
we first describe the physical conditions of the ism that embeds grbs on pc scales .
the left ( right ) panel of figure [ fig : nt ] shows the distribution of grb rate in the density - temperature ( density - metallicity ) parameter space .
we see two separate concentrations of grbs in the @xmath30 plane , with @xmath31 equal to @xmath32 and @xmath33 , respectively . it must be made clear that the density and temperature are defined on the local gas cell of scale of a few pc that a grb sits .
the appearance of grb afterglow spectra depends , in most case , more strongly on the properties of gas along the line of sight rather than the gas immediately embedding them , as will be shown later . it is seen that most of the grbs reside in the low density , high temperature peak , contributing to 87% of grbs .
it is easy to identify two corresponding concentrations in the @xmath34 plane in the right panel : the low density , high temperature peak corresponds the high metallicity peak in the range @xmath35 $ ] in solar units , while the high density , low temperature peak corresponds the low metallicity peak in the range @xmath36 $ ] .
we note that super solar metallicities in hot winds driven by type ii supernova explosions in starburst galaxies in conditions similar to those of our simulated galaxies are locally observed .
for example , @xcite observe metallicity of hot x - ray gas of @xmath37 times the solar value in m82 .
@xcite find that the best fit model for the hot x - ray gas metallicity in dwarf starburst galaxy ngc 1569 is solar , although a metallicity as high as 5 times solar still gives @xmath38 value only about 0.1% larger than the best fit model ; on the contrary , the model with 0.25 times solar has much worse @xmath38 value .
the extant observations of grb afterglow spectra do not have the capability to detect the metallicity of the x - ray absorbing medium of relatively low column density .
metallicities of lower temperature gas phases are observed and are predicted to be substantially sub solar , as will be shown in figure [ fig : zg ] later .
-0.0 cm -0.0 cm -0.0 cm the @xmath39 peak coincides with the cores of dense gas clouds in the simulation , where star formation is centered .
the temperature of @xmath40 is mostly produced by atomic hydrogen cooling and the lower temperature extension due to metal cooling included in the simulation . as a numerical example , for a gas parcel of density @xmath41@xmath1 , temperature of @xmath41k and metallicity of 1% of solar value , the cooling time is about @xmath42 yrs , which is much shorter than relevant dynamic time scales .
it is thus clear that the cold density phase seen our simulation is easily understandable .
however , due to lack of treatment for molecular hydrogen cooling , gas is unable to cool significantly below @xmath43k .
had we included molecular hydrogen cooling and low temperature metal cooling , we expect the gas to cool approximately isobarically to about @xmath44k .
thus , we prefer not to infer any observable properties of grbs that would depend strongly on the nature of this cold gas phase , such as molecular clouds
. it is more appropriate to treat @xmath39 as bounds : @xmath45 .
this is also why we present our results for the high - density low - temperature regions as bounds in the abstract and conclusions sections as well as places where clarification is helpful .
the noticeable sub - dominance of grbs residing in very high density ( @xmath46@xmath1 ) , star - forming regions suggests that a large number of stars are displaced from their birth clouds .
this may be achieved by a substantial relative motions between stars and their birth clouds due to hydrodynamic interactions of the later or dynamical effects of stars . as a numerical illustration for the former possibility ,
a relative motion of @xmath47 between the birth cloud and the star would yield a displacement of @xmath48pc in a lifetime of @xmath49myr .
we note that the runaway ob stars in our simulation have typical velocities relative to the birth clouds of @xmath50 .
thus the runaway ob stars have contributed significantly to the displacing grbs from their birth clouds .
the grbs being in hot low density environment is also a result of supernova heating by earlier supernovae exploding in the birth clouds .
we estimate that these two effects are responsible about equally for placing most of the grbs in low - density high temperature regions .
while it is not possible to locate grbs within the host galaxies at high redshift at this time , observations of low redshift grbs may still be instructive .
@xcite show that grb 980425 occurring in a nearby ( @xmath51 ) sbc - type dwarf galaxy appears to be displaced from the nearest h ii region by @xmath52kpc , which is in fact significantly larger than the displacement distances for the vast majority of our simulated grbs in high redshift galaxies .
interestingly , the optical afterglow luminosity has a bimodal distribution at 12 hours after trigger @xcite .
the bimodal distribution of volumetric density seen in the left panel of figure [ fig : nt ] alone should produce a bimodal distribution of the afterglows with respect to break frequencies , luminosities , and break times , etc ( e.g. , * ? ? ?
we can not make detailed comparisons , because the circumburst density of the high @xmath53 grb subset is underestimated due to our limited resolution and because it remains uncertain if the appearance of grb afterglows would also depend strongly on intervening material ( dust obscuration , etc ) .
it is suggestive that the complex situations seen in simulations may account for the observed bimodality of afterglows without having to invoke intrinsic bimodality of grbs .
one of the most important points that this paper hopes to highlight and convey is that the appearances of the afterglows of grbs are not solely determined by the circumburst medium in their immediate vicinity ( e.g. , the physical conditions shown in figure [ fig : nt ] are on pc scales centered on grbs ) .
they also strongly depend on the line of sight beyond the immediate circumburst medium through the ism in the host galaxy , which we now quantify .
let us first give the meaning for our chosen value of the intervening neutral hydrogen column density @xmath54@xmath55 , which is used in figure [ fig : nt ] to separate grbs into separate groups .
figure [ fig : nhitoge ] shows the distribution of neutral hydrogen column density integrated along the line of sight for all grbs , separately for halos in five mass ranges .
a bimodal distribution of @xmath56 is seen , peaked at @xmath57@xmath55 and @xmath58@xmath55 , respectively , well separated by @xmath59@xmath55 .
it is clear that the bimodality exists for all halo masses surveyed .
the low @xmath56 peak is rather broad , extending all the way to @xmath60@xmath55 , suggesting some locations of grbs well into the diffused hot ism .
there is a noticeable dip in the neutral hydrogen column density distribution at @xmath61@xmath55 for the most massive galaxies of @xmath62 .
we attribute this to more significant shock heating in the most massive halos .
-0.0 cm -0.0 cm returning to figure [ fig : nt ] , it is now easy to see that the grbs in the low @xmath56 ( @xmath63@xmath55 ) peak in figure [ fig : nhitoge ] is composed of only one set of grbs situated in low density environment around @xmath32 , seen as the red contours in figure [ fig : nt ] .
the high @xmath56 ( @xmath64@xmath55 ) peak in figure [ fig : nhitoge ] , on the other hand , consists of a combination of two separate populations with distinctly different circumburst medium , which correspond to two separate loci of the blue contours at @xmath32 and @xmath33 in figure [ fig : nt ] .
the apparently two different groups of grbs situated around @xmath65 - one with low @xmath56@xmath63@xmath55 ( red ) and the other with low @xmath56@xmath64@xmath55 ( blue ) - are due entirely to the line of sight through the ism of the host galaxy .
overall , we find that 38% of grbs have @xmath56@xmath66@xmath55 ( i.e. , optically thin to lyman continuum ) , whereas 44% have @xmath56@xmath67@xmath55 ( i.e , containing a damped lyman - alpha system ) .
it is clear that various properties of the afterglows of grbs , even sitting in the same very local environment on pc scales , may appear different due to different intervening interstellar gas and dust along the line of sight through the host galaxy . in summary
so far , there are three separate populations of grb afterglows are expected , if our model is correct .
one might classify them in the following simple way : ( 1 ) hnhn=(high volumetric density @xmath46@xmath1 , high neutral column density @xmath56@xmath64@xmath55 ) , ( 2 ) lnhn=(low volumetric density @xmath68@xmath1 , high neutral column density @xmath56@xmath64@xmath55 ) , ( 3 ) lnln=(low volumetric density @xmath68@xmath1 , high neutral column density @xmath56@xmath63@xmath55 ) .
again , types ( 2 ) and ( 3 ) are a result of different viewing angles , where type ( 2 ) is due to viewing angles through largely hot ionized gas and type ( 3 ) viewing angles going through cold and dense gas in addition .
@xcite analyze multi - wavelength observations of the afterglow of grb 120521c ( @xmath69 ) and re - analyze two previous grbs at @xmath70 ( grb 050904 and 090423 ) , and conclude that the circumburst medium has a volumetric density of @xmath71@xmath1 that is constant .
the grbs in the lnhn or lnln group provide the right match to the observations .
while the statistic is still small , it is expected that about 87% of grbs should arise in in either the lnhn or lnln group .
observations of grb 050904 at @xmath72 reveal that it contains a damped lyman alpha systems ( dlas ) system in the host galaxy of column density @xmath73@xmath55 and metallicity of @xmath74 to @xmath75 @xcite . based on x - ray observations , @xcite conclude that @xmath76 for grb 050904 .
the evidence thus suggests that grb 050904 likely resides in a dense environment , although it can not be completely sure because the metallicity range of the low - density peak ( right panel of figure [ fig : nt ] ) overlaps with the observed range .
it is useful at this juncture to distinguish between the metallicity of the local environment of a grb and that of absorbers in the grb afterglow spectrum .
cm -0.0 cm let us now turn to the expected metallicity of uv / optical absorbers in the grb afterflow spectra .
figure [ fig : zg ] shows the pdfs of total hydrogen - column - density - weighted metallicity of gas along the line of sight , excluding gas hotter than @xmath77k , for the three sub - populations of grbs .
we see that for all three grb groups the metallicity of the absorbers in the grb spectra peaks in the range @xmath78 to @xmath75 .
thus , it is now clear that our model can easily account for the observed properties of grb 050904 .
the additional evidence that , based on the analysis of the equivalent width ratio of the fine structure transition lines si ii *
@xmath791264.7 and si ii @xmath791260.4 , infers the electron density of @xmath80 .
furthermore , the magnitude of the optical afterglow at 3.4 days after the burst favors a high density circumburst medium . in combination , it appears that grb 050904 is likely in a dense environment being to the hnhn group .
this appears to be at some minor odds with our model , since we only expect that 13% of grbs to arise in the hnhn group
. it would be highly desirable to obtain a larger sample of high - z grbs to provide a statistically firmer test .
analyses of observations of grb 130606a at @xmath81 indicate that it likely contains a sub - dla system of @xmath82@xmath55 in the host galaxy @xcite .
the inferred low metallicity of @xmath83 to @xmath84 in solar units @xcite and @xmath85 to @xmath86 @xcite , in conjunction with the @xmath56 , suggests that grb 130606a may reside in a low density environment with a foreground sub - dla system in the host galaxy .
this proposal is consistent with the evidence of detection of highly ionized species ( e.g. , n v and si iv ) @xcite .
it seems likely that grb 130606a belongs to the lnhn group .
it is easy to see that in our model the metallicity distribution of uv / optical absorbers in grb afterglow spectra is wide , which itself is due to the very inhomegeneous metallicity distributions in the ism of the host galaxies .
thus , it would be a rather chancy practice trying to infer the metallicity of the host galaxy solely based on a small number of ( typically one ) grb afterglow absorption spectra .
the reader has already seen clearly that the distributions of all concerned physical quantities , including metallicity , density , total and neutral hydrogen column density , are wide .
we will add yet one more quantity and show the cumulative distributions of the ratio of neutral hydrogen to total hydrogen column density for the three groups in the right panel of figure [ fig : nhn ] .
we see that for grbs in the lnln group the neutral hydrogen to total hydrogen column ratio is significantly less than unity . even for the hnhn and lnhn groups , ( 10%,14% ) of grbs
have the ratio less than @xmath87 .
in other words , it is generally a pretty bad assumption that the apparent absorbers in the grb afterglow spectra are mainly neutral .
this indicates that the so - called missing gas problem " ( e.g. , * ? ? ?
* ) may be accommodated in this model .
0.5 cm -0.0 cm the left panel of figure [ fig : nhn ] shows the pdfs of the total hydrogen column density for the three sub - populations of grbs , which is most relevant for probing grb x - ray afterglows and hence a useful test of our model .
one expectation from our model is that the vast majority of grbs sitting in low density circumburst medium ( lnhn + lnln ) do not have compton thick obscuring gas .
this prediction is verifiable with a combination of afterglow light curves and x - ray observations . on the other hand ,
one expects from our model that a significant fraction of the grbs sitting in high density circumburst medium ( hnhn ) have an extended high @xmath88 tail and dominate the grbs with @xmath89@xmath55 .
quantitatively , we find that ( 45% , 3% ) of grbs have @xmath90@xmath55 ; it is noted that these two numbers are likely lower bounds due to possible numerical resolution effects .
as already noted earlier , it is seen from the right panel that the grbs in the lnln group are intervened by highly ionized gas peaking at an average neutral fraction of @xmath91 , with no cases having a neutral fraction exceeding @xmath92 .
in contrast , for grbs in both the hnln and hnhn groups , more than 50% of them have an average neutral fraction greater than @xmath93 .
0.5 cm -0.0 cm -0.0 cm we now turn to the issue of dust obscuration .
the left panel of figure [ fig : av ] shows the pdfs of visual extinction @xmath94 .
it is noted that the simulation does not follow dust formation explicitly .
thus , we have adopted the well known empirical relation between metal column density and visual extinction : @xmath95 @xcite . while the applicability of the relation derived from local observations is uncertain , detailed analysis of galaxy colors at eor
suggest that the simulated galaxies based on this relations give rise to self - consistent results when comparing to observations @xcite .
moreover , direct observations of dust suggest that this relation holds well in other galaxies locally , and galaxies and damped lyman alpha systems at moderate to high redshift ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . nevertheless , it is possible that the normalization factor in front of the relation is probably uncertain to order of unity .
it is evident that a significant fraction of grbs in the high hnhl group ( blue solid curve ) are heavily dust obscured , with ( 53%,16% ) of grbs in the hnhl group have @xmath96 . at the other extreme
, we see that the grbs in the lnln group ( red dashed curve ) have negligible dust columns with no case of @xmath97 ; nevertheless , it is worth pointing out that , even for this set of grbs 12% has an @xmath98 due largely to dust in high temperature gas .
the grbs in the lnhn group ( green dotted curve ) situates inbetween the above two groups , with a small but non - negligible fraction ( 7% ) at @xmath99 .
observationally , the issue of dust in high - z grb hosts is less than settled .
@xcite , based on a re - analysis of the multi - epoch data of the afterglow of grb 050904 at @xmath72 , conclude that there is no evidence of dust .
given that the neutral column density of the in situ dla for grb 050904 is @xmath73@xmath55 and low metallicity @xmath74 to @xmath75 @xcite , an @xmath100 is possible , if one adopts the lower metallicity value that is statistically allowed in our model ( see the right panel of figure [ fig : nt ] ) .
thus , both the low extinction and a standard ratio of dust to metals are still consistent with the observations .
our model indicates that 9% of grbs have @xmath101 .
thus , with a @xmath3 grb sample of size @xmath102 , one expects to see one grbs with @xmath103 that is significantly obscured by dust with @xmath99
. this may be testable with swift data relatively soon .
the right panel of figure [ fig : av ] shows the pdfs of average gas temperature weighted by @xmath104 , excluding gas with temperature greater than @xmath77k ( the exclusion is a crude way to say that dust in gas hotter than @xmath77k is destroyed ) .
the purpose of this plot is to provide an indication the diversity of intervening gas with dust .
one notes that the lines of sight of hnhl grbs contain dust in cold medium ( @xmath105k ) , whereas those of high lnhn grbs are dominated by dust residing in gas @xmath106k , and the lnln grbs are intervened by dust in hotter gas of @xmath107k . under the assumption that the hotter gas is presumably produced by shocks , which are more destructive to larger dust grains
, one might suggest that dust becomes increasingly grayer from hnhl to lnhn to lnln .
one expectation is that some lines of sight , especially those for the hnln and hnhn groups the total dust arise from multiple , different temperature regions .
this may provide a physical explanation for the observational indications of multiple dust components ( e.g. , * ? ? ?
* ) . as a side note on the sfr of grb hosts .
@xcite place the star formation rate ( sfr ) of grb 080913 at @xmath108 to be less than @xmath109/yr .
@xcite obtain an upper bound on the star formation rate of grb 050904 at @xmath72 less than @xmath110/yr . from the simulations we find that ( 42%
, 57% , 66% ) of grbs occur in galaxies with sfr less than ( 0.3 , 1.0 , 3.0)@xmath111/yr .
thus , while simulations and observations are in good agreement , larger data sets are needed to place the comparisons on a solid statistical ground .
finally , we must stress that the analysis performed here has focused on the ism embedding the grbs at eor .
the exact details of the state of the igm at eor are uncertain both observationally and theoretically .
the theoretical difficulty is in part computational , because we do not have the capability to simulate a large enough volume to capture of the reionization of the igm self - consistently , while still having enough resolution for the ism .
the goal of this work is to present the signatures of the ism theoretically , which is lacking .
it may be argued that the properties of ism in galaxies are somewhat detached from the properties of the igm on large scales ; in other words , the observed spectra of grb afterglows at eor may be considered to be imprinted by both ism and igm as a linear superposition .
consequently , a proper understanding of the ism will not only aid in the interpretation of the ism of galaxies at eor but also is highly needed for proper interpretation of the properties ( neutral fraction , topology , etc ) of the igm at eor .
we perform an analysis to quantify the physical condition of the ism embedding grbs as well as intervening gas in the host galaxies at high redshift @xmath3 .
our analysis is based on a zoomed - in cosmological radiation hydrodynamics simulation of @xmath7 mpc@xmath112 box ( comoving ) with 731 halos of mass @xmath27 at @xmath28 at high spatial ( @xmath113 4 pc , physical ) and stellar mass resolution of 49 @xmath111 .
the following are new findings . on parsec scales ,
grbs are concentrated in two regions in density - temperature - metallicity space , with @xmath114 being @xmath115 and @xmath116 , consisting of 87% and 13% of grbs , respectively .
the appearance of grb afterglows , however , also strongly depends on the line of sight thanks to varying physical properties of intervening gas and dust in the host galaxy , which in turn splits the low density peak into two subsets . as a result , three separate apparent groups of grb afterglows composing of ( 13%,37%,50% )
are expected to arise with the following physical properties : ( 1 ) a cold neutral circumburst gas of hydrogen density of @xmath117@xmath1 , neutral hydrogen column density 90% range of @xmath118@xmath55 , total hydrogen column density 90% range of @xmath119@xmath55 , with ( 53%,16% ) having @xmath94 greater than ( 1,10 ) , ( 2 ) a hot circumburst gas of hydrogen density of @xmath120@xmath1 , neutral hydrogen column density 90% range of @xmath121@xmath55 , total hydrogen column density 90% range of @xmath122@xmath55 , with ( 7%,1% ) having @xmath94 greater than ( 1,10 ) , ( 3 ) a hot circumburst gas of hydrogen density of @xmath120@xmath1 , neutral hydrogen column density 90% range of @xmath123@xmath55 , total hydrogen column density 90% range of @xmath124@xmath55 , with ( 0.2%,0% ) having @xmath94 greater than ( 1,10 ) .
common among all three groups of grbs is that the metallicity of proximity optical / uv absorbers in the afterglow spectra is expected to be in the range of @xmath125 to @xmath75 .
the strikingly diverse physical properties - metallicity , neutral hydrogen column density , total column density , gas temperature , dust column - of intervening absorbers of grb afterglows as well as the bimodal physical properties of local ( parsec - scale ) environment indicates that a solid statistical comparison between the model predictions and observations needs to await a large observational sample of grbs at the eor .
it is obvious that the available small sample of grb afterglows complicates the task of interpretating the ism of a small number of grb host galaxies and the additional task of inferring the state of the igm at eor .
nevertheless , utilizing high - z grbs to probe interstellar medium and intergalactic medium must be undertaken properly taking into consideration the physical diversities of the interstellar medium .
the analysis presented will provide a physical framework to be confronted by future observations statistically .
@xcite find that grbs are more concentrated in the very brightest regions of their host galaxies than are the core - collapse supernovae .
our analysis so far has assumed that the two populations are just proportional to one another .
one may turn the argument around and use the observed distributions of grbs that depend on both embedding environment and intervening material to test the connection between grbs and star formation .
needless to say , larger samples will be able to shed light on this extremely important issue , which will have strong bearings on relating grb rates to cosmological reionization .
we would like to thank omer bromberg for discussion .
computing resources were in part provided by the nasa high- end computing ( hec ) program through the nasa advanced supercomputing ( nas ) division at ames research center .
the research is supported in part by nsf grant ast-1108700 and nasa grant nnx12af91 g . , e. , chary , r. , cowie , l. l. , price , p. a. , schmidt , b. p. , fox , d. b. , cenko , s. b. , djorgovski , s. g. , soderberg , a. m. , kulkarni , s. r. , mccarthy , p. j. , gladders , m. d. , peterson , b. a. , & barger , a. j. 2007 , , 665 , 102 , a. j. , snchez - ramrez , r. , ellison , s. l. , jelnek , m. , martn - carrillo , a. , bromm , v. , gorosabel , j. , bremer , m. , winters , j. m. , hanlon , l. , meegan , s. , topinka , m. , pandey , s. b. , guziy , s. , jeong , s. , sonbas , e. , pozanenko , a. s. , cunniffe , r. , fernndez - muoz , r. , ferrero , p. , gehrels , n. , hudec , r. , kubnek , p. , lara - gil , o. , muoz - martnez , v. f. , prez - ramrez , d. , v strobl , j. , lvarez - iglesias , c. , inasaridze , r. , rumyantsev , v. , volnova , a. , hellmich , s. , mottola , s. , castro cern , j. m. , cepa , j. , g , e. , gver , t. , nal tac s , . , park , i. h. , sabau - graziati , l. , & tejero , a. 2013 , arxiv e - prints , a. , levan , a. j. , fox , d. b. , tanvir , n. r. , ukwatta , t. n. , berger , e. , krhler , t. , kpc yolda , a. , wu , x. f. , toma , k. , greiner , j. , olivares , f. e. , rowlinson , a. , amati , l. , sakamoto , t. , roth , k. , stephens , a. , fritz , a. , fynbo , j. p. u. , hjorth , j. , malesani , d. , jakobsson , p. , wiersema , k. , obrien , p. t. , soderberg , a. m. , foley , r. j. , fruchter , a. s. , rhoads , j. , rutledge , r. e. , schmidt , b. p. , dopita , m. a. , podsiadlowski , p. , willingale , r. , wolf , c. , kulkarni , s. r. , & davanzo , p. 2011
, , 736 , 7 , b. t. , dale , d. a. , bendo , g. , gordon , k. d. , smith , j. d. t. , armus , l. , engelbracht , c. w. , helou , g. , kennicutt , jr . , r. c. , li , a. , roussel , h. , walter , f. , calzetti , d. , moustakas , j. , murphy , e. j. , rieke , g. h. , bot , c. , hollenbach , d. j. , sheth , k. , & teplitz , h. i. 2007 , , 663 , 866 , a. s. , levan , a. j. , strolger , l. , vreeswijk , p. m. , thorsett , s. e. , bersier , d. , burud , i. , castro cern , j. m. , castro - tirado , a. j. , conselice , c. , dahlen , t. , ferguson , h. c. , fynbo , j. p. u. , garnavich , p. m. , gibbons , r. a. , gorosabel , j. , gull , t. r. , hjorth , j. , holland , s. t. , kouveliotou , c. , levay , z. , livio , m. , metzger , m. r. , nugent , p. e. , petro , l. , pian , e. , rhoads , j. e. , riess , a. g. , sahu , k. c. , smette , a. , tanvir , n. r. , wijers , r. a. m. j. , & woosley , s. e. 2006 , , 441 , 463 , j. , krhler , t. , fynbo , j. p. u. , rossi , a. , schwarz , r. , klose , s. , savaglio , s. , tanvir , n. r. , mcbreen , s. , totani , t. , zhang , b. b. , wu , x. f. , watson , d. , barthelmy , s. d. , beardmore , a. p. , ferrero , p. , gehrels , n. , kann , d. a. , kawai , n. , yolda , a. k. , mszros , p. , milvang - jensen , b. , oates , s. r. , pierini , d. , schady , p. , toma , k. , vreeswijk , p. m. , yoldac s , a. , zhang , b. , afonso , p. , aoki , k. , burrows , d. n. , clemens , c. , filgas , r. , haiman , z. , hartmann , d. h. , hasinger , g. , hjorth , j. , jehin , e. , levan , a. j. , liang , e. w. , malesani , d. , pyo , t .- s . , schulze , s. , szokoly , g. , terada , k. , & wiersema , k. 2009 , , 693 , 1610 , n. , kosugi , g. , aoki , k. , yamada , t. , totani , t. , ohta , k. , iye , m. , hattori , t. , aoki , w. , furusawa , h. , hurley , k. , kawabata , k. s. , kobayashi , n. , komiyama , y. , mizumoto , y. , nomoto , k. , noumaru , j. , ogasawara , r. , sato , r. , sekiguchi , k. , shirasaki , y. , suzuki , m. , takata , t. , tamagawa , t. , terada , h. , watanabe , j. , yatsu , y. , & yoshida , a. 2006 , , 440 , 184 , e. , smith , k. m. , dunkley , j. , bennett , c. l. , gold , b. , hinshaw , g. , jarosik , n. , larson , d. , nolta , m. r. , page , l. , spergel , d. n. , halpern , m. , hill , r. s. , kogut , a. , limon , m. , meyer , s. s. , odegard , n. , tucker , g. s. , weiland , j. l. , wollack , e. , & wright , e. l. 2011 , , 192 , 18 , n. r. , fox , d. b. , levan , a. j. , berger , e. , wiersema , k. , fynbo , j. p. u. , cucchiara , a. , krhler , t. , gehrels , n. , bloom , j. s. , greiner , j. , evans , p. a. , rol , e. , olivares , f. , hjorth , j. , jakobsson , p. , farihi , j. , willingale , r. , starling , r. l. c. , cenko , s. b. , perley , d. , maund , j. r. , duke , j. , wijers , r. a. m. j. , adamson , a. j. , allan , a. , bremer , m. n. , burrows , d. n. , castro - tirado , a. j. , cavanagh , b. , de ugarte postigo , a. , dopita , m. a. , fatkhullin , t. a. , fruchter , a. s. , foley , r. j. , gorosabel , j. , kennea , j. , kerr , t. , klose , s. , krimm , h. a. , komarova , v. n. , kulkarni , s. r. , moskvitin , a. s. , mundell , c. g. , naylor , t. , page , k. , penprase , b. e. , perri , m. , podsiadlowski , p. , roth , k. , rutledge , r. e. , sakamoto , t. , schady , p. , schmidt , b. p. , soderberg , a. m. , sollerman , j. , stephens , a. w. , stratta , g. , ukwatta , t. n. , watson , d. , westra , e. , wold , t. , & wolf , c. 2009 , , 461 , 1254
|
analysis is performed on ultra - high resolution large - scale cosmological radiation - hydrodynamic simulations to , for the first time , quantify the physical environment of long - duration gamma - ray bursts ( grbs ) at the epoch of reionization .
we find that , on parsec scales , 13% of grbs remain in high density ( @xmath0@xmath1 ) low - temperature star - forming regions , whereas 87% of grbs occur in low - density ( @xmath2@xmath1 ) high temperature regions heated by supernovae .
more importantly , the spectral properties of grb afterglows , such as the neutral hydrogen column density , total hydrogen column density , dust column density , gas temperature and metallicity of intervening absorbers , vary strongly from sightline to sightline . although our model explains extant limited observationally inferred values with respect to circumburst density , metallicity , column density and dust properties , a substantially larger sample of high - z grb afterglows would be required to facilitate a statistically solid test of the model .
our findings indicate that any attempt to infer the physical properties ( such as metallicity ) of the interstellar medium of the host galaxy based on a very small number of ( usually one ) sightlines would be precarious . utilizing high - z grbs to probe interstellar medium and intergalactic medium should be undertaken properly taking into consideration the physical diversities of the interstellar medium .
= 1
|
shanon proved existence of maximum data transmission rate , called channel capacity @xcite . since then
, different capacity - approaching codes have been designed , like turbo codes @xcite and ldpc codes @xcite .
the first provable capacity - achieving codes , polar codes , were recently invented by arikan @xcite .
polar codes are considered to be a major breakthrough in coding theory , since they are the first family of codes known to achieve channel capacity with explicit construction . besides achieving the capacity for binary - input symmetric memoryless channels @xcite
, polar codes were also proved in @xcite to be able to achieve the capacity for any discrete and continuous memoryless channel . moreover ,
an explicit construction method for polar codes was provided and it was shown that they can be efficiently encoded and decoded with complexity @xmath0 , where @xmath1 is the code length . since then , polar codes have become one of the most popular topics in information theory and have attracted a lot of attention .
several decoding methods are available for decoding polar codes @xcite-@xcite , scd and its variants and bpd are two popular methods .
sc decoders suffer from long latency due to the serial nature of the sc algorithm .
however , the sc algorithm requires less computation as compared to bpd .
based on this property , several high - throughput low - cost sc decoders were reported in @xcite@xcite .
another advantage of the sc algorithm is its ability to achieve good error - correcting performance for long code lengths . for short code length , based on the scd
, the list - decoding or stack decoding method also achieve good error - correcting performance @xcite@xcite . on the other hand , polar bp decoders @xcite-@xcite have the intrinsic advantage of parallel processing .
therefore , compared with their sc counterparts , polar bp decoders are more attractive for low - latency applications . for iterative decoders ( such as polar bp decoders ) , the required latency and energy dissipation increase linearly with the number of iterations .
however , the need for a large number of iterations makes bp decoders suffer from high computation complexity , and hence polar bp decoders are still not as attractive as their sc counterparts . to this end , another decoding method , called soft cancellation ( scan ) decoding , is proposed in @xcite . by restricting the soft information propagation schedule in the decoding process
, the computational complexity of scan is much lower than that of bpd .
however , different from bpd , the scan operation is serial in nature , leading to a much longer decoding latency .
hence , aiming at the low - latency polar codes decoder , we concentrate on the bpd in this work . to address the issues of the large number of iterations and high computation complexity inherent in bp decoders , yuan et al .
@xcite proposed a g - matrix - based early stopping scheme , which is based on the fact that iterative decoders normally converge earlier than reaching a fixed maximum number of iterations . the g - matrix - based stopping criterion
can then be used to stop the computation if convergence has been reached . to further reduce the computation complexity , in this paper ,
we propose a method based on the convergence of the sub - factor - graphs , which is reached at a much earlier stage .
borrowing the idea from scd , some of the sub - factor - graphs are checked during each iteration and if they have converged , they are frozen and do not need to be computed in the subsequent iterations .
also the freezing of these sub - factor - graphs will help to improve the convergence of the decoding process over rest of the factor graph . as a result , the computation complexity and also the average number of iterations are reduced .
experimental results show that our proposed method results in about 40 ~ 46 % lower computation complexity , as well as lower latency , when compared to the previously proposed early stopping scheme @xcite . in this paper , the following notation conventions are used .
matrices are denoted in boldface capital letters , and vectors in boldface lowercase letters .
the subscript @xmath2 of a matrix represents an @xmath3 square matrix and @xmath4 denotes an @xmath5 vector .
@xmath6 $ ] stands for the @xmath7 element of vector @xmath8 , @xmath9 stands for vector @xmath8 at the @xmath10 iteration and @xmath11 represents the sub - vector of @xmath8 with the starting and ending index of @xmath12 and @xmath13 .
the transpose of a vector @xmath8 is denoted by @xmath14 .
polar codes are based on the phenomenon of channel polarization .
more precisely , by recursively combining and splitting individual channels , some of these channels become essentially error - free , while others become completely noisy .
furthermore , the fraction of the noiseless channels tends towards the capacity of the underlying binary symmetric channels @xcite .
therefore , an @xmath15 polar code can be generated in two steps .
first , an @xmath1-bit message @xmath16 is constructed by assigning the @xmath17 reliable and @xmath18 unreliable positions as information bits and 0 bits , respectively .
the @xmath18 unreliable positions , which are forced to 0 , are called the frozen bits ( also known as the frozen set @xmath19 ) .
then , the @xmath1 -bit @xmath16 is multiplied with the generator matrix @xmath20 to generate an @xmath1 -bit transmitted codeword @xmath8 , where @xmath21 is the @xmath22 kronecker power of @xmath23 $ ] and @xmath24 .
1 shows the encoding signal flow graph for @xmath25 polar codes , where the @xmath26 sign represents the xor operation . as presented in @xcite ,
similar to ldpc codes , polar codes can be decoded by applying the belief propagation ( bp ) algorithm over their factor graphs . for an @xmath15 polar code @xmath27 ,
the factor graph is an @xmath28-stage network consisting of @xmath29 nodes , where each node is associated with a right - to - left and a left - to - right likelihood message denoted by @xmath30 and @xmath31 , respectively .
@xmath32 denotes the right to left likelihood message of the @xmath7 node at the @xmath33 stage and the @xmath10 iteration .
fig.2 ( a ) shows an example of a 3-stage factor graph for @xmath25 polar codes . here
each stage consists of @xmath34 processing elements ( pes ) . during the bp decoding procedure ,
these messages are propagated and updated among adjacent nodes using the min - sum updating rule , as shown by the following equations @xcite : @xmath35 @xmath36 @xmath37 @xmath38 @xmath39 is a scaling parameter introduced in @xcite for the improvement of the decoding performance of a bp decoder . according to the decoding procedure of bp algorithm ,
pes are activated stage - by - stage from left to right in each iteration .
after the number of iterations reaches the specific maximum number ( max_iter ) , node @xmath40 will output the decoded information bit @xmath41 based on the hard decision of the messages @xmath42 .
3(a ) shows the scheduling tree of the successive cancellation decoding ( scd ) of the ( 8,4 ) polar code @xcite , and fig .
3(b ) depicts the equivalent bpd factor graph of the same ( 8,4 ) polar code . at each stage
the scd scheduling tree is split into a number of sub - trees , each of which is responsible for decoding a corresponding constituent code .
the size of the sub - tree varies at each level and is reduced by half when moving from one stage to another stage . before presenting the details of our proposed scheme , we first introduce the notion of the _ connected sub - factor - graph_. a connected sub - factor graph ( csfg ) is defined as a sub - factor - graph which has the same number of inputs and outputs and where the output nodes are at the stage @xmath43 and each input is connected to each output through some pes in the sub - factor - graph .
3(b ) shows two examples of csfgs .
it can be seen that each csfg has a corresponding sub - tree in the scheduling tree of scd .
3 ( a ) and ( b ) show examples of the corresponding sub - trees and the connected sub - factor - graph of the ( 8,4 ) polar code . the number of csfgs at each stage is given by @xmath44 , where @xmath45 is the stage number .
for the ( 8,4 ) polar code , as shown in fig .
3 ( c ) and ( d ) , the numbers of csfgs at stages 1 and 2 are 2 and 4 , respectively . at each iteration @xmath46 , the nodes at stage @xmath45 in the bpd factor graph output left - to - right llr - based propagating messages @xmath47 , and these are the inputs to the @xmath44 csfgs at stage @xmath45 .
@xmath48 are the inputs to the first csfg , while @xmath49are those for the @xmath50 csfg .
each csfg is responsible for the decoding of the corresponding constituent code from its respective input messages .
the proposed scheme borrows the idea of successive cancellation decoding ( scd ) , where the results of the previous - decoded bits are used for the decoding of the current bit .
here we introduce a csfg freezing concept for a low complexity bpd . at a particular iteration @xmath46 ,
when the message passing reaches a certain stage @xmath45 , if a csfg at that stage can correctly decode its corresponding constituent code ( i.e. the csfg has reached convergence ) , it is frozen and no message passing or updating within the csfg will be needed in the subsequent iterations .
the details of how to check whether a csfg can be frozen will be presented later .
one important thing is the checking order for the freezing of the csfg .
a csfg can only be frozen if all the previous csfgs ( in the order of the decoding bits ) at that stage have been frozen .
if a csfg is not frozen , it means the message values inside it will still be changed in the subsequent iterations .
similar to the scd operation , the message values of this csfg will be used for the decoding of the constituent codes of the subsequent csfgs .
therefore the freezing of the csfgs at a stage has to follow an order based on the decoded bit . when a csfg at a certain stage is checked for freezing ,
if it can not correctly decode its constituent code , then it can not be frozen and the message passing and updating have to be executed for pes at that stage .
after that , we move to the next stage and check the convergence of the corresponding csfgs .
when we move to the next stage , the number of csfgs will be doubled .
this freezing - checking procedure will continue from stage to stage until the end of the bpd factor graph is reached .
next we will present how we can freeze a csfg . as discussed above
, a csfg corresponds to a sub - tree in the scd scheduling tree , which can also be viewed as a constituent code of the original polar code . at the @xmath10 iteration and stage @xmath45 ,
the left - to - right propagation messages @xmath49 connected to the @xmath50 csfg can be viewed as the llr inputs to decode the corresponding constituent code .
we can apply maximum - likelihood decoding ( mld ) on this constituent code with @xmath49as input to obtain a decoded output vector @xmath51 , which is a sub - vector of the source word ( @xmath52 ) of the original polar code . as will be shown later ,
if the freezing of the csfgs follows the proposed order , the input messages of csfg @xmath49 are reliable enough and mld @xmath51 , based on these input messages , can be taken as the decoded result of the constituent code .
the freezing order of the csfg has to follow the decoded bit order , and the top csfgs at each stage will be frozen first .
4 shows the scd scheduling tree and the factor graph of the ( 8,4 ) polar code .
we can see that the top csfgs are actually corresponding to the first few sub - trees that follow the depth - first traversal of the scd scheduling tree .
at the first iteration , the input messages to these csfgs are the same as the input llr messages of the corresponding scd sub - trees .
hence if we can decode the input messages of these csfgs using mld , the decoding performance on the corresponding constituent code will achieve or even exceed that of scd .
if the csfgs can not be frozen at this iteration , and need further iteration to converge , due to the nature of the iterative decoding , the reliability of the input messages to these csfgs will become better and hence the input llr messages of these csfgs will be more reliable than the input messages to the scd sub - tree . as a result the mld performance will not be worse than that of scd .
mld is based on an exhaustive search and hence it has a huge complexity . to reduce the complexity ,
novel checking criterion is suggested to efficiently find the mld result of the constituent code .
let @xmath48 be the left - to - right propagation messages of a csfg at stage @xmath45 .
we obtain a hard decision vector @xmath53 $ ] for these messages where @xmath54 given @xmath55 as input to the csfg , the decoded bit vector at its output @xmath56 , which is also a sub - vector of the source word of the original polar code @xmath57 , is obtained by the inverse operation of polar code encoding that is given as @xmath58 @xmath59 fig .
5(a ) shows an example of hard decision decoding .
the csfg can be frozen if the sub - source - word vector @xmath56 satisfies the following frozen set criteria : @xmath60 the following lemma shows that if the frozen set criteria ( 4 ) are satisfied , the sub - source - word vector @xmath56 obtained by ( 3 ) is indeed the decoding results of the mld on the corresponding constituent code of the csfg .
let @xmath61and @xmath55 be the input llr messages and hard decision vector based on ( 2 ) for the corresponding csfg at the @xmath33 stage .
if @xmath56 is obtained from @xmath55 based on ( 3 ) and it satisfies the frozen - set criteria of ( 4 ) , then @xmath56 is the maximum likelihood detection ( mld ) result of the corresponding constituent code with input messages @xmath61 .
the csfg at the @xmath33 stage represents a short polar ( constituent ) code of length @xmath62 .
its input and output are related by @xmath63 . from @xcite and @xcite ,
given the input llr @xmath61 , the likelihood value of an arbitrary source word @xmath64 is given by @xmath65)r_{1:2^{m - j},j+1}[i]$ ] , where @xmath63 .
if no source word bit is a frozen bit , i.e. , @xmath41 can assume both 0 and 1 for @xmath66 , the source word @xmath56 obtained from @xmath55 has a maximum likelihood value which is equal to @xmath67|.$ ] if a certain source word bit is a frozen bit , the searching space of the valid source word is smaller and @xmath67|$ ] may not be achieved . however , if @xmath56 satisfies ( 4 ) , this likelihood value is achievable and the source word @xmath56 is a valid source word .
hence , @xmath56 is the mld result .
when a csfg at stage @xmath45 is frozen , the corresponding computations and message updating are not needed for the rest of the iterations .
we can also fix its right - to - left feedback propagating messages @xmath68 for the rest of the iterations based on its @xmath55 since the output decoding decision for this csfg has already been made and we have @xmath69 in one iteration , propagating messages from left to right , for any csfg , if the frozen set criteria ( 4 ) is not satisfied then we can not freeze this csfg .
we then update the messages at this stage using equation ( 1 ) , move to the next stage and repeat the same procedure .
5(b ) shows an example . at the second iteration
, we check the bottom csfg at stage 1 .
@xmath70 does not satisfy the frozen bit criteria ( 4 ) and we can not freeze this csfg .
so the messages are updated at stage 1 and we move to the next stage ( stage 2 ) to check whether the first un - frozen csfg can be frozen at stage 2 , as shown in fig .
a csfg can only be considered for freezing if all the preceding csfgs at the same stage have been frozen .
this procedure is repeated until all the csfgs at a stage are frozen or we reach the maximum number of iterations , which corresponds to the completion of the decoding process . with the freezing of csfgs , computations and message updating operations
do not need to be executed for rest of the iterations .
therefore the overall computation complexity , and hence the energy consumption , are reduced .
moreover the right - to - left feedback propagating messages @xmath68 are fixed to either -@xmath71 or + @xmath71 depending on the value of the hard - decision bit when a csfg is frozen .
this boosts the reliability of the feedback messages and will help the rest of the unfrozen csfgs to converge faster in the subsequent iterations , thus helping to reduce the overall number of iterations for the decoding and hence the average latency .
to verify the error correcting performance and complexity saving for the proposed frozen - csfg - based bpd scheme , we carry out a simulation on a polar code of length 1024 and rate and compare the result with the original bpd scheme in @xcite ( which we denote as the baseline bpd ) and the bpd using a g - matrix - based stopping criterion in @xcite . fig .
6 shows the simulation results over an awgn channel with bpsk modulation . for a fair comparison ,
we use the same set of parameters as @xcite , where min - sum approximation with scaling parameter @xmath72 and max_iter = 40 were used . as seen in fig .
6(a ) , the proposed method has no performance degradation compared with the other two existing bpd schemes .
the average number of iterations required for decoding a code word are compared in fig .
it can be seen that the proposed method requires the least number of iterations , resulting in lower latency and higher throughput compared to @xcite . at snr
= 3db , the average number of iterations is reduced by 46% and 17% when compared to the baseline bpd and the g - matrix - based early stop method , respectively .
we also compare the overall computation complexity of the three bpd schemes . for each pe in the factor graph , we count the number of iterations until its operation is frozen in the proposed scheme .
we then sum the number of iterations , for which that pe is active , for all the pes . for the other two schemes , since every pe needs to be executed in every iteration , the computation complexity just depends on the average number of iterations .
6(c ) shows the normalized average number of computations required for all three schemes .
it can be observed that the proposed scheme requires the least number of computations , which translates directly to lower power consumption and latency for the overall decoding process .
it can be seen that at snr = 3db , the average computation complexity is reduced by 65% and 46% when compared with the baseline scheme and the early - stopping scheme , respectively .
as state of the are bpd , computaion savings for the proposed method are comaperd with g - matrix based early stopping method in fig .
in this work we have presented a novel scheme to reduce the average number of computations as well as average latency in belief propagation decoding ( bpd ) for polar codes based on the concept of a frozen connected sub - factor - graph .
simulation results show that there is no performance degradation of the proposed scheme when compared with the original belief propagation algorithm and the g - matrix - based early stopping criterion , while the scheme enjoys a 46 ~ 65 % reduction in computation complexity , and 17 ~ 46% reduction in latency at snr = 3db . in future work , the vlsi architecture and a hardware implementation will be developed . c. leroux , i. tal , a. vardy , and w. j. gross , hardware architectures for successive cancellation decoding of polar codes , _ in proc .
acoust . , speech , signal process .
( icassp ) _ , pp .
16651668 , may 2011 .
y. fan , j. chen , c. xia , c.y .
tsui , j. jin , h. shen and b. li low - latency list decoding of polar codes with double thresholding _ in proc .
speech , signal process .
( icassp)_,in press , arxiv:1504.03437 .
|
since their invention , polar codes have received a lot of attention because of their capacity - achieving performance and low encoding and decoding complexity . successive cancellation decoding ( scd ) and belief propagation decoding ( bpd ) are two approaches for decoding polar codes .
scd is able to achieve good error - correcting performance and is less computationally expensive as compared to bpd .
however scd suffers from long latency due to the serial nature of the successive cancellation algorithm .
bpd is parallel in nature and hence is more attractive for low latency applications .
however , since it is iterative , the required latency and energy dissipation increases linearly with the number of iterations . in this work
, we borrow the idea of scd and propose a novel scheme based on sub - factor - graph freezing to reduce the average number of computations as well as the average number of iterations required by bpd , which directly translates into lower latency and energy dissipation .
simulation results show that the proposed scheme has no performance degradation and achieves significant reduction in computation complexity over the existing methods .
= 1 belief propagation decoding ( bpd ) ; successive cancellation decoding ( scd ) ; energy efficiency ; iterative decoders ; factor graph ; polar codes
|
the lattice simulations are presently considered as the only first principle tool to investigate the deconfinement phase transition ( pt ) in quantum chromodynamics ( qcd ) .
such a pt is also expected in gluodynamics ( gd ) which is a pure non - abelian gauge theory .
the svetitsky - jaffe hypothesis @xcite relates the deconfinement pt in ( d+1)-dimensional su(n ) gd to the magnetic pt in z(n ) symmetric spin model in d - dimensions .
the local polyakov loops in gd are playing the role of spins in z(n ) symmetric spin model .
for the ( d+1)-dimensional lattice having the size @xmath2 the local polyakov loop is defined by the temporal gauge links @xmath3 as @xmath4 a high level of understanding of the spin systems along with the svetitsky - jaffe hypothesis led to a significant progress in studying the su(n ) gd properties in the pt vicinity .
formation of geometrical clusters composed of the polyakov loops is an important feature of gd @xcite .
a similar phenomenon is well known in spin systems and it is responsible for percolation of clusters , which already was studied in gd with paying a special attention to the largest and the next to the largest clusters @xcite .
however , already from the famous fisher droplet model ( fdm ) @xcite it is well - known that some essential features of the liquid - gas pt are encoded in the properties of smaller clusters .
a principally important feature of the fdm ( and all its followers ) is that at the critical point the size distribution of physical clusters obeys a power law which is controlled by the fisher topological exponent @xmath0 .
hence the value of @xmath0 maybe a key element both for developing a consistent cluster model of liquid - gas pt for qcd and for experimental localization of the qcd critical point with its help .
therefore , in this work we study the geometrical clusterization in su(2 ) gd and analyze the properties of clusters of all possible sizes .
this approach allows us to explain the deconfinement of color charges as the liquid - gas pt @xcite of special kind .
in case of su(2 ) gauge group @xmath5 has the real values from @xmath6 to @xmath7 . similarly to spin models , for a given lattice configuration
the polyakov loops being the nearest neighbors can be attributed to the same cluster , if they have the same sign .
the boundaries of clusters with opposite signs of @xmath5 are characterized by strong fluctuations .
therefore , similarly to refs .
@xcite we introduced the minimal absolute value of the polyakov loop attributed to the clusters , i.e. a cut - off @xmath8 .
all space points @xmath9 with @xmath10 are attributed to `` auxiliary '' or `` confining '' vacuum which volume fraction is independent of the inverse lattice coupling @xmath11 @xcite , where @xmath12 is the lattice coupling constant .
the above definition allows us to define the monomers , the dimers , etc . as the clusters made of a corresponding number of gauge spins " of the same sign which are surrounded either by the clusters of opposite `` spin '' sign or by a vacuum @xcite
. obviously , there are clusters of two types related to two signs of the local polyakov loop .
we introduce a formal definition of anticlusters , if their sign coincides with the sign of the largest n - mer existing at a given lattice configuration .
the largest anticluster is the
anticluster droplet " and the other n - mers ( n = 1 , 2 , 3 , ... ) of the same sign correspond to the gas of anticlusters " .
the clusters are defined to have an opposite sign of the polyakov loop and the largest of them is called the cluster droplet " .
( left panel ) and @xmath13 ( right panel ) are shown for the cut - off @xmath14 . ] [ ivan_fig1 ]
the described scheme of the ( anti)cluster identification was realized numerically .
the polyakov loops were obtained at each spatial point of 3 + 1 dimensional lattice with the spatial and temporal extents @xmath15 and @xmath16 , respectively . the simulations were performed for 13 values of the inverse lattice coupling @xmath17 inside the interval @xmath18 $ ] .
the physical temperature @xmath19 is defined via two - loop @xmath17 dependence of the lattice spacing @xmath20 as @xmath21 @xcite .
the @xmath17 points were distributed not uniformly .
they were concentrated in the pt region which is of principal interest for this study .
the identification of ( anti)clusters was performed for two values of the polyakov loop cut - off @xmath22 and @xmath14 . for most of @xmath17 values the number of ( anti)clusters of each size was averaged over the ensemble of 800 and 1600 gauge field lattice configurations .
the distributions obtained in this way were the same within the statistical errors .
for the gaseous anticlusters at three largest values of @xmath17 the statistics was increased to 2400 configurations .
the right hand side vicinity of pt was also analyzed with such a statistics in order to exclude the effects of numerical fluctuations which we observed for @xmath23 and @xmath24 .
the typical size distributions of ( anti)clusters for @xmath14 are shown in fig .
[ ivan_fig1 ] . for @xmath17 below the critical value @xmath25 in an infinite system @xcite the distributions of ( anti)clusters are identical due to existing global z(2 ) symmetry . if @xmath17 is even slightly above @xmath26 then the symmetry between ( anti)clusters breaks down ( the left panel of fig . [ ivan_fig1 ] ) . in this case
the size of the cluster ( anticluster ) droplet decreases ( increases ) .
it is remarkable , that the corresponding gaseous clusters behave contrary to their droplets .
therefore , the deconfinement pt in su(2 ) gd can be considered as an evaporation of the cluster droplet into the gas of clusters and a simultaneous condensation of the gas of anticlusters into the anticluster droplet . for large values of @xmath17 ( the right panel of fig .
[ ivan_fig1 ] ) the cluster droplet becomes indistinguishable from its gas whereas the size of anticluster droplet becomes comparable to the system size .
also we found that the gas and liquid branches of anticluster distributions are always well separated from each other . for several values of @xmath27 and for a few values of @xmath17 found by the 4-parametric fit of the ldm formula .
all these results refer to the cut - off @xmath14 . ]
[ fig3 ] the found distributions closely resemble the ones discussed for the nuclear fragments and for the ising spin clusters @xcite . hence , we analyzed whether the liquid droplet formula ( ldf ) @xcite @xmath28 is able to reproduce the size distributions of clusters ( @xmath29 ) and anticlusters ( @xmath30 ) . here
@xmath31 is the bulk part of free energy of k - mer , @xmath32 denotes its surface free energy with the surface proportional to @xmath33 , @xmath0 is the fisher topological constant and @xmath34 is the normalization factor . since the ldm is valid only for ( anti)clusters which are sufficiently large , then the question which has to be clarified first was a determination of the minimal size @xmath27 to which eq .
( [ eqii ] ) can be applied .
the standard procedure of @xmath35 minimization with respect to @xmath34 , @xmath36 , @xmath37 , @xmath0 and @xmath27 allowed us to answer this question .
our analysis shows that the ldf is already able to describe the dimers .
indeed , we found that @xmath38 for all @xmath39 , whereas for @xmath40 we got @xmath41 which corresponds to the low quality of data description . within the statistical errors we found that @xmath0 is independent on @xmath17 for @xmath42 ( see fig .
[ fig3 ] ) .
this remarkable result is in line with the predictions of cluster models @xcite .
moreover , we found that @xmath43 both for clusters and for anticlusters , which agrees with the exactly solvable model of the nuclear liquid - gas pt @xcite and contradicts to the fdm @xcite .
thus , from a four parametric fit of the ldf we found that @xmath42 and @xmath44 both for clusters and for anticlusters . for fixed values @xmath42 and @xmath44 we performed a three parametric fit of the ( anti)cluster size distributions to define @xmath34 , @xmath36 and @xmath37 with high precision .
the typical value of @xmath45 was obtained for any @xmath17 , which signals about high quality of the data description .
the @xmath17-dependences of @xmath36 and @xmath37 are shown in fig .
[ figfig ] for @xmath14 .
for @xmath22 the results are similar . obtained for @xmath14 .
the curves shown in the right panel represent eq .
( [ vi ] ) . ] [ figfig ]
from fig . [ figfig ]
it is seen that the behavior of the reduced surface tension coefficient drastically changes , if @xmath17 exceeds the critical value @xmath46 .
indeed , for @xmath47 this quantity is constant and it is identical for clusters and anticlusters , while @xmath48 monotonically decreases and @xmath49 monotonically increases with @xmath17 for @xmath50 .
the qualitatively different behavior of @xmath48 and @xmath49 allows us to treat them as an order parameter of the deconfinement pt in su(2 ) gd . on the other hand ,
the average value of polyakov loop @xmath51 is traditionally considered as an order parameter of the deconfinement pt in su(2 ) gd . in @xcite we demonstrated that @xmath51 is mainly defined by the largest ( anti)cluster droplets whose average size ( see fig . [ fig5 ] ) is given by @xmath52 the @xmath17 dependence of @xmath37 and @xmath53 in the right hand side vicinity of @xmath54 is parametrized as @xmath55 + where the signs + " and - " correspond to a = acl and a = cl , respectively , and @xmath56 and @xmath57 are the critical exponents , whereas @xmath58 and @xmath59 are the normalization factors .
values of these parameters found from the fit are shown in table [ table ] .
it is remarkable that the found exponents @xmath56 are close to the critical exponent @xmath60 of the 3-dimensional ising model @xcite and to the critical exponent @xmath61 of simple liquids @xcite .
[ table ] .the fit parameters according to eq . ( [ vi ] ) and eq . ( [ vii ] ) [ cols="^,^,^,^,^,^,^,^ " , ] ( left panel ) and for @xmath14 ( right panel ) .
the curves represent eq .
( [ vii]).,title="fig : " ] ( left panel ) and for @xmath14 ( right panel ) .
the curves represent eq .
( [ vii]).,title="fig : " ] [ fig5 ]
in this contribution we present a novel approach to study the deconfinement pt in the su(2 ) gd in terms of the geometrical clusters composed of the polyakov loops of the same sign .
we justify the separation of ( anti)clusters into liquid " droplet and gas " of smaller fragments and investigate their physical properties .
we also explain the deconfinement pt as a special kind of the liquid - gas transition between two types of liquid whose behavior is drastically different in the region of broken global z(2 ) symmetry . above pt the cluster liquid droplet evaporates , whereas the anticluster liquid droplet condensates the accompanying gas of anticlusters .
a successful application of the ldf to the description of the size distributions of all gaseous ( anti)clusters excluding the monomers is the main result of this study .
this approach allowed us to determine the @xmath17-dependences of the reduced chemical potential and the reduced surface tension coefficient .
while in the phase of unbroken global z(2 ) symmetry these quantities are identical for the fragments of both kinds , their behavior is drastically different in the deconfined phase .
another important finding of this study is a high precision determination of the fisher topological constant @xmath62 which has the same value both for clusters and for anticlusters .
this result is in line with the exactly solvable model of the nuclear liquid - gas pt @xcite and disproves the fdm prediction that @xmath63 @xcite .
we argue that the reduced surface tension coefficient and the mean size of the largest ( anti)cluster can be used as the new order parameters of deconfinement pt in su(2 ) gd .
the authors thank a. v. taranenko for the fruitful discussions , valuable comments and kind help in the icppa 2016 participation .
this work was supported in part by the national academy of sciences of ukraine and by the nas of ukraine grant of grid simulations for high energy physics .
|
the liquid droplet formula is applied to an analysis of the properties of geometrical ( anti)clusters formed in su(2 ) gluodynamics by the polyakov loops of the same sign . using this approach
, we explain the phase transition in su(2 ) gluodynamics as a transition between two liquids during which one of the liquid droplets ( the largest cluster of a certain polyakov loop sign ) experiences a condensation , while the droplet of another liquid ( the next to the largest cluster of the opposite sign of polyakov loop ) evaporates .
the clusters of smaller sizes form two accompanying gases , which behave oppositely to their liquids .
the liquid droplet formula is used to analyze the size distributions of the gaseous ( anti)clusters .
the fit of these distributions allows us to extract the temperature dependence of surface tension and the value of fisher topological exponent @xmath0 for both kinds of gaseous clusters .
it is shown that the surface tension coefficient of gaseous ( anti)clusters can serve as an order parameter of the deconfinement phase transition in su(2 ) gluodynamics .
the fisher topological exponent @xmath0 of clusters and anticlusters is found to have the same value 1.806 @xmath1 0.008 .
this value disagrees with the famous fisher droplet model , but it agrees well with an exactly solvable model of the nuclear liquid - gas phase transition .
this finding may evidence for the fact that the su(2 ) gluodynamics and this exactly solvable model of nuclear liquid - gas phase transition are in the same universality class .
|
our starting point is the hubbard model @xmath165 we implement a rotation invariant decoupling of the hubbard interaction as follows .
first , one can write @xmath166 where @xmath167 is the charge density , @xmath168 is the local electron spin operator , and @xmath169 is an _
arbitrary unit vector_. the partition function of the hubbard model is @xmath170e^{-s } \cr s & = & \int_{0}^{\beta}d \tau { \cal l}(\tau ) \cr { \cal l } & = & \sum_{i\sigma } \bar{c}_{i\sigma}(\tau)\partial_{\tau}c_{i\sigma}(\tau ) + h(\tau ) \nonumber\end{aligned}\ ] ] we can introduce two space - time varying auxiliary fields for a hubbard - stratonovich transformation : ( i ) @xmath171 coupling to charge density , and ( ii ) @xmath172 coupling to electron spin density ( @xmath173 is real positive ) .
this allows us to define a su(2 ) invariant hs transformation ( see ref .
@xcite ) , @xmath174 the partition function now becomes : @xmath175 \nonumber\end{aligned}\ ] ] as discussed in the text , to make progress we need two approximations : ( i ) neglect the time ( @xmath176 ) dependence of the hs fields , ( ii ) replace the field @xmath39 by its saddle point value @xmath177 , since the important low energy fluctuations arise from the @xmath38 .
substituting these , and simplifying the action , one gets the effective hamiltonian @xmath178 where @xmath179 . for convenience
we redefine @xmath180 , so that the @xmath38 is dimensionless .
this leads to the effective hamiltonian used in the text : @xmath181 the partition function can be written as : @xmath182 for a given configuration @xmath183 the problem is quadratic in the fermions , while the configurations themselves are obtained by a monte carlo as discussed in the text .
the conductivity of the two dimensional system is first calculated as follows ( ref.@xcite ) , using the kubo formula : @xmath184 where , the current operator @xmath185 is @xmath186 \nonumber\end{aligned}\ ] ] the d.c conductivity is the @xmath187 limit of the result above .
@xmath188=@xmath189 , the scale for two dimensional conductivity , has the dimension of _
@xmath190 is the fermi function , and @xmath191 and @xmath192 are respectively the single particle eigenvalues and eigenstates of @xmath56 in a given background \{@xmath38}. the results we show in the text are averaged over equilibrium mc configurations .
the experimental results are quoted as _
resistivity _ of a three dimensional material .
if we assume that the planes are electronically decoupled , as we have done in the text , then the three dimensional resistivity @xmath193 can be estimated from the resistance of a cube of size @xmath194 .
if the 2d resistivity is @xmath195 , the resistance of a @xmath196 sheet is just @xmath197 itself .
a stacking of such sheets , with spacing @xmath85 in the third direction , implies that the resistance of the @xmath194 system would be @xmath198 . by definition
this also equals @xmath199 . equating the two , @xmath200 .
we extract the thermal and spin averaged spectral function @xmath148 as follows .
first , the retarded greens function @xmath201 which can be simplified to @xmath202 are the single particle eigenstates and @xmath191 are eigenvalues in a given @xmath203 background . in frequency domain , this becomes @xmath204 from this : @xmath205 im@xmath206 is simply @xmath207 .
we average this over spin orientations , @xmath208 , and over thermal configurations . the @xmath149 dependent weight at @xmath146 is shown in fig.5 in the text .
the full spectral function at the ` cold spot ' and ` hot spot ' , where @xmath209 is maximum and minimum , are shown , respectively , in the top and bottom panels in the figure [ akf ] .
we have averaged the spectrum over the four @xmath149 neighbours of the nominal ` cold ' and ` hot ' points of our @xmath19 lattice .
this averaging _ reduces _ the anisotropy , so the true anisotropy would be greater than what we show here . also note that at @xmath210 , where the @xmath151 in the text is shown , the spectral function has no _ peak _ at @xmath146 either in the cold or hot spot .
the pseudogap feature is visible all over the fs even at @xmath211 .
99 weng , z. y. and ting , c. s. and lee , t. k. , phys .
b * 43 * 3790 ( 1991 ) k. borejsza and n. dupuis , europhys
63 * 722 ( 2003 ) p. b. allen in _ conceptual foundation of materials v.2 _ , edited by steven g. louie , marvin l. cohen , elsevier ( 2006 ) .
|
we use a novel monte carlo method to study the mott transition in an anisotropic triangular lattice . the real space approach , retaining extended spatial correlations , allows an accurate treatment of non trivial magnetic fluctuations in this frustrated structure . choosing the degree of anisotropy to mimic the situation in the quasi - two dimensional organics , @xmath0(bedt - ttf)@xmath1cu[n(cn)@xmath1]-x
, we detect a wide pseudogap phase , with anomalous spectral and transport properties , between the ` ungapped ' metal and the ` hard gap ' mott insulator .
the magnetic fluctuations also lead to pronounced momentum dependence of quasiparticle damping and pseudogap formation on the fermi surface as the mott transition is approached .
our predictions about the ` bad metal ' state have a direct bearing on the organics where they can be tested via tunneling , angle resolved photoemission , and magnetic structure factor measurement .
the mott metal - insulator transition ( mit ) , and the proximity to a mott insulator in doped systems , are crucial issues in correlated electron systems @xcite .
the mott transition on a bipartite lattice is now well understood , but the presence of triangular motifs in the structure brings in geometric frustration @xcite .
this promotes incommensurate magnetic fluctuations whose nature , and impact on the mit , remain outstanding problems .
the organic salts provide a concrete testing ground for these effects @xcite .
the @xmath0(bedt - ttf)@xmath1cu[n(cn)@xmath1]-x salts are quasi two dimensional ( 2d ) materials where the bedt - ttf dimers define a triangular lattice with anisotropic hopping @xcite .
the large lattice spacing , @xmath3 , leads to a low bandwidth , enhancing electron correlation effects , while the triangular motif disfavours neel order .
the x@xmath4cl@xmath5br@xmath6 family shows a mit as @xmath7 drops below @xmath8 @xcite .
the metallic state is _ very incoherent _ above @xmath9k : the resistivity @xcite is large , @xmath10m@xmath11 cm , the optical response has non drude character @xcite , and nmr @xcite suggests the presence of a pseudogap ( pg ) .
how these properties arise in response to magnetic fluctuations , and the crucial low energy spectral features in the vicinity of the mott transition , remain to be clarified .
we use a completely new approach to the mott transition , using auxiliary fields , that emphasizes the role of spatial correlations near the mit .
our principal results , based on monte carlo ( mc ) on large lattices are the following .
( i ) the interaction @xmath12-temperature @xmath13 phase diagram that we establish has a striking correspondence with @xmath14-bedt in terms of magnetic transition and re - entrant insulator - metal crossovers .
( ii ) at intermediate temperature , in the magnetically disordered regime , we obtain a strongly non drude optical response in the metal , and predict a pseudogap ( pg ) phase over a wide interaction and temperature window .
( iii ) the electronic spectral function @xmath15 is _ anisotropic _ on the fermi surface , with both the damping rate and pg formation showing a clear angular dependence arising from coupling to incommensurate magnetic fluctuations .
@xmath16 phase diagram of the hubbard model at @xmath17 .
the phases are paramagnetic metal ( pm ) , paramagnetic insulator ( pi ) , antiferromagnetic metal ( afm ) and antiferromagnetic insulator ( afi ) .
the afm and afi are not simple neel ordered .
pg indicates a pseudogap phase , metallic or insulating .
there is no genuine magnetic transition in two dimensions so our @xmath18 indicates the temperature where the magnetic correlation length becomes larger than the system size @xmath19 .
the mit line is determined from change in sign of the temperature derivative of resistivity , _ i.e _ , @xmath20 .
, width=257,height=188 ] to provide a quick background , there have been several studies of the single band hubbard model on a triangular lattice@xcite to model organic physics .
dynamical mean field theory ( dmft ) has been the method of choice @xcite , usually used in its cluster variant ( c - dmft ) @xcite to handle short range spatial correlations .
the results depend on the degree of frustration and the specific method but broadly suggest the following : ( i ) the ground state is a pm fermi liquid at weak coupling , a ` spin liquid ' pi at intermediate coupling , and an afi at large coupling@xcite , ( ii ) the qualitative features in optics @xcite and transport @xcite are recovered , ( iii ) there could be a re - entrant insulator - metal - insulator transition with increasing temperature for a certain window of frustration @xcite , ( iv ) the low temperature sc state could emerge @xcite from hubbard physics , although there is no consensus @xcite . surprisingly , there seems to be limited effort on the nature of _ spatial fluctuations _ , which could be significant in this low dimensional frustrated system . to clarify this aspect we study the single band hubbard model on the anisotropic triangular lattice : @xmath21 we use a square lattice geometry but with the following anisotropic hopping : @xmath22 when @xmath23 or @xmath24 , where @xmath25 is the lattice spacing , and @xmath26 when @xmath27 .
we will set @xmath28 as the reference energy scale .
@xmath29 corresponds to the square lattice , and @xmath30 to the isotropic triangular lattice .
we have studied the problem over the entire @xmath31 window @xmath32 $ ] , but focus on @xmath17 in this paper .
@xmath33 controls the electron density , which we maintain at half - filling , @xmath34 .
@xmath35 is the hubbard repulsion .
we use a hubbard - stratonovich ( hs ) transformation that introduces a vector field @xmath36 and a scalar field @xmath37 at each site @xcite to decouple the interaction .
we need two approximations to make progress .
( i ) we will treat the @xmath38 and @xmath39 as classical fields , _ i.e _
, neglect their time dependence .
( ii ) while we completely retain the thermal fluctuations in @xmath38 , we treat @xmath39 at the saddle point level , _ i.e _ , @xmath40 at half - filling .
with this approximation the half - filled problem is mapped on to electrons coupled to the field @xmath38 ( see supplement ) .
@xmath41 where @xmath42 .
we can write @xmath43 , where @xmath44 . for a _ given _
configuration @xmath45 one just needs to diagonalise @xmath46 , but the @xmath45 themselves have to be determined from the distribution @xmath47 equation ( 2 ) describes electron propagation in the @xmath38 background , while equation ( 3 ) describes how the @xmath38 emerge and are spatially correlated due to electron motion .
the neglect of dynamics in the @xmath38 reduces the method to unrestricted hartee - fock ( uhf ) mean field theory at @xmath48 .
however , the exact inclusion of classical thermal fluctuations quickly improves the accuracy of the method with increasing temperature .
we will discuss the limitations of the method further on .
temperature dependence of the resistivity for different @xmath49 in the neighbourhood of @xmath50 .
the normalising scale is @xmath51 ( see text ) .
this is @xmath52 cm for the organics .
the @xmath53 values are indicated on the curves .
the inset shows the experimental transport gap in the cl@xmath5br@xmath6 family ( in kelvins ) , and our estimated transport gap .
we used a fit @xmath54 ( see text ) to reproduce the transport gap@xcite estimated from the experiments .
, width=264,height=238 ] due to the fermion trace , @xmath55 is not exactly calculable . to generate the equilibrium @xmath45 we use mc sampling @xcite
computing the energy cost of an attempted update requires diagonalising @xmath46 . to access large sizes within limited time
, we use a cluster algorithm @xcite for estimating the update cost . rather than diagonalise the full @xmath56 for every attempted update , we calculate the energy cost of an update by diagonalizing a small cluster ( of size @xmath57 ,
say ) around the reference site .
we have extensively benchmarked this cluster based mc method@xcite .
the mc was done for lattice of size @xmath58 , with clusters of size @xmath59 .
we calculate the thermally averaged structure factor @xmath60 at each temperature .
the onset of rapid growth in @xmath61 at some @xmath62 , say , indicates a magnetic transition .
the electronic properties are calculated by diagonalising @xmath46 on the full lattice for equilibrium @xmath45 configurations .
fig.[phd ] shows the @xmath16 phase diagram at @xmath17 .
our low temperature result is equivalent to uhf and leads to a transition from an uncorrelated paramagnetic metal to an incommensurate af metal with wavevector @xmath63 at @xmath64 . at @xmath65
there is a transition to an af ` mott ' insulator with @xmath66 .
the magnitude @xmath67 is small in the afm and grows as @xmath53 increases in the mott phase .
the existence of the af metal , and the nature of order in the intermediate @xmath53 mott phase , could be affected by the neglected quantum fluctuations of the @xmath38 .
finite temperature brings into play the low energy fluctuations of the @xmath38 .
the effective model has the @xmath68 symmetry of the parent hubbard model so it can not sustain true long range order at finite @xmath69 . however , our annealing results suggest that magnetic correlations grow rapidly below a temperature @xmath18 , and weak interplanar coupling would stabilise in plane order below @xmath18 .
this scale increases from zero at @xmath70 , reaches a peak at @xmath71 , and falls beyond as the virtual kinetic energy gain reduces with increasing @xmath49 .
we classify the finite @xmath69 phases as metal or insulator based on @xmath72 , the temperature derivative of the resistivity . the dotted line indicating the mit corresponds to the locus @xmath73 .
in addition to the magnetic and transport classification we also show a window around the mit line where the electronic density of states ( dos ) has a pseudogap . to the right of this region
the dos has a ` hard gap ' while to the left it is featureless .
the mit line shows re - entrant insulator - metal - insulator behavior with increasing @xmath69 near @xmath74 .
we can attempt a quick comparison of the phase diagram with that in the @xmath14-bedt family .
the primary hopping is @xmath75mev , and @xmath76 @xcite ( a recent _ ab initio _ estimate suggests @xmath77 for @xmath14-cl ) .
fitting the transport gap in @xmath14-cl ( see fig.[phd ] inset ) suggests that @xmath78 at @xmath79 . from our results
this would indicate that @xmath80 , at @xmath79 , _ i.e _ , @xmath81k , not too far from the nmr inferred @xmath82k .
fig.[rho ] shows the resistivity @xmath83 computed via the kubo formula for varying @xmath53 .
we first compute the planar resistivity ( which has the dimension of resistance ) and then compute the effective three dimensional resistivity of ` decoupled ' layers ( see supplement ) by using the @xmath84-axis spacing , @xmath85 . in the cl@xmath5br@xmath6 family
it is observed that the transport gap can be fitted to @xmath86 kelvin@xcite .
we match this to the @xmath49 dependence of our calculated gap , @xmath87 , and infer @xmath88 .
the mit occurs at @xmath89 and for us at @xmath90 . fitting a quadratic form to @xmath91 to capture the measured transport gap , we get @xmath92 . the @xmath53 range in fig.[rho ] corresponds roughly to @xmath93 $ ] . since @xmath94mev
, @xmath95 is approximately @xmath96k .
optical conductivity at @xmath97 and @xmath98 for @xmath49 varying across @xmath50 . at these temperatures
the @xmath99 is non drude even in the ` weakly correlated ' case @xmath100 .
the finite frequency peak evolves into the hubbard transition at large @xmath53 . for a rough comparison to organics , @xmath10180k , @xmath102@xmath103 , and @xmath104@xmath103 .
, width=309,height=151 ] our resistivity is in units of @xmath105 . using @xmath106 , @xmath107 cm , yields @xmath108m@xmath11 cm at @xmath109 , while experimental value is @xmath110m@xmath11 cm .
the difference could come from electron - phonon scattering absent in our model .
limelette _
et al_@xcite presented dmft based resistivity result that compares favourably with experiments , but , apparently , involves an arbitrary scale factor .
our re - entrant window @xmath111=0.4@xmath112 near @xmath50 , inferred from thermally driven i - m - i crossover , is equivalent to @xmath113=0.2 .
this is consistent with @xmath114 in the cl@xmath5br@xmath6 family@xcite .
the c - dmft estimates of the re - entrant window varies widely , from @[email protected]@xmath112@xcite to @xmath1151.2@xmath112@xcite .
density of states at @xmath116 for @xmath49 varying across @xmath50 .
the dip in the dos deepens with increasing @xmath69 for @xmath117 . for larger @xmath53
the system slowly gains spectral weight with increasing @xmath69 .
, width=317,height=181 ] fig.[optics ] shows the optical conductivity @xmath99 at @xmath118 and @xmath119 as @xmath53 varied across the mott crossover .
our first observation is the distinctly non drude nature of @xmath99 in the metal , @xmath120 , with @xmath121 .
the low frequency hump in the bad metal evolves into the interband hubbard peak in the mott phase .
the change in the lineshape with increasing @xmath69 is more prominent in the metal , with the peak location moving outward , and is more modest deep in the insulator . in the organics the experimenters
have carefully isolated the mott - hubbard features in the spectrum by removing phononic and intra - dimer effects @xcite .
since we have already fixed our @xmath122 we have no further fitting parameter for @xmath99 .
the measured spectrum at @xmath123 and @xmath124k has a peak around @xmath125@xmath103 . using @xmath126 and @xmath127 we get @xmath128 , which translates to @xmath129@xmath103 .
the magnitude of our @xmath99 at @xmath130 is @xmath131@xmath103 , since @xmath132@xmath103 .
this is remarkably close to the measured value , @xmath133@xmath103(ref .
@xcite fig.3 ) .
while the characteristic scales in @xmath99 match well with experiments , the experimental spectrum has weaker dependence on temperature and composition @xmath7 .
this could arise from the subtraction process and the presence of other interactions in the real material .
our result differs from dmft @xcite , and agrees with the experiments , in that we do not have any feature at @xmath134 .
we have verified the _
f_-sum rule numerically .
the crossover from the bad metal to the insulator involves a wide window with a pseudogap in the electronic dos , @xmath135 .
one may have guessed this from the depleting low frequency weight in @xmath99 , fig.[dos ] makes this feature explicit .
we are not aware of tunneling studies in the organics , but our results indicate a wide window , @xmath136 $ ] , where there is a distinct pseudogap in the global dos .
this suggests that the entire @xmath137 $ ] window in the organics should have a pg . for @xmath138
the dip feature deepens with increasing @xmath69 , we have @xmath139 ( compare panels ( a ) and ( b ) , fig.[dos ] ) , while for @xmath140 we have a weak @xmath141 .
the pg arises from the coupling of electrons to the fluctuating @xmath38 .
a large @xmath142 at all sites would open a mott gap , independent of any order among the moments .
weaker @xmath142 , thermally generated in the metal near @xmath143 and with only short range correlations , manages to deplete low frequency weight without opening a gap .
since the typical size @xmath144 increases with @xmath69 in the metal , we see the dip deepening at @xmath145 . while the _ size _ of the @xmath38 determine the overall depletion of dos near @xmath146 and the opening of the mott gap , the _ angular correlations _ dictate the momentum dependence of the spin averaged electronic spectral function @xmath147 ( see supplement ) . within ` local self energy ' picture , as in dmft , @xmath148 should have no @xmath149 dependence on the fermi surface ( fs ) . in that case
we should have @xmath149 independent suppression of @xmath150 with increasing @xmath53 .
fig.[aksq ] , top row , shows maps of @xmath151 for @xmath152 $ ] , at @xmath97 , as increasing @xmath53 transforms the bad metal to a mott insulator .
the first panel at @xmath153 ( roughly a br sample ) shows weak anisotropy on the nominal fs while fig.[dos].(a ) suggests that a weak pg has already formed . at @xmath154 , next panel ,
the weak anisotropy is much amplified and the weight in the @xmath155\rightarrow [ \pi,\pi]$ ] direction is distinctly larger .
the next three panels basically show insulating states but without a hard mott gap .
overall , the ` destruction ' of the fs seems to start near @xmath156 $ ] , the ` hot ' region , and ends with the region near @xmath157 $ ] , the ` cold spot ' .
we show data on the full @xmath148 in the supplement that indicates that with increasing @xmath49 a pg feature forms at the hot spot while the cold spot still has a quasiparticle peak .
second row in fig.[aksq ] shows the @xmath61 of the auxiliary fields at @xmath97 for the same @xmath53 as in the upper row . while there is no magnetic order we can see the growth of correlations centered around @xmath158 $ ] as @xmath53 increases .
the dominant electron scattering would be from @xmath149 to @xmath159 , and the impact would be greatest in regions of the fs in the proximity of minima in @xmath160 .
the location of the hot spots on the fs , and the momentum connecting them , indeed correspond to this scenario .
while we have a method that captures non trivial spatial correlations and its impact on electronic properties , we need to be cautious about some shortcomings .
( i ) neglecting the dynamics of the @xmath38 misses correlation effects in the ground state of the metal and underestimates @xmath161 .
( ii ) it also misses the ` fermi liquid ' physics in the low @xmath69 metal , but should be reliable in the @xmath162 regime that we have focused on .
( iii ) there is potentially a ` spin liquid ' insulator @xcite at intermediate @xmath53 and @xmath163 .
we do not know of such results at @xmath17 , but would prefer to emphasize our finite @xmath69 results rather than the nature of the ground state . _
conclusion : _ we introduced and explored in detail a method which retains the spatial correlations that are crucial near the mott transition on a frustrated lattice . using electronic parameters that describe the @xmath14-bedt based organics
we obtain a magnetic @xmath164 , metal - insulator phase diagram , and optical response that reproduces the key experimental scales .
we uncover a wide pseudogap regime near the mit , and predict distinct signatures of the incommensurate magnetic fluctuations in the angle resolved photoemission spectrum .
we acknowledge use of the hpc clusters at hri .
pm acknowledges support from the dst india ( athena ) , a dae - src outstanding research investigator grant , and a discussion with t. v. ramakrishnan .
99 n. f. mott , _ metal - insulator transitions _ , taylor and francis ( 1990 ) .
a. georges , g. kotliar , w. krauth and m. j. rozenberg , rev .
mod .
phys . * 68 * 13 ( 1996 ) .
m. imada , a. fujimori and y. tokura , rev .
mod .
phys . * 70 * 1039 ( 1998 ) e. dagotto , rev .
mod .
phys . * 66 * , 763 ( 1994 ) .
n. p. ong and r. j. cava , science * 305 * 52 ( 2005 ) .
l. balents , nature * 464 * 199 ( 2010 ) .
k. kanoda and r. kato , annu .
. condens
. matter physics .
* 2 * 167 ( 2011 ) .
b. j. powell and r. h. mckenzie rep .
. phys . * 74 * , 056501 ( 2011 ) .
h. c. kandpal , _ et al .
_ , phys .
rev .
lett .
* 103 * 067004 ( 2009 ) , y. shimizu , m. maesato and g. saito , j. phys .
. jpn . * 80 * 074702 ( 2011 ) .
s. lefebvre , _ et al .
_ , phys .
rev .
lett . * 85 * 5420 ( 2000 ) , f. kagawa , t. itou , k. miyagawa , and k. kanoda , phys .
rev .
b. * 69 * 064511 ( 2004 ) .
s. yasin , _ et al .
_ , eur .
phys .
j. b. * 79 * 383 ( 2011 ) .
j. merino , _ et al .
_ , phys
. rev .
lett .
* 100 * 086404 ( 2008 ) .
m. dumm , _ et al .
_ , phys .
rev .
b. * 79 * 195106 ( 2009 ) .
b. j. powell , e. yusuf , and r. h. mckenzie , phys .
rev . b. * 80 * 054505 ( 2009 ) .
a. kawamoto , k. miyagawa , y. nakazawa , and k. kanoda , phys .
rev .
b. * 52 * 15522 ( 1995 ) .
h. morita , s.watanabe and m. imada , j. phys .
soc .
j * 71 * 2109 ( 2002 ) .
t. watanabe , h. yokoyama , y. tanaka , and j. inoue , phys .
rev .
b * 77 * 214505 ( 2008 ) .
k. inaba , a. koga , s. suga , and n. kawakami , j. phys .
: conf .
ser . * 150 * 042066 ( 2009 ) , t. yoshioka , a. koga and n. kawakami , phys .
rev .
lett . * 103 * 036401 ( 2009 ) .
l. f. tocchio , a. parola , c. gros , and f. becca , phys .
rev .
b * 80 * 064419 ( 2009 ) .
p. limelette , _ et al .
_ , phys .
rev .
lett . * 91 * 016401 ( 2003 ) .
k. aryanpour , w. e. pickett , and r. t. scalettar , phys .
rev .
b. * 74 * 085117 ( 2006 ) .
d. galanakis , t. d. stanescu and p. phillips , phys .
rev .
b. * 79 * 115116 ( 2009 ) .
o. parcollet , g. biroli and g. kotliar , phys .
rev .
lett . * 92 * , 226402 ( 2004 ) .
t. ohashi , t. momoi , h. tsunetsugu , and n. kawakami , phys .
rev .
lett . * 100 * 076402 ( 2008 ) .
a. liebsch , h. ishida , and j. merino , phys .
rev .
b. * 79 * 195108 ( 2009 ) .
t. sato , k. hattori and h. tsunetsugu , j. phys .
. jpn . * 81 * 083703 ( 2012 ) .
p. sahebsara and d. senechal , phys .
rev .
lett . * 100 * , 136402 ( 2008 ) .
h. y. yang , a. m. lauchli , f. mila and k. p. schmidt , phys .
rev .
lett . * 105 * , 267204 ( 2010 ) .
t. watanabe , h. y. okoyama , y. tanaka and j. inoue , j. phys .
. jpn . * 75 * 074707 ( 2006 ) .
b. kyung and a .- m .
s. tremblay , phys .
rev .
lett . * 97 * 046402 ( 2006 ) .
j. liu , j. schmalian , and n. trivedi , phys .
rev .
lett . * 94 * , 127003 ( 2005 ) .
r. t. clay , h. li , and s. mazumdar , phys .
rev .
lett .
* 101 * , 166403 ( 2008 ) .
j. hubbard , phys .
rev .
lett . * 3 * 77 ( 1959 ) .
h. j. schulz , phys .
rev .
lett . * 65 * 2462 ( 1990 ) .
m. mayr , g. alvarez , a. moreo , and e. dagotto , phys .
rev .
b. * 73 * 014509 ( 2006 ) .
e. dagotto and t. hotta and a. moreo , phys . rep . * 344 * 1 ( 2001 ) .
y. dubi , y. meir and y. avishai , nature , * 449 * , 876 ( 2007 ) .
s. kumar and p. majumdar , eur .
phys .
j. b , * 50 * , 571 ( 2006 ) .
|
uvsipm is a stand - alone portable photon detector instrument designed to measure electromagnetic radiation in the 320900 nm wavelength range .
it has been developed in the framework of astri ( astrofisica con specchi a tecnologia replicante italiana ) project @xcite a `` flagship project '' financed by the italian ministry of education , university and research ( miur ) and led by the italian national institute of astrophysics ( inaf ) .
the astri project is focused , in its first phase , on the design and construction of an end - to - end small size scale telescope prototype for the cherenkov telescope array ( cta ) @xcite , the international next generation ground - based observatory for very high energy gamma - rays ( from a few tens of gev up to several hundred of tev ) .
the astri project will cover the highest part of the energy spectrum adopting innovative and challenging solutions never used in the framework of cherenkov telescopes : the optics system will be based on a double mirror in schwarzschild - couder configuration instead of the standard single mirror ; the camera will be composed by an array of monolithic multipixel silicon photo - multipliers ( sipm ) instead of the usual photo multiplier tubes ( pmt ) .
uvsipm , developed at iasf - palermo and based on the same sipm used for the astri prototype , has the aim of characterizing the device and evaluating its performance in lab ; the final goal is to use uvsipm as tool for the on field absolute calibration of the astri prototype .
the instrument is designed around a single multipixel ( 4@xmath04 pixels ) sipm sensor coupled to an electronic chain working in single photon counting ( spc ) mode , capable of 10 ns double pulse resolution .
[ fig1 ] shows the uvsipm detector and electronics . for on field measurements
, a collimator will be mounted in front of the uvsipm detector to regulate the angular aperture of the instrument .
silicon photo - multipliers are a family of light sensors with very interesting characteristics .
they are basically avalanche photo diodes working in geiger - mode , in which the reverse bias voltage is set beyond the breakdown voltage ( overvoltage ) . in this way
, a single photon absorbed in silicon develops a saturated current avalanche with a gain of the order of 10@xmath1 .
there are many advantages in using sipms compared to the traditional pmts : excellent single photon resolution , high quantum efficiency ( qe ) , no hv ( bias voltages of the order of 3090v ) , no damage when exposed to ambient light , insensitive to magnetic fields , small size .
the drawbacks however are : high dark counts , afterpulses , optical crosstalk , gain strongly dependent on temperature .
sipms are the sensor that will be used to fill the camera at the focal plane of the astri prototype .
the device chosen is the hamamatsu s11828 - 3344 m , a monolithic sipm array in a configuration of 4x4 square pixels ( see fig .
[ fig2 ] ) with a geometrical filling factor of about 74% .
each pixel , 3x3 mm in size , is made up of 3600 elementary diodes of 50 @xmath2 m pitch .
the uvsipm instrument is basically composed by the sipm sensor hamamatsu s11828 - 3344 m , coupled to an electronic chain working in spc mode , with a double pulse resolution of 10 ns . the total photon detection efficiency ( pde ) of sipms is defined as : @xmath3 where @xmath4 is the quantum efficiency , @xmath5 the pixel filling factor and @xmath6 the geiger avalanche probability . a typical pde of an hamamatsu sipm is plotted in fig . [ fig3 ] .
the sensor response ranges from uv to infrared wavelength due to the _ p - on - n _ technology used by hamamatsu .
normally the pde is measured from the mean value of the output current , when the device is exposed to a known photons flux . in this way ,
pulses coming from crosstalk or afterpulses are included in the total pde leading to an overestimate of the real efficiency .
evaluating the level of crosstalk and afterpulses is then one of the task of the sipm sensors characterization .
when a photon is absorbed in silicon a current avalanche produces an output pulse with a sharp rise - time ( the discharge ) followed by a very long tail ( the recovery from the breakdown ) as shown in fig .
typical rise - time is of the order of few hundreds of picoseconds , whereas the recovery time is in the range of tenths of nanoseconds . from fig .
[ fig4 ] , it is also evident the sipm capability of distinguishing the number of detected photoelectron from the level of the pulse height . at high pulse rate
pile - up effects occur and afterpulses also contribute to rise the pile - up effects . the most efficient way to count pulses with the peculiar shape produced by the sipm , i.e. fast rise followed by a long tail , is to use the derivative of the signal .
the electronics that performs this function is composed by an input amplifier buffer ac coupled to the sipm , followed by a shaping circuit that gives the derivative of the signal cutting the negative part ( pole - zero cancellation ) ; an inverting amplifier drives a comparator whose threshold is set through a 12-bit dac .
[ fig5 ] presents a snapshot from the oscilloscope that shows how a long tail pulse ( sipm - like ) is changed into a short one .
[ fig6 ] shows the block diagram of the electronics and a snapshot from the oscilloscope in which two piled up pulses taken from a pulse generator are resolved in two separate pulses .
a fpga is in charge to manage the pulse counting and the housekeepings collection . the communication with an external pc to control the instrument and the data exchange is managed by the delphin , a proprietary interface board designed at iasf palermo .
uvsipm has been designed as support instrumentation of the astri prototype .
both of them use the same model of sipm sensor and several kind of measurements , both in lab and on field , are foreseen for the characterization of the sensors and the calibration of the whole camera at the focal plane of the astri prototype .
the characterization of the sipm will be carried out at the cold detector laboratory in catania ( oact / inaf ) . among all the measurements ,
the most important are : pde , dark counts , level of crosstalk and afterpulses evaluated as function of wavelength , overvoltage and temperature .
measurements of the detector response versus of the photon incidence angles are also planned .
the design and implementation of uvsipm derive from the well - tested uvscope instrument @xcite , developed at iasf - palermo , whose sensor is a calibrated multi - anode pmt ( mapmt ) working in spc mode .
the simultaneous usage of the two instruments on field , both provided with a calibrated john - b-25 johnson / bessel filter and equipped with collimators of proper length , will allow us to test the performance of uvsipm under real observing conditions .
the measurements will be based on the evaluation of the night sky background with the two instruments ( uvsipm and uvscope ) .
the comparison will allow us to test the end - to - end performances of sipm vs mapmt and then to use uvsipm for the absolute calibration of astri telescope .
r. canestrari , et al . , `` the italian astri program : an end - to - end dual - mirror telescope prototype for cherenkov light imaging above few tev '' , 32nd international cosmic ray conference beijing ( 2011 ) the cta consortium , `` design concepts for the cherenkov telescope array cta : an advanced facility for ground - based high - energy gamma - ray astronomy '' , experimental astronomy 32 issue 3 ( 2011 ) 193 .
maccarone , et al .
, `` performance and applications of the uvscope instrument . '' , nim a 659 ( 2011 ) 569
|
uvsipm is a light detector designed to measure the intensity of electromagnetic radiation in the 320900 nm wavelength range .
it has been developed in the framework of the astri project whose main goal is the design and construction of an end - to - end small size class telescope prototype for the cherenkov telescope array .
the uvsipm instrument is composed by a multipixel silicon photo - multiplier detector unit coupled to an electronic chain working in single photon counting mode with 10 nanosecond double pulse resolution , and by a disk emulator interface card for computer connection .
the detector unit of uvsipm is of the same kind as the ones forming the camera at the focal plane of the astri prototype .
eventually , the uvsipm instrument can be equipped with a collimator to regulate its angular aperture .
uvsipm , with its peculiar characteristics , will permit to perform several measurements both in lab and on field , allowing the absolute calibration of the astri prototype .
detector : sipm
|
in this supplemental material , we present several numerical details about the spin-1 chain , @xmath126 .
\end{aligned}\ ] ] for @xmath63 and @xmath127 , from the divergent behavior of the correlation length in fig.1(a ) of the main text , we expect a gaussian transition at @xmath72 . to confirm this , we calculate the von neumann entanglement entropy @xmath128 for a bipartition of the system into two half chains as a function of the correlation length @xmath129 . from conformal field theory ,
the entanglement entropy is known to scale as @xcite @xmath130 where @xmath131 is the central charge , @xmath132 is the lattice spacing ( we set @xmath133 ) , and @xmath134 is a nonuniversal constant . as shown in fig .
[ fig : eeandmx ] ( a ) , the entanglement entropy @xmath128 at @xmath135 is well fitted by a linear function of @xmath136 , and the central charge is estimated as @xmath137 .
this is close to the expected value @xmath138 at the gaussian transition . for @xmath65 and @xmath139 ,
a gaussian transition at @xmath64 splits into two ising transitions , and we have an intermediate phase between these transitions .
in fact , between the two peaks in fig.1(b ) of the main text , the @xmath57 spin reversal symmetry in @xmath61 axis ( @xmath140 ) is spontaneously broken .
then the staggered magnetizations along @xmath61 axis , @xmath141 , take finite expectation values in the intermediate phase , as shown in fig .
[ fig : eeandmx ] ( b ) .
such an intermediate nel phase between two distinct gapped symmetric phases is also observed in a two - leg spin-@xmath142 ladder @xcite when the @xmath143 symmetry is explicitly broken . to see a clear signature of the topologically trivial phases , we check the degeneracy in the entanglement spectra . for @xmath144 ,
the haldane phase is protected by time reversal , bond - centered inversion , and dihedral group of the spin rotations , as observed in ref .
@xcite . from fig .
[ fig : entspec ] ( a ) , the whole entanglement spectrum is two - fold degenerate in a region @xmath145 . on the other hand , as shown in fig .
[ fig : entspec ] ( b ) , once we introduce a finite staggered magnetic field @xmath12 , this two - fold degeneracy is lifted , and the haldane phase is merged with the imposed af ordered phase which is topologically trivial .
a nonzero dimerization @xmath81 breaks the site - centered inversion symmetry but still preserves the two - site translational invariance . in the main text
, we discuss an adiabatic continuity between the two trivial states @xmath74 and @xmath73 in the absence of the combined symmetry , @xmath146 .
we first show the correlation length as a function of @xmath2 for @xmath147 and @xmath64 in fig .
[ fig : cl_dimer ] . compared with fig.1(a ) in the main text , the correlation length around @xmath148 becomes shorter and exhibits a saturating behavior by increasing @xmath66 .
this indicates the absence of the phase transition due to the dimerization which breaks the combined symmetry @xmath149 .
in fact , this is easily and rigorously seen from the perfectly dimerized limit @xmath77 . in this limit
, the hamiltonian with @xmath64 is reduced to the sum of independent two - site hamiltonians , @xmath152 + h_z(s^z_1-s^z_2 ) .
\end{aligned}\ ] ] again we can obtain the two trivial states @xmath153 and @xmath154 in the limits @xmath155 and @xmath156 , respectively . therefore , the continuity between @xmath153 and @xmath154 is confirmed by finding a path on which no level crossing occurs in the lowest energy spectrum between these limits . in fig .
[ fig : energy_jd ] , we plot energy spectra of the two - site hamiltonian . the singlet state at @xmath157 is adiabatically connected to both the states @xmath158 and @xmath159 . from our numerical data
we thus conclude that , by breaking the site - centered inversion symmetry , the two states @xmath153 and @xmath154 are adiabatically connected and thus no longer distinguished .
4ifxundefined [ 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 http://stacks.iop.org/1742-5468/2004/i=06/a=p06002 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.235106 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.195122 [ * * , ( ) ] link:\doibase 10.1103/physrevb.81.064439 [ * * , ( ) ]
|
the ground state of the @xmath0 antiferromagnetic heisenberg chain belongs to the haldane phase a well known example of symmetry - protected topological phase . a staggered field applied to the @xmath0 antiferromagnetic chain breaks all the symmetries that protect the haldane phase as a topological phase , reducing it to a trivial phase .
that is , the haldane phase is then connected adiabatically to an antiferromagnetic product state .
nevertheless , as long as the symmetry under site - centered inversion combined with a spin rotation is preserved , the phase is still distinct from another trivial phase .
we demonstrate the existence of such distinct symmetry - protected _ trivial _ phases using a field - theoretical approach and numerical calculations .
furthermore , a general proof and a non - local order parameter are given in terms of an matrix - product state formulation . _
_ introduction.__while symmetry broken phases can be completely classified using the landau theory , there still exists no exhaustive understanding of topological quantum phases .
topological quantum phases are gapped phases of matter that are distinct from trivially disordered states but can not be characterized by any local order parameter . over the past few years
, new theoretical frameworks have been developed to understand and classify many different topological phases .
for example , topological phases of noninteracting fermions are now completely classified using k - theory @xcite .
more generally , two gapped ground states belong to the same phase if and only if they can adiabatically connected with respect to local hamiltonians @xcite .
even when different states are connected by a general adiabatic process , it is possible that they can no longer be adiabatically connected if we impose symmetries on the hamiltonian .
these are either states with spontaneous symmetry breaking or belong to the class of symmetry - protected topological ( spt ) phase @xcite .
examples of spt phases include topological insulators @xcite , which are protected by time reversal symmetry , and the haldane phase @xcite in one dimension , which is protected by either one of the time reversal , bond - centered inversion , or the dihedral group of the spin rotations @xcite . in these examples ,
spt phases are nontrivial in the sense that they can not be adiabatically connected to a trivial product state , once an appropriate symmetry is imposed .
they also support gapless edge states and/or nontrivial degenerate structures in the entanglement spectrum .
while the notion of the spt phases is now established and widely recognized , in this work we demonstrate that site - centered inversion symmetry allows to distinguish different trivial one - dimensional phases .
that is , there are multiple `` symmetry - protected _ trivial _ '' ( spt ) phases , i.e. , symmetric phases connected adiabatically to product states , which are still distinct in the presence of the imposed symmetry .
we note that , the word `` symmetry - protected trivial phase '' is sometimes used in place of the standard terminology of spt ( symmetry - protected topological ) phase , because the entanglement in such a phase is short - ranged and is removable in an adiabatic process if the symmetry is disregarded .
in contrast , in what is called an spt phase in this work , the entanglement can be completely removed adiabatically , even in the presence of the imposed symmetry , to reduce the state to a product state .
however , it is still distinct from another trivial phase .
the spt phases introduced here represent a new class of 1d quantum phases that are protected by a point - group symmetry but not captured by the cohomology classification . while each of these phases is trivial by itself , the quantum phase transition between them is experimentally detectable ( e.g. , by a divergence of some susceptibilities ) .
moreover , we derive non - local order parameters that could be used to characterize each phase .
while the concept of spt phases is very general , we illustrate it for clarity with a simple model of spin-1 chain in the following .
_ _ the model.__in order to make the discussion concrete , let us first consider the following simple model of @xmath0 chain : @xmath1 .
\label{eq.ham}\end{aligned}\ ] ] the first term is the standard heisenberg model with antiferromagnetic exchange interactions which stabilize the celebrated haldane gap @xcite .
the @xmath2 term is the uniaxial single - ion anisotropy , which is commonly present in magnetic ions with @xmath0 such as ni@xmath3 .
the model with uniaxial anisotropy @xmath2 and @xmath4 is well understood @xcite . for small @xmath5 ,
the system is in the haldane phase and undergoes a quantum phase transition into the `` large-@xmath6 '' phase at @xmath7 .
both phases are gapped and have the full symmetry of the hamiltonian .
the haldane phase is a well - known example of spt phases @xcite as discussed in the introduction .
the large-@xmath6 phase is a trivial phase which is adiabatically connected to the product state @xmath8 , where @xmath9 represents the local spin state with @xmath10 .
this state is the exact ground state of the hamiltonian in the limit @xmath11 .
the @xmath12 term represents a staggered field , which occurs in many quasi - one - dimensional materials , including haldane gap systems , with an alternating crystal structure under an applied ( uniform ) field . for simplicity
, we only include the staggered field term without the uniform one . in the limit @xmath13 , the spins are fully polarized along the staggered field , and the ground state is reduced to another trivial product state @xmath14 , where @xmath15 and @xmath16 represent the local spin states with @xmath17 and @xmath18 , respectively . it was recognized earlier that there is no phase transition for @xmath19 ( for @xmath20 ) @xcite .
that is , the haldane phase is adiabatically connected to the nel state @xmath21 with imposed antiferromagnetic ( af ) order . in the spt framework , this is naturally understood ; since the staggered field breaks all the symmetries that protect the haldane phase as an spt phase , it reduces the haldane phase to a trivial phase which also includes the nel state @xmath21 .
now let us discuss the model with both @xmath2 and @xmath12 .
in fact , this model has been studied in ref .
@xcite by a field theory and numerical methods where a quantum phase transition between the large-@xmath6 phase and the imposed af phase was found .
this is rather surprising , since both phases are trivial and are adiabatically connected to product states @xmath22 and @xmath21 that have the full symmetry of the hamiltonian .
it is perhaps even more surprising in the light of the recent concept of the spt phases , where the existing classification scheme @xcite would not distinguish them . while the nature of the phase transition were studied in ref .
@xcite , why ( and when ) these two trivial phases are distinguished was not completely clarified . in the remainder of this paper , we demonstrate that this is an example of distinct spt phases and identify the symmetry which protects them . _
bosonization. _ the standard bosonization procedure of @xmath0 chains starts from two coupled @xmath23 chains , and the low - energy effective field theory for @xmath24 is given by the hamiltonian @xcite @xmath25 \notag \\ & + { g_\textrm{eff}}\int dx \cos{(2\phi ) } , \label{eq : effham}\end{aligned}\ ] ] where @xmath26 and @xmath27 are dual field of each other , satisfying @xmath28 = i(\pi/2)[\textrm{sgn } ( x - x')+1]$ ] .
the hamiltonian represents the so - called sine - gordon field theory , which is ubiquitous in many problems in @xmath29 dimensions .
its properties essentially depend on the coupling constant @xmath30 .
when @xmath31 , the coupling @xmath32 is irrelevant under the renormalization group ( rg ) , and the system is renormalized in the low - energy limit into the free boson theory with @xmath33 , which is nothing but a gapless tomonaga - luttinger liquid ( tll ) . on the other hand ,
if @xmath34 and @xmath32 is non - vanishing , the interaction is rg - relevant , and the system acquires an excitation gap . in the absence of the staggered field , the haldane and the large-@xmath6 phases correspond to @xmath35 and @xmath36 respectively , both with @xmath34 .
it is easy to see that , within the effective hamiltonian , the two phases with @xmath35 and @xmath36 are always separated by the critical point @xmath33 .
this actually comes from the fact that @xmath37 is the only interaction compatible with the symmetry and the compactification , @xmath38 and @xmath39 , up to subleading terms @xmath40 @xmath41 . in general
, the effective theory can also have the @xmath42 term , which can be combined with the @xmath43 term as @xmath44 with a phase shift @xmath45 .
it is clear that , by changing @xmath45 from @xmath46 to @xmath47 , the two phases with @xmath35 and @xmath36 are adiabatically connected without closing the gap @xcite .
thus , for the two phases to be distinct , @xmath42 has to be forbidden by some symmetry . [ cols="<,^,^,^",options="header " , ] in fact , in the framework of bosonization , this is how a symmetry protects the haldane phase as an spt phase which is distinct from the trivial large-@xmath6 phase .
any of the three symmetries , which are known to protect the haldane phase , forbids the @xmath42 interaction @xcite . here
, for brevity , among these three symmetries , we only show the representation of the bond - centered inversion @xmath48 in terms of the bosonic field @xmath26 , @xmath27 in table [ table : symop ] .
the action of @xmath48 , @xmath49 , forbids @xmath50 , which leads to the distinction of the two phases with @xmath35 and @xmath36 .
on the other hand , table [ table : symop ] shows that the site - centered inversion @xmath51 also has the same action on @xmath26 , forbidding @xmath42 .
however , @xmath51 by itself is not sufficient to keep the distinction between the two phases ; they are adiabatically connected without gap closing , because of the vertex operators @xmath52 allowed in the absence of the u(1 ) symmetry of spin rotation about @xmath53 axis . according to table [
table : symop ] , the combined operation @xmath54 ( @xmath55 is the global @xmath47-rotation about @xmath53 axis ) , as well as @xmath48 , forbids @xmath52 . they still allow the next leading ones @xmath56 .
nevertheless , they just replace the direct transition between the @xmath35 and @xmath36 phases by an intermediate phase with spontaneous @xmath57-symmetry breaking .
thus @xmath58 alone should still maintain the distinction between two spt phases .
_ numerical results.__the above bosonization analysis suggests that we can introduce microscopic models with less symmetries than eq . , but with the symmetry under @xmath58 , to maintain the two distinct phases . as an example
, we consider the following hamiltonian : @xmath59 the new term @xmath60 represents the ( uniform ) dzyaloshinskii - moriya ( dm ) interaction with the dm vector parallel to @xmath61 axis .
this term breaks not only the u(1 ) spin - rotational symmetry about @xmath53 axis , but also both @xmath51 and @xmath55 as individual symmetries .
however , @xmath62 with a non - vanishing @xmath60 still preserves the symmetry @xmath58 under the composite operation . .
the parameters are varied as ( a ) @xmath63 , @xmath64 , and ( b ) @xmath65 .
each color and symbol denotes the different number of kept states @xmath66 from @xmath67 to @xmath68 .
insets show the nonlocal order parameters @xmath69 for @xmath70 and @xmath68 ( see text).,scaledwidth=45.0% ] we numerically study hamiltonian using infinite density - matrix renormalization group ( idmrg ) @xcite .
the correlation lengths as functions of @xmath2 are plotted in fig . [ fig : corrlength ] for different numbers @xmath66 of kept states and parameters of the model .
a divergent correlation length with increasing @xmath66 indicates a critical point . for @xmath63 and @xmath64 , we find that the haldane phase and the nel state ( @xmath71 ) are adiabatically connected since all of the three symmetries protecting the haldane phase are broken .
however , as found in refs .
@xcite , the transition at @xmath72 still exists ( see fig . [
fig : corrlength ] ( a ) ) .
this indicates that there is a phase transition between two trivial phases connected to @xmath73 and @xmath74 . to confirm that this transition is protected by @xmath58 alone
, we further introduce @xmath60 in fig .
[ fig : corrlength ] ( b ) . a single transition in fig .
[ fig : corrlength ] ( a ) is now split into two transitions , but the two phases are still separated by ( two ) transitions and thus are distinct . in the intermediate phase , an af order along @xmath61 axis occurs and thus @xmath58 is spontaneously broken .
further details about this calculation are shown in .
once an explicit dimerization in introduced , e.g. , by adding a term @xmath75 with @xmath76 , @xmath58 is broken without affecting any other symmetries in @xmath62 , and there is only one trivial phase . numerically , we observe that the correlation length remains finite for all values of @xmath2 when @xmath64 .
in fact , this can also be shown analytically by considering the limit of @xmath77 with @xmath78 . here ,
the entire chain is decomposed into isolated dimers . in particular , at @xmath79 , the ground state is simply given by a product of spin singlet states on each dimer .
it can be shown , by solving the two - spin problem explicitly , that this dimerized state is connected adiabatically to both @xmath80 and @xmath13 limits .
thus , the two trivial product states @xmath22 and @xmath21 can be adiabatically connected through the dimerized limit , and belong to a single phase , in the presence of @xmath81 .
this fact rules out the possibility that the two trivial phases are distinct under the two - site translation invariance and some on - site symmetry , as indicated in refs .
@xcite .
_ _ matrix - product state formulation.__matrix - product states ( mps ) can represent gapped ground states of local hamiltonians in one dimension faithfully .
thus , the classification of gapped phases in one dimension , including the spt phase proposed in the present work , can be proven rigorously within the mps formalism .
let us begin with the general mps @xcite , without assuming any translation invariance : @xmath82}_{m_{n-1 } } \lambda^{[n-\frac{1}{2 } ] } \gamma^{[n]}_{m_n } \lambda^{[n+\frac{1}{2 } ] } \gamma^{[n+1]}_{m_{n+1 } } \cdots \nonumber \\ & & \times \left| \cdots m_{n-1 } m_n m_{n+1 } \cdots \right\rangle,\end{aligned}\ ] ] where @xmath83}$ ] is a @xmath84 positive diagonal matrix , @xmath85}$ ] is a @xmath86 matrix , and @xmath87 represents the physical degrees of freedom on site @xmath88 .
an mps representation is not unique for a given state but we can always choose the canonical mps @xcite satisfying @xmath89}\right)^2\right ] } = 1 $ ] , and @xmath90}(\mathbb{i}_{\chi_{n+1/2 } } ) = \mathbb{i}_{\chi_{n-1/2 } } , \ ; \ ; \bar{{\mathcal{e}}}^{[n ] } ( \mathbb{i}_{\chi_{n-1/2 } } ) = \mathbb{i}_{\chi_{n+1/2 } } , \label{eq.canonical}\ ] ] where @xmath91 is the @xmath92 identity matrix , and @xmath93}$ ] and @xmath94}$ ] are completely positive maps defined by @xmath95}(x ) & \equiv \sum_m \gamma_m^{[n ] } \lambda^{[n+\frac{1}{2 } ] } x \lambda^{[n+\frac{1}{2 } ] } \left ( \gamma_m^{[n ] } \right)^\dagger , \\ \bar{{\mathcal{e}}}^{[n ] } ( y ) & \equiv \sum_m \left ( \gamma_m^{[n ] } \right)^\dagger \lambda^{[n-\frac{1}{2 } ] } y \lambda^{[n-\frac{1}{2 } ] } \gamma_m^{[n]}. \end{split}\ ] ] by introducing the metric @xmath96})^2 x^\dagger]$ ] in the vector space of @xmath84 matrices , we can introduce a singular value decomposition of @xmath93}$ ] and @xmath94}$ ] .
the canonical condition eq
. means that the identity matrices are left / right `` eigenvectors '' of @xmath93}$ ] and @xmath94}$ ] belonging to the largest singular value @xmath97 . in the following
we assume that the mps is pure , that is the largest singular value @xmath97 is nondegenerate . in order to consider the symmetry @xmath58
, we define @xmath98 so that @xmath51 can be identified with @xmath99 with the inversion center at site @xmath100 . following refs .
@xcite , if @xmath101 is invariant under the combined symmetry @xmath58 , it satisfies @xmath102 } \right)^t = e^{i\theta_{{\mathcal{i}}'}^{[n ] } } \left ( u_{{\mathcal{i}}'}^{[-n-\frac{1}{2 } ] } \right)^\dagger \gamma_m^{[-n ] } u_{{\mathcal{i}}'}^{[-n+\frac{1}{2 } ] } , \label{eq.fe_calip}\end{aligned}\ ] ] where @xmath103 is the representation of @xmath55 acting on the physical hilbert space of each site , @xmath104}$ ] is a phase , and @xmath105}$ ] is a @xmath84 unitary matrix commuting with @xmath83}$ ] .
@xmath106 also implies that @xmath83 } = \lambda^{[-a]}$ ] and @xmath107 . using the above relation twice ,
we obtain @xmath108 } \left ( a^{[n-\frac{1}{2 } ] } \right ) & = e^{- i ( \theta^{[n]}_{{\mathcal{i}'}}+ \theta^{[-n]}_{{\mathcal{i } ' } } ) } a^{[n+\frac{1}{2 } ] } , \label{eq.cmpea}\end{aligned}\ ] ] where @xmath109 } \equiv \left ( u^{[-a]}_{{\mathcal{i}'}}\right)^t \left ( u^{[a]}_{{\mathcal{i}'}}\right)^\dagger$ ] . since @xmath110}|^2=1 $ ] , eq
. implies @xmath111}$ ] are left / right eigenvectors of @xmath94}$ ] belonging to the singular value @xmath97 .
the assumption of the pure mps , namely non - degeneracy of the singular value @xmath97 of @xmath94}(x)$ ] , implies @xmath112 } = e^ { i \phi^{[a]}_{{\mathcal{i } ' } } } \mathbb{i}_{\chi_a}$ ] , where @xmath113}_{{\mathcal{i}'}}+ \phi^{[-a]}_{{\mathcal{i}'}}= 0\mod{2\pi}$ ] . combining these with eq . and the canonical condition
, we obtain @xmath114}_{{\mathcal{i}'}}+ \theta^{[-n]}_{{\mathcal{i}'}}= \phi^{[n+1/2]}_{{\mathcal{i}'}}- \phi^{[n-1/2]}_{{\mathcal{i}'}}\mod{2\pi}$ ] .
in particular , for @xmath100 , we find @xmath115}_{{\mathcal{i}'}}- \phi^{[1/2]}_{{\mathcal{i } ' } } ) = 0 \mod{2\pi}.\end{aligned}\ ] ] as a consequence , @xmath116}_{{\mathcal{i}'}}-\phi^{[1/2]}_{{\mathcal{i}'}}$ ] is quantized to either @xmath46 or @xmath47 ; it can not change unless the system undergoes a quantum phase transition .
this implies that , in the presence of the @xmath106 symmetry , there are two distinct phases corresponding to @xmath116}_{{\mathcal{i}'}}-\phi^{[1/2]}_{{\mathcal{i}'}}=0 $ ] and @xmath47 .
let us now consider the limits of the trivial product states @xmath22 and @xmath21 . here
, all the matrices @xmath85}$ ] , @xmath117}$ ] , and @xmath118}_{{\mathcal{i}'}}$ ] are reduced to scalars ( @xmath119 matrices ) and thus commute with each other .
then the fundamental relation for @xmath100 reads @xmath116}_{{\mathcal{i}'}}-\phi^{[1/2]}_{{\mathcal{i}'}}= 0 $ ] for @xmath22 and @xmath116}_{{\mathcal{i}'}}-\phi^{[1/2]}_{{\mathcal{i}'}}= \pi$ ] for @xmath21 .
this establishes that , under the @xmath58 symmetry , the two product states @xmath22 and @xmath21 indeed belong to distinct phases , which are always separated by a quantum phase transition . as in the case of spt phases ,
no local order parameter can distinguish spt phases .
however , using the mps framework , we can directly derive _ non - local _ order parameters @xcite which are sensitive to the phase factor @xmath116}_{{\mathcal{i}'}}-\phi^{[1/2]}_{{\mathcal{i}'}}$ ] .
in particular , we can define an operator @xmath120 which inverts a block of @xmath121 consecutive sites . for @xmath122
much larger than the correlation length , we find that @xmath123}_{{\mathcal{i}'}}-\phi^{[1/2]}_{{\mathcal{i}'}})}. \end{aligned}\ ] ] the insets in fig .
[ fig : corrlength ] show that the different spt phases are indeed distinguished by @xmath124 while @xmath125 when @xmath58 is broken .
_ conclusion and discussion. _ we demonstrated that there exists two distinct spt phases in the presence of the symmetry under the site - centered inversion combined with a spin rotation .
we showed the existence of such phases by field - theoretical arguments based on bosonization and presented a general proof based on the mps formalism .
while it is known that distinct trivial phases can exist in translation - invariant systems @xcite , it is surprising that only point - group symmetries can stabilize distinct trivial phases in simple 1d systems .
our finding implies that more studies are needed for complete classification of quantum phases in one dimension under symmetries .
we derived non - local order parameter that could be measured in optical lattice realizations @xcite .
moreover , even without any measurement of the non - local order parameter , the quantum phase transition separating the distinct spt phases can be detected in standard experimental measurements , such as a divergence of the low - temperature specific heat when the gap is closing .
this , in fact , is more in line with the operational definition of the spt phases .
the notion of the spt phases is not restricted to one - dimensional systems .
in fact , what appear as examples of spt phases in two dimensions were discussed in ref .
@xcite .
the lack of universal theoretical description of quantum many - body systems in higher dimensions makes a systematic analysis of spt phases more difficult than in one dimension .
nevertheless , it would be certainly an interesting direction for the future .
[ [ acknowledgment ] ] acknowledgment + + + + + + + + + + + + + + + yf thanks p. lecheminant , s. nishimoto , and k. totsuka for fruitful discussions .
yf was supported in part by the program for leading graduate schools , mext , japan .
this work was supported in part by jsps grant - in - aid for scientific research ( kakenhi ) no .
25400392 and jsps strategic international networks program no .
r2604 `` toponet . ''
numerical calculations were performed on supercomputers at the institute for solid state physics , university of tokyo .
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 link:\doibase 10.1063/1.3149495 [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.085103 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.94.140601 [ * * , ( ) ] link:\doibase 10.1103/physrevb.82.155138 [ * * , ( ) ] link:\doibase 10.1103/physrevb.84.165139 [ * * , ( ) ] link:\doibase 10.1103/physrevb.80.155131 [ * * , ( ) ] link:\doibase 10.1103/physrevb.81.064439 [ * * , ( ) ] link:\doibase 10.1103/physrevb.83.035107 [ * * , ( ) ] link:\doibase 10.1103/physrevb.84.235128 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.155114 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.95.146802 [ * * , ( ) ] link:\doibase 10.1016/0375 - 9601(83)90631-x [ * * , ( ) ] link:\doibase 10.1103/physrevlett.50.1153 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.59.799 [ * * , ( ) ] http://projecteuclid.org/euclid.cmp/1104161001 [ * * , ( ) ] link:\doibase 10.1103/physrevb.40.4709 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.66.798 [ * * , ( ) ] link:\doibase 10.1103/physrevb.67.104401 [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.075125 [ * * , ( ) ] link:\doibase 10.1143/jpsj.67.302 [ * * , ( ) ] link:\doibase 10.1103/physrevb.57.r8087 [ * * , ( ) ] link:\doibase 10.1103/physrevb.34.6372 [ * * , ( ) ] link:\doibase 10.1103/physrevb.77.245119 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.106.110405 [ * * , ( ) ] http://arxiv.org/abs/1410.4211 [ ] link:\doibase 10.1103/physrevlett.69.2863 [ * * , ( ) ] http://arxiv.org/abs/0804.2509 [ ] link:\doibase 10.1103/physrevb.87.235106 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.195101 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.91.147902 [ * * , ( ) ] link:\doibase 10.1103/physrevb.78.155117 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.100.167202 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.125441 [ * * , ( ) ] link:\doibase 10.1126/science.1209284 [ * * , ( ) ] link:\doibase 10.1103/physrevb.76.161104 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.166402 [ * * , ( ) ] http://stacks.iop.org/1742-5468/2004/i=06/a=p06002 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.195122 [ * * , ( ) ]
|
weyl semimetals ( wsm ) are three dimensional topological electronic systems whose low energy band structure is described by pairs of weyl fermions separated in momentum space .
they were first proposed to occur in a class of iridate materials@xcite , but it was only recently that they were experimentally realized@xcite .
the topological properties of wsms are encoded in a chern - simons ( cs ) term that stems from the axial anomaly .
the anomaly is responsible of a plethora of phenomena like the axial magnetic effect@xcite , chiral separation effect@xcite , chiral magnetic waves@xcite and , more recently , a new mechanism for the phonon hall viscosity@xcite . the experimental evidence of the wsm phase is based on arpes and transport experiments@xcite , and on optical conductivity experiments@xcite (
optical measurements have been done in dirac semimetals as well@xcite ) .
the cs term induces birefringence@xcite , and circular dichroism@xcite .
it has been also proposed that chiral electromagnetic waves exist and propagate along domain walls in magnetic wsms@xcite .
the interaction of light with magnons in wsms has been addressed in@xcite , while plasmons in weyl metals have also been studied@xcite .
more recently , a new type of helicon mode under applied magnetic fields has been proposed@xcite at finite doping .
here we address the problem of light propagation in a wsm at the neutrality point ( i.e. , the fermi level lies at the weyl nodes ) by computing the poles of the full photon propagator .
a similar approach was done for quantum electrodynamics ( qed ) under an external magnetic field@xcite . for the case of the solid state - realized wsm
there are two key aspects that are central : 1 ) nontrivial topology is present even in the absence of an external magnetic field and 2 ) contrary to qed there are two velocity scales in wsms : the fermi velocity of the electrons and the velocity of light in the material .
we will see how these two features combine so that novel collective excitation behavior is realized .
we will resort ourselves on a rpa treatment of the problem .
this implies that we will effectively consider a non - self interacting weyl fermions interacting with an external electromagnetic field .
the validity of a rpa treatment has been extensively studied in the literature for two dimensional dirac materials@xcite and more recently in the general case of interacting three dimensional dirac systems@xcite . following these references ,
some of the results presented here might be extended to the self - interacting case under particular conditions ( @xmath0 expansions and so ) .
the rest of the paper is structured as follows . in section
ii we describe the model for wsm that we use and set notations . in section iii , we give the neccessary details for the computation of the polarization function and explain it . in section iv , we compute the spectrum of the electromagnetic collective modes through the spectral function , and describe the most salient features observed . in section v , we discuss the results and mention some future lines to follow from this work .
our starting point is the low energy action of a wsm coupled to an electromagnetic field : @xmath1 where @xmath2 is the fermi velocity and @xmath3 is the speed of light in the corresponding material .
this action describes two weyl fermions separated in momentum space by a vector @xmath4 so that time reversal symmetry is broken . without loss of generality will fix @xmath5 .
quantum corrections to ( [ clasaction ] ) are well known to generate a nonzero odd part of the polarization function of the photon in the form of a cs term @xcite .
the coefficient accompanying the cs term is ill defined , as it depends on the routing of internal momenta in the loop integrals as well as on the regularization method . in a solid state realization of a wsm
this coefficient can be fixed by means of the bulk - boundary correspondence relating the theory in the bulk with the fermi arcs at the edges@xcite .
another way is by considering the 3d band structure of a wsm as a set of independent 2d band structures , each of them with a well defined chern number@xcite .
in addition to the action for the fermionic sector ( [ clasaction ] ) , the action for the electromagnetic field is : @xmath6 where @xmath7 and @xmath8 are the dielectric permittivity and magnetic permeability of the material respectively .
the electric and magnetic fields are related to the gauge field @xmath9 as : @xmath10
in what follows we will obtain the 1-loop order polarization function in terms of @xmath9 .
the computation has been performed and deeply examined by several authors , both the parity - even@xcite and parity - odd@xcite parts of the polarization function .
we can write it as : @xmath11 where @xmath12 is the contribution at zero order in @xmath13 , and @xmath14 is the first order correction .
in@xcite it was shown that the polarization function is analytic in @xmath13 , and the correction to the even part is second order in @xmath13 , breaks gauge invariance and gives a mass to the photon . imposing gauge invariance by using a regularization which preserves it , as dimensional or pauli - villars regularizations ,
one obtains a zero correction@xcite . regarding the odd part ,
it is known to be linear in @xmath13@xcite and no corrections appear beyond the first order .
( [ polarization ] ) is exact , i.e. there are not higher order corrections in @xmath13 .
the computation of @xmath12 is standard . using dimensional regularization and in the minimal subtraction renormalization scheme : @xmath15 @xmath16 where the four momentum is @xmath17 , and where the space components of @xmath12 have a relative weight @xmath18 with respect to the time components , as can be read from eq .
( [ clasaction ] ) . from now on we will redefine @xmath19 as @xmath20 so that @xmath21 . the parameter @xmath22 is an energy scale that appears in the process of dimensional regularization ( see for example @xcite ) .
the appearance of this parameter reflects the formal absence of any characteristic scale in the linear electronic spectrum .
it could be fixed by either the experiment or by invoking a standard renormalization program ( here we are dealing with a modification of standard quantum electrodynamics , that is well known to be renormalizable ) or related to the lattice cutoff of the underlying bandstructure by considering a full lattice model of the weyl semimetal , keeping in mind that there are several materials with very different lattice structures proposed to be weyl semimetals . for typical values of lattice spacings of the order of ngstrms , @xmath23 can be estimated to be @xmath24 ev , however , the reader should keep in mind that the low energy / long wavelength aspects of the results presented here do not depend on the particular value of @xmath22 as long as all the frequencies and momenta involved in the problem are presented in a _ dimensionless _ fashion ( that is , divided by @xmath23 ) no matter which regularization scheme is used .
note that the function @xmath25 in eq .
( [ polfunczeroorder2 ] ) has a constant imaginary part for @xmath26 which defines the electron - hole continuum threshold .
a normal procedure would be to take the local limit @xmath27 in @xmath12 .
as we know for graphene , at the neutrality point and zero temperature no well defined plasmon mode exists within the random phase approximation ( rpa ) .
however , it was shown that a plasmon pole appears beyond rpa when the momentum dependence in the polarization tensor is maintained@xcite , which highlights the importance of not considering the local limit in some circumstances . here
we will limit ourselves to the rpa approximation but we will keep the momentum dependence , and we will see that staying away from the local limit is crucial for capturing the physics near the electron - hole continuum threshold .
@xmath14 is precisely the cs term .
we invoke the boundary - bulk correspondence to fix its value@xcite . on the @xmath28 gauge @xmath14
has only one independent component : @xmath29 the presence of this non - diagonal term in the polarization tensor induces gyrotropy in the wsm , so that left- and right - handed polarized light propagate at different speeds@xcite .
since we are working under the free electron approximation , eq.([oddpol ] ) is exact . in an interacting theory
, @xmath30 acquires dynamically generated corrections ( see @xcite for an explicit discussion when interactions at constant external magnetic field and finite densities are considered , or @xcite for corrections in the strong coupling regime ) .
the inverse full photon propagator can be written as : @xmath31 where the inverse free propagator can be obtained from eqs .
( [ actionphoton],[actionphoton2 ] ) . to obtain the collective excitations we look for the poles of the full propagator .
the equation @xmath32=0 $ ] on the @xmath28 gauge reads : @xmath33 where @xmath34 represents the momentum in the plane perpendicular to @xmath35 , @xmath36 is the fine structure constant , @xmath37 gives the squared ratio between the speed of light and the fermi velocity , and : @xmath38 @xmath39 we will take the fermi velocity of the wsm similar to that of graphene ( @xmath40 ) and assume a coupling constant @xmath41 , which is reached with a relative permittivity of the material of @xmath42 .
the speed of light of the wsm is given by @xmath43 ( assuming a relative permeability @xmath44 ) and the value of the squared ratio between the speed of light and the fermi velocity is @xmath45 .
since we are not taking into account interactions between weyl electrons , a value of the effective coupling @xmath41 is not problematic , and this precise value has been chosen for convenience . in realistic solid - state wsm , however , it is not expected to have these values for @xmath46 and/or larger values of @xmath47 , resulting effectively in an smaller value of @xmath48 .
there is a zero mode @xmath49 corresponding to the unphysical longitudinal gauge field , arising from the residual gauge freedom , as @xmath28 does not completely fix the gauge . for the remaining part , there are two physical transverse gauge fields .
the dispersion relation of the unscreened field can be obtained from : @xmath50 instead of directly solving the equation , we will define a spectral function for this field : @xmath51.\ ] ] the spectral function is plotted in fig . [ fig 1](a ) .
we see that there is a peak with finite width , which corresponds to a damped mode , so that the delta function corresponding to a free photon is replaced in a wsm by an attenuated excitation .
hence our first finding is that light is attenuated when propagating through a wsm .
this behavior is rooted to the existence of two different velocity scales : the value of the fermi velocity is much smaller than the speed of light .
this mismatch is encoded in the function @xmath52 appearing in eq .
( [ eq unscreened mode ] ) , which fulfills @xmath53 .
this means that in the limit where the value of the fermi velocity matches the value of the speed of light in the material , we have a non attenuated free photon traveling at the speed of light with dispersion relation @xmath54 . although the gapless nature of wsms means that there is no gap to overcome for electron - hole pair creation , we just showed that this is not enough for light attenuation , and that damping is generated for a system of gapless electrons only when there exist two non - equal competing velocity scales .
this happens also in dirac semimetals since this property is not related to the cs term . *
( a ) * * ( b ) * the nontrivial topology of the wsm phase shows up inducing a screening in the other transverse gauge field
. the dispersion relation can be obtained from the second line of eq .
( [ eq det ] ) . instead of solving the equation
, we will again introduce a spectral function for this field : @xmath55.\end{aligned}\ ] ] in the analysis following we will consider two separate cases : @xmath56 and @xmath57 .
\i ) @xmath56 . in fig .
[ fig 1](b ) the spectral function is plotted as a function of @xmath58 and the momentum @xmath59 in the direction of @xmath35 .
two branches of attenuated collective excitations coexist , one of them clearly gaped .
the gaped nature of the screened field is expected in wsms , as the cs term acts as a topological mass for the photon@xcite .
however we see that the second branch corresponds to a new excitation which seems to be gapless and considerably less attenuated than its gaped companion .
in fact , at a certain large momentum @xmath60 there is a crossover where the gaped excitation enters into an overdamped regime , as can be seen from the absence of a well defined peak in fig . [ fig 1](b ) . to highlight the behavior of the novel gapless excitation at low frequencies near the electron - hole continuum threshold @xmath54 , we plot in fig . [ fig 2](a ) the spectral function as a function of the frequency in a logarithmic scale , for different values of @xmath60 .
we see that there is a low energy crossover also for the novel excitation from an underdamped to an overdamped regime as the wavelength increases .
overdamping appears for frequencies near the electron - hole continuum threshold .
this suggests a gapless nature of this new excitation with dispersion relation @xmath61 at low energies .
we see also that increasing the momentum @xmath60 , the gapless branch deviates from the linear dispersion behavior and approaches the gaped one . from fig .
[ fig 1](b ) it is clear that the gapless branch starts to behave as a massive excitation for large enough momentum @xmath60 .
\ii ) @xmath57 .
[ fig 1](c , d ) show the spectral function as a function of @xmath58 and @xmath62 for the values @xmath63 and @xmath64 respectively .
it can be seen that increasing the perpendicular momentum produces a merging of the two branches into a single peak .
this is a result of a second , high energy crossover for the gapless excitation which again becomes overdamped , but now at big enough @xmath65 . for energies near the electron - hole continuum threshold
, we can see in fig .
[ fig 2](b ) that small fluctuations in @xmath65 drive the gapless excitation across the low energy crossover from the overdamped to the underdamped regime , so in a realistic situation when the propagation is not perfectly aligned with the @xmath35 direction the gapless branch behaves as a well defined attenuated excitation at low energies .
it is important to note that the behavior near the electron - hole continuum would have never been captured in the local limit @xmath66 of the polarization function .
in particular the low energy crossover into the overdamped regime of the massless branch would have never been captured without considering the full momentum dependence .
we have studied light propagation in a neutral wsm where the fermi level lies at the weyl nodes . in this situation
there are two degrees of freedom corresponding to two transverse gauge fields , while the plasmon longitudinal mode that appears at finite chemical potential is absent . besides the generation of a plasmon mode ,
the presence of a plasma frequency would also induce a gap to both transverse gauge fields , however at the neutrality point there is only one gaped mode , whose mass is a consequence of the nontrivial topology .
our first finding is that light is attenuated when propagating through both weyl and dirac semimetals . in the case of wsms ,
nontrivial topology shows up as a screening effect in one of the two transverse gauge fields , for which we find two branches of attenuated collective excitations .
in addition to the known topologically gaped photon mode , a novel massless attenuated excitation appears , which is our second and most important finding .
this splitting of a transverse field into two branches was obtained before for qed in the presence of an external magnetic filed@xcite , where the ciclotron frequency defines the mass threshold for the electron - hole continuum . in the case of a wsm
there is no mass threshold as nontrivial topology is present in the absence of magnetic fields .
furthermore for the case of the solid state - realized wsm the fermi velocity of the electrons is much smaller than the speed of light , which translates into a completely novel collective behavior of the two branches .
in particular we find that in a realistic situation of light collimated along the direction of @xmath35 but with small deviations in the perpendicular direction , the massless branch shows up as a well defined and damped peak in the spectral function . at low energies
its dispersion relation is given by @xmath67 , which defines an excitation traveling coherently with the electrons at the fermi velocity .
at the other limit of the spectrum ( @xmath68 ) the gapless branch behaves as a massive damped photon with velocity close to the speed of light .
in the situation where light propagation is not aligned with @xmath35 and the momentum in the transverse direction is large enough , a crossover exists and the massless branch enters into an overdamped regime , corresponding to a peak with divergent width in the spectral function
. the novel optical excitation could provide a new and clean signature for wsms .
it could be detected in experiments as an additional resonance on the transmission coefficient of light .
there are also indirect proofs , like electron energy - loss spectroscopy ( eels ) or subluminal cherenkov radiation spectroscopy@xcite stemming from the small phase velocity of the novel excitation .
in addition , due to the matching of the propagation velocity of the novel mode with the fermi velocity of the electrons , a counter intuitive enhancement of the effective dressed electron - photon interaction is expected .
this is qualitatively different from the effect of the coulomb interaction enhancement due to the renormalization of the fermi velocity that appears , for instance , in graphene@xcite .
it should be noted that , although their effect is negligible on the bulk propagation of light , the 1d fermi arc chiral states at the edges of a wsm would give additional resonances as surface electromagnetic modes .
however , the polarization tensor of 1d chiral electrons has already been computed in the literature@xcite and therefore the fermi arc contribution can be distinguished from bulk propagation .
we thank m. a. h. vozmediano for enlightening discussions and encouragement . this research is partially supported by the spanish mecd grant no .
fis2011 - 23713 , the european union structural funds and the comunidad de madrid mad2d - cm program ( s2013/mit-3007 ) , and by the european union seventh framework programme under grant agreement no .
604391 graphene flagship .
|
we study light propagation in a neutral weyl semimetal with the fermi level lying at the weyl nodes in the weak self - interacting regime .
the nontrivial topology induces a screening effect in one of the two transverse gauge fields , for which we find two branches of attenuated collective excitations .
in addition to the known topologically gaped photon mode , a novel massless and slightly damped excitation appears .
strikingly , at low energies this new excitation has a linear dispersion and it propagates with the same velocity than the electrons , while at energies well above the electron - hole continuum threshold it behaves as a massive attenuated photon with velocity similar to the speed of light in the material .
there is a crossover at certain momentum in the direction perpendicular to the separation of the weyl nodes above which the novel gapless mode enters into an overdamped regime .
regarding the unscreened gauge field we show that it is also attenuated , which is a non - topological property shared by dirac semimetals as well .
|
the sn impostor sn 2009ip is an lbv star @xcite that experienced a series of outbursts starting in 2009 ( e.g. , @xcite ) .
there were two outbursts in 2012 : the 2012a outburst that was similar to the previous outbursts , and the 2012b outburst in september 2012 that was much more energetic than the previous outbursts ( e.g. , @xcite ) .
there is no consensus on the nature of the 2012a and 2012b outbursts ( e.g. @xcite ) , and not even on whether the lbv primary star survived the 2012 outbursts or exploded as a core collapse supernova ( ccsn ) .
the peak luminosity of the 2012b outburst and the p - cygni absorption wings that extend to @xmath2 brought @xcite to suggest that the 2012a outburst was a weak sn event , while the major 2012b outburst is a result of the collision of the sn ejecta with previously ejected gas ( also @xcite ) .
however , @xcite noted that the photosphere expansion velocity of @xmath3 during the 2012b outburst implies that the gas that accelerated the photosphere must have originated long after the peak of the 2012a event .
namely , the gas was ejected long after the star has ceased to exist according to the scenario proposed by @xcite .
@xcite suggest that the star did survive the 2012 outbursts due to the fact that peaks and troughs are present in the declining light curve of the 2012b outburst .
@xcite noted that the september 2011 outburst already had such wide p - cygni wings , and raised the possibility that the 2012b outburst was not a sn after all .
the early outbursts of sn 2009ip were compared with the 18371856 great eruption of @xmath4 carinae @xcite .
@xmath4 carinae also had a small fraction of the mass ejected during the great eruption moving at high velocities @xcite of up to @xmath5 @xcite .
@xcite , for a terminal periastron passage , and @xcite , for repeated periastron passages , went further and argued that the 2012b outburst was powered by mass accretion onto a main sequence ( ms ) or a wolf - rayet ( wr ) stellar companion , much like their model for powering the great eruption of @xmath4 carinae @xcite .
in the scenarios considered by @xcite the major 2012b outburst of @xmath6 was powered by @xmath7 accreted onto the companion during a periastron passage ( the first passage ) of the binary system when the companion was inside the extended primary envelope .
the major open question in the binary scenarios for sn 2009ip is how such an interaction can give rise to ejecta with velocities of @xmath8 . in this first study
we limit ourselves to outbursts with low - mass high - velocity ejecta , such as the 2011 outburst of sn 2009ip and the great eruption of @xmath4 carinae .
the numerical setup is described in section [ sec : numerics ] .
our results are described in section [ sec : results ] , and our summary is in section [ sec : summary ] .
the simulations are performed using the hydrodynamics code @xcite .
we employ a full 3d uniform grid ( equal - size cells ) with cartesian @xmath9 geometry . in all
runs the grid is a cube of size @xmath10 composed of @xmath11 cubical cells .
a high - resolution run with @xmath12 cubical cells shows no difference in the results .
the temporary common envelope has a general oblate structure @xcite , that allows us to use a rectangular grid and a simple plane - parallel structure for the shell , schematically presented in figure [ fig : setup ] . plane .
the blue dashed line marks the computational domain .
the solid red line marks the initial location of the cloud with which the jet interacts .
lower panel : the density profile along the @xmath13 axis through the initial jet launching cells and the shell at @xmath14 ( before jets expansion took place ) for the @xmath15 run .
the simulation results for this density profile are presented in figures [ fig : results_3000 ] and [ fig : vel_hist ] .
note that @xmath16 is not the equatorial plane , but rather the lower boundary of the computational domain .
, title="fig : " ] + plane .
the blue dashed line marks the computational domain .
the solid red line marks the initial location of the cloud with which the jet interacts .
lower panel : the density profile along the @xmath13 axis through the initial jet launching cells and the shell at @xmath14 ( before jets expansion took place ) for the @xmath15 run .
the simulation results for this density profile are presented in figures [ fig : results_3000 ] and [ fig : vel_hist ] .
note that @xmath16 is not the equatorial plane , but rather the lower boundary of the computational domain .
, title="fig : " ] we simulate the general flow structure proposed by @xcite , where the companion launches two opposite collimated outflows while inside or close to the primary star .
the @xmath17 plane is the lower boundary of the computational domain , just above the outskirts of the temporary common envelope , @xmath18 above the equatorial plane .
the symmetry of the flow allows us to simulate the jet - shell interaction only in the @xmath19 half space .
the jet is injected along the @xmath20 direction and from the bottom three cells of the computational grid . for the 2012b outburst of sn 2009ip @xcite suggest that the interaction lasted for @xmath21 days during which the companion accreted @xmath7 .
we consider the less energetic 2011 outburst and simulate only @xmath22 hours of jets launching episode , but assume a similar accretion rate .
we also take most of the gravitational energy released by the accretion process to be channelled to the jets , such that the kinetic power in each jet is @xmath23 .
the jets velocity is taken to be @xmath15 , similar to the wind velocity of the secondary in @xmath4 carinae @xcite .
we also simulate a case with the same power but with @xmath24 , similar to the escape velocity from very massive ms and wr stars .
the total ejected mass in the two jets in our simulation is @xmath25 ( @xmath26 ) for the jets velocity of @xmath27 at 8 hours simulation time ( @xmath28 at 11 hours simulation time ) .
the secondary is a ms star , and we take the jets radius to be @xmath29 , @xmath30 times the secondary radius . as the companion accretes only @xmath31 , and by our assumption most of the released gravitational energy is carried by the jets , we do nt expect the radius of the secondary star to change by much for the duration of the simulation .
we assume that the jets interact with a shell , that might actually be an extended envelope .
such an extended envelope might be formed around unstable lbv and agb stars experiencing high mass loss rates ( e.g. , @xcite ) .
the lbv star @xmath4 carinae had a huge atmosphere during its great eruption , when the photosphere was at @xmath32 @xcite .
the radius of pulsating agb stars can vary by a factor of @xmath33 ( e.g. , @xcite ) .
such an extended envelope is not a wind , and it is expected to be rather smooth and not clumpy as winds might be . in the proposed model
the extended envelope results both from the unstable phase of the lbv primary star and from the tidal disturbance caused by the companion .
we assume that this expansion of the envelope results in a steep density profile .
one might take the steep density profile as a requirement of our model to obtain high velocities .
the density profile of the shell is taken to be @xmath34 } \qquad { \rm for } \qquad 20 r_{\odot } < z < 100 r_{\odot } , \label{eq : density}\ ] ] where @xmath35 , @xmath36 is the location of the lower boundary of the shell and @xmath37 ( @xmath38 ) for a jet velocity of @xmath27 ( @xmath28 ) .
the initial temperature of the shell is set to @xmath39 .
we let the jet interact with a ` cloud ' ( see figure [ fig : setup ] ) , namely , part of the shell .
when we took a full shell ( extended in the entire @xmath40 directions ) we obtained identical results .
for numerical stability we allow a low density gas , @xmath41 , to fill the space between the initial jet and the shell ( cloud ) at @xmath42 , extended in the entire @xmath43 plane between @xmath44 and @xmath45 .
the ambient density outside the shell is set to be @xmath46 at a temperature of @xmath47 . as the material in the shell heats up to high temperatures , @xmath48 , radiation pressure dominates and we take the adiabatic index to be @xmath49 .
as the jet hits the shell a shock wave is formed . because of the steeply declining density the shock accelerates outward , and with it the shell material , up to velocities of @xmath50
this is demonstrated in figure [ fig : results_3000 ] for the @xmath51 case , and in figure [ fig : results_2000 ] for the @xmath24 case .
after the jet breaks out of the shell the acceleration ceases .
however , in reality the secondary star orbits the primary star and so the jets source moves , allowing the jet to encounter additional shell material . during the several hours jet - activity period the secondary will move @xmath52 near periastron @xcite , so that the total accelerated mass can be @xmath33 times larger than the mass we obtain here . + to quantify the mass distribution at different velocities we plot in figure [ fig : vel_hist ] the value of @xmath53 as function of velocity , as well as the total mass above a certain velocity @xmath54 . in panel
[ subfigure : vel_hist1 ] , we present @xmath53 for the @xmath51 case at two times as indicated , and the same at a late time for the @xmath24 case .
although reaching lower velocities than in the @xmath15 case , jets of @xmath24 can also accelerate gas to very high velocities . in panel
[ subfigure : vel_hist2 ] we show the distribution of the original jet and shell media constituents for the @xmath51 case .
the material at velocities larger than @xmath55 is predominantly shell material .
panel [ subfigure : vel_hist3 ] depicts the results for @xmath51 cases where the shell mass was either 3 times larger or 3 times lower than our fiducial case presented in the lower panel of figure [ fig : setup ] . when the density of the shell is 3 times larger than in our fiducial case , the inner boundary of the shell is denser than the jet .
we find that such an encounter does not lead to efficient acceleration to very high velocities . in panel
[ subfigure : vel_hist4 ] we plot @xmath54 for the @xmath51 case . for the @xmath55 case
the total mass ( two jets ) of gas having a velocity of @xmath56 is @xmath57 .
the photosphere during the pre-2012b outbursts is at @xmath58 @xcite , so that the hydrogen column density and density are @xmath59 and @xmath60 , respectively .
is this gas sufficient to account for the absorption observed by @xcite in the 2011 outburst of sn 2009ip ?
interestingly , these values are very similar to those of the absorbing gas in the broad absorption lines ( bal ) quasar lbqs 1206 + 1052 , where clear balmer absorption lines are detected @xcite .
further study that includes calculation of the ionization stages of various elements is required and planned for the future , but it seems that the amount of very high - velocity gas obtained in our simulations can in principle account for the observations of the 2011 outburst of sn 2009ip .
we briefly comment on two of our assumptions , that the secondary wind plays no role in the interaction , and that the radiative cooling is small .
we are confident that the wind from the secondary star plays no role during the unstable phases of the primary star , when the primary stellar envelope expands and its mass loss rate is high .
the accretion onto the secondary star is triggered by dense clumps formed in the post - shock region of the lbv dense envelope or wind , as was calculated and demonstrated for the accretion onto the companion of @xmath4 carinae near periastron passages ; neither the wind nor the radiation pressure of the accreting secondary star can prevent accretion @xcite . in @xmath4 carinae the lbv wind density , as used in the above calculations , is much lower than the wind density in the bloated envelope of the lbv primary in our model of sn2009ip , and the secondary wind and radiation pressure play an even smaller role than in present @xmath4 carinae . as well , as in the case of the winds interaction in @xmath4 carinae @xcite , the wind from the lbv star bends the wind of the companion , being an o star or a wr star , such that the wind from the secondary star does not influence at all the interaction region near the lbv star that is studied here . to explore the role of radiative cooling
we compare the photon diffusion time with expansion time . in figure
[ fig : time_evolution ] we show the evolution of the value of @xmath53 as function of velocity .
the main acceleration occurs within @xmath61h around @xmath62h .
we take the results presented in figure [ fig : results_3000 ] at time @xmath63hours . at point
@xmath64 the velocity is @xmath65 , and the density is @xmath66 .
the distance from this region outward is @xmath67 , and for electron scattering the optical depth is @xmath68 .
the photon diffusion time out of this point is @xmath69 hours , much longer than the diffusion time . at points
@xmath70 and @xmath71 the velocities are @xmath72 and @xmath73 , and the diffusion times are @xmath74h and @xmath75h , respectively .
these simple calculations show that radiative cooling is important and must be considered in future calculations . for the parameters used here radiative cooling proceeds
much faster than acceleration for @xmath76 . for gas at these velocities we overestimate the mass , but our results hold for lower velocities .
we expect that a more compact interaction region will give higher velocities
. this can be the case for narrower jets from a wr secondary star that interact closer to the launching region , as the flow time is shorter and the diffusion time is longer .
it is possible that these results hint to a wr secondary star if sn 2009ip is indeed a binary star .
there is an ongoing debate regarding the nature of the sn impostor sn 2009ip and the several outbursts that followed its 2009 eruption .
most controversial is the nature of the 2012a+2012b outbursts @xcite , with three main types of scenarios : ( @xmath77 ) single terminal explosion of the star ( either in the 2012a or 2012b outburst ) .
namely , a sn explosion which can be followed by interaction with the csm @xcite or by a quark - nova @xcite .
( @xmath78 ) non - terminal single star eruption @xcite with one outburst in 2012a followed by shell collision for the 2012b outburst , or two outbursts in 2012a and 2012b .
( @xmath79 ) binary interaction @xcite . in the binary model the lbv primary enters an unstable phase that triggers binary interaction near periastron passages @xcite .
the main energy source is the launching of two opposite jets by the secondary star as it accretes mass from the primary envelope .
as the jets are launched at @xmath80 , where @xmath81 is the escape velocity from the secondary star , one of the challenges for the binary model is to account for gas moving at @xmath82 as observed in the 2011 outburst @xcite and in the 2012a outburst @xcite . in the present study we addressed this challenge .
our flow structure that is depicted schematically in figure [ fig : setup ] has two basic assumptions .
( @xmath77 ) a binary companion launches jets in the outskirts of the primary envelope .
this is taken from the binary model @xcite .
( @xmath78 ) gas with a steep density profile resides within @xmath83 around the binary system .
the shell might actually be composed of clouds and segments .
the total mass in such a shell is @xmath84 .
we expect such an extended envelope to be formed from the unstable primary phase and the strong binary interaction .
our assumption of the existence of this shell ( extended envelope ) , will be examined in a future work .
we numerically studied the propagation of the jets through the extended envelope . in the present demonstrative paper
we simulate only two types of jets , with velocities of @xmath85 and @xmath86 , and vary only the density of the shell , not its size or profile .
the results of our simulations for initial jet velocities of @xmath15 and @xmath24 are presented in figures [ fig : results_3000 ] and [ fig : results_2000 ] , respectively . while the jet is well collimated , the interaction with the shell creates an outflowing gas with a large opening angle .
although the flow remains non - spherical , i.e. , has a bipolar structure when both jets are considered , no fine tuning of the observing angle is required in order to observe the absorption lines . after the gas expands to more than several @xmath87
, the absorption lines should be detected from most directions .
we found that for the parameters used here radiative cooling will reduce the amount of gas with velocities of @xmath88 .
jets launched by a wr companion will be narrower and denser than the jets simulated here , with shorter flow time and longer photon diffusion time .
this will allow acceleration to higher velocities .
we take this as a hint that in the binary model of sn 2009ip the secondary star should be a wr star @xcite . in section [ sec :
results ] we estimated that there is sufficient gas to explain the observation of @xcite of absorption lines at @xmath50 in the 2011 outburst .
the energy and duration of the 2012a+b outbursts are @xmath89 times those of the 2011 outburst . scaling from several hours to several days
, we deduce that our flow structure can account for @xmath90 of gas with @xmath91 in the 2012b outburst .
it is desired to explore the parameters space to find a more efficient acceleration flow structure , e.g. , taking a wr companion , such that more mass will be accelerated to very fast velocities .
this is the task of an ongoing research .
our present results could offer an explanation to the nature of the 2011 outburst , and further support the claim made by @xcite and @xcite regarding the binary nature of sn 2009ip system .
|
using hydrodynamical numerical simulations we show that high - velocity ejecta with @xmath0 in the outbursts of the supernova impostor sn 2009ip and similar luminous blue variable ( lbv ) stars can be explained by the interaction of fast jets , having @xmath1 , with a circumbinary shell ( extended envelope ) .
the density profile in the shell is very steep such that the shock wave , that is excited by the jets interaction with the shell , accelerates to high velocities as it propagates outward .
the amount of very fast ejecta is small , but sufficient to account for some absorption lines .
such an extended envelope can be formed from the binary interaction and/or the unstable phase of the lbv primary star .
the jets themselves are launched by the more compact secondary star near periastron passages .
|
in this contribution we extend the study in ref .
@xcite of an extension of the standard model in which a real scalar @xmath0 triplet with zero hypercharge is added to the usual scalar @xmath0 doublet . for details
the scalar potential of the model in terms of the usual standard model higgs , @xmath1 , and the new triplet , @xmath2 , reads @xmath3 where @xmath4 are the pauli matrices .
the expansion of the field components is @xmath5 with @xmath6 and @xmath7 is the goldstone boson which is eaten by the @xmath8 .
the model violates custodial symmetry at tree level giving a prediction of @xmath9 for the @xmath10-parameter . in the neutral higgs sector
we have two cp - even states which mix with angle @xmath11 .
the mass eigenstates @xmath12 are defined by @xmath13 in the charged higgs sector the mass eigenstates @xmath14 are @xmath15 the @xmath16 are the goldstone bosons corresponding to @xmath17 and , at tree level , the mixing angle is @xmath18 .
the precision electroweak data constrain @xmath19 to be smaller than about @xmath20 @xcite .
in ref . @xcite the beta functions for the couplings
were calculated using the one - loop effective potential @xcite with @xmath21 renormalization in t hooft - landau gauge and the anomalous dimensions for @xmath22 and @xmath23 , the results read @xmath24 in the gauge and top quark sector the beta functions for the @xmath25 , @xmath26 and yukawa couplings are the same as in the standard model and only the @xmath0 coupling is modified due to the extra higgs triplet in the adjoint representation , i.e. @xmath27 . working with the tree - level effective potential with couplings evolved using the one - loop @xmath19 and @xmath11 functions we are able to resum the leading logarithms to all orders in the effective potential . to carry out the rg analysis we first introduce the parameter @xmath28 , related to @xmath29 through @xmath30 .
we perform evolution starting at @xmath31 .
the rg equations are coupled differential equations in the set @xmath32 .
we choose rather to use the set @xmath33 . within the accuracy to which we are working
, the values of the couplings at @xmath31 can be obtained from the input set using the appropriate tree - level expressions . making use of the vacuum conditions and the notation @xmath34 and @xmath35 we can write the tree level masses as @xmath36 with @xmath37 .
inverting these relations we can fix the @xmath31 boundary conditions for the subsequent evolution . to ensure that the system remains in a local minimum we impose the condition that the squared masses should remain positive .
we also impose that the couplings remain perturbative , insisting that @xmath38 for @xmath39 and @xmath40 .
we run the evolution from @xmath41 to @xmath42 , with @xmath43 tev .
in the non - decoupling regime the triplet can not be arbitrarily heavy .
for this case in fig . 1 we show the range of higgs masses allowed when there is no mixing in the neutral higgs sector , @xmath44 , for @xmath45 .
such a value is interesting because it allows a rather heavy lightest higgs ( e.g. for @xmath46 gev and for @xmath47 gev ) @xcite .
the strong correlation between the @xmath48 and @xmath49 masses arises in order that @xmath50 remain perturbative .
the upper bound on the triplet higgs masses ( @xmath51 gev ) comes about from the perturbativity of @xmath52 whilst that on @xmath53 ( @xmath54 gev ) comes from the perturbativity of @xmath55 .
the hole at low masses is due to vacuum stability .
[ b004g0 ] the @xmath56 case has been considered in ref . @xcite and the bounds are similar to the no - mixing case . in the maximal mixing case of @xmath57 the bounds are more democratic .
the largeness of @xmath58 can be arranged either by tuning @xmath59 or by having small enough @xmath55 and @xmath60 . in the former case ,
all masses are approximately degenerate . in the latter case , which corresponds to light masses ,
the degeneracy is lifted .
the bounds for @xmath61 are very similar to those for @xmath62 on interchanging @xmath49 and @xmath53 . for @xmath63 and small @xmath11 but away from the decoupling regime
the allowed regions are very similar to those for @xmath64 . for larger @xmath11 ,
the mass bounds are again as for larger @xmath19 .
for @xmath65 there is no doublet - triplet mixing and no bound on the triplet mass .
this is a special case of the more general decoupling scenario , which occurs when @xmath66 , where the triplet decouples from the doublet . for small mixing angles ,
the triplet higgs has mass squared @xmath67 and it is possible to have @xmath68 by keeping @xmath69 large . in this case @xmath70
this is the decoupling limit in which the triplet mass lies far above the mass of the doublet and the low energy model looks identical to the standard model .
we have computed the one - loop beta functions for the scalar couplings in an extension to the standard model which contains an additional real triplet higgs . through considerations of perturbativity of the couplings and vacuum stability we have identified the allowed masses of the higgs bosons in the non - decoupling regime @xcite . in the decoupling regime
, the model tends to resemble the standard model .
the near degeneracy of the triplet higgs masses ensures that , at least for small @xmath11 , the quantum corrections to the @xmath71 parameter are negligible ( the @xmath72 parameter vanishing since the triplet has zero hypercharge ) @xcite .
this means that the lightest higgs boson can be heavy as a result of the compensation arising from the explicit tree - level violation of custodial symmetry and it is possible to be in a regime where all the higgs bosons are heavy without any dramatic deviation from the physics of the standard model .
i would like to thank jeff r. forshaw and ben e. white , my collaborators in this work , and the participants of dis 2003 , in particular , wilfried buchmueller and stefan schlenstedt for their interest in our results .
j. r. forshaw , d. a. ross , b. e. white , jhep * 0110 * ( 2001 ) 007 .
j. r. forshaw , a. sabio vera , b. e. white , to be published in the journal of high energy physics , hep - ph/0302256 . s. r. coleman and e. weinberg , phys .
d * 7 * ( 1973 ) 1888 ; r. jackiw , phys . rev .
d * 9 * ( 1974 ) 1686 ; j. iliopoulos , c. itzykson and a. martin , rev .
* 47 * ( 1975 ) 165 ; b. kastening , phys . lett .
b * 283 * ( 1992 ) 287 ; c. ford , d. r. jones , p. w. stephenson and m. b. einhorn , nucl .
b * 395 * ( 1993 ) 17 ; m. quiros , helv .
acta * 67 * ( 1994 ) 451 .
|
we study the renormalization group ( rg ) evolution of the couplings in a model with a real su(2 ) triplet in the higgs sector . insisting that the model remain valid up to 1 tev we show that it is possible for there to be no light higgs bosons without any otherwise dramatic deviation from the physics of the standard model .
|
recently properties of systems far from equilibrium have been studied by many people .
although mean field techniques may give qualitatively correct results for higher dimensions , for low - dimensional systems fluctuations have important roles .
different methods have been used to study reaction - diffusion models , including analytical and approximational methods . among them
, there are methods to obtain some quantities can be obtained exactly .
for example in @xcite , imposing some constraints on the reaction rates leads to a closed set of equations for average number densities in each site .
the empty interval method , is another method , which has been also used to analyze the one dimensional dynamics of diffusion - limited coalescence @xcite .
the most general one - dimensional reaction - diffusion model with nearest - neighbor interactions that can be solved exactly through empty interval method , has been introduced in @xcite .
empty interval method has been also generalized in @xcite .
different methods has been introduced to calculate different quantities exactly .
however , exactly solvable models are only a special class of reaction - diffusion models , and so people are motivated to use also approximate methods to understand the role played by fluctuations . in @xcite
a two species model has been considered . in this model
there are three competing reactions @xmath4 , @xmath5 , and @xmath6 .
asymptotic density decay rates of the two type of species for a special choice of parameters have been studied using the smoluchowski approximation and also field theoretic renormalization group techniques . a similar model focusing on the same diffusion rates for the two type of species has been studied in @xcite .
field theoretic renormalization group analysis suggest that contrary to the ordinary mean - field technique , the large time density of the minority species decays at the same rate as the majority ones in one - dimensional case .
although ordinary mean - field technique , generally do not give correct results for low - dimensional systems , its generalizations such as cluster mean - field may give correct results .
any how , in the mean field approximation at most one - point functions may be obtained . to obtain more - point functions one should use other methods .
one possible way is , to use a generalization of mean field known as the cluster mean field approximation .
one of the topics , which have absorbed many interests in recent years , is non - equilibrium phase transitions .
there are examples , in which mean field ( mf ) solutions are not valid , but its generalization , cluster mean field ( cmf ) gives qualitatively correct results @xcite .
a coagulation - production model is recently considered in @xcite .
although mf equations do not give correct results , cmf approximation predicts phase transition , supported also by monte carlo simulations .
steady state properties in the absorbing phase of 1*d * pair contact process model are also investigated using monte carlo simulations and the cluster approximation .
the cluster approximation qualitatively confirms the numerical results @xcite .
the scheme of the paper is as follows . in section 2 ,
the mean field equations for general parameters have been solved exactly .
it is seen that , the large time behavior of the average densities depend both on initial average densities and reaction rates , and are independent of diffusion rates . in section 3 ,
the cluster mean field equations for one- and two - point functions have been solved numerically .
it is shown that the general large time behavior is determined by the diffusion rates .
the model addressed in this article is a two - species exclusion reaction - diffusion model .
that is , each site is a vacancy ( @xmath7 ) or at most occupied by a particle @xmath8 or @xmath9 .
the interaction is between nearest sites , and the adjacent sites interact according to the following interactions with the indicated rates .
@xmath10 we consider translationally invariant initial conditions . in the mean - field approximation
, diffusion rates do not have any effect on the evolution equations of average number densities .
the mean - field equations for the average densities @xmath11 and @xmath12 are @xmath13 the large time behaviors of these equations for special choices of parameters have been studied in @xcite .
now , we want to solve these equations exactly and then we will show that there are cases which are not considered in @xcite , and give qualitatively correct result for large time behaviors , although the exponent of the decay rate is not correct . consider the following cases .
\i ) @xmath14 . the evolution equation for @xmath15 , is @xmath16 using ( [ 2],[3 ] )
, it is seen that @xmath17 which can be integrated to @xmath18 where @xmath19 and @xmath20 are the initial values of @xmath21 and @xmath22 , respectively .
now we can obtain the large time behavior of the average densities .
it is seen that the large time behavior of @xmath21 depends on the ratio @xmath23 .
0.5 cm i.1 ) @xmath24 at large times , obviously @xmath25 , so it is seen from ( [ 5 ] ) that depending on the initial value @xmath19 , two case may occur at large times @xmath26 at large times @xmath27 assuming an imbalance in the initial average densities , for example @xmath28 ( @xmath29 ) , ( [ 2 ] ) gives the large time behavior of @xmath30 , and @xmath31 as @xmath32 which means that for @xmath24 , in the mean - field approximation the minority species dies out earlier than the majority one , and the decay exponent of @xmath31 is independent of diffusion rates .
0.5 cm i.2 ) @xmath33 as a consequence of the large time behavior of @xmath22 , @xmath34 , it is seen from ( [ 5 ] ) , that at large times @xmath35 . defining @xmath36 , @xmath37 to obtain the large time behavior of @xmath22 and @xmath21 , we should use again ( [ 2 ] , [ 3 ] ) , which give @xmath38 which means that at large times both the minority and the majority species decay with the same rate .
the exponent of decay rate does not depend on the diffusion rates .
\ii ) @xmath39 .
for this case one arrives at @xmath40u\over ( \lambda + \delta u)a},\ ] ] which after integration gives , @xmath41 now , it is easy to obtain large time behavior of the average densities . generally , there are three cases , 0.5 cm ii.1 ) @xmath42 depending on the initial average densities , the large time behavior of the average densities ratio is @xmath43 ( @xmath44 ) or @xmath45 ( @xmath46 ) , which means that one kind of species decays faster
. 0.5 cm ii.2 ) @xmath47 defining @xmath48 , at large times @xmath49\over [ \lambda \lambda'-\delta^2]}$ ] . in this case two
kind of species decays with the same rate . 0.5 cm ii.3 )
@xmath50 at large times
the average densities ratio @xmath43 , and @xmath51 . 0.5 cm as it is seen , in the mf approximation only for a special choice of parameters , which is independent of diffusion rates , the two types of species decay with the same rate .
the case with @xmath52 , and @xmath53 has been considered in @xcite .
using field theoretic renormalization group analysis , it is shown that in one - dimension both type of species decay with the same rate .
monte carlo data also supports the field theory predictions in the one - dimensional model .
now , we want to use cluster mean - field approximation .
if the diffusion rate for both type of species is the same , @xmath54 cluster mean - field approximation gives the same value for the decay rates for both type of species , even if there is an imbalance in the initial average densities .
if two type of species diffuse with different rates , irrespective of the initial values , at large times particles with greater diffusion rates decay more rapidly . for the nearest - neighbor interactions , the evolution equation of @xmath55-point functions @xmath56
contains at most @xmath57-point functions .
so , generally this set of equations will be a hierarchy , which can not be solved exactly .
one way to overcome this difficulty is to impose constraints on the reaction rates that leads to disappearance of @xmath57-point functions from the evolution equation of @xmath55-point functions .
this method has been used to calculate some correlators exactly in @xcite .
another possible way is to use the cluster approximation . in the @xmath55-site cluster approximation , the set of evolution equations truncates and one encounters with a closed set of equations which may be solvable , at least numerically .
any how , for a two - site cluster approximation , a three site joint probability for a sequence of nearest - neighbor sites is approximated by @xmath58 where @xmath59 is the conditional probability . in the mean - field approximation
there are three variables , @xmath60 , @xmath61 , and @xmath62 , only two of them are generally independent . in the two - site cluster approximation , or pair approximation , the variables are @xmath63 , @xmath61 , @xmath62 , @xmath64 , @xmath65 , @xmath66 , @xmath67 , among them there are six independent variables which we choose to be @xmath68 , @xmath61 , @xmath69,@xmath70 , @xmath71 , and @xmath72 .
in fact in the pair approximation , besides the average densities the two - point functions can also be obtained .
the equation of motion for the average densities when @xmath73 are @xmath74 similar to the mean - field approximation , the diffusion rates do not appear in the evolution equations of the average densities .
but in fact they affect the average densities through the evolution equations of two - point functions , which are , @xmath75 @xmath76 @xmath77 @xmath78 to solve these equations in the cluster approximation , one should first approximate three - point functions and then all the equations should be expressed in terms of independent variables .
for example @xmath79 can be written as @xmath80 and then using probability conservation @xmath65 should be expanded , @xmath81 figure 1 and 2 show results for @xmath60 , @xmath61 , and the density ratios @xmath82 obtained using numerical solutions of equations ( [ 12]-[16 ] ) . as it is seen both types of species decay with the same rate irrespective of equality or inequality of reaction rates @xmath83 and @xmath84 . in the mf approach
, @xmath85is not an independent quantity and is @xmath86 .
but in the cmf approach it is an independent one and the numerical result obtained for it is plotted in figure 3 .
as it is seen , in the cmf approximation it approaches a constant value at large times , means that both @xmath69 and @xmath72 decay with the same rate .
equality of their decay rates is independent of equality or inequality of reaction rates @xmath83 and @xmath87 .
as mf equations are independent of diffusion rates , their solutions remain unaltered .
but in the pair approximation , only equations ( [ 13 ] ) remains unaltered .
the diffusion rate @xmath89 in ( [ 14],[15 ] ) should be changed properly to @xmath90 or @xmath91 , and the equations ( [ 16],[17 ] ) become .
these set of equations has been solved numerically , and the numerical results for the average densities has been plotted in figures 4 and 5 . as it is seen , at large times , species with greater diffusion rate dies out faster .
if species with greater diffusion rate are majority initially , there is a cross over , as it is seen from figure 4 .
the general behavior of the average densities ratio is independent of @xmath83 , @xmath87 , and @xmath94 , and the general form of large time behavior is determined by the diffusion rates .
see figure 6 .
the numerical results for @xmath95 have been summarized in figure 7 .
99 g. m. schtz ; j. stat . phys . * 79*(1995 ) 243 .
a. aghamohammadi , a. h. fatollahi , m. khorrami , & a. shariati ; phys .
rev . * e62 * ( 2000 ) 4642 .
a. aghamohammadi & m. khorrami ; j. phys .
* a34 * ( 2001 ) 7431 .
m. a. burschka , c. r. doering , & d. ben - avraham ; phys .
63 * ( 1989 ) 700 .
d. ben - avraham ; mod .
b9 * ( 1995 ) 895 .
d. ben - avraham ; in `` nonequilibrium statistical mechanics in one dimension '' , v. privman ( ed . ) , pp 29 - 50 ( cambridge university press,1997 ) .
d. ben - avraham ; phys . rev .
* 81 * ( 1998 ) 4756 .
m. alimohammadi , m. khorrami & a. aghamohammadi ; phys .
* e64 * ( 2001 ) 056116 .
m. mobilia & p. -a . bares ; phys .
* e64 * ( 2001 ) 066123 .
a. aghamohammadi , m. khorrami & m. alimohammadi ; cond - mat/0207175 .
m. howard ; cond - mat/9510053 .
z. konkoli , h. johannesson ; phys . rev .
* e62 * ( 2000 ) 3276
. g. odor ; phys .
* e63 * ( 2001 ) 067104 .
m. c. marques , m. a. santos , & j. f. f. mendes ; phys . rev .
* e65 * ( 2002 ) 016111 .
r. dickman , w. r. m. rabelo , g. odor ; phys . rev . *
e65 * ( 2002 ) 016118
. g. odor , m. c. marques , m. a. santos ; phys .
* e65 * , ( 2002 ) 056113 .
* figure captions * figure 1- average densities @xmath60 , and @xmath61 as a function of time .
the rates are @xmath96 , @xmath97 , and @xmath98 .
0.5 cm figure 2- ratio of average densities , @xmath99 , as a function of time . for the dashed line ,
the rates are @xmath96 , @xmath100 , @xmath101 , and @xmath98
|
a two species reaction - diffusion model , in which particles diffuse on a one - dimensional lattice and annihilate when meeting each other , has been investigated .
mean field equations for general choice of reaction rates have been solved exactly .
cluster mean field approximation of the model is also studied .
it is shown that , the general form of large time behavior of one- and two - point functions of the number operators , are determined by the diffusion rates of the two type of species , and is independent of annihilation rates .
* cluster approximation solution of a two species annihilation model * 2 cm f. tabatabaee@xmath0 , a. aghamohammadi@xmath1 5 mm _ @xmath2 department of physics , alzahra university , tehran 19834 , iran .
_ _ @xmath3 institute for applied physics , tehran 15857 - 5878 iran . _
* pacs numbers : * * keywords : * reaction - diffusion , cluster mean field
|
the recent use of three - dimensional doppler tomography to investigate the algol binaries u crb and rs vul has unveiled a significant amount of detail regarding the accretion process in these systems . as well as confirming the presence of a number of features seen in two - dimensional tomographic data , the new data have shown an unexpectedly large amount of gas located outside of the orbital plane in both systems .
u crb also displays a transient accretion disc inclined with respect to the orbital plane , which is not expected if mass transfer occurs as a result of gravitational and centrifugal forces alone .
the companions in short - period algols are rapidly rotating late - type stars that are likely to be magnetically active .
magnetic behavior occuring at or near the first lagrange point ( l1 ) may offer an explanation for the out - of - plane material .
the notion that large - scale magnetic effects may drastically influence the morphology and time - evolution of the accretion processes is evidenced by the observations of @xcite , who noted the similarity in the out - of - plane velocities located at l1 and magnetically active regions on the donor star in rs vul .
moreover , they found evidence of a loop prominence on the cool star close to the l1 point in the 3d tomogram of rs vul , which suggests that the magnetic field may have deflected the gas stream relative to the central plane .
such a deflection would cause gas to be introduced into the system with additional ( non - z ) angular momentum that could then propagate toward the primary and persist as a tilted disc .
this deflection could be attributed to a variety of mechanisms .
starspots on the donor in the prototype algol system @xmath4 persei were observed by @xcite , and coronal ejections in eight other algol systems , including rs vul , have been detected by @xcite .
more recently , a large - scale persistent coronal loop in algol was fully resolved by @xcite , suggesting the possibility of an extended magnetic field structure between the two stars .
while the likelihood of magnetic activity in u crb and rs vul is high , the lack of information presently available regarding the exact magnetic structure of these systems makes it difficult to determine the degree to which magnetic effects are responsible for the transport of gas out of the orbital plane .
further complicating the mass transfer process in algols is the fact that their short orbital period ( @xmath5 days ) combined with the close orbit and relatively large radius of the primary results in the accretion stream directly impacting or grazing the surface of the accreting star .
direct - impact systems display a permanent or transient keplerian disc and sometimes a partial disc or localized region characterized by an asymmetric mass distribution and sub - keplerian velocities @xcite . in these systems ,
the localized region is associated with gas that has circled the accreting star and has slowed down because of the interaction with the incoming gas stream @xcite .
gas is expected to leave the orbital plane due to interactions at the site where the stream impacts the surface of the primary , the stream - disc interaction , and the localized region @xcite .
both u crb and rs vul are short - period binaries , and both display a number of h@xmath2 emission features characteristic of direct - impact systems .
u crb has been classified as an alternating system , since it has been observed in both a streamlike accretion state and a disclike state @xcite .
the stream state is characterized by strong emission along the predicted ballistic trajectory of the accretion stream against a lack of circumprimary emission . in the disc mode ,
emission from the accretion disc is the most prominent feature .
the mechanism responsible for the transition between these two states has yet to be explained , although hydrodynamic simulations of @xmath4 per showed that the morphology of the accretion flow depended strongly on the density of the inflowing gas .
the hydrodynamic simulations of @xcite also measured a large relative increase in the h@xmath2 emissivity of the stream relative to the disc when stream density was varied .
the similarity of the stellar components and orbital parameters of rs vul to those of u crb ( rs vul = b5v + g1 iii - iv , p=4.48 days ; u crb = b6v + g0 iii - iv , p=3.45 days ) makes it likely that a disc mode is possible in rs vul , and it has been classified as an alternating algol by @xcite despite having not yet been observed in a disc mode .
three - dimensional tomographic imaging of u crb has allowed for a more thorough analysis of the accretion features by enabling the creation of 3d velocity space plots of the h@xmath2 emission sources . mapping this data to cartesian coordinates
is not possible due to the inability to resolve the exact cartesian structure of the system , but the theoretical analysis of key features of the system provides reference points from which the gas dynamics of the system can be inferred . features commonly used as landmarks include the motion of the stellar components , the ballistic trajectory of the accretion stream , and the locus of the accretion disc as defined by the keplerian velocities at the surface of the primary and its outermost roche surface .
the first 3d tomographic analysis of u crb was performed by , and showed the system in a disc mode .
an additional series of h@xmath2 profiles were later analyzed in three dimensions and displayed a stream mode @xcite .
notable features of these two modes include : ( 1 ) a gas stream transferring mass from the donor to the primary star with velocities as high as 450 km s@xmath6 in the orbital plane and negligible z - velocities .
( 2 ) circumprimary emission associated with an accretion disc that displays both positive and negative z - velocities ( @xmath3 = @xmath7 km s@xmath6 ) .
the different signs of the z - velocities on opposite sides of the primary star suggest that the disc may be inclined with respect to the orbital plane .
( 3 ) a stream - disc interaction region , or `` hotspot '' ( @xmath3 = + 180 km s@xmath6 to + 540 km s@xmath6 ) , as well as a star - stream impact site ( @xmath3 = -300 km s@xmath6 to + 60 km s@xmath6 ) . ( 4 ) a localized region ( @xmath3 = + 200 km s@xmath6 to + 540 km s@xmath6 ) .
in contrast to the two accretion modes displayed by u crb , rs vul has only been observed in a stream state with a much stronger out - of - plane flow .
a three - dimensional tomographic analysis of rs vul was performed by @xcite , and displays many of the same features as u crb .
major differences from u crb include : a truncated gas stream ( @xmath8 km s@xmath6 ) with no indication of a star - stream hotspot , asymmetric circumprimary emission with no well - defined disc , a localized region diametrically opposite the circumprimary emission with opposite z - velocities ( suggesting that the gas orbiting the accreting star is flowing along a tilted path with @xmath9 to @xmath10 km s@xmath6 ) , and emission from a magnetically active region on the donor that suggests the presence of a loop prominence ( @xmath11 to @xmath12 km s@xmath6 ) .
the fact that the h@xmath2 emission at l1 possesses z - velocities equal to those of magnetically active regions on the donor strongly suggests that the stream is magnetically deflected at or near l1 .
the relatively high velocities associated with the stream also suggest that the stream velocity may be boosted to many times the speed that would occur via roche lobe overflow alone .
our goal was to determine whether a deflected stream is responsible for introducing a sufficient amount of tilt to the accretion disc , and to what extent the deflection influences the presence or absence of other out - of - plane accretion features . to investigate this we performed hydrodynamic simulations of mass transfer in rs vul in which magnetic effects were mimicked by manually deflecting and boosting the stream at l1 .
the resulting steady - state solutions were analyzed to determine the effect of the deflection on the morphology of the circumstellar material , the presence of out - of - plane flow , and the presence or absence of the accretion features noted in the tomographic data .
the hydrodynamic model we used is described in section 2 , results from runs with undeflected and deflected streams are given in section 3 , an analysis of the tilt of the disc with comparisons to analytical estimates is given in section 4 , and our results summarized in section 5 .
our simulations were performed with the three - dimensional hydrodynamics code vh-1 , which was developed by the numerical astrophysics group at the virginia institute for theoretical astrophysics .
vh-1 solves the euler equations for the flow of an inviscid fluid characterized by an adiabatic index @xmath13 . to solve the system
, vh-1 utilizes operator splitting to reduce the grid into a number of 1-d segments .
the fluid equations are solved along each segment using the piecewise parabolic method , an explicit high - order godunov method employed on an eulerian ( fixed ) grid @xcite .
the ppm can be used for both 2d and 3d geometries without necessitating significant changes to the algorithm other than the extension from 2d polar to 3d spherical coordinates .
we adapted vh-1 to model our system in spherical coordinates on a corotating grid located in the reference frame of the binary system and centered upon the primary star .
the inner boundary of the grid coincides with the surface of the primary at a radius of @xmath14 , and the outer boundary extends to twice the distance from the centre of the primary to l1 ( @xmath15 ) . since the system does not vary widely in scale between the two stars , we space our zones equally in the radial direction .
the parameters of u crb and rs vul are given in table 1 , and were taken from @xcite and @xcite .
we note that while we use the parameters corresponding to rs vul in our simulations , both the orbital parameters and stellar components of u crb are similar to those of rs vul .
this enables us to interpret our results in the context of either system . to avoid grid convergence and the coordinate singularities intrinsic to spherical coordinates ,
we employ a yin - yang overset grid @xcite .
the yin - yang grid consists of two separate spherical polar grids oriented 90@xmath16 with respect to each other , so that the polar region of one grid aligns with the equatorial region of the other .
each grid extends @xmath17 azimuthally , and @xmath18 in the polar direction ( figure [ fig : yinyang ] ) .
this grid configuration allows for reasonable limits on the courant number and ensures that our runtimes do not become excessive . as a result , we are able to extend our outer boundary to a sufficiently large radius without imposing limits on the the azimuthal extent of the grid .
we define our coordinates such that the positive x - axis of the yin grid lies along the binary axis and points toward the secondary .
the positive y - axis points in the direction of motion of the secondary , and the z - axis is perpendicular to the orbital plane .
each piece of the grid has dimensions of @xmath19 , which results in an angular resolution of @xmath20 and @xmath21 = 0.018 at the inner boundary .
since the inner boundary corresponds to the surface of the accreting star , we must choose a boundary condition that allows gas to accrete while still adequately modeling the effects of the stellar atmosphere .
the fact that the stream penetrates only a small portion of the stellar atmosphere ( * ? ? ? * @xmath22 for algol , ) allows us to approximate the surface of the primary as a solid surface .
interactions between the atmosphere and the stream could have a non - negligible effect on the structure of the circumstellar gas ( especially in the case of a rotating and precessing primary ) , but for the purposes of this paper we neglect the structure of the atmosphere .
instead , we employ a partially reflecting boundary condition that sets the radial velocity of the gas at the innermost radial zone to zero every timestep .
this approximates the gas encountering a shock at the surface of the star and accreting shortly afterward . at the outer boundary
, gas should be allowed to freely exit the grid , so we implement a modified zero - gradient boundary condition with the added requirement that gas may not be allowed to flow inward once it has passed off the grid . [ cols="<,<,<",options="header " , ] the 2d simulations of algol by @xcite found that the morphology of the circumprimary flow depended strongly on the initial stream density due to the effects of radiative cooling .
the stream density s influence on the flow is seen primarily in the stream - disc interaction .
a denser stream will cool more rapidly at the shock , negating the effects of heating within the shock .
this results in a flow that behaves similarly to a compressible isothermal gas .
a lower - density stream will not cool to the same extent , and will behave adiabatically . to investigate the relative effect of cooling in u crb and rs vul
, we compare the cooling time at the shock to a characteristic dynamical time .
the dynamical time can be found using the gas speed at the outer edge of the disc where the stream - disc shock is located . assuming a keplerian velocity at the outer edge of the disc ( @xmath23 km s@xmath24 ) allows us to calculate a dynamical time of : @xmath25 in both u crb and rs vul , the dynamic timescale is calculated to be @xmath26 .
if a cooling function is known , we can calculate the characteristic cooling time at the shock to be @xmath27 here , @xmath28 is the electron number density in the post - shock flow , and @xmath29 is the cooling function evaluated at the temperature of the shocked gas , @xmath30 .
we calculate @xmath31 with the cooling function for a low - density plasma as given by @xcite . from the rankine - hugoniot jump conditions
, we can calculate the post - shock temperature to be @xmath32 , where @xmath33 is the average mass per particle .
we assume a gas comprised largely of ionized hydrogen , in which case @xmath34 .
we assume the velocity of the shocked gas , @xmath35 to be on the order of the keplerian velocity at the shock site . for both u crb and rs vul , we calculate @xmath36 to be approximately @xmath37 k , which results in a cooling coefficient of @xmath38 ergs @xmath39 s@xmath6 .
an analytical estimate of the density at the shock site is difficult to obtain , so we note that our simulations show densities in the shock region on the order of @xmath40 . converting this density to physical units
requires a measurement of the mass transfer rate .
since data on the exact mass transfer rates for u crb and rs vul are not available , we use the lower limit of the transfer rate in algol , @xmath41 yr@xmath6 @xcite .
comparing the characteristic dynamical time to the cooling time , we find that @xmath42 .
this is a conservative estimate , as we have used a small value of the mass transfer rate to calculate the cooling time .
any increase in the mass transfer rate will serve to decrease the cooling time further , since @xmath43 varies as @xmath44 .
in general , the mass transfer rates in short - period algols vary widely between systems , and are not likely to remain constant in time . for algol
, @xmath45 could range from @xmath46 to @xmath47 yr@xmath6 @xcite . with these factors taken into account ,
it is likely that cooling at the shock is sufficiently rapid to counter the heating that occurs in the stream - disc interaction .
this rapid cooling ensures that the shock - heated gas will quickly fall to some minimum temperature below @xmath48 k , at which point the recombination of h will diminish the cooling rate .
the cooled gas will be heated by the strong photoionizing uv flux from the primary , which is sufficient to maintain the temperature at a value close to @xmath48 k. taking this information into account , we assume a constant temperature of @xmath49k in our simulations .
this value falls midway between the effective temperatures of the primary and secondary in both u crb and rs vul .
this allows us to expedite our calculations by employing an isothermal equation of state ( @xmath50 ) .
* hereafter ls ) conducted an extensive analysis of the gas dynamics at the l1 point , and we use their prescription to determine the initial conditions of the accretion stream .
ls parametrized the flow in terms of the ratio of isothermal sound speed to the system s orbital speed : @xmath51 , for which we obtain @xmath52 .
this gives a local sound speed of @xmath53 km s@xmath6 at l1 .
ls calculate that the stream will be accelerated to supersonic speeds as it travels through l1 , so we assume an unboosted speed of mach 1 .
the angle the stream makes with the binary axis can also be calculated using @xmath54 , and for rs vul we obtain @xmath55 .
assuming an ideal gas , the pressure at l1 should be initialized to a value on the order of @xmath56 .
for this work we require control over the direction and magnitude of the gas velocity at l1 in order to manually deflect the stream .
to achieve this , we calculate the position of l1 on the grid and hold the stream parameters in the corresponding zones fixed at the values given above .
this eliminates the need for the donor star to be modeled hydrodynamically , and we remove it from the grid while retaining its term in the gravitational potential . a small amount of flow into the region corresponding to the secondary occurs , but it remains confined to the volume that the secondary would normally occupy and does not affect the behavior of the accretion flow .
in addition to giving us more control over the stream at l1 , this strategy has the benefit of allowing for a more economic grid size , since the full secondary need not be present on the grid .
ls calculate the width of the stream at l1 to be on the order of @xmath57 , which requires an angular resolution of at least @xmath58 .
the angular resolution of our grid is small enough to accomodate this with two zones ; each has an angular extent of @xmath59 .
symmetry requires the gas stream at l1 to extend equal distances above and below the orbital plane as well as an equal distance to either side of the binary axis .
this necessitates the use of four zones to represent l1 . in reality
the density profile of the gas at l1 is gaussian , but we lack the resolution to resolve this
. instead we hold the density of the gas fixed to a constant at our l1 zones .
ultimately , our system exhibits behavior similar to that of @xcite in that the morphology of the accretion flow is relatively insensitive to the geometry of the zones at l1 .
our simulations do not assume any gas to be initially present around the primary , and the system requires many orbital periods to allow the full formation of a disc .
a steady - state solution in which a disc is present and has stopped gaining mass ( indicated by @xmath60 ) requires @xmath61 p@xmath62 according to preliminary low - resolution runs . to expedite the disc formation
, we multiply the density in the disc by a constant factor after two orbital periods have elapsed .
since the mass accretion rate is proportional to the density at the inner boundary , this results in the mass accretion rate being boosted to a value closer to that at which it stabilizes .
to ensure that our model behaved appropriately in the absence of a deflected stream , we performed a fiducial run with the stream left unmodified ( @xmath63 , @xmath64 ) .
figure [ fig : notilt_topside ] shows the stream following the ballistic trajectory as it falls toward the primary and narrowing as it approaches the surface .
the stream impacts the surface after approximately @xmath65 .
we allow gas to accrete until @xmath66 , during which a thin accretion disc begins to form . at @xmath66 , we boost the disc density .
the boost instigates some radial density oscillations due to increased pressure along the density discontinuity that occurs at the radius of the boosted region .
these oscillations propagate with the flow around the disc , but fade within the next few orbital periods . following the density boost , the radius of the disc continues to increase for several orbital periods , and a steady state is achieved at roughly @xmath67 .
when the system has stabilized , we measure the disc to have an outer radius of @xmath68 ( where @xmath69 is the binary separation as measured from the center of the accreting star ) . examining the density profile averaged over @xmath70 ( figure [ fig : density_v_radius ] ) shows that the density of the innermost portion of the disc ( @xmath71 ) is of the same order of magnitude as the density at l1 ( @xmath72 ) .
the density decreases sharply to just above @xmath73 at the outer edge of the disc .
the disc is symmetric about the orbital plane to approximately @xmath74 above and below the plane . beyond this region
the density decreases rapidly , and is measured to be less than @xmath75 .
the velocities within the disc are oriented largely within the plane .
as the gas accelerates from l1 , its speed increases from @xmath76 km s@xmath6 to a final velocity of @xmath77 km s@xmath6 before encountering the stream - disc shock .
( all velocities are measured with respect to a corotating frame . ) after crossing the shock and entering the higher - density regions of the disc , the gas assumes velocities which fall within the range of @xmath78 km s@xmath6 , with small velocities corresponding to the outer edge of the disc , and higher velocities located near the surface of the accretor .
the velocities within the disc remain relatively constant in direction and are sub - keplerian throughout ( @xmath79 percent at the outer and inner edges of the disc , respectively ) .
these outer velocities are less uniform near the localized region and stream - disc interaction site , and reach values up to @xmath80 km s@xmath6 in the vicinity of the stream - disc shock . within the orbital plane at @xmath81 .
densities are normalized to 1 at l1 .
the solid line depicts the density for the undeflected stream , while the dashed line corresponds to a stream tilted to @xmath0 and boosted to mach 30 .
l1 is located at @xmath82 , and the surface of the primary at @xmath83.,width=317 ] to quantify the angle with which the disc is inclined , we integrate the angular momentum present in the disc and measure its angle with respect to the z - axis .
we limit our integration to the region occupied by the high - density portions of the disc , using figure [ fig : density_v_radius ] as a guideline .
the result is summed angular momentum values for each value of @xmath84 in the orbital plane .
each value of @xmath85 defines a ring within the disc for which we are able to calculate the tilt .
the sign of the tilt is set equal to the sign of the x - component of the net angular momentum .
this has the effect of assigning positive tilt angles to rings that are tilted toward the donor , and negative angles for rings that are tilted away . for the undeflected stream ,
the disc is largely symmetric about the orbital plane and remains essentially untilted over time .
examining the tilt as a function of radius at @xmath81 ( figure [ fig : tilt_v_radius ] ) shows that the high - density portions of the disc from the inner boundary to @xmath86 only possess tilt on the order of @xmath87 .
this is a fraction of the angular resolution of by one grid zone .
the tilt varies gradually as the radius increases , and the sign changes three times along the radius , indicating that some warping occurs about the orbital plane .
the tilt becomes slightly larger in the low - density portions of the disc and reaches a maximum value of @xmath88 .
this tilt is smaller than the amount able to be resolved with one zone .
the presence of this tilt is likely a result of a comparatively large amount of non - z angular momentum created in the stream - disc interaction through the splashing of low - density gas .
turbulence in the gas flowing along the outer low - density edge of the disc may also contribute to this measurement . by summing the angular momentum within the disc over @xmath70 for each value of @xmath85 .
the solid line depicts tilt for the undeflected stream , while the dashed line corresponds to a stream tilted to @xmath0 and boosted to mach 30.,width=317 ] the disc s orientation with respect to the orbital plane can be found by examining a plot of the x and y components of the angular momentum . in cartesian space ,
the projection of the angular momentum vector into the orbital plane points toward the direction that the disc is azimuthally inclined .
( this is analogous to treating the disc as a rigid body with a rod through its axis of rotation .
the projection of the angular momentum vector into the orbital plane is the direction you would push the rod to tilt the disc . )
figure [ fig : orientation ] shows a @xmath89 vs. @xmath90 plot for both runs , with @xmath90 and @xmath89 components shown as a fraction of the total angular momentum .
the data points for the undeflected stream lie clustered around the origin , with magnitudes of less than @xmath91 percent of the total angular momentum .
this supports the qualitative observation that the disc lies almost entirely in the orbital plane throughout the length of the simulation .
the disc appears to precess at a period of @xmath92 , which is a by - product of measurements of the angular momentum taken from within a co - rotating reference frame . .
the undeflected stream results in an disc that remains largely in the orbital plane , with minor variations over time that remain symmetric about the origin .
the deflected and boosted stream produces a disc biased toward @xmath93 due to the stream impacting the far side of the disc from below and tilting it toward the donor star.,width=317 ] to simulate a system with a deflected and a boosted velocity , we modified the stream parameters at the zones corresponding to l1 by directing the stream velocity @xmath0 below the orbital plane and boosting it to thirty times the local sound speed ; this is comparable to gas flows within magnetic loops . these values were chosen to maximize the likelihood that a tilted disc would form . in the early stages of evolution ( figure [ fig : tilt_topside ] ) , the stream falls along the predicted ballistic trajectory , which extends below the orbital plane to just below the bottom of the accreting star .
as it nears the primary , it is pulled back toward the orbital plane and wraps around the accretor .
the boosted velocity is large enough to cause the stream to miss the surface of the star , resulting in the absence of a star - stream collision and a gas stream that extends above the orbital plane to nearly @xmath94 .
the stream then circles the accretor and interferes with the gas stream from the secondary as an accretion disc begins to form near the orbital plane .
the flow during the first few orbital periods displays a large degree of out - of - plane flow as the system approaches a steady state . at @xmath66 the density within the disc
is boosted , inducing some minor radial oscillations as in the undeflected case .
as the disc gains mass , the stream - disc interaction continues to drive gas upward above the orbital plane .
the system stabilizes over the next @xmath67 , after which the accretion disc remains present along with a large plume of gas deflected above the orbital plane by the stream - disc interaction .
investigating the morphology of the accretion disc shows some radial enlargement and warping due to the deflected stream .
averaging the density in the central plane over @xmath70 ( figure [ fig : density_v_radius ] ) shows that the density falls off more slowly than in the undeflected case , reaching @xmath95 at @xmath96 .
the inner density of the disc remains on the same order of magnitude as in the undeflected case , and is slightly larger in radius .
the top view of figure [ fig : tilt_topside ] at @xmath81 shows that the disc is also asymmetric , with a slightly enlarged portion in the upper right portion of the disc .
this large density region originates as the disc forms and propagates around the disc with time .
it does not appear to originate due to the artificial density boost that occurs at @xmath66 .
the vertical extent of the disc is similar to that of the undeflected case , neglecting slight enlargement above and below the orbital plane due to warping that occurs in the outer edges of the disc .
the majority of the disc still lies in the orbital plane , but with lower density gas ( @xmath97 ) deflected above the plane due to the stream - disc interaction . this gas is pulled back into the plane , where further dissipation of angular momentum due to interactions with the disc causes a portion of the material to join the disc and eventually accrete .
the deflected stream begins at a speed of @xmath98 km s@xmath6 ( mach 30 ) .
as the circumprimary material begins to build up into a disc , gas is blown upward in the collision with z - velocities of up to @xmath99 km s@xmath6 . within the disc
we observe a circular orbit with velocities that remain sub - keplerian in the same range as the undeflected case ( @xmath78 km s@xmath6 for the high - density portion , @xmath100 km s@xmath6 for the low - density portion ) .
we also measure z - velocities on the order of @xmath101 10 km s@xmath6 due to the warping and tilting in the disc .
examining the tilt within the disc as a function of radius after @xmath81 has elapsed ( figure [ fig : tilt_v_radius ] ) shows that the warping within the disc is more pronounced than in the undeflected stream .
the inner portion of the disc is tilted at @xmath102 near the the surface of the primary , and increases to @xmath103 at @xmath104 .
warping is indicated by the variations in tilt angle with increasing radius .
the tilt continues to increase at radii greater than @xmath104 , mostly due to the high z - velocities in the material blown upward in the stream - disc interaction .
the tilt angle calculated for this low - density gas reaches values as high as @xmath105 at the outer edge of the disc .
all the tilt measurements for this timestep result in positive values , indicating that the disc is tilted consistently in the @xmath106 direction ( toward the secondary ) .
the orientation of the disc changes noticeably over time when we deflect and boost the stream .
the plot of @xmath89 vs. @xmath90 ( figure [ fig : orientation ] ) shows a much wider spread of values than that of the undeflected stream .
the fraction of angular momentum that is distributed in the @xmath107 and @xmath108 components remains between 2 and 3 percent at all times , and grows to 5 percent when the disc is oriented toward the donor star .
when this occurs , the effects of the stream s impact on the disc are enhanced due to the upward force of the stream on the already upwardly - moving gas in the disc . as in the undeflected case ,
the apparent precession is due to measurements from within the corotating reference frame .
additional information regarding the out - of - plane velocities and accretion features are more easily visible if we mimic the tomographic results with a pseudo - emissivity plot in velocity space .
this was first done for algol systems by @xcite . in our simulations , emissivity scales as @xmath109 and lacks any dependence on the temperature or ionization state of the gas .
we slice our velocity space plots into slabs along @xmath3 , sum the emissivity along @xmath3 within each slab , and scale the emissivity to the maximum value observed in the slab .
this maximum value is also listed in terms of the maximum emissivity on the entire grid .
the velocities measured in our hydrodynamic simulations take place in a corotating frame , and we transform these into an inertial frame for the sake of comparison with tomographic data . since we use the inner boundary to represent the primary and we have removed the secondary from the grid , emission features associated with stellar activity occuring on the surface of either star ( e.g. circumprimary emission , chromospheric emission from the secondary ) are absent from the plot . as a result , the plots appear more sparse than the original tomographic data presented in @xcite and @xcite .
we also note that the strong density dependence causes the emissivity in the central plane to appear many orders of magnitude greater than that of other slices .
the emissivity plots for the undeflected stream ( figure [ fig : halpha_nt ] ) clearly show the presence of the disc in the central slice ( @xmath110 to @xmath111 km s@xmath6 ) .
since the inner portion of the disc circles the primary with a higher velocity than the outer portion , the disc is turned inside - out on the velocity space plot .
the bulk of the disc possesses velocities that fall within the keplerian predictions ( 400 - 515 km s@xmath6 ) , although we also measure sub - keplerian velocities as small as 370 km s@xmath6 .
the accretion stream is also visible in the central slice , beginning near @xmath112 km s@xmath6 and extending further in the -x direction as it gains speed and falls along the binary axis .
the slightly stronger emissivity region located along the inside region of the``ring '' at roughly @xmath113 = ( @xmath114 ) km s@xmath6 corresponds to the stream - disc interaction .
emission from this feature extends up to @xmath115 km s@xmath6 .
we also note that the emissivity is slightly enhanced near ( @xmath116 ) km s@xmath6 due to the overlap between the yin and yang grids .
some emission is present in the slab corresponding to @xmath3 = 30 to 90 km s@xmath6 , and is located primarily within the locus of the accretion disc .
the intensities of this emission are seven orders of magnitude below the maximum intensity on the grid , and distinct emission features are not immediately apparent in this slice . for gas wtih
downward velocities of @xmath3 = @xmath117 to @xmath118 km s@xmath6 , some emission is present in the locus of the disc again , though at intensities ten orders of magnitude below the maximum grid intensity . for velocities greater than @xmath3 = @xmath11990 km s@xmath6
, the emission lacks any distinct features and occurs at intensities on the order of @xmath120 of the maximum intensity .
the emissivity plots of the tilted disc ( figure [ fig : halpha_tilt ] ) show much higher emissivity measurements in slices above and below @xmath121 .
the disc is again the dominant feature in the central slice , and is larger in radius than in the undeflected case , as expected from the density profiles .
this results in an emissivity region that extends inward to a slightly greater extent .
the majority of the velocities lie within the keplerian predictions ( @xmath122 km s@xmath6 ) , with some gas possessing velocities as low as 350 km s@xmath6 .
since the stream has been deflected below the orbital plane , it appears in the upper left quadrant of the @xmath123 slices .
the stream - disc interaction is visible in the slices corresponding to positive z - velocities , and appears most prominently near @xmath113 = ( @xmath124 ) km s@xmath6 in the slice of @xmath125 to @xmath126 km s@xmath6 .
the high @xmath3 material on the far side of the disc is also apparent in the emissivity plots .
as it rises and wraps around the secondary ( with @xmath127 ) , it can be seen in the lower left quadrant .
as it falls back towards the orbital plane ( @xmath123 ) , it shifts toward the lower right quadrant , indicating motion back toward l1 .
the gas collides with the orbital plane again , and a portion is driven below the orbital plane once again .
this is visible in the @xmath127 slices as a feature in the upper right quadrant . as a whole
, the emissivity for the deflected and boosted stream reveals values above and below @xmath121 with intensities much closer to the values measured in the central slice .
this indicates a much stronger out - of - plane flow produced by the deflected and boosted stream .
it should also be noted that the maximum intensity on the grid is nearly five times larger for the deflected and boosted stream than for the undeflected stream .
this is to be expected , since the densities within the disc are slightly larger when the stream was deflected , and any changes in density will appear in the emissivity as @xmath128 . examining the emissivity in the early stages of our run ( @xmath66 , just before the density is boosted ) shows the intensity more uniformly distributed throughout the @xmath3 slices .
no notable differences in the structure of the emissivity features are seen .
the main features visible on our emissivity plots for both the undeflected and deflected streams are the disc emission , accretion stream , and stream - disc interaction . in the undeflected case
we also measure non - localized `` clouds '' with z - velocities up to @xmath129 km s@xmath6 , but with magnitudes smaller than the emission features located in the central plane by a minimum of eight orders of magnitude .
for the deflected stream , the main sources of emissivity are the disc and the out - of - plane gas located on the far side of the secondary ( extending from @xmath130 km s@xmath6 to @xmath131 km s@xmath6 ) .
the disc structure is similar to that of the undeflected case , but displays some radial enlargement and slight warping .
the out - of - plane gas appears with much stronger intensities relative to the gas in the orbital plane , and measures @xmath132 , as opposed to @xmath133 .
the early stages of our deflected stream simulation are comparable to the tomographic data of rs vul @xcite . the emission seen in the tomographic data of rs vul
is not confined to the keplerian locus and is found primarily along the stream , near the mass donor and accretor , and in several other localized high - velocity regions . in our simulations ,
prior to the formation of the accretion disc ( i.e. before the density boost occurs at @xmath134 ) we observe a similar structure .
the stream is visible in slices with negative velocities as low as @xmath135 km s@xmath6 , and the high - velocity features corresponding to the stream material wrapping around the primary are visible from @xmath11 km s@xmath6 to @xmath136 km s@xmath6 . as in the tomographic data ,
the star - stream interaction is not visible , since the deflected stream misses the star entirely .
emission from the keplerian locus is non - uniform and occurs at intensities lower than those corresponding to the stream by an order of magnitude . despite similarities with the morphology of our simulations ,
the tomographic data from rs vul show a more feature - rich display , with out - of - plane velocities as high as @xmath137 km s@xmath6 .
these are much larger than the @xmath129 km s@xmath6 maximum we measure in our emissivity plots , but the magnitude of the tomogram velocities may be somewhat exaggerated due to the data s relatively low resolution in the z - direction .
another important difference from our model is that the intensities of the emission features in the tomograms remain on the order of those in the central plane even at very high values of @xmath3 .
the primary explanation for our lower intensity measurements is that the emission of low - density regions is suppressed by our use of an isothermal gas .
since the emissivity calculation depends strictly on density in the isothermal case , the magnitude of the emission in the simulation data appears smaller than it otherwise would in regions containing hot low - density gas .
since we have neglected the influence on the ionization state and temperature of the gas in our emissivity plots , we expect our intensities to fall drastically in the lower - density portion of the flow . in the early stages of the run , this results in the maximum intensities occurring at the location of the stream .
the intensity elsewhere is smaller by a factor of @xmath138 .
the late stages of our simulation display the accretion disc in a steady - state with a morphology more akin to the tomographic data of u crb in its disc mode @xcite .
this tomographic data shows more concentrated emission in the keplerian locus than the tomograms of rs vul , as well as both a star - stream and stream - disc interaction region .
our late - stage emissivity plots are also dominated by emission from the keplerian locus due to the presence of a disc , but the material driven away from the orbital plane in the stream - disc interaction remains visible .
due to the strong density dependence of the emissivity , the disc dominates the emissivity plots .
intensities elsewhere range from @xmath139 below the maximum measurement on the central slice .
our measurements of the tilt angle of the disc show that deflecting and boosting the stream introduces warping of up to @xmath1 , but no significant tilt with respect to the orbital plane . alternately , the lower - density gas blown upward above the orbital plane does display significantly larger tilt that measures up to @xmath105 at the outer edge of the disc . obtaining a tilt measurement from the tomographic data for comparison
is difficult , particularly for rs vul , which does not display a well - defined accretion disc . in u crb , z - velocities of @xmath7 km s@xmath6
were observed within the predicted locus of the keplerian accretion disc .
a tilted disc was inferred from the observation that emission in tomogram slices corresponding to negative z - velocities was stronger on the left side of the @xmath140 velocity - space plots , while positive z - velocity slices showed stronger emission in the lower - right side of the @xmath140 plot . quantifying this tilt by assuming a z - velocity of @xmath141 km s@xmath6 and keplerian velocities at the surface of the primary gives a tilt angle of @xmath142 . performing the same calculation at the outer keplerian orbit produces a tilt of @xmath143 , which is comparable to the tilt measured in our simulations for the plume of gas on the far side of the accretor .
it should also be noted that the @xmath3 values are uncorrected for the lower resolution in the @xmath3 direction . @xcite
claim the correction factor will be small , but any correction factor will serve to reduce the predicted tilt . assuming a factor as large as @xmath144 would reduce the maximum z - velocities to @xmath145 km s@xmath6 , which is close to the maximum @xmath3 we measure .
this correction results in tilts of @xmath146 and @xmath147 at the inner and outer radii .
our simulations indicate that it is unlikely that the disc will be uniformly tilted , since the stream - disc interaction acts primarily on the outer edge of the disc .
the tomographic data that depicts asymmetric intensities measured on opposite sides of the accreting star could be interpreted as warping in the disc , as opposed to a uniformly inclined disc .
a large part of the out - of - plane flow in our simulations also occurs within the keplerian locus , and could contribute to this measurement .
the primary difference between the deflected and undeflected stream in our simulations is that the modified stream enables large amounts of flow above and below the orbital plane .
boosting the stream allows gas to circle the primary without striking its surface , which in turn means the system retains angular momentum that would normally be destroyed in the stream - disc and star - stream collisions . the additional ( non - z ) angular momentum manifests in the out - of - plane flow , and is eventually dissipated in the accretion disc .
the disc s dimensions appear larger and the densities slightly higher due to the absence of the star - stream interaction as an angular momentum dissipation mechanism .
the bulk of the disc remains primarily within the orbital plane , although it is more dense and more uniform than the disc shown in the tomographic measurements of u crb .
we can reconcile the differences between the simulated and observed discs by noting that our isothermal approximation results in a gas that is more easily compressible .
this will lead to a more compact accretion disc and the tendency for less gas to be blown away in the interaction regions .
it is also possibile that the tomographic images of u crb in a disc mode show the early stages of the disc , in which case the relative intensity of the disc would be lower , and the disc would have a more nonuniform structure .
we see no evidence of a transition between a stream and disc accretion mode in our simulations .
this is likely a result of our use of a constant mass transfer rate through l1 , and potentially a byproduct of our isothermal equation of state .
examining our emissivity plots at @xmath148 shows a morphology very similar to that at @xmath81 , but with the central slice intensity @xmath149 orders of magnitude higher than that on the remainder of the grid . while the disc is present , it is not fully formed , and appears less uniform than late in the run . a reduced rate of mass transfer could offer the disc material a chance to accrete without being replenished by material from the stream
a non - isothermal gas would contribute to this effect by resulting in a less compressible flow that would impede the formation of a stable disc . during this time
, we would expect to observe the system in its stream phase . since the time necessary for a full disc to form is large ( preliminary runs showed @xmath150
was achieved after @xmath151 ) , variations in the mass transfer rate occurring across smaller timescales would prevent the full formation of a stable disc .
our use of an isothermal gas also increases the compressibility , and subsequently the density of our disc .
relaxing the isothermal approximation could result in a lower - density , less uniform disc .
our simulations of rs vul have shown that boosting the velocity of the accretion stream and deflecting it below the orbital plane at l1 introduces significant changes in the structure of the accretion flow . using an isothermal equation of state
, we simulate the formation of an accretion disc with both an undeflected stream and a stream boosted to mach 30 and deflected by @xmath0 . in the undeflected case ,
we measure the formation of an accretion disc that is symmetric about the orbital plane and displays less than one angular grid zone ( @xmath88 ) of warping . modifying the stream causes some radial enlargement of the disc , along with warping as large as @xmath1 .
in addition , a low - density plume of gas is ejected above the orbital plane on the far side of the accreting star with densities on the order of @xmath152 and tilt angles as high as @xmath105
. the intensities we measure in our pseudo - emissivity plots are strongest in the orbital plane due to our use of an isothermal gas , which causes the emissivity to depend solely on @xmath128 .
the increased out of plane flow that occurs due to the deflected stream increases the emissivity above and below the central plane by a minimum of five orders of magnitude .
we anticipate that relaxing the isothermal condition would cause higher emissivities to be measured outside of the central plane , as in the original tomographic data .
our results are consistent with the hypothesis that stream deflection is responsible for the `` tilted '' out - of - plane material that is measured in the tomograms of u crb , and also for the extreme out - of - plane flow observed in rs vul .
we have shown that a strong velocity boost and deflection at l1 are sufficient to change the direct - impact nature of the system , which increases the amount of out - of - plane flow drastically .
we expect that a variable mass transfer rate and potentially more extreme boosting would only serve to enhance the features observed in the out - of - plane flow .
this research was supported by an allocation of advanced computing resources provided by the national science foundation .
the computations were performed on kraken at the national institute for computational sciences .
this research was also supported in part by a north carolina space grant award .
the author would like to thank dr .
john blondin for his assistance with this work , and the anonymous referee for the helpful suggestions regarding the manuscript .
|
recent observations have shown that the direct - impact algol systems u crb and rs vul possess gas located outside of the orbital plane , including a tilted accretion disc in u crb .
observations of circumstellar gas surrounding the mass donor in rs vul suggest magnetic effects could be responsible for deflecting the accretion stream out of the orbital plane , resulting in a tilted disc .
to determine whether a tilted disc is possible due to a deflected stream , we use three - dimensional hydrodynamic simulations of the mass transfer process in rs vul . by deflecting the stream @xmath0 out of the orbital plane and boosting the magnitude of the stream s velocity to mach 30 , we mimic the effects of magnetic activity near l1 .
we find that the modified stream parameters change the direct - impact nature of the system .
the stream misses the surface of the star , and a slightly warped accretion disc forms with no more than @xmath1 of disc tilt .
the stream - disc interaction for the deflected stream forces a large degree of material above the orbital plane , increasing the out - of - plane flow drastically . plotting the h@xmath2 emissivity in velocity space allows us to compare our results with tomographic observations . deflecting and boosting the stream increases the emissivity in each @xmath3 slice of the out - of - plane flow by at least three and up to eight orders of magnitude compared to the undeflected case .
we conclude that a deflected stream is a viable mechanism for producing the strong out - of - plane flows seen in the tomographic images of u crb and rs vul .
accretion , accretion discs hydrodynamics stars : binaries : close
|
d - branes @xcite have recently emerged as a key ingredient driving the non - perturbative dynamics of string theory .
these objects provide an exact description of string theory solitons , and have led to remarkable developments towards understanding the nature of stringy black holes @xcite .
d - branes also provide a fascinating connection between string theory and gauge theories .
the low - energy effective dynamics of @xmath0 parallel d - branes is precisely that of supersymmetric yang - mills theory with gauge group @xmath1 @xcite ; such a low - energy effective theory has in turn been conjectured to provide a reformulation of string theory itself @xcite .
the goal of this paper is to extend further the range of phenomena in d - brane physics which can be described precisely in the language of yang - mills field theory or its extension to born - infeld theory . in particular , we consider the correspondence between the perturbative fluctuation spectra of d - brane configurations wrapped on a torus and gauge theories with a constant background field on the dual torus .
constant background field configurations in yang - mills theory on the torus are t - dual to d - branes which are either diagonally wrapped ( `` tilted '' ) @xcite or intersecting at an angle @xcite .
related recent work on branes at an angle can be found in @xcite .
the fluctuation spectra of gauge theories in a constant background field have been investigated previously in @xcite .
we compare these results to what is expected from the d - brane side .
the constant background field strength in the yang - mills theory gives rise to a new scale in the problem in addition to the string scale and the radius of the torus . in an appropriate scaling limit
, we show that yang - mills theory is insufficient to reproduce the spectrum of open strings on the d - branes and must be replaced by the full born - infeld action .
we find that in many cases the full born - infeld action reproduces in a field theory context the exact spectrum of low - energy string excitations around the d - brane background .
the story is slightly more subtle in the non - abelian case .
the excitation spectra for intersecting d - branes and the corresponding dual gauge fields were discussed in @xcite and @xcite respectively .
however , these spectra agree only to lowest order in the angle of intersection .
we investigate the resolution of this discrepancy by including higher order terms in the non - abelian born - infeld action .
although the non - abelian generalization of the born - infeld action has yet to be understood in full , a formulation involving a symmetrized trace was proposed recently by tseytlin @xcite .
we find that this prescription for non - abelian born - infeld fails to reproduce the exact fluctuation spectrum seen in string theory .
nonetheless , we encounter remarkable relations encoded in the general structure of tseytlin s non - abelian born - infeld action which provide a hint indicating how this discrepancy might ultimately be resolved . since the relation which emerges from the analysis of the action is highly non - trivial , we believe that there is a strong element of truth in tseytlin s proposal
. however , we do not completely resolve the discrepancy at finite angles , and we leave this as a problem for future investigations . a particularly interesting class of systems of the type considered here are those corresponding to non - supersymmetric d - brane configurations .
on @xmath2 constant gauge field backgrounds are stable if and only if the gauge field is ( anti)-self - dual . in @xcite
it was argued that the ( anti)-self - dual condition is equivalent to the condition that the dual d - brane configuration preserve enough supersymmetry to be a bps state .
considering non - supersymmetric d - brane configurations , we find that the tachyonic instability of the string ground state appears naturally in the field theory language as an instability of a saddle point configuration in the yang - mills / born - infeld theory .
this provides a natural framework for studying aspects of brane - anti - brane interactions in the context of field theory .
the organization of this paper is as follows : in section 2 we consider d - brane configurations which are tilted with respect to the torus ; these configurations correspond to gauge theory backgrounds with constant field strengths in the central @xmath3 part of @xmath1 .
the necessity for replacing the yang - mills action with the born - infeld action in order to exactly reproduce the fluctuation spectrum emerges from this analysis .
we also discuss the issue of fractional quantization from the gauge theory point of view in this section . in section 3
we consider systems of intersecting d - branes which correspond to reducible connections in the dual gauge theory .
we discuss the discrepancy between the spectra and its possible resolution by the non - abelian born - infeld action .
we also discuss here the d - brane / born - infeld correspondence for non - supersymmetric d - brane configurations and the associated tachyonic instability .
we begin with a brief review of some salient features of d - branes and their behavior under t - duality ; for more background see @xcite . in this paper
we are concerned with d - branes in type ii string theory . in type iia ( iib )
string theory , d - branes with even ( odd ) dimension appear .
when the theory is compactified on a torus @xmath4 there is a t - duality symmetry in each compactified direction which exchanges neumann and dirichlet string boundary conditions and which therefore changes the dimension of a d - brane by one .
the dynamics of a single d - brane are controlled by the born - infeld action , which reduces in the case of small field strength and flat background metric to a supersymmetric @xmath3 yang - mills action with adjoint scalars corresponding to transverse fluctuations @xcite . when @xmath0 d - branes become coincident , the yang - mills action is extended to a supersymmetric non - abelian @xmath1 yang - mills theory @xcite .
the generalization of the full born - infeld action to the non - abelian case has not yet been completely understood ; however , recent progress in this direction was made by tseytlin @xcite .
when d - branes are living in a space which has been compactified in a transverse direction , there are winding strings which wrap around the compactified direction an arbitrary number of times .
under t - duality , string winding and string momentum are exchanged so that these winding strings naturally become momentum modes of the gauge field under duality in the compact direction . at the level of the yang - mills action
, this duality can be made manifest .
as discussed in @xcite , the winding strings can be packaged in an infinite matrix which corresponds to the operator @xmath5 in the dual gauge theory . under this correspondence , the fluctuation spectrum of strings on a d - brane which is unwrapped in some set of compact dimensions
is precisely captured by the dual gauge theory . when we consider d - branes which are not parallel to the generators of a perpendicular torus , however , or systems of intersecting d - branes , we must be more careful .
it was pointed out in @xcite that nontrivial gauge configurations can carry d - brane charge .
this follows from the chern - simons term in the d - brane action which couples higher powers of the gauge field strength @xmath6 to rr fields .
when we have a system of tilted or intersecting d - branes on a torus , t - dualizing to a gauge theory on the torus will generally lead to a topologically nontrivial bundle whose chern classes determine the d - brane charges on various subtori .
such bundles were discussed in the @xmath7 case in @xcite , and in the context of d - branes in @xcite .
the approach we will take in this paper is to consider fixed configurations of flat d - branes on the torus which may be tilted or intersecting .
these d - brane configurations map under t - duality to constant field strength backgrounds in the dual gauge theory .
we will compare the fluctuation spectra of the d - brane configurations as calculated in string theory to the fluctuation spectra of the gauge fields around the constant background .
as we shall see , in many situations the action must be generalized to the full non - abelian born - infeld action to achieve an exact correspondence between these two spectra .
let us begin with the simplest example of a gauge field configuration corresponding to the t - dual of a tilted d - brane . consider a 2-brane wrapped on @xmath8 with periods @xmath10 and @xmath11 .
the fluctuation spectrum of this theory is precisely that of susy @xmath3 yang - mills theory in 2 + 1 dimensions around a vanishing background field .
there are gauge fields corresponding to the regular bosonic @xmath3 theory , adjoint scalars corresponding to transverse fluctuations , and various fermionic fields . throughout this paper
we will restrict attention to the fluctuation spectra of the bosonic modes , although a similar analysis could easily be performed for the fermion modes .
consider turning on a background @xmath12 the flux quantization condition requires that @xmath13 be an integer multiple of the flux quantum : @xmath14 such a configuration corresponds to a @xmath3 bundle with first chern class @xmath15 . for concreteness ,
let us take @xmath16 .
from the point of view of @xmath3 gauge theory , it is straightforward to compute the spectrum of fluctuations around this background . in terms of the fluctuation @xmath17 defined through @xmath18
the lagrangian is simply ( up to an overall constant ) @xmath19 we can ignore the additive constant and the total derivative terms , so one recovers the original yang - mills lagrangian @xmath20 whose physical spectrum is simply that of the quantized momenta @xmath21 the spectrum for the transverse bosonic fields is clearly the same , as they obey the same periodic boundary conditions .
let us now consider the d - brane configuration given after t - duality in the @xmath22 direction .
in the gauge theory configuration described above , we have a single unit of 2-brane charge corresponding to the @xmath3 gauge group .
the flux @xmath13 carries a single unit of 0-brane charge . under t - duality
, we get a d - brane configuration with one unit of 1-brane charge in each direction @xmath23 .
t - duality inverts the radius of the torus along the @xmath22 direction , so the dimensions of the dual torus are @xmath24 .
the background field given by ( [ background ] ) can be interpreted under t - duality as the transverse coordinate of a 1-brane on @xmath25 , @xmath26 this shows that the 1-brane is wound diagonally on the dual torus , as expected of a ( 1 , 1 ) bound state ( see figure [ tilt1 ] ) .
the field theory limit of this configuration corresponds to taking @xmath10 and @xmath11 to be of the same order of magnitude while sending @xmath27 to zero . in this limit ,
the dual torus becomes long and thin .
the low - lying states in the t - dual description are therefore given by momentum modes along the 1-brane and winding modes in the direction perpendicular to the 1-brane .
the winding modes can also be thought of as strings stretching between adjacent 1-branes in the covering space of the dual torus . from the point of view of string theory in a background given by such a d - brane configuration
, we expect the spectrum of low - lying excitations to be given by @xmath28 where @xmath29 is the length of the 1-brane along the diagonal @xmath30 and @xmath31 is the distance between adjacent 1-branes @xmath32 in terms of the original variables , ( [ tmpspec ] ) becomes @xmath33 comparison of eq .
( [ gaugespec ] ) and ( [ dbranespec ] ) indicates that they disagree by a factor of @xmath34 .
we will now show that the spectrum of excited states arising from the born - infeld action in 2 + 1 dimensions gives rise to the correct factor of @xmath35 , correcting ( [ gaugespec ] ) to precisely match the string theory result found in equation ( [ dbranespec ] ) .
the born - infeld action in 2 + 1 dimensions is given by ( up to an overall constant ) @xmath36 where @xmath37 is the minkowski metric @xmath38 , which will henceforth be used to raise and lower the lorentz indices .
just as in the case of the original yang - mills theory , the dynamics of small fluctuations is described by expanding the action around a fixed background .
substituting @xmath39 into ( [ born - infeld ] ) and expanding to quadratic order in @xmath17 , one finds that up to an overall scaling factor the quadratic term in the action is given by @xmath40 where @xmath41 is the fluctuation .
it will be convenient to separate @xmath42 into its symmetric and antisymmetric components : @xmath43 using this decomposition and ignoring the overall factor of @xmath44 , the action takes the form @xmath45 further computation shows that the terms containing @xmath46 vanish , and @xmath47 is a diagonal matrix @xmath48 with @xmath49 compared to the dynamics of small fluctuations in yang - mills theory given by equation ( [ original ] ) , the born - infeld action has effectively rescaled the spatial coordinates by a factor of @xmath50 .
this properly rescales the spectrum of the momentum modes so that there is a precise agreement with the spectrum of low - lying open string states ( [ dbranespec ] ) seen from the d - brane point of view .
this exercise of establishing the necessity for considering the born - infeld action over the yang - mills action in order to exactly match the string theory spectrum might seem purely academic in light of the fact that in the field theory limit @xmath51 goes to zero and the factor in question , @xmath52 , is identically 1 to leading order in @xmath27 .
however , there is a way to take a field theory limit keeping @xmath51 constant .
consider the same background as before @xmath53 with @xmath54 not equal to 1 . from the d - brane point of view
, this corresponds to a 1-brane with wrapping number @xmath55 .
( see figure [ tilt2 ] . )
again , we expect the spectrum to be given by ( [ tmpspec ] ) , and just as before , we obtain @xmath56 now , we can take the field theory limit by keeping @xmath10 , @xmath11 and @xmath57 constant and sending @xmath54 to infinity .
this time , the difference between born - infeld and yang - mills theory is non - trivial even at leading order in @xmath27 .
the fact that the born - infeld action captures the dual string spectrum more correctly than yang - mills theory is very natural in light of the fact that the dynamics on the world - volume of a d - brane is indeed described by the born - infeld action and yang - mills theory is only its leading approximation in the field theory limit for a fixed background .
in fact , the same type of system as the one we have discussed in this section was analyzed in @xcite from the point of view of string theory in a background gauge field , giving essentially the same results .
the interesting point about this type of configuration is that by scaling the background as above it is possible to retain the structure of born - infeld dynamics even in the field theory limit .
the extra terms arising from the born - infeld action are usually subleading in the @xmath58 expansion and are associated with the non - locality of string theory .
we can in fact see that the winding strings of minimum length attached to the tilted branes are not local objects from the world - volume observer s point of view , as is illustrated in figure [ nonlocal ] .
the results from this section indicate that it is possible to enhance the non - locality of string theory even in the field theory limit by scaling up the background field strength . in the presence of such a background , the interaction terms in the born - infeld action
will induce a mass shift in the spectrum of excited states to precisely reproduce the string theory spectrum . in the previous subsection we considered only configurations with a single 2-brane in the gauge theory picture , and all our analysis
was done in the context of the abelian born - infeld action . in this subsection
we discuss a more general configuration which is t - dual to a d1-brane on @xmath8 with wrapping numbers @xmath60 with @xmath61 relatively prime .
to describe such a configuration we need to define a @xmath62 bundle over @xmath8 whose connections have total flux @xmath63 .
the flux is quantized in units of @xmath64 since the gauge group is @xmath65 .
we can describe a @xmath62 bundle in the fundamental representation by giving the boundary overlap functions for sections @xmath66 the condition that these boundary conditions give a well - defined @xmath62 bundle is @xmath67 to have a total flux of @xmath63 we can decompose these boundary conditions into abelian and non - abelian components .
the abelian component should have flux @xmath63 , while the non - abelian component should have an t hooft @xcite twist giving a flux @xmath68 where @xmath69 and @xmath70 .
a standard choice of gauge for defining a bundle of this type is to pick @xmath71 where @xmath72 and @xmath73 are the diagonal and shift matrices @xmath74 choosing this set of boundary conditions , however , complicates the discussion of t - duality in the @xmath22 direction .
we will therefore instead choose the boundary conditions @xmath75 with @xmath76 being the diagonal matrix @xmath77 where @xmath78 is the integral part of @xmath79 , and where the multiplicities of the diagonal elements of @xmath76 are @xmath80 and @xmath54 respectively .
since @xmath76 has integral diagonal elements , we have @xmath81 so ( [ eq : bundlecondition ] ) is clearly satisfied .
the abelian part of @xmath76 is just @xmath82 so this bundle has the correct flux .
since the boundary conditions are trivial in the @xmath22 direction , we can discuss t - duality in this direction without encountering undue complications .
the gauge fields @xmath83 on this bundle will satisfy the boundary conditions @xmath84 the adjoint scalar fields @xmath85 will satisfy the same boundary conditions as @xmath86 .
the constant curvature background corresponding to these boundary conditions is @xmath87 where @xmath88 the t - dual along @xmath22 gives rise to the type of d1-brane configuration illustrated in figure [ tilt3 ] , with @xmath89 it is straightforward to derive the spectrum of open strings around this d - brane background : @xmath90 in order to reproduce this spectrum in field theory , we need the non - abelian generalization of the born - infeld action .
although there are some ambiguities in defining the non - abelian born - infeld action , a procedure for defining this action was recently proposed by tseytlin in @xcite . in the following , we will simply follow this proposal , which gives an action of the form @xmath91 where `` str '' indicates that the trace is to be taken after we symmetrize all the non - commuting products .
our analysis is simplified drastically by the fact that the background gauge field is proportional to the identity and commutes with everything . expanding the non - abelian born - infeld action to second order will therefore lead to @xmath92 \label{abelian}\ ] ] with the same @xmath47 we encountered previously .
just as in the previous case , we see that the effect of including the correction arising from the full born - infeld type interaction is the rescaling of the spectrum by a factor of @xmath93 .
thus , we have a spectrum given by @xmath94 where @xmath95 is the eigenvalue of the covariant derivative operator @xmath96 acting on the modes satisfying the boundary conditions ( [ bcond ] ) . due to the twisting in the boundary conditions ,
the quantization conditions of these modes are complicated slightly , and the @xmath97-fold degeneracy corresponding to the number of independent components of matrices in the adjoint of @xmath1 is lifted .
let us now consider the effects of the boundary conditions on the periodicity of the various modes .
the inhomogeneous term in the boundary condition for @xmath98 does not affect the periodicity structure of the fluctuation modes since it is taken care of by the background field . thus , the fluctuation spectrum will be the same for each of the gauge fields and the transverse scalars .
each of these fields is described by a matrix containing @xmath97 elements .
it is straightforward to show that the homogeneous part of the boundary condition ( [ bcond ] ) is satisfied by modes given by @xmath99 where @xmath100 is a @xmath101 matrix @xmath102 where @xmath103 for @xmath104 .
the boundary condition ( [ bcond ] ) further implies that @xmath105 indicating that @xmath106 takes on values quantized in units of @xmath107 .
since @xmath108 vanishes , the covariant derivative in the @xmath25 direction is simply the partial derivative , and it follows that @xmath109 the boundary condition in the @xmath22 direction is trivial , indicating that @xmath110 . due to the presence of the background , however , the covariant derivative is modified by the commutator term .
one finds that @xmath111 = \frac{2 \pi}{l_2}\left(m_2-\frac{r}{p}\right ) \delta a ( x_1,x_2).\ ] ] since @xmath112 could be any integer between 0 and @xmath113 , we find that @xmath114 is also quantized in units of @xmath107 : @xmath115 thus , in general we find that the @xmath97 degrees of freedom in the matrix appear as modes with momentum quantized in units of @xmath107 in both the @xmath25 and @xmath22 directions . substituting the expressions for @xmath116 and @xmath114 in equations ( [ k1 ] ) and ( [ k2 ] ) back into the spectrum ( [ back ] ) , we recover @xmath117 in agreement with the string theory answer .
it is interesting to note that the fractional quantization of momentum in the @xmath25 direction appears naturally as a periodicity on a @xmath118-fold covering , whereas the fractional quantization in the @xmath22 direction appears because there are @xmath118 different `` types '' of excitations , labeled by which diagonal of the matrix they appear in .
these two distinct mechanisms for fractional quantization correspond naturally to the string geometry in the d - brane picture , where the @xmath25 excitations correspond to momentum on a @xmath118-fold wrapped brane while the @xmath22 excitations correspond to strings connecting branes separated by a distance proportional to @xmath119 .
this explicit construction in the context of gauge theory gives an interesting point of view on fractional momentum states , which have been considered in a number of contexts@xcite .
there is a close correspondence between this discussion and the description of fractional momentum states as arising from the twisted sector in an orbifold by the center of the gauge group @xcite .
in the previous section , we examined the d - brane / gauge theory correspondence for tilted d - branes , corresponding to a gauge theory background in the central @xmath3 of @xmath1 . in this section ,
we examine systems of intersecting d - branes which correspond to constant curvature backgrounds with reducible connections .
when the constant background field can be expressed in terms of a background connection @xmath120 with components which all commute , the t - dual system corresponds to a system of intersecting branes with well - defined positions . for concreteness
, we will concentrate on systems corresponding to gauge theories with @xmath121 .
the constant curvature @xmath122 backgrounds which describe dual configurations of intersecting d - branes are precisely those background fields which can be described in terms of reducible connections .
thus , in the gauge theory language , we will be considering connections which decompose as a direct sum of @xmath3 connections .
this restriction simplifies the analysis somewhat .
a complete classification of constant curvature connections on @xmath123 bundles over @xmath2 was given by van baal in @xcite ; in this work he also calculated the yang - mills fluctuation spectra around these backgrounds .
in the d - brane language , the corresponding backgrounds describe a pair of branes intersecting at angles ; the spectrum of low - lying open strings in such backgrounds have been studied in @xcite .
although the spectra computed in @xcite and @xcite have striking qualitative similarities , they are distinct in certain details . we will investigate the possibility of resolving this discrepancy by considering the full non - abelian born - infeld action .
we also discuss the dynamics of non - supersymmetric backgrounds .
we begin our discussion by reviewing the d - brane construction of @xcite .
one of the main points made in @xcite is the fact that there exist supersymmetric configurations of branes intersecting at angles .
here , we consider a general class of configurations of intersecting branes which includes non - supersymmetric brane configurations .
the configurations we are interested in correspond to intersecting d - branes living on tori .
as discussed above , these configurations are dual to gauge theory backgrounds with reducible connections
. we will here restrict attention to pairs of intersecting branes , concentrating in particular on the case of two 2-branes on @xmath2 .
there are several kinds of string excitations around such a background .
there are strings which stretch from each of the two 2-branes back to itself .
these strings carry momentum and winding numbers .
these excitations correspond to modes living in one of the reducible components of the dual bundle and are essentially equivalent to those discussed in the previous section .
thus , we will not consider these modes further here but will concentrate on the strings connecting the two different branes . unlike the strings connecting a brane to itself , the strings connecting the different branes are unaffected by the compactification of the space in which the branes are living .
this can be seen by going to the covering space , where all strings connecting the two branes are homotopically connected ( as long as they are associated with the same intersection point ) .
of course , at higher order there are diagrams connecting these strings with winding strings connecting a brane to itself
. however , these diagrams will not affect the fluctuation spectrum .
thus , to understand the fluctuation spectrum of the strings connecting the distinct 2-branes , it is sufficient to understand the spectrum of strings connecting two 2-branes in noncompact space .
this is precisely the situation considered in @xcite , and we recapitulate their discussion here with the additional feature that we do not insist on the supersymmetry of the d - brane configuration . a sufficiently general configuration of two 2-branes in @xmath124 can be defined by beginning with two 2-branes oriented along the 14 axis , and rotating one of them by an angle @xmath125 in the 12 plane and by an angle @xmath126 in the 34 plane . for @xmath127 ,
this configuration was shown to be supersymmetric in @xcite .
following @xcite , we use complex coordinates @xmath128 and @xmath129 instead of @xmath130 and @xmath131 , respectively . in world - sheet language ,
strings stretching between the two branes are described by open strings with boundary conditions @xmath132 the mode expansion for the complex bosonic field is given by @xmath133 where @xmath134 and @xmath135 .
the vacuum energy @xmath136 can be computed either from the @xmath137-function regularization of the sum @xmath138 or from the dimensions of the world sheet operators needed to twist the boundary condition @xcite .
the low - lying open string excitations in the ns sector described in equation ( 3.5 ) of @xcite generalize to @xmath139 and their contribution to the masses are , respectively , @xmath140 this gives a complete characterization of the fluctuation spectra of strings stretching between two 2-branes at angles @xmath141 , including cases where the brane configuration is not bps .
note that when @xmath142 the spectrum contains a tachyon . we will now discuss a gauge theory configuration whose dual corresponds to two 2-branes on @xmath2 . as stated
, we will restrict attention to @xmath122 bundles with connections which are reducible to a direct sum of two @xmath3 connections .
the discussion could be generalized in a straightforward way to arbitrary reducible @xmath1 bundles .
it was pointed out by t hooft @xcite that for @xmath143 there are constant curvature connections which do not satisfy abelian boundary conditions ; these correspond to d - brane configurations with fewer than @xmath0 d - branes , with some of the d - branes wrapped multiple times around the base space as in section [ sec : pq ] .
we will consider here a constant background field given by a @xmath122 connection @xmath144 on @xmath2 with the following components @xmath145 where @xmath146 are the dimensions of the 4-torus and @xmath147 is an anti - selfdual matrix with with @xmath148 and zero otherwise .
@xmath149 is the su(2 ) generator @xmath150 .
this configuration has an instanton number @xcite .
@xmath151 because the field strength is proportional to @xmath149 the first chern class vanishes .
thus , this configuration corresponds to a ( singular ) point in the moduli space of @xmath122 instantons on @xmath2 with instanton number @xmath152 .
( note that while a single instanton on @xmath2 shrinks to a point , there is a regular moduli space of instantons with @xmath152 . ) performing a t - duality transformation along directions @xmath153 and 3 will invert the periods along these directions , and will give rise to a pair of 2-branes embedded according to the equations ( see figure [ inter1 ] ) @xmath154 the field strength @xmath155 can immediately be related to the angles @xmath156 between the branes : @xmath157 the connection @xmath144 is gauge equivalent to one considered in @xcite , @xmath158 van baal investigated the spectrum of @xmath123 yang - mills fluctuations around backgrounds of this type .
he explicitly expanded the yang - mills action around @xmath120 using @xmath159 to second order in @xmath160 and @xmath161 . here ,
@xmath160 and @xmath161 are real and complex fields respectively ; in @xmath123 , we would have @xmath162 .
there are also ghost fields @xmath163 living in the adjoint representation .
the action , expressed in these variables , takes the form @xmath164 with @xmath165 as discussed in section [ branes.at.angles ] , the fields @xmath166 correspond to strings in the dual picture connecting a brane with itself , and have spectra corresponding to the discussion in section [ sec : tilted ] .
thus , we concentrate now on the spectrum for the field @xmath167 corresponding to strings stretching between the two branes . the computation of this spectrum is described in detail in @xcite , so we simply summarize the results of that investigation here . the boundary conditions on @xmath167 indicate that it is a theta function on @xmath2 .
the operator @xmath168 acts on the appropriate space of theta functions as a laplace - beltrami operator up to a constant .
the eigenfunctions of this operator were classified in @xcite , and are given by a sum over the lattice of harmonic oscillator eigenfunctions rotated by linear phases .
the low - lying eigenfunctions are localized in the vicinity of the string intersection points .
we will discuss a specific example of such eigenfunctions more explicitly in section [ sec : unstable ] .
the spectrum of eigenvalues for the four - dimensional system of interest here is given in equation ( 3.25 ) of @xcite : @xmath169 where @xmath170 we therefore see that the spectrum of fluctuations around this constant field strength background has precisely the structure of the oscillator excitations ( [ eq3 ] ) found from the string theory point of view in @xcite up to the identification @xmath171 thus , the fluctuation spectra computed from these two points of view agree up to a discrepancy between @xmath172 and @xmath173 . for small @xmath174 these spectra
agree exactly .
the difference in the spectra resulting from the discrepancy between @xmath172 and @xmath175 might seem minor in light of the fact that they are in perfect agreement at small angles , and that in the field theory limit keeping the background constant , @xmath174 goes to zero .
however , just as in section [ sec : tilted ] , one can take a field theory limit while scaling the background to keep @xmath174 finite but small compared to 1 . to be more specific
, we can consider the background @xmath176 and take the limit @xmath177 keeping the ratios of @xmath146 s and @xmath178 constant .
after a gauge transformation to set @xmath179 , this will lead in the t - dual picture to 2-branes wrapping multiply around the 23 cycles as illustrated in figure [ inter2 ] .
we saw in section [ sec : tilted ] that scaling the field strength enhances the terms ordinarily subleading in @xmath58 and that the full born - infeld action was needed to recover the exact correspondence in the case of tilted branes .
it is therefore natural to suspect that the same mechanism underlies the resolution of the @xmath172 vs.@xmath180 discrepancy .
we investigate this possibility in this subsection .
the analysis is complicated by the fact that the background field does not commute with all fluctuations .
the field theory analysis therefore depends on the full structure of the non - abelian born - infeld action .
recently , a concrete formulation of the non - abelian born - infeld action was proposed by tseytlin @xcite .
using tseytlin s born - infeld action , we compute the spectrum of fluctuations around the same background as was considered in the previous subsection .
a remarkable relation emerges in the analysis , bringing the field theory spectrum tantalizingly close to the string theory spectrum of section [ branes.at.angles ] .
however , we fall slightly short of reproducing the string theory spectrum in its entirety .
we interpret this situation in the following manner : due to the highly non - trivial nature of the partial success in resolving the @xmath172 vs. @xmath175 discrepancy , there must be a strong element of truth in tseytlin s formulation of the non - abelian born - infeld action , but in as much as it fails to reproduce the string theory spectrum precisely , it is still in some sense incomplete .
reformulating the born - infeld action to resolve this discrepancy is likely to be a rather tricky enterprise .
nonetheless , we feel that reproducing the spectrum of a system of d - branes intersecting at an angle will serve as a useful test and could possibly be used as a guiding principle in formulating the elusive non - abelian born - infeld action .
the key feature of tseytlin s non - abelian born - infeld action is its resolution of product ordering ambiguities through symmetrization @xmath181 our strategy here is to perform formal manipulations of quantities inside the symmetric trace as if they were abelian . at the last stage of the analysis
, we will re - insert the non - commuting factors and perform the symmetrized trace , thereby promoting the abelianized action back up to its original non - abelian counterpart . in our analysis we consider the symmetrized trace term by term in an expansion with respect to the background as well as the fluctuation .
for example , a term in the expansion corresponding to an abelianized expression , @xmath182 where @xmath183 and @xmath184 are in the adjoint of @xmath123 , is replaced by @xmath185 \label{expansion}\ ] ] upon reinsertion of non - commuting factors .
let us consider in particular the case where @xmath186 and @xmath187 are restricted to the range of values @xmath188 .
simple algebraic manipulations of the pauli matrices show that @xmath189 at this point , it is convenient to consider the cases of @xmath190 even and @xmath190 odd separately .
for @xmath190 odd , @xmath191 whereas for @xmath190 even , @xmath192 therefore , if in the abelian born - infeld action the terms quadratic in fluctuations had an expansion consisting solely of terms with even powers in the background : @xmath193 then its non - abelian generalization for @xmath194 and @xmath195 fields will give rise to an expansion of the form @xmath196 these two expansions are related by @xmath197 the goal at this point is to examine the fluctuations of @xmath198 in a background defined by @xmath199 and @xmath200 -i [ a^0_\mu,\delta a_\nu ] -i [ \delta a_\mu , \delta a_\nu]\ ] ] we expand the action to quadratic order using the standard identity @xmath201 + \frac{1}{8 } ( { { \rm tr}}[m_0^{-1 } \delta m])^2 - \frac{1}{4 } { { \rm tr}}[m_0^{-1 } \delta m m_0^{-1 } \delta m ] \right)\ ] ] in terms of our variables , the quadratic action reads @xmath202 \label{quadaction}\end{aligned}\ ] ] where now , @xmath203 with @xmath204 and @xmath205 -g^ f _ ^.@xmath206 the term linear in @xmath207 contributes at quadratic order because the commutator term in @xmath207 is quadratic in the fluctuations .
this time , the terms quadratic in @xmath46 turn out not to vanish but give a topological contribution @xmath208 which can be ignored in the discussion of the fluctuation spectrum . in order to compute the fluctuation spectrum
we now must apply the integral prescription ( [ relation ] ) to the action in order to deal with the lack of commutativity between the fluctuations and the background .
tseytlin s symmetrized trace prescription indicates that we should associate a single @xmath122 generator with each factor of the field strength @xmath6 .
this gives a well - defined formulation of a non - abelian action . with this prescription ,
the action as a function of the components of the field strength and its off - diagonal fluctuations should be computed by applying ( 20 ) to the terms quadratic in @xmath209 in ( 21 ) .
the term linear in @xmath209 is not modified , since it is only nonzero when @xmath209 commutes with the background .
unfortunately , this does not seem to give exactly the results that we would expect from string theory .
nonetheless , there is some indication that this prescription comes close to doing the right thing . to see this
let us consider a special case where the field strength is anti - self - dual , corresponding to a bps brane configuration .
in this case we have @xmath210 . the determinant factor in the action
gives a single factor of @xmath211 .
this cancels with the @xmath212 factors in the terms quadratic in @xmath213 , so that the kinetic terms are canonically normalized .
the potential terms quadratic in @xmath209 carry two factors of @xmath212 so that they are multiplied by an overall factor of @xmath214 compared to the pure yang - mills theory . at this point
, we encounter a pleasant surprise . applying the relation ( [ relation ] ) to the metric leads to @xmath215 this is precisely the rescaling necessary to match the field theory spectrum with the string theory result ! in fact , in this case we find that the spacing of the fluctuation spectrum corresponds precisely to that predicted by string theory .
unfortunately , however , the energy of the lowest fluctuation is also affected by the term linear in @xmath209 .
the factors of @xmath212 cancel in this term so that the calculation of standard yang - mills theory is not modified by the symmetrized trace form of the born - infeld action .
thus , even in the special case where the field is anti - self - dual , we do not get exactly the spectrum predicted by string theory .
we would expect that even in the more general case where the field is not anti - self - dual , the spectra computed from string theory and from the non - abelian born - infeld theory should agree . in the more general case , however , we find that even the spacing of the energy levels fails to be reproduced exactly by the symmetrized trace prescription .
thus , it seems clear that this specific formulation of the non - abelian born - infeld theory is not sufficiently general to capture all the relevant physics of the situation we are considering here .
there are several possible explanations for why the symmetrized trace prescription does not exactly match the string theory calculation .
one possibility is that the association of a single generator to each factor of the field strength should be modified . as an example of this sort of modification
, we might choose to normalize the fluctuations @xmath216 so that the kinetic terms in the action always have a canonical normalization .
for cases where the field strength is not anti - self - dual this will incorporate extra factors into the potential terms .
it can be shown that these extra terms guarantee that the scaling factor associated with the quadratic term in @xmath209 will give fluctuations with a level spacing precisely that predicted by string theory , as we showed above explicitly in the anti - self - dual case . unfortunately , however
, it does not seem to be possible to arrange the normalization so that the linear term in @xmath209 also scales correctly .
another possible reason why the symmetrized trace prescription does not reproduce the string theory results correctly is that covariant derivative terms @xmath217 or commutator terms @xmath218 $ ] may be relevant in this situation . in tseytlin
s discussion of the non - abelian born - infeld action , terms of these forms were explicitly dropped .
since the fluctuations we are concerned with here are theta functions which give nontrivial spatially dependent fluxes , it is quite possible that the effects of covariant derivative terms must be incorporated into the non - abelian born - infeld action in order to correctly reproduce the string theory results .
a discussion of a related discrepancy was given in @xcite , where it was shown that it was necessary to add commutator and covariant derivative terms to the born - infeld action already at order @xmath219 to reproduce results from string theory .
in any case , we do not have a complete understanding of why the prescription we have described corrects some , but not all , of the terms appearing in the action , and as such we have not provided a satisfactory resolution to the problem of the @xmath172 vs. @xmath173 discrepancy . nonetheless , the relation between @xmath172 and @xmath175 is transcendental , and subtle details must work out just right in order for the correct transformation to emerge from this kind of analysis .
the fact that one can naturally derive from the symmetrized trace prescription the precise scaling factor needed to achieve agreement with string theory seems to provide some evidence that there is an element of truth in this description .
we believe that when it is correctly defined , the full non - abelian born - infeld action should resolve the discrepancy we have discussed here .
we therefore pose the resolution of the @xmath172 vs. @xmath173 discrepancy as a test which should be satisfied by any candidate for the full non - abelian born - infeld action .
we have discussed the correspondence between fluctuation spectra in a variety of situations corresponding to tilted and intersecting d - branes . as mentioned in section [ branes.at.angles ] , the formulae for the fluctuation spectra around intersecting d - branes extend to cases where the configuration of branes breaks supersymmetry . however , in such situations the state is no longer bps and generally admits a tachyonic instability @xcite . in this subsection
we discuss the simplest example of such an instability in slightly more explicit terms .
consider a pair of 1-branes wrapped on @xmath8 with winding numbers @xmath220 and @xmath221 .
these 1-branes intersect at two points .
this configuration is unstable , since a simple change in the topology at one of the crossing points turns the configuration into a single 1-brane with winding @xmath222 ( see figure [ f : unstable ] ) .
we can understand this instability in the system from both the string theory and the yang - mills / born - infeld point of view . in the gauge theory picture
, we have a pair of 2-branes on @xmath8 , one carrying a unit of 0-brane charge , and the other carrying a negative 0-brane charge , corresponding to a single anti-0-brane .
thus , the tachyonic instability in this picture corresponds to the instability of a brane - anti - brane configuration . to be specific , we will consider a @xmath122 gauge theory on a @xmath8 with sides of length @xmath223 .
we will take the boundary conditions to be given by @xmath224 although these seem like nontrivial boundary conditions , they define a trivial @xmath122 bundle over @xmath8 in an unusual choice of gauge ; this gauge choice allows us to write the background connection in a form which manifestly corresponds to the desired t - dual d - brane configuration .
the boundary conditions on the gauge fields are given by @xmath225 we will be considering the constant curvature background corresponding to the connection @xmath226 clearly , through t - duality this corresponds to a pair of 1-branes whose transverse positions are given by @xmath227 just as shown in figure [ f : unstable ] .
the tachyonic instability of this system is easy to see in the d - brane language .
the string excitation spectrum is essentially given by ( [ eq3 ] ) where we set @xmath228 and where @xmath229 is related to the angle @xmath174 between the 1-branes by @xmath230 .
as mentioned in section [ branes.at.angles ] , this spectrum contains a tachyon .
we now discuss the instability of this system from the gauge theory point of view . as we have seen in previous sections ,
the born - infeld action is necessary to get the precise normalization of the fluctuation spectra .
however , the sign of the various fluctuations are not dependent upon such a precise calculation , so we should be able to detect the instability within the simple yang - mills framework . to find the unstable modes , we can use essentially the same analysis as that of van baal @xcite , simply setting to zero one of the two curvatures @xmath231 .
we find that indeed there are modes with negative eigenvalues corresponding to unstable directions in the system .
to describe these modes more precisely , let us consider the boundary conditions on the off - diagonal modes of the system .
decomposing as above @xmath159 we find that each @xmath161 satisfies the boundary conditions @xmath232 these conditions imply that @xmath233 is a section of a @xmath3 bundle with @xmath234 .
we can express @xmath161 as a theta function satisfying the condition @xmath235 where @xmath236 by transforming @xmath237 a general discussion of the theta functions of this type which are eigenfunctions of the laplace - beltrami operator is given in @xcite . in the particular case of interest here , the description of the states in terms of the boundary conditions ( [ eq : boundary ] )
is fairly straightforward .
we are looking for eigenfunctions of the operator @xmath238 with the minimal eigenvalue , subject to the condition that the eigenfunctions must satisfy the boundary conditions ( [ eq : boundary ] ) .
there are precisely two such eigenfunctions with the minimum eigenvalue @xmath239 .
these functions are given by @xmath240 + 2 \pi i ( \frac{x_1}{l_1 } m-\frac{x_2}{l_2 }
n ) -2 \pi i \frac{x_1 x_2}{l_1 l_2 } \right).\ ] ] and @xmath241 \right.\\ & & \hspace{1.3in}\left.+ 2 \pi i ( \frac{x_1}{l_1 } \hat{m}-\frac{x_2}{l_2 } \hat{n } ) -2 \pi i \frac{x_1 x_2}{l_1 l_2 } \right).\end{aligned}\ ] ] these solutions are given by superimposing copies of the gaussian harmonic oscillator ground state centered around points on a lattice .
as @xmath11 becomes small , these functions become localized around the values @xmath242 and @xmath243 , which are precisely the @xmath25 values of the two points of intersection for the dual d - branes . in the corresponding d - brane picture ,
as @xmath11 becomes small the dual radius becomes large and the angle between the branes increases , which as discussed in @xcite should localize the string modes in the vicinity of the intersection points .
furthermore , since the unstable modes are associated with the @xmath167 entries of the @xmath122 matrix , which affect the commutativity between the two brane positions , it is natural to interpret these modes geometrically in terms of a change in topology localized to the brane intersection points .
this corresponds perfectly with what we would expect from the geometrical picture shown in figure [ f : unstable ] .
how this explicit description of the unstable mode relates to the d - brane instability is difficult to make completely precise , however , due to the nonlocality of the t - duality transformation in the @xmath22 direction .
a better understanding of this relationship might lead to interesting results about brane - anti - brane interactions which could be understood from a simple field theory perspective .
in this paper , we examined the spectra of low - energy fluctuations around various d - brane configurations in string theory and the t - dual spectra of fluctuations around constant background fields in gauge theory .
background fields in the central @xmath3 of a @xmath1 gauge theory correspond to tilted branes on the dual torus . by taking the field theory limit while keeping the tilt angle constant
, we encountered a discrepancy between the field theory spectrum and the string theory spectrum .
the discrepancy was resolved by computing the fluctuation spectrum of the full born - infeld action instead of the yang - mills action .
this is a natural result in light of the fact that the scaling limit which keeps the tilt constant corresponds to scaling up the background field strength keeping @xmath244 constant .
this forces the terms ordinarily subleading in @xmath58 to play a relevant role .
we also examined the spectrum of excitations around constant background fields from reducible connections , corresponding to branes intersecting at an angle on the dual torus .
we found that the results of @xcite can be extended to non - supersymmetric configurations and that the dynamics of tachyons can be studied naturally in the context of gauge theories . in the small angle limit ,
the d - brane and yang - mills spectra are in agreement .
the correspondence breaks down , however , when the angle of intersection is fixed and finite .
we investigated the possibility of making the correspondence exact even at finite angles by considering the full born - infeld action . here
, a non - abelian version of the born - infeld action is necessary to make the appropriate correspondence .
we analyzed the gauge theory spectrum using tseytlin s recent proposal for the symmetrized non - abelian born - infeld action @xcite .
although we were unable to resolve the discrepancy at finite angle in full , we found a strong hint in the general structure of the action suggesting how this discrepancy might be resolved .
it seems rather remarkable that the subtle transcendental relations required to reproduce the correct spectrum of strings attached to the d - branes are encoded in the symmetrized born - infeld action .
perhaps one could turn this type of argument around and use the correspondence with string theory through t - duality as a guiding principle for resolving various difficulties which plague the non - abelian born - infeld action .
an intriguing aspect of the correspondence described here is related to the spatial dependence of the fluctuations in the gauge theory and string theory pictures for intersecting branes . in the gauge theory description ,
the off - diagonal fluctuations correspond to sections of nontrivial @xmath3 bundles over the torus , and are described in @xcite in terms of theta functions .
these theta functions were described in @xcite in terms of linear superpositions along the lattice of harmonic oscillator eigenfunctions .
this description essentially coincides with the characterization of states in a constant magnetic field on the torus in terms of landau levels . on the other hand , as argued by berkooz , douglas and leigh in @xcite , the strings connecting intersecting d - branes are essentially living in a harmonic oscillator potential well and are localized to the intersection of the branes .
it is natural to expect that the spatial dependence of the second quantized string excitations in the intersecting d - brane picture will correspond to gaussian and higher excited harmonic oscillator eigenfunctions .
this agrees nicely with the theta function description in the gauge theory language ; however , there is a subtlety in this relationship because a t - duality transformation has been performed in some of the directions on the torus .
making this correspondence more precise might lead to a better understanding of the action of t - duality as a map between the hilbert spaces of quantized born - infeld theory and second quantized string theory .
although in this paper we have concentrated on stationary backgrounds , the configurations of tilted and intersecting branes which we have considered here are closely related to boosted branes and systems of branes scattering from one another . for these systems , a convenient dynamical quantity to compute is the phase shift due to scattering @xcite . a discrepancy in the relativistic corrections to d - brane scattering processes as computed in a quantum mechanics model derived from yang - mills theory was discussed in @xcite .
this discrepancy seems similar in nature to the @xmath172 vs. @xmath175 discrepancy we have described here .
the authors of @xcite also consider the possibility of resolving this discrepancy by considering the full born - infeld action , but encounter similar difficulties in making the exact correspondence at subleading order in backgrounds .
in particular , they report a discrepancy between the phase shift computed from the relativistic 0-brane quantum mechanics ( equation ( 63 ) of @xcite ) @xmath245 and from the semiclassical phase shift in the eikonal approximation ( equation ( 64 ) of @xcite ) @xmath246 it is interesting to note that @xmath247 and @xmath248 are related by a relation similar to ( [ relation ] ) @xmath249 although it is not completely clear why the integral prescription ( [ relation ] ) would apply to the phase shift in resolving the discrepancy between @xmath250 and @xmath251 , this observation provides an indication that by analyzing the combinatorics of tseytlin s symmetrized trace prescription @xcite , relativistic quantum mechanics of 0-branes might be made to agree with the semiclassical calculation .
another aspect of the d - brane / gauge theory correspondence which we have not pursued here , but which is an interesting direction for further study , is the relationship between the moduli spaces of vacua in the two pictures and the corresponding 0-modes of the system .
it was pointed out by van baal in @xcite that the physical 0-modes in the gauge theory picture can be determined explicitly , and that their number corresponds to the expected dimension of the moduli space of instantons .
it has been argued @xcite that the moduli space of vacua for the intersecting d - brane system is precisely given by the appropriate instanton moduli space .
the approach we have discussed here may provide a concrete framework in which this connection can be explored in more detail .
the web of relations between string theories and gauge theories has become an active area of investigation in light of our increased understanding of the role d - branes play in the non - perturbative dynamics of string theory and in reformulating string theory itself though m(atrix ) theory .
backgrounds with constant fluxes have played an important role in recent studies of extended objects in m(atrix ) theory @xcite .
although there is clearly more remaining to be said regarding the exact definition of the non - abelian born - infeld theory , it seems that when this theory is properly defined it should provide a mechanism for calculating some rather nontrivial results for systems of interacting d - branes within a field theory context .
we thank v. balasubramanian , c. callan , o. ganor , i. klebanov , g.lifschytz , s. ramgoolam , a. tseytlin , and p. van baal for useful discussions .
the work of ah is supported in part by doe grant de - fg02 - 91er40671 , the nsf presidential young investigator award phy-9157482 and the james s. mcdonnell foundation grant no .
the work of wt is supported in part by the national science foundation ( nsf ) under contract phy96 - 00258 .
|
we consider the spectra of excitations around diagonal and intersecting d - brane configurations on tori .
these configurations are described by constant curvature connections in a dual gauge theory description .
the low - energy string fluctuation spectrum is reproduced exactly by the gauge theory in the case of vanishing field strength ; however , this correspondence breaks down for fixed nonzero field strength .
we show that in many cases the full born - infeld action correctly captures the low - energy spectrum in the case of non - vanishing field strength .
this gives a field theory description of the low - energy physics of systems of diagonally wound branes and branes at angles as considered by berkooz , douglas and leigh .
this description extends naturally to non - supersymmetric configurations , where the tachyonic instability associated with brane - anti - brane systems appears as an instability around a saddle point solution of the corresponding yang - mills / born - infeld theory . in some cases
, the field theory description requires a non - abelian generalization of the born - infeld action .
we follow tseytlin s recent proposal for formulating such an action . in the case of intersecting branes ,
the non - abelian born - infeld theory produces a transcendental relation which comes tantalizingly close to reproducing the correct spectrum ; however , a discrepancy remains which indicates that a further clarification of the non - abelian born - infeld action may be necessary .
= msbm10 at 12pt = msbm7 = msbm5 = = = # 1#1 = 6.5 in pupt-1693 + hep - th/9703217 * fluctuation spectra of tilted and intersecting d - branes from the born - infeld action * akikazu hashimoto and washington taylor iv _ department of physics _ + _ joseph henry laboratories _ + _ princeton university _ + _ princeton , new jersey 08544 , u.s.a . _
+ aki , wati @princeton.edu march 1997
|
let @xmath3 be a smooth manifold of dimension @xmath1 , possibly with a smooth boundary @xmath4 .
recall that two riemannian metrics @xmath5 and @xmath6 on @xmath3 are _ conformal _ if there exists a smooth function @xmath7 on @xmath3 such that @xmath8 pointwise as quadratic functions on the fibers of @xmath9 .
conformality is obviously an equivalence relation , with the equivalence class of a metric @xmath5 called the _
conformal class _ of @xmath5 : @xmath10 = \left\{u^{\frac{4}{n-2 } } g : u > 0 \text { is a smooth function on } m\right\}.\ ] ] the conformal class is an indispensable object in geometric analysis . in the proof of the riemannian penrose inequality @xcite ,
bray observed the following fact : the relation on riemannian metrics @xmath11 defined by @xmath12 is an equivalence relation .
here , @xmath13 is the laplacian operator on functions with respect to @xmath5 .
reflexivity of @xmath11 is clear ; symmetry and transitivity follow from the formula for conformal metrics @xmath8 : @xmath14 for any smooth function @xmath15 on @xmath3 , where @xmath16 is the laplacian with respect to @xmath17 ( c.f .
lemma 2.1 of @xcite ) .
formula ( [ eqn_conf_laplacian ] ) also explains the exponent @xmath18 : no such equivalence relation exists for other values of this exponent . the @xmath11 equivalence class of a metric @xmath5 is called the _ harmonic conformal class _ of @xmath5 ( although later we will slightly refine this definition ) .
the harmonic conformal class is intimately connected with scalar curvature . for ,
if @xmath8 are conformal metrics , the scalar curvatures @xmath19 and @xmath20 are related by @xmath21 so we see that @xmath22 if and only if @xmath23 .
in particular , metrics @xmath22 have the same pointwise sign of scalar curvature .
if @xmath0 is compact and without boundary , the harmonic conformal class consists only of the constant rescalings of @xmath5 , as follows from the maximum principle . in the context of general relativity , a natural class of manifolds to study are asymptotically flat manifolds of nonnegative scalar curvature that possess a compact boundary .
the harmonic conformal class is well - adapted to studying such spaces , since it preserves both asymptotic flatness ( under suitable restrictions ) and the nonnegativity of scalar curvature . in the literature ,
much emphasis is placed on manifolds whose boundary consists of minimal surfaces ( c.f .
@xcite , for instance ) , but we make no such restriction here .
our purpose is to define and study objects canonically associated to the harmonic conformal class of an asymptotically flat manifold with boundary .
the outline is as follows : in section [ sec_definitions ] we standardize our definitions of asymptotic flatness , adm mass , and the harmonic conformal class .
section [ sec_numerical ] defines two real number invariants of the harmonic conformal class , called @xmath24 and @xmath25 . in section [ sec_mass_profile ]
we motivate a constrained optimization problem for the adm mass and apply it to define a function @xmath26 that depends only on the harmonic conformal class .
the first main result , theorem [ thm_mu ] , is that @xmath27 is given by an explicit formula involving @xmath24 and @xmath25 .
we introduce a function @xmath28 in section [ sec_area_profile ] , also a harmonic conformal invariant , that is substantially more subtle than its counterpart , @xmath27 .
roughly , the value of @xmath29 is determined by maximizing the least area needed to enclose the boundary among metrics in the harmonic conformal class that measure the boundary area to be at most @xmath30 .
this minimax - type definition is somewhat delicate no simple formula for @xmath31 is expected , and even showing that the maximum is attained seems to require enlarging the harmonic conformal class to allow for weak boundary regularity of the conformal factors . with this additional flexibility ,
we study @xmath31 and the properties of the metrics which optimize it in section [ sec_properties ] .
the main result here is that assuming good regularity for these optimal metrics , the resulting manifolds are such that the boundaries are enclosed by an area - minimizing surface that `` almost '' has zero mean curvature .
the final two sections consist of examples and some conjectured applications of the techniques developed herein . assuming a certain extension of the riemannian penrose inequality
, we establish inequalities between the numerical invariants @xmath24 and @xmath25 and the functions @xmath27 and @xmath31 .
these estimates are particularly relevant for the study of zero area singularities , which we recall . in closing , we discuss a possible generalization of the positive mass theorem that allows for metrics with certain types of singularities .
most of the content of this paper was part of my thesis work , and i am very indebted to my advisor hugh bray for countless discussions and suggestions .
i would also like to thank bill allard , graham cox , michael eichmair , george lam , and mark stern for helpful discussions .
we will consider asymptotically flat manifolds , which are spaces that geometrically approach euclidean space in a precise sense .
[ def_af ] a smooth , connected , riemannian manifold @xmath0 ( possibly with compact boundary ) of dimension @xmath32 is * asymptotically flat * ( with one end ) if a. there exists a compact subset @xmath33 and a diffeomorphism @xmath34 ( where @xmath35 is a closed ball ) , and b. in the coordinates @xmath36 on @xmath37 induced by @xmath38 , the metric obeys the decay conditions : @xmath39 for @xmath40 sufficiently large and all @xmath41 , where @xmath42 , @xmath43 , and @xmath44 are constants , @xmath45 is the kronecker delta , @xmath46 , and @xmath20 is the scalar curvature of @xmath5 .
such @xmath47 are called * asymptotically flat coordinates*. several other inequivalent definitions of asymptotic flatness appear in the literature .
next , we recall the definition of the adm mass @xcite , a number associated to any asymptotically flat manifold , which provides some measure of the rate at which the metric becomes flat near infinity .
the * adm mass * of an asymptotically flat manifold @xmath0 is the number @xmath48 where @xmath47 are asymptotically flat coordinates , @xmath49 is the coordinate sphere @xmath50 , and @xmath51 is the area of the unit sphere in @xmath52 . other conventions for the normalizing constant appear in the literature .
the fundamental work of bartnik establishes that the limit exists , is independent of the choice of asymptotically flat coordinates , and is therefore a geometric invariant of @xmath0 @xcite .
for instance , let @xmath53 , and consider the following metric on @xmath52 minus the open ball of radius @xmath54 about the origin , equipped with the metric @xmath55 where @xmath56 is the standard flat metric .
@xmath57 is called the _ schwarzschild metric of mass @xmath58 _ and is asymptotically flat with adm mass equal to @xmath58 . to define the harmonic conformal class , we slightly abuse our previous terminology by requiring the harmonic functions @xmath59 to approach one at infinity .
( allowing @xmath59 to approach any positive constant at infinity only introduces constant rescalings of the metric . ) the * harmonic conformal class * of an asymptotically flat metric @xmath5 on a manifold @xmath3 of dimension @xmath32 is the set of riemannian metrics @xmath10_h
= \left\ { u^{\frac{4}{n-2 } } g : u > 0 \text { is smooth , } \delta u = 0 ,
\text { and } u \to 1 \text { at infinity}\right\}.\ ] ] for example , the schwarzschild metric of mass @xmath58 and the flat metric on @xmath52 minus the ball @xmath60 belong to the same harmonic conformal class . in the case
that @xmath0 is a complete asymptotically flat manifold _ without _ boundary , @xmath2_h$ ] consists of the single element @xmath5 .
( the maximum principle and the boundary condition @xmath61 forces @xmath62 . ) throughout the rest of this paper , @xmath0 is an asymptotically flat manifold of dimension @xmath1 with compact , smooth , nonempty boundary @xmath63 . to be explicit , in the above definition
we require that @xmath59 extends smoothly to the boundary .
it is straightforward to check that every metric in @xmath2_h$ ] is asymptotically flat , using the existence of an expansion of @xmath59 into spherical harmonics near infinity @xcite . to conclude this section
, we remark that @xmath0 has a unique poisson kernel , namely a smooth , positive function @xmath64 , where @xmath65 , and @xmath66 , that is harmonic with respect to @xmath5 in the @xmath67-variable and satisfies the following property : if @xmath68 is a smooth , harmonic function on @xmath3 that approaches zero at infinity , then @xmath69 where @xmath70 is the hypersurface measure on @xmath71 induced by @xmath5 .
the dirichlet problem for laplace s equation can be uniquely solved by prescribing continuous boundary data on @xmath71 and a constant at infinity .
the existence of the poisson kernel follows from the existence of a green s function , which in turn follows from asymptotic flatness . to simplify notation later ,
we define the following constants : @xmath72
in this section , we construct two _ numerical _ invariants of the harmonic conformal class of a fixed asymptotically flat manifold @xmath0 with compact boundary @xmath71 .
the motivation for these natural invariants will come later in the paper .
first , recall that the _ capacity _ of @xmath71 in @xmath0 , denoted @xmath73 , is the coefficient @xmath42 in the spherical harmonic expansion @xmath74 of the unique harmonic function @xmath75 that vanishes on @xmath71 and approaches one at infinity .
an explicit formula for the capacity is @xmath76 where the last equality defines the notation @xmath77 . here
, @xmath78 is the unit normal to the indicated surface pointing toward infinity , @xmath79 is the directional derivative , and @xmath70 is the area form on the indicated surface , all with respect to @xmath5 .
since @xmath75 is harmonic , the divergence theorem shows @xmath80 for all coordinate spheres @xmath49 .
we follow the convention that the normal @xmath78 to @xmath71 also points toward infinity ( into the manifold ) . by the maximum principle , @xmath81 on @xmath71 or @xmath49 .
[ lemma_i1 ] if @xmath0 is asymptotically flat , then the number @xmath82 is an invariant of the harmonic conformal class @xmath2_h$ ] .
that is , if @xmath83_h$ ] , then @xmath84 note that @xmath24 , the difference of adm mass and twice the capacity , can be positive , negative , or zero . from the definition of adm mass and asymptotic flatness
, it readily follows that the adm masses of @xmath85 and @xmath5 ( where @xmath86 ) are related by @xmath87 the fact that the adm mass minus the capacity is an invariant of @xmath2_h$ ] then follows from the formula @xmath88 which we now prove .
let @xmath75 and @xmath89 be the harmonic functions ( with respect to @xmath5 and @xmath17 , respectively ) that vanish on @xmath71 and approach one at infinity . using formula ( [ eqn_conf_laplacian ] ) , one can check that @xmath90 is harmonic with respect to @xmath17 , is zero on @xmath71 , and approaches one at infinity .
therefore by uniqueness , @xmath91 , so @xmath92 where @xmath93 and @xmath94 are hypersurface measure and the unit normal with respect to @xmath17 . since @xmath61 and @xmath95 at infinity
, we have @xmath96 proving ( [ eqn_conf_capacity ] ) .
[ lemma_i2 ] suppose @xmath0 is asymptotically flat , of dimension @xmath1 .
let @xmath75 be the harmonic function ( with respect to @xmath5 ) that vanishes on @xmath71 and approaches one at infinity .
then the number @xmath97 is an invariant of the harmonic conformal class of @xmath5 .
that is , if @xmath83_h$ ] with hypersurface measure @xmath93 , unit normal @xmath94 to the boundary , and harmonic function @xmath89 that vanishes on @xmath71 and approaches one at infinity , then @xmath98 suppose @xmath99 belongs to @xmath2_h$ ] .
as explained in the proof of lemma [ lemma_i1 ] , we have the equality @xmath91 .
since lengths with respect to @xmath17 and @xmath5 are related by a factor of @xmath100 pointwise , we see @xmath101 and using the fact that @xmath75 vanishes on @xmath71 , @xmath102 next , the hypersurface measures are related by @xmath103 , and it readily follows that @xmath104 as measures on @xmath71 . in particular , the integrals over @xmath71 of these measures agree .
in this section and the next , we fix @xmath0 as above and consider some problems that involve minimizing or maximizing certain geometric quantities within the harmonic conformal class of @xmath5 . working with asymptotically flat manifolds ,
it is natural to consider the adm mass as a geometric quantity to be optimized .
recall that if @xmath85 belongs to @xmath2_h$ ] , then formula ( [ eqn_conf_mass_infinity ] ) relates the adm masses of @xmath17 and @xmath5 .
using the fact that @xmath59 is harmonic with respect to @xmath5 , we also see @xmath105 for any coordinate sphere @xmath49 .
the last term , including the minus sign , can be interpreted as twice the coefficient @xmath106 in the expansion of @xmath59 into spherical harmonics for @xmath107 large : @xmath108 from ( [ eqn_conf_mass ] ) , it is not difficult to see that the adm mass can be made arbitrarily large for metrics @xmath109 in @xmath2_h$ ] by choosing a harmonic conformal factor @xmath59 that is large on @xmath71 . for such @xmath59 , @xmath71 clearly has large @xmath17-area .
this motivates the question : how large can the adm mass be made among metrics in @xmath2_h$ ] that have a fixed upper bound @xmath30 on the area of the boundary @xmath71 ?
let @xmath110 denote the area ( hypersurface measure ) of @xmath71 with respect to a metric @xmath5 .
we have the following definition . [ def_mu ]
given a number @xmath111 , define @xmath112_h } \{m_{adm}({\overline}g ) : \ ; |\sigma|_{{\overline}g } \leq a\}.\ ] ] the number @xmath113 is the largest possible value of the adm mass among metrics in @xmath2_h$ ] that measure the boundary area to be at most @xmath30 .
below we will see that @xmath113 is finite for each value of @xmath30 , so in particular @xmath27 is well - defined as a function @xmath114 .
but first we make the observation : the function @xmath26 is independent of the choice of metric in @xmath2_h$ ] .
so we say that @xmath27 is an invariant of the harmonic conformal class of @xmath5 and call @xmath27 the _ mass profile function _ of @xmath2_h$ ] .
the proof of the lemma is trivial : @xmath27 is formed by maximizing a geometric quantity ( adm mass ) subject to a geometric constraint ( upper bound for area ) over the whole harmonic conformal class . before moving on
, we remark that @xmath27 may be interpreted purely in terms of the behavior of harmonic functions on @xmath3 : @xmath115 where @xmath116 .
that is , @xmath113 is essentially found by maximizing the coefficient of the @xmath117 term in the expansion at infinity of positive harmonic functions @xmath59 with an @xmath118 upper bound for @xmath119 .
however , from this viewpoint it is not transparent that @xmath27 is an invariant of @xmath2_h$ ] . in the following theorem
we give a complete understanding of @xmath113 by proving an explicit formula in terms of the numerical invariants @xmath24 and @xmath25 introduced above .
in the course of the proof , we show that given @xmath30 , there exists a unique metric in @xmath2_h$ ] attaining the supremum for @xmath113 .
[ thm_mu ] for @xmath111 , we have the formula @xmath120 in particular , @xmath121 is a smooth , increasing function that is bounded below and unbounded above .
moreover , the function @xmath27 is completely determined by the numerical invariants @xmath24 and @xmath25 defined in lemmas [ lemma_i1 ] and [ lemma_i2 ] .
fix @xmath111 , and suppose initially that @xmath122 attains the supremum in the definition of @xmath113 and satisfies the area bound @xmath123 we claim that @xmath124 . for ,
if @xmath125 , then by adding a small constant to boundary data for @xmath126 , we could construct a metric in @xmath2_h$ ] with boundary area equal to @xmath30 . by the maximum principle and formula ( [ eqn_conf_mass_infinity ] ) , the adm mass of this new metric would exceed that of @xmath17 , so that the latter could not attain the supremum for @xmath113 .
so @xmath124 .
we show that @xmath126 satisfies a variational principle . for @xmath127 ,
let @xmath128 be a smoothly - varying path in the space of positive harmonic functions on @xmath3 , passing through @xmath126 at @xmath129 , such that a. for each @xmath130 , @xmath128 approaches one at infinity , and b. @xmath131 for all @xmath130 .
this is equivalent to stating that @xmath132 is a smooth path of metrics in @xmath2_h$ ] that fixes the boundary area at the value @xmath30 . since @xmath133 maximizes @xmath113 among metrics with boundary area @xmath30 , the smooth function @xmath134 has a local maximum at @xmath129 .
using formula ( [ eqn_conf_mass ] ) , we have @xmath135 where @xmath136 also , since the boundary area is constant in @xmath130 , @xmath137 observe that @xmath138 is a smooth harmonic function on @xmath3 that approaches zero at infinity .
given any such @xmath138 satisfying @xmath139 , it is possible to construct a smooth family of harmonic functions @xmath128 with @xmath140 satisfying properties ( i ) and ( ii ) above .
we now see that the unknown harmonic function @xmath126 ( if it exists ) satisfies the statement : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ if @xmath138 is any harmonic function on @xmath3 that approaches zero at infinity , and if @xmath141 , then @xmath142 as well . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ to make this more concrete , lemma [ lemma_compute_term ] below shows how to compute @xmath143 solely from the boundary data for @xmath138 : there exists a smooth , positive function @xmath144 on @xmath71 such that for all harmonic functions @xmath138 on @xmath3 that approach zero at infinity , we have @xmath145 writing @xmath146 we see that @xmath126 satisfies the property : @xmath147 the trick is to utilize this observation to determine what @xmath126 should be : we will _ define _ @xmath126 to be harmonic , one at infinity with boundary data given by a constant times @xmath148 , so that @xmath126 satisfies ( [ eqn_critical_property ] ) automatically .
specifically , let @xmath149 be the function on @xmath71 given by : @xmath150 and let @xmath126 be harmonic , one at infinity , with boundary data @xmath149 .
then the metric @xmath151_h$ ] has boundary area equal to @xmath30 and is , by the above computations , a critical point for the adm mass among metrics in @xmath2_h$ ] with boundary area equal to @xmath30 .
the next step is to show @xmath133 indeed attains the supremum for @xmath113 .
we do so by showing the adm mass satisfies a concavity property on paths of metrics in @xmath2_h$ ] that fix the boundary area .
let @xmath152 be any smooth , positive function on @xmath71 distinct from @xmath149 that serves as boundary data for a harmonic function @xmath153 that approaches one at infinity .
assume @xmath154 . to consider a path between @xmath149 and @xmath152 that fixes the @xmath155 norm , we define for @xmath156 $ ] : @xmath157.\ ]
] let @xmath128 be the harmonic function , one at infinity , with boundary data @xmath158 , so @xmath132 is a path in @xmath2_h$ ] for @xmath159 $ ] that has boundary area @xmath30 for all @xmath130 . by convexity of the @xmath155 norm
, we have @xmath160 then certainly @xmath161 applying lemma [ lemma_compute_term ] to each of these three terms , adding @xmath162 to both sides , and using formula ( [ eqn_conf_mass_infinity ] ) , we have @xmath163 note that the inequality is strict for @xmath164 ( because ( [ eqn_convexity ] ) is ) and equality holds at the endpoints . if @xmath133 were not a maximizer for @xmath113 , then there exists @xmath165_h$ ] with boundary area @xmath30 and @xmath166 differentiating ( [ eqn_concavity ] ) at @xmath129 we see that @xmath167 contradicting the fact that @xmath133 is a critical point for the adm mass among metrics in @xmath2_h$ ] with the same boundary area .
this shows that @xmath133 is indeed a maximizer for @xmath113 , and ( [ eqn_concavity ] ) also shows that the maximizer is unique . finally , we compute @xmath113 as the adm mass of @xmath168 and simplify . using formula ( [ eqn_conf_mass_infinity ] ) again , as well as lemma [ lemma_compute_term ] , @xmath169 since @xmath170 is harmonic , approaching zero at infinity .
the integral @xmath171 can be simplified based on our definition of @xmath172 .
it is straightforward to check that @xmath173 also , applying lemma [ lemma_compute_term ] to the harmonic function @xmath174 that is one on @xmath71 and zero at infinity , we see @xmath175 putting it all together , we have @xmath176 using the definition of @xmath24 and @xmath25 , as well as the fact that @xmath177 ( see the proof of lemma [ lemma_compute_term ] ) , we obtain @xmath178 [ lemma_compute_term ] suppose @xmath0 is asymptotically flat with boundary @xmath71
. then there exists a unique smooth , positive function @xmath144 on @xmath71 such that for all harmonic functions @xmath68 that approach zero at infinity , @xmath179 in other words , @xmath144 represents the linear functional on @xmath180 that maps the boundary values of any harmonic function @xmath68 vanishing at infinity to the coefficient of the @xmath117 term in the expansion of @xmath68 into spherical harmonics . note that since @xmath68 is harmonic , the integral over @xmath77
may be replaced with the integral over any coordinate sphere @xmath49 .
uniqueness is readily verified ; we proceed to derive a formula for @xmath144 .
let @xmath68 be a harmonic function with smooth boundary data that approaches zero at infinity .
let @xmath75 be harmonic , zero on @xmath71 and one at infinity .
then by harmonicity we have @xmath181 dv,\end{aligned}\ ] ] where @xmath182 and @xmath183 are the divergence , gradient , and volume measure with respect to @xmath5 . by the divergence theorem , this becomes @xmath184 using the fact that @xmath68 approaches zero at infinity and @xmath75 vanishes on @xmath71 and approaches one at infinity , we conclude @xmath185 choosing @xmath186 , the proof is complete .
after having studied the function @xmath113 , it is natural to consider the problem of _ minimizing _ the adm mass within @xmath2_h$ ] among metrics having a fixed _ lower bound _ on the boundary area : @xmath187_h } \{m_{adm}({\overline}g ) : \ ; |\sigma|_{{\overline}g } \geq a\}.\ ] ] this infimum defines a function of @xmath30 that is an invariant of @xmath2_h$ ] .
we claim that @xmath188 is the constant function equal to @xmath24 . first
, suppose @xmath189_h$ ] .
from formula ( [ eqn_conf_mass ] ) and the maximum principle , we see @xmath190 which equals @xmath24 , so @xmath191 . to prove equality ,
it is enough to find a sequence @xmath192 in @xmath2_h$ ] such that each metric in the sequence has boundary area @xmath30 and such that @xmath193 converges pointwise to @xmath75 on the interior of @xmath3 .
then by harmonicity , this would then show that @xmath194 we leave it to the reader to construct such a sequence by letting @xmath193 have boundary data @xmath195 , where @xmath196 approximates a dirac delta of measure @xmath30 as @xmath197 .
in this section we introduce a function @xmath198 whose definition formally appears similar to that of the mass profile function @xmath27 ; the idea is still to maximize a geometric quantity over @xmath2_h$ ] subject to a boundary area constraint .
we will see , however , that @xmath31 is much more subtle than @xmath27 in a number of respects .
given an asymptotically flat manifold @xmath0 , a _ surface _
@xmath199 _ enclosing _ the boundary @xmath71 is a set that differs ( in the sense of integral currents ) from @xmath71 by the boundary of a bounded region @xmath200 with @xmath201-rectifiable boundary of finite @xmath202 measure : @xmath203 here , @xmath202 is hausdorff @xmath201-measure with respect to @xmath5 ; we also use the notation @xmath204 , which we call the _ area _ of @xmath199 .
we define the _ minimal enclosing area _ of the boundary @xmath71 with respect to @xmath5 to be the number : @xmath205 from standard results in geometric measure theory , this infimum is attained by at least one surface ( this uses asymptotic flatness and the federer fleming compactness theorem for integral currents ; see @xcite for instance ) .
such a surface @xmath199 is called a _ minimal area enclosure _ of @xmath71 with respect to @xmath5 . in low dimensions
@xmath206 , @xmath199 has @xmath207 regularity , and @xmath208 , if non - empty , is a @xmath209 minimal ( zero mean curvature ) surface .
we remark that there exists a unique _ outermost _ minimal area enclosure that we denote by @xmath210 . for more details on existence , uniqueness , and regularity see section 1 of @xcite , which uses the terminology of minimizing hulls .
the minimal enclosing area of @xmath71 is a geometric quantity , and we consider the problem of optimizing it within the harmonic conformal class . like the adm mass ,
the number @xmath211 takes on arbitrarily large values within @xmath2_h$ ] but this is not true if we restrict to metrics in @xmath2_h$ ] with an upper bound for the boundary area .
this motivates the preliminary definition : for @xmath111 , let @xmath212_h } \{\min(\sigma , { \overline}g ) : \
; |\sigma|_{{\overline}g } \leq a\}.\ ] ] in words , @xmath213 is the maximum possible value of the minimal enclosing area for metrics in @xmath2_h$ ] that have boundary area at most @xmath30 . in sections [ sec_properties ] and [ sec_applications ]
we will give more geometric motivation for why optimizing the minimal area enclosure is of interest .
it is clear that @xmath214 defines a function @xmath215 that is nondecreasing and satisfies @xmath216 .
( to see the last point , note that if the boundary area is at most @xmath30 , then the least area needed to enclose the boundary is certainly no more than @xmath30 . )
what is far from clear is that the supremum for @xmath213 is attained .
a few essential differences between the functions @xmath214 and @xmath27 now come to the surface : a. suppose @xmath217_h$ ] attains the supremum for @xmath213 . consider a smooth path @xmath218 in @xmath2_h$ ] passing through @xmath17 at @xmath129 and preserving the condition that the boundary area is at most @xmath30 . in general , particularly in the case in which there exist multiple minimal area enclosures of @xmath71 , it is entirely possible that the function @xmath219 is not differentiable at @xmath129 .
in other words , a maximizing metric @xmath17 does not obviously satisfy a variational property .
b. even in the case in which @xmath220 , the resulting variational statement is not particularly useful : it leads to a statement on the behavior of harmonic functions restricted to the surface @xmath221 , rather than the boundary .
moreover , it seems highly unlikely that a formula for @xmath213 could be given in terms of the numerical invariants @xmath24 and @xmath25 .
this all contrasts sharply with the case of @xmath113 ( c.f .
theorem [ thm_mu ] ) .
c. concavity : in the proof of theorem [ thm_mu ] , we saw the adm mass satisfies a concavity property that allowed us to show that a critical point was necessarily a global maximum .
the minimal enclosing area evidently satisfies no such property , essentially for the reason that the area of a surface @xmath199 with respect to @xmath222 is a _
function of @xmath59 . despite these difficulties
, we would still like to maximize the minimal enclosing area within @xmath2_h$ ] in the same spirit as in ( [ eqn_alpha_0 ] ) . to carry this out
, we will enlarge the space @xmath2_h$ ] so as to obtain a space with a useful compactness property , a step that was unnecessary for @xmath113 . for the purposes of maximizing @xmath223 as in equation ( [ eqn_alpha_0 ] ) , we enlarge the set @xmath2_h$ ] as follows by allowing metrics with weaker boundary regularity .
let @xmath224 belong to @xmath225 ( with respect to the hypersurface measure @xmath70 induced by @xmath5 ) .
the reason for considering @xmath155 with @xmath116 is that for smooth conformal metrics @xmath226 , the area measures on hypersurfaces are related by @xmath227 for @xmath67 in the interior of @xmath3 , define @xmath228 where @xmath64 is the poisson kernel for @xmath0 ( where @xmath229 , @xmath230 and @xmath66 ) , and @xmath231 is the unique @xmath5-harmonic function that vanishes on @xmath71 and approaches one at infinity ( c.f .
section [ sec_definitions ] ) .
in particular , @xmath59 is @xmath5-harmonic in the interior of @xmath3 and tends to one at infinity .
we call @xmath59 the _ harmonic function associated to _ @xmath232 .
since @xmath232 is determined uniquely by @xmath59 ( up to almost - everywhere equivalence ) , we also say @xmath232 is the function in @xmath225 _ associated to _
@xmath59 .
[ remarks_harmonic ] the function @xmath59 defined in ( [ eqn_assoc_harmonic ] ) is smooth and positive in @xmath233 , but need not extend continuously to @xmath71 .
also , while it is not clear that the trace of @xmath59 onto @xmath71 is defined ( in the sense of sobolev spaces ) , it is the case that for almost all @xmath230 , given a path @xmath234 with @xmath235 and @xmath236 transverse to @xmath71 , @xmath237 converges to @xmath238 as @xmath239 .
this is true essentially because for @xmath240 , the function @xmath241 forms an approximation to the identity on @xmath71 , based at @xmath242 , in the limit @xmath243 . on @xmath244
, @xmath222 is a smooth riemannian metric , and we make the following definition .
[ def_ghcc ] the * generalized harmonic conformal class * of @xmath5 is the set @xmath245_h}$ ] of all riemannian metrics @xmath222 on @xmath244 , where @xmath59 is the harmonic function associated to some nonnegative @xmath246 as in ( [ eqn_assoc_harmonic ] ) .
we may think of @xmath245_h}$ ] as the closure of @xmath2_h$ ] with respect to the @xmath225 norm .
we remark that both of the sets @xmath245_h}$ ] and @xmath225 are unchanged if @xmath5 is replaced by some other metric in @xmath2_h$ ] , and that @xmath245_h}$ ] is ( non - canonically ) bijective to the set of nonnegative functions in @xmath225 via ( [ eqn_assoc_harmonic ] ) .
sometimes for emphasis we will refer to @xmath2_h$ ] as the _ smooth _ harmonic conformal class of @xmath5 .
we also point out that every metric in @xmath245_h}$ ] is asymptotically flat as in definition [ def_af ] , modulo the implicit assumption of smoothness up to the boundary .
to define the area of any surface @xmath199 enclosing @xmath71 with respect to @xmath247_h}$ ] , we decompose @xmath199 into the pieces @xmath248 and @xmath208 : @xmath249 where @xmath232 is the function in @xmath225 associated to @xmath59 , and @xmath202 is hausdorff @xmath201-measure on @xmath3 with respect to @xmath5 . equation ( [ eqn_def_area ] ) is well - defined ( dependent only on @xmath17 and not on @xmath5 ) and reproduces the usual notion of area in the case that @xmath17 is smooth .
using the generalized harmonic conformal class , we now define the area profile function .
[ def_alpha ] given numbers @xmath111 and @xmath250 , define @xmath251_h } } \left\{\min(\sigma , { \overline}g ) : \ ; |\sigma|_{{\overline}g } \leq a \text { and } u \leq c\right\},\ ] ] and let @xmath252 we call @xmath29 the * area profile function*. in this definition , we use ( [ eqn_def_area ] ) to define area with respect to metrics @xmath17 and define the minimal enclosing area @xmath223 exactly as in ( [ eqn_min_enclosing_area ] ) .
the basic idea here is to first maximize @xmath253 with an area upper bound @xmath30 and requiring the conformal factors to be bounded by @xmath254 , then let @xmath254 go to infinity .
observe that the limit defining @xmath29 exists , since for fixed @xmath30 , the quantity @xmath255 viewed as a function of @xmath254 is bounded above by @xmath30 ( by the definition of minimal enclosing area ) and non - decreasing ( by definition of supremum ) .
we require @xmath256 since the function @xmath59 must approach the value of 1 at infinity .
we remark that it is possible to define an area profile function by replacing @xmath2_h$ ] with @xmath245_h}$ ] in ( [ eqn_alpha_0 ] ) , obtaining a harmonic conformal invariant , but we opt not to do so working directly with conformal metrics with merely @xmath155 ( rather than @xmath257 ) boundary data is difficult , and our definition of @xmath29 allows us to bypass this technical detail .
evidently @xmath255 is not an invariant of @xmath2_h$ ] , since the pointwise upper bound @xmath258 is not a geometric statement .
however , invariance is restored by taking the limit @xmath259 .
the function @xmath198 depends only on the harmonic conformal class @xmath2_h$ ] .
let @xmath260 be two metrics in the same harmonic conformal class , determining functions @xmath261 and @xmath262 , respectively , as in definition [ def_alpha ] .
say @xmath263 , where @xmath264 is smooth , positive , and harmonic with respect to @xmath265 , approaching one at infinity . there exist positive constants @xmath266 such that @xmath267 on @xmath3 , since @xmath4 is compact and @xmath268 at infinity .
let @xmath269 be a valid test metric in @xmath270_h}$ ] for @xmath271 , meaning @xmath272 then the same metric @xmath17 can be written as @xmath273 , an element of @xmath274_h}$ ] ( by formula ( [ eqn_conf_laplacian ] ) ) that satisfies @xmath275 therefore @xmath17 is a valid test metric for @xmath276 .
since @xmath265 and @xmath277 determine the same generalized harmonic conformal class , we see that the set of test metrics for @xmath261 is a subset of the set of test metrics for @xmath276 . in particular ,
@xmath278 taking the limit @xmath259 , then applying the same argument with the roles of @xmath265 and @xmath277 ( and @xmath106 and @xmath279 ) swapped , we see @xmath280 proving that @xmath29 depends only on the harmonic conformal class . our first goal is to show that for given values of @xmath30 and @xmath254 , the supremum in the definition of @xmath255 is attained . this fact is the advantage of considering the enlarged space @xmath245_h}$ ] over @xmath2_h$ ] our sole motivation for introducing the generalized harmonic conformal class was to prove the existence of a maximizer . [ thm_maximizer ] given @xmath281 , there exists @xmath282_h}$ ] satisfying @xmath283 that attains the supremum in definition [ def_alpha ] : @xmath284 moreover , @xmath285 , provided @xmath254 is sufficiently large ( @xmath286 ) . from now on , such @xmath17
will be called a _
maximizer for _ @xmath255 without further comment . unlike the case of @xmath113 ,
we make no claim that a maximizer for @xmath255 is unique .
fix @xmath281 .
let @xmath287 be a maximizing sequence for @xmath255 in @xmath245_h}$ ] .
that is , assume @xmath288 let @xmath195 be the function on @xmath71 associated to @xmath193 , so @xmath289 almost - everywhere on @xmath71 .
since @xmath290 the sequence @xmath291 is bounded in @xmath225 , and thus has a weakly convergent subsequence ( of the same name , say ) with limit @xmath246 ( by the banach
alaoglu theorem @xcite ) .
recall this means that for all continuous functions @xmath15 on @xmath71 , @xmath292 redefining on a set of measure zero , we may assume that @xmath293 .
let @xmath59 be the harmonic function associated to @xmath232 , and let @xmath99 , an element of @xmath245_h}$ ] . by the maximum principle , @xmath258 .
since the @xmath155 norm is lower semi - continuous with respect to weak convergence , we have that @xmath294 , so that @xmath295 .
in other words , @xmath17 is a valid test metric for @xmath255 .
we claim that @xmath17 is a maximizer for @xmath255 , i.e. , @xmath17 has minimal enclosing area equal to @xmath255 .
let @xmath199 be a surface enclosing @xmath71 that is disjoint from @xmath71 . from the definition of the minimal enclosing area
, we have that for all @xmath296 , @xmath297 the left hand side converges to @xmath255 by assumption .
we also see from ( [ eqn_assoc_harmonic ] ) , the fact that the poisson kernel @xmath298 is continuous for @xmath299 , and the definition of weak convergence ( [ eqn_weak_convergence ] ) that @xmath300 pointwise in the interior of @xmath3 as @xmath197 . by harmonicity , this convergence is uniform on compact sets disjoint from @xmath71 .
the surface @xmath199 is such a set , so @xmath301 now , taking the limit @xmath197 of ( [ eqn_min_test ] ) , we obtain for all @xmath199 disjoint from @xmath71 : @xmath302 lemma [ lemma_push_out ] below shows that to compute the minimal enclosing area , it is sufficient to consider only surfaces @xmath199 disjoint from the boundary .
it then follows that @xmath303 from the definition of @xmath255 , the reverse inequality holds as well , so @xmath17 is the desired maximizer for @xmath255 . finally , we argue that we may assume @xmath17 has boundary area equal to @xmath30 , for @xmath254 large enough .
if not , suppose that @xmath304 these inequalities preclude the possibility that @xmath305 almost - everywhere .
for @xmath156 $ ] , consider the family of functions @xmath306 that interpolate between @xmath232 and @xmath254 .
note that @xmath307 for each @xmath130 . by the intermediate value theorem and ( [ eqn_area_c ] )
, there exists @xmath308 such that @xmath309 let @xmath310 be the harmonic function associated to @xmath311 . by construction
, @xmath312 is a valid test metric for @xmath255 . by the maximum principle , since @xmath313 , we have that @xmath314 pointwise
. then we have an inequality for the minimal enclosing areas : @xmath315 the left - hand side is at most @xmath255 , and the right - hand side was already shown to equal @xmath255 .
it follows that @xmath316 is a maximizer for @xmath255 , and moreover its boundary area equals @xmath30 .
now we prove a lemma used in the above construction of a maximizer for @xmath255 .
[ lemma_push_out ] for the purposes of computing the minimal enclosing area ( [ eqn_min_enclosing_area ] ) with respect to @xmath247_h}$ ] , with @xmath258 , it is sufficient to consider only surfaces @xmath199 that are disjoint from the boundary
. let @xmath199 be any surface enclosing the boundary .
let @xmath317 be a smooth , compactly supported vector field on @xmath3 such that @xmath318 equals @xmath78 , the unit normal vector field to @xmath71 pointing into @xmath3 .
for @xmath319 , let @xmath320 be the flow generated by @xmath317 ; note that @xmath321 is a diffeomorphism onto its image , is the identity map outside a compact set , and maps @xmath322 into the interior of @xmath3 for @xmath240 .
in particular , for @xmath240 , @xmath323 is a surface enclosing @xmath71 that is disjoint from the boundary .
to prove the lemma , we need only show that the area of @xmath323 with respect to @xmath17 varies continuously in @xmath130 .
observe that @xmath324 , since @xmath258 and @xmath325 is finite by our definition of surface .
we decompose @xmath199 into the disjoint , @xmath202-measurable sets @xmath248 and @xmath208 .
first , consider @xmath248 .
reparametrizing the integral , @xmath326 where we have formed the pullback measure @xmath327 on @xmath71 of the measure @xmath202 on @xmath328 : @xmath329 for @xmath330 measurable . since @xmath321 is a diffeomorphism , @xmath327 is absolutely continuous with respect to @xmath202 , and we may write @xmath331 for a measurable function @xmath332 on @xmath71 that converges uniformly to 1 as @xmath243 ( since @xmath333 identity smoothly as @xmath243 ) . next , @xmath334 converges pointwise almost - everywhere to @xmath232 ( see the remarks preceding definition [ def_ghcc ] ) . since @xmath258 , the dominated convergence theorem allows us to evaluate the limit : @xmath335 the left - hand side is @xmath336 by ( [ eqn_reparam_integral ] ) ; the right - hand side is @xmath337 .
the proof for @xmath208 is essentially the same : on @xmath208 , @xmath338 converges pointwise to @xmath59 and is dominated by the integrable function @xmath254 .
we have established that the maximum for @xmath255 is attained , but we reiterate the point that the maximizer does not seem to satisfy a variational principle that would allow us to automatically deduce regularity of this maximizer ( see points ( i)(iii ) near the beginning of section [ sec_area_profile ] ) . in the next section , we study maximizers in a regular case . for now , we close this section by showing some nice properties satisfied by the function @xmath29 .
@xmath339 [ prop_alpha_continuous ] a. @xmath198 is nondecreasing , lipschitz continuous , satisfying for @xmath111 .
b. there exists @xmath111 such that @xmath340 .
the second statement rules out the possibility that @xmath31 is the identity function @xmath215 .
@xmath341 + _ ( i ) _ the bounds @xmath342 follow immediately from the definitions of @xmath343 and @xmath31 , as does the fact that @xmath31 is nondecreasing .
( one can show that @xmath31 is strictly increasing , but we do not carry this out here . )
let @xmath344 , and set @xmath345 , a number in @xmath346 .
fix a constant @xmath254 so that @xmath347 . by theorem [ thm_maximizer ] ,
there exists a maximizer @xmath348_h}$ ] for @xmath349 such that @xmath350 and @xmath351 .
let @xmath352 be the function on @xmath71 associated to @xmath353 , and note that @xmath354 by our choice of @xmath355 .
let @xmath153 be the harmonic function associated to @xmath356 , and let @xmath357 .
in particular , @xmath358 measures the boundary area to be @xmath359 , and @xmath360 ( where the first inequality follows from the maximum principle ) .
then @xmath358 is a valid test metric for @xmath361 , so @xmath362 by definition .
next , for @xmath67 in the interior of @xmath3 , by ( [ eqn_assoc_harmonic ] ) @xmath363 since @xmath364 and @xmath365 .
in particular , the minimal enclosing area for @xmath366 is at least @xmath367 times that for @xmath368 : @xmath369 but we chose @xmath368 to have minimal enclosing area equal to @xmath370 . putting our inequalities together ,
we have @xmath371 taking @xmath372 of both sides and rearranging , we have @xmath373 for all @xmath374 .
it follows that the function @xmath375 is non - increasing .
combined with the fact that @xmath29 is non - decreasing and at most equal to @xmath30 , one can readily check that @xmath31 is lipschitz continuous with lipschitz constant at most 1 . _
( ii ) _ given @xmath376 , we will produce a number @xmath377 and a surface @xmath199 enclosing @xmath71 that has the property that @xmath378 in particular , letting @xmath17 be a maximizer for @xmath255 , then letting @xmath259 , we deduce to complete the proof , we proceed to establish ( [ eqn_suff_cond ] ) . without loss of generality ( by rescaling ) ,
assume @xmath379 .
observe that a harmonic function @xmath59 that is one at infinity with @xmath118 boundary data on @xmath71 can be uniquely written as @xmath380 where @xmath381 is a parameter , @xmath264 is a @xmath5-harmonic function on @xmath3 tending to zero at infinity with @xmath382 boundary data of @xmath118 norm equal to one . as usual , @xmath75 is harmonic , zero on @xmath71 and one at infinity .
we will also use the letter @xmath264 to denote the boundary data for @xmath264 .
the point is that any metric in the generalized harmonic conformal class @xmath245_h}$ ] with boundary area @xmath30 can be uniquely written in the form @xmath383 for some @xmath264 as above and @xmath384 .
for now , take @xmath381 and @xmath264 with @xmath385 to be arbitrary .
the poisson kernel @xmath64 is harmonic as a function of @xmath67 , approaching zero at infinity . identifying @xmath67 with
an asymptotically flat coordinate chart , @xmath64 is @xmath386 in @xmath67 for large @xmath387 .
( this decay is independent of @xmath230 , since @xmath71 is compact . )
then any harmonic function @xmath264 as above satisfies : @xmath388 where @xmath42 is a constant depending only on @xmath0 but not on @xmath264 .
let @xmath376 be given .
the @xmath5-area of a coordinate sphere @xmath49 in @xmath3 is asymptotic to @xmath389 and is therefore less than @xmath390 for @xmath391 sufficiently large . in particular ,
the quantity @xmath392 can be made less than @xmath393 by choosing @xmath391 sufficiently large , _ independently of _ @xmath264 .
fix such a value of @xmath391 , and let @xmath394 . by construction ,
the area of @xmath71 with respect to any @xmath395 as above is @xmath396 .
let us compute the area of @xmath199 in the metric @xmath395 : @xmath397 factoring out @xmath396 , applying the minkowski inequality , and using @xmath398 we have @xmath399 then for some @xmath400 sufficiently large , @xmath401 for all choices of @xmath264 .
since every metric @xmath402_h}$ ] can be written as @xmath395 for some @xmath400 and @xmath264 , we have shown ( [ eqn_suff_cond ] ) with @xmath403 , completing the proof .
[ cor_alpha_a ] suppose that @xmath340 for some value of @xmath30 .
then @xmath404 for all @xmath405 .
moreover , @xmath406 both statements follow from the proof of proposition [ prop_alpha_continuous ] . for the first , we showed that @xmath407 is non - increasing as a function of @xmath30 .
for the second , we also showed that given @xmath408 , there exists @xmath111 large so that @xmath409 .
continuing the previous section , suppose that @xmath410 and that @xmath411_h$ ] is a maximizer for @xmath255 as in theorem [ thm_maximizer ] . at this point , we know only that the boundary data @xmath232 for the conformal factor @xmath59 is a nonnegative function in @xmath412 . in @xcite
, we worked directly with the poor regularity to prove results on the geometry of @xmath17 under some additional technical hypotheses , and with a slightly different definition of @xmath29 .
rather than pursue this approach here , our present purpose is to prove a result regarding @xmath17 in the case where the maximizing metric is _ a priori _ assumed to be smooth .
this suggests a heuristic for what we expect to occur in general ( c.f .
conjecture [ conj_rpi ] ) .
the main idea of the following theorem is that if @xmath17 is regular , then the outermost minimal area enclosure touches the boundary only on a set of small measure and has uniformly small mean curvature with respect to @xmath17 .
let @xmath0 be asymptotically flat of dimension @xmath206 with compact boundary @xmath71 .
suppose @xmath410 and @xmath413 .
let @xmath99 be a maximizer for @xmath255 given by theorem [ thm_maximizer ] , and assume that @xmath59 is smooth and positive on @xmath71 .
let @xmath414 be the outermost minimal area enclosure of @xmath71 with respect to @xmath17 .
then a. @xmath415 ( where @xmath202 is hausdorff @xmath201-measure on @xmath416 ) ) , and b. the mean curvature @xmath417 of @xmath418 with respect to @xmath17 is bounded ( pointwise almost - everywhere ) between @xmath419 and @xmath420 , for some constant @xmath421 depending only on @xmath5 .
[ thm_mean_curv ] note that @xmath414 exists because @xmath59 , and therefore @xmath17 , is smooth by assumption ( see the beginning of section [ sec_area_profile ] ) .
moreover , @xmath414 has nonnegative mean curvature with respect to @xmath17 , or else an outward variation would produce a surface of less @xmath17-area .
since @xmath422 is a minimal surface for @xmath17 , the theorem states that @xmath414 has uniformly small mean curvature ( depending on @xmath254 ) .
the significance of @xmath340 is that it prevents @xmath71 from being its own outermost minimal area enclosure : @xmath71 has area @xmath30 and @xmath414 has area @xmath423 . in the case that @xmath418 is empty or has zero @xmath202-measure , we are done .
assume otherwise ; we aim to show that @xmath424 supposing that ( [ eqn_u_equiv_c ] ) holds , we complete the proof .
first , @xmath425 but by ( [ eqn_u_equiv_c]),this reduces to @xmath426 , proving ( i ) . recall that @xmath414 is a @xmath207 surface and has mean curvature @xmath417 ( with respect to @xmath17 ) defined almost - everywhere and nonnegative .
the set @xmath422 is a smooth hypersurface with @xmath427 .
( these regularity assertions require @xmath428 . ) on the set @xmath418 , @xmath417 agrees with the mean curvature @xmath429 of @xmath71 with respect to @xmath17 ( see ( 1.15 ) of @xcite ) .
then for almost all @xmath430 , @xmath431 having used the law for the transformation of mean curvature under conformal changes . here
@xmath432 is the mean curvature of @xmath71 with respect to @xmath5 . using ( [ eqn_u_equiv_c ] ) , we have @xmath433 ; moreover , by the maximum principle , @xmath434 , since @xmath59 attains its global maximum value of @xmath254 at @xmath242 .
let @xmath435 , so that @xmath436 for almost all @xmath430 , proving ( ii ) . to complete the proof
, we use the failure of ( [ eqn_u_equiv_c ] ) to construct a valid `` variation '' within @xmath245_h}$ ] that increases the minimal enclosing area , contradicting the assumption that @xmath17 is a maximizer for @xmath255 .
if ( [ eqn_u_equiv_c ] ) fails , there exists @xmath437 of positive @xmath202-measure and a constant @xmath438 such that @xmath439 let @xmath440 be the characteristic function of @xmath441 . for @xmath442 , consider the following family of functions on @xmath71 : @xmath443 where @xmath106 is the constant @xmath444 .
for all @xmath130 sufficiently small , say @xmath445 , @xmath158 is positive and bounded above by @xmath254 ( using the definition of @xmath441 and @xmath446 ) .
we consider @xmath158 as the boundary data for harmonic functions @xmath128 that approach one at infinity .
a formula for @xmath128 is @xmath447 where @xmath68 is harmonic , zero at infinity , with boundary data @xmath448 on @xmath71 .
then @xmath449 is a smooth path of metrics in @xmath245_h}$ ] passing through @xmath17 at @xmath129 .
these metrics are valid test metrics for @xmath255 , modulo the fact that the boundary area @xmath450 of @xmath71 is not necessarily @xmath451 ; we address this issue later . for now , observe that @xmath452 is stationary to first order at @xmath129 by our choice of the constant @xmath106 : @xmath453 we proceed to show that the minimal enclosing area @xmath454 is increasing near @xmath129 .
let @xmath199 be any minimal area enclosure of @xmath71 with respect to @xmath17 ( possibly @xmath414 itself ) .
since @xmath414 is the _ outermost _ minimal area enclosure , we see that @xmath455 we estimate the rate of change of the area of the fixed surface @xmath199 along the path of metrics @xmath456 : @xmath457 on the third line , we used ( [ eqn_omae_contains ] ) and the fact that @xmath458 and @xmath75 are positive on the set @xmath208 ( by the maximum principle ) . on the last two lines , we used the fact that @xmath446 is supported in @xmath437 , the definition of @xmath106 , and the fact that @xmath459 . the last term is positive , since @xmath410 by hypothesis .
this shows that the rate of change of areas of all minimal area enclosures is uniformly positive at @xmath129 , proving that @xmath454 is increasing near @xmath129 . to remedy the fact that @xmath456 does not fix the boundary area for @xmath240 , define @xmath460 , and consider the harmonic functions @xmath461 with boundary data @xmath462 .
then by construction , the metrics @xmath463 have boundary area equal to @xmath30 , and are valid test metrics for @xmath255 .
since @xmath464 , first derivative computations at @xmath129 agree for @xmath456 and @xmath465 ; by the above , @xmath466 is increasing near @xmath129 .
this contradicts the assumption that @xmath467 was a maximizer of @xmath255 . in section [ sec_applications ] , we propose a conjecture regarding maximizers @xmath17 of @xmath255 in general , without _ a priori _ assumptions on regularity .
suppose @xmath3 is @xmath52 minus the unit open ball centered at the origin , equipped with the flat metric @xmath5 .
the function @xmath468 is harmonic , vanishes on @xmath322 , and approaches one at infinity .
now it is straightforward to compute that @xmath469 and @xmath470 , so that @xmath471 by theorem [ thm_mu ] .
recall from section [ sec_definitions ] that @xmath0 is in the same harmonic conformal class as the schwarzschild metric of mass 2 .
see figure [ fig_mu_plot ] for a plot of @xmath27 for @xmath472 .
[ fig_mu_plot ] next , to address @xmath29 for harmonic conformal class of @xmath0 , let @xmath473 be the spherically - symmetric harmonic function that is one at infinity such that @xmath474 has boundary area equal to @xmath30 , given explicitly by : @xmath475 to aid with the discussion , we recall that formula ( [ eqn_schwarz_metric ] ) extends to define a metric on @xmath476 with two asymptotically flat ends , with a reflection symmetry across the euclidean sphere of radius @xmath477 , called the horizon .
the horizon is a minimal surface for the schwarzschild metric .
let @xmath478 , so we see that @xmath479 is isometric to a subset of the two - ended schwarzschild manifold of mass @xmath58 . if @xmath480 the boundary @xmath71 ( the euclidean unit sphere ) is its own outermost minimal area enclosure , and therefore has area @xmath30 , essentially because @xmath481 excludes the horizon .
if @xmath482 , @xmath481 includes the horizon of the schwarzschild manifold , which is the outermost minimal area enclosure @xmath71 and has area @xmath483 . in the borderline case @xmath484
, the boundary of @xmath3 agrees with the horizon .
these observations allow us to give a formula for the minimal enclosing area as a function of @xmath30 : @xmath485 we conjecture that the spherically symmetric metric @xmath486 is a maximizer for @xmath255 , for all @xmath487 .
this would immediately imply that @xmath29 is given by the above formula for @xmath488 . without proof , we remark that the above spherically - symmetric metrics @xmath486 are _ local _ maxima for the minimal enclosing area among metrics in @xmath245_h}$ ] that have boundary area @xmath30
. see figure [ fig_alpha_plot ] for a plot of this conjectured form of @xmath31 .
[ fig_alpha_plot ] if @xmath29 does indeed have the above form in this example , then we remark that we have equality @xmath489 finally , we point out that the metrics @xmath486 behave consistently with theorem [ thm_mean_curv ] : in the case @xmath410 , the surface @xmath414 for @xmath486 is disjoint from @xmath71 and therefore has zero mean curvature .
in this section we present a natural conjecture regarding the function @xmath29 . assuming this conjecture , we deduce statements relating the functions @xmath113 , @xmath29 and the numerical invariants @xmath24 and @xmath25 .
one consequence is a general estimate of the adm mass of an asymptotically flat manifold of nonnegative scalar curvature with compact boundary .
first we recall the riemannian penrose inequality , proved as stated below by bray @xcite ( for dimension @xmath472 ) and later by bray and lee @xcite ( for @xmath490 ) .
huisken and ilmanen gave a proof for @xmath472 , with @xmath30 replaced by the area of the largest connected component of @xmath4 @xcite .
let @xmath491 be asymptotically flat of dimension @xmath206 with nonnegative scalar curvature .
suppose the boundary @xmath63 has area @xmath30 , zero mean curvature , and every surface enclosing @xmath71 has area strictly greater than @xmath30 .
then @xmath492 moreover , if equality holds and @xmath3 is a spin manifold ( or if @xmath472 ) , then @xmath0 is isometric to the schwarzschild manifold of mass @xmath493 .
we remark that the harmonic conformal class was a crucial element in both the bray and bray lee proofs . for the remainder of the article we assume , unless noted otherwise , that @xmath0 has dimension @xmath206 and nonnegative scalar curvature ( but not necessarily minimal boundary ) .
consider a maximizer @xmath494 for @xmath495 with @xmath254 large .
if we assume @xmath494 is smooth , then based on theorem [ thm_mean_curv ] , we see that the outermost minimal area enclosure @xmath496 is `` close to '' a minimal surface : the mean curvature is uniformly bounded by a constant times @xmath497 . by construction , @xmath496 has less area than any surface that encloses it . together with the riemannian penrose inequality ,
this behavior suggests that the adm mass of @xmath494 ought to be bounded from below in terms of @xmath498 in the limit @xmath259 .
[ conj_rpi ] let @xmath0 be an asymptotically flat @xmath499-manifold , @xmath206 , of nonnegative scalar curvature , with nonempty , smooth , compact boundary @xmath71 .
fix @xmath377 for which @xmath410 .
let @xmath494 be a maximizer for @xmath255 .
then @xmath500 in other words , the riemannian penrose inequality holds for @xmath501 in the limit @xmath259 .
suppose @xmath340 .
for all @xmath254 sufficiently large , @xmath503 by theorem [ thm_maximizer ] . since @xmath504 , conjecture [ conj_rpi ] can be written : @xmath505 we claim the left - hand side is at most @xmath113 . to see this , note that a harmonic function @xmath59 with @xmath155 boundary data can be approximated in @xmath225 norm by a harmonic function @xmath506 with smooth boundary data so that the adm masses of @xmath222 and @xmath507 differ by less than @xmath508 .
thus , the value of @xmath113 in definition [ def_mu ] is unchanged if the supremum is taken over the generalized harmonic conformal class . in particular
, @xmath494 can be viewed as a valid test metric for @xmath113 , so we have @xmath509 for each @xmath254 . taking the limit
@xmath259 completes the proof .
we emphasize the point that while @xmath113 is determined solely from the numerical invariants @xmath24 and @xmath25 , @xmath29 involves much more of the global geometry of @xmath0 the areas of hypersurfaces .
one interesting immediate consequence is the following upper bound for the minimal enclosing area . if @xmath510 ( which holds for all @xmath111 sufficiently large by proposition [ prop_alpha_continuous ] and corollary [ cor_alpha_a ] ) , then for all metrics @xmath511_h$ ] with boundary area at most @xmath30 : @xmath512 as follows from the previous proposition and the definition of @xmath31 . by theorem [ thm_mu ] ,
the left - hand side can be computed explicitly in terms of @xmath24 , @xmath25 , and @xmath30 .
this is an example of @xmath24 and @xmath25 giving control on the geometry of metrics in @xmath2_h$ ] .
let @xmath408 be given .
as a consequence of proposition [ prop_alpha_continuous ] , corollary [ cor_alpha_a ] , and the intermediate value theorem , there exists @xmath111 such that @xmath515 by ( [ eqn_mass_c ] ) ( which uses the conjecture ) , there exists @xmath516 sufficiently large so that a maximizer @xmath85 of @xmath255 satisfies : @xmath517 the left - hand side may be computed using formula ( [ eqn_conf_mass ] ) , and the right hand side with ( [ eqn_a_epsilon ] ) : @xmath518 applying lemma [ lemma_compute_term ] ( which still holds for @xmath155 boundary data ) , we have @xmath519 where @xmath232 is the function on @xmath71 associated to @xmath59 . by taking the infimum of the above over all nonnegative functions @xmath232 in @xmath225 and letting @xmath520
, we obtain : @xmath521 the term inside the braces can be viewed as a strictly convex functional on functions @xmath246 . using an euler
lagrange approach , one can compute that the unique global minimum of the functional is attained by @xmath522 a smooth function on @xmath71 ( see chapter 5 of @xcite for the full details in the case @xmath472 )
. using the formula @xmath177 and our expression for the minimizer @xmath232 , we deduce from ( [ eqn_mass_ineq_prelim ] ) that @xmath523 recalling the definitions of the numerical invariants @xmath24 and @xmath25 from lemmas [ lemma_i1 ] and [ lemma_i2 ] , we see that the above inequality is equivalent to the statement @xmath524 .
one interpretation of ( [ eqn_mass_estimate ] ) is a general estimate for the adm mass of an asymptotically flat manifold of nonnegative scalar curvature with compact boundary . in the case
that the riemannian penrose inequality applies to @xmath0 , inequality ( [ eqn_mass_estimate ] ) is weaker .
however , we emphasize that ( [ eqn_mass_estimate ] ) requires no assumptions on the boundary geometry , such as minimality .
[ cor_rzi ] assume conjecture [ conj_rpi ] .
let @xmath75 be the @xmath5-harmonic function that vanishes on @xmath71 and approaches one at infinity .
then the asymptotically flat metric @xmath525 ( which is singular on @xmath71 ) satisfies the mass estimate : @xmath526 the proof follows from equation ( [ eqn_mass_estimate ] ) and the observation that equals @xmath527 by formula ( [ eqn_conf_mass_infinity ] ) .
the reason that @xmath6 is singular on the boundary is that the conformal factor @xmath75 vanishes there .
this type of metric singularity is an example of a _ zero area singularity _ , or zas , which we now describe .
following @xcite , in a manifold with smooth metric on the interior ( but not necessarily on the boundary ) , a boundary component @xmath199 is said to be a zero area singularity if for all sequences of surfaces @xmath528 converging in the @xmath529 sense to @xmath199 , the areas of the @xmath530 converge to zero .
metrics @xmath531 as in the corollary have a zas on each boundary component . also note that @xmath6 has nonnegative scalar curvature because @xmath5 does and @xmath75 is harmonic .
the motivating example of a manifold with a zero area singularity is the schwarzschild metric of negative mass : the metric given in equation ( [ eqn_schwarz_metric ] ) with @xmath532 on @xmath52 minus a ball of radius @xmath533 .
it is true but not immediately obvious that the right - hand side of ( [ eqn_rzi ] ) is intrinsic to the singularities , in that it depends only on the geometry of @xmath6 in any neighborhood of @xmath71 ( and not on the data @xmath534 ) .
this number is suggestively called the _ mass _ of @xmath71 ( or zas mass ) , and equals @xmath58 for the schwarzschild metric of mass @xmath535 . in general , for zas that do not necessarily arise from metrics of the form @xmath525 , it is still possible to define a meaningful notion of zas mass , a number in @xmath536 $ ] .
corollary [ cor_rzi ] implies the statement : in manifolds of nonnegative scalar curvature that contain zero area singularities @xmath71
, the adm mass is bounded below by the zas mass : @xmath537 for a more thorough discussion , we refer the reader to @xcite .
there are two cases in which inequality ( [ eqn_adm_zas ] ) is firmly established without the use of conjecture [ conj_rpi ] .
first , if @xmath472 and @xmath322 is connected , robbins @xcite proved the inequality using weakly - defined inverse mean curvature flow as developed by huisken and ilmanen @xcite .
if @xmath71 is disconnected , however , inverse mean curvature flow yields no such inequality , not even a weaker version .
the second case for which corollary [ cor_rzi ] is known is that in which the harmonic conformal class @xmath2_h$ ] contains a metric for which the hypotheses of the riemannian penrose inequality hold .
if such a metric exists , bray showed inequality ( [ eqn_rzi ] ) directly from the riemannian penrose inequality ( c.f .
however , in general , @xmath2_h$ ] need not contain such a metric ( see chapter 2 of @xcite ) . in closing
, we make a connection with the positive mass theorem ( pmt ) of schoen and yau @xcite , proved also for spin manifolds by witten @xcite .
note that the pmt was a key ingredient in the bray and bray
lee proofs of the riemannian penrose inequality .
[ thm_pmt ] let @xmath0 be a complete , asymptotically flat riemannian @xmath499-manifold without boundary , with either @xmath206 or @xmath3 a spin manifold . if @xmath0 has nonnegative scalar curvature , then the adm mass is nonnegative , and zero if and only if @xmath0 is isometric to @xmath52 with the flat metric
. we can view ( [ eqn_adm_zas ] ) as a generalization of the pmt : metrics with zas are generally incomplete , so theorem [ thm_pmt ] does not apply . if we interpret the zas mass as quantifying the defect due to the presence of singularities in terms of their local geometry , then inequality ( [ eqn_adm_zas ] ) gives a lower bound for the adm mass as the size of this defect .
we emphasize that ( [ eqn_adm_zas ] ) is unproven in general . a case of particular interest
is when @xmath0 contains only zas of zero mass : inequality ( [ eqn_adm_zas ] ) would establish nonnegativity of the adm mass .
.... \section{notes } % --can you get anything viewing $ \mu$ as a map of af metrics mod equivalence to continuous functions?\\ % then $ \mu$ gives an upper bound for $ \alpha$. ( unproven ) % --what can you get out of robbin 's result in the case that $ \sigma$ is connected ?
can we deal with that here ?
% if that can be adapted , we get that maximizer is spherically symmetric for ss case .
what the rzi shows is essentially % that at largest $ a$ for which $ \alpha(a)=a$ , we have $ \mu(a ) \geq ... $ .
can you show that persists for larger $ a$?\\ --another interpretation : consider the map which sends the bv of a harmonic function to the coefficient of the $ 1/r$ term .
( this data is equivalent to $ w$. ) what can this tell us about the geometry of the manifold ?
well , the rzi gives a lower bound for the adm mass , assuming nonnegative scalar curvature .
the inequality $ \mu(a ) \geq ... $ tells us something about the minimal enclosing area for the maximizer and thus for for any metric with an upper bound on the boundary area , purely in terms of $ a$ and $ w$.\\ % --knowing $ \alpha(a)=a$ for $ a$ small + rzi tells you , by continuity that there exists $ a$ such that $ \mu(a ) \geq \sqrt{a/16\pi}$. given % the explicit formula for $ \mu$ , what does that tell you?\\ % --include argument that maximizer for $ \alpha(a)$ is spherically symmetric , knowing the conjecture.\\ % --note that numerical invariants alone should distinguish schwarzschild , assuming cc .
did robbins prove the case of equality?\\ --$\alpha_{\circ}$ looks like $ \alpha_c$ with $ c=0$.\\
% --took out section on how invariants behave under perturbations to hcc ( see below)\\ --notation\\ --look in stein 's book for stuff on approximation to identity ....
|
consider an asymptotically flat riemannian manifold @xmath0 of dimension @xmath1 with nonempty compact boundary .
we recall the harmonic conformal class @xmath2_h$ ] of the metric , which consists of all conformal rescalings given by a harmonic function raised to an appropriate power .
the geometric significance is that every metric in @xmath2_h$ ] has the same pointwise sign of scalar curvature .
for this reason , the harmonic conformal class appears in the study of general relativity , where scalar curvature is related to energy density ( c.f .
@xcite ) . our purpose is to introduce and study invariants of the harmonic conformal class .
these invariants are closely related to constrained geometric optimization problems involving hypersurface area - minimizers and the adm mass . in the final section , we discuss possible applications of the invariants and their relationship with zero area singularities and the positive mass theorem .
|
all observed cp violation can be described by the complex parameter @xmath4 , which describes an imaginary contribution to the @xmath8 mixing of the neutral @xmath9 mesons .
such a mixing implies the existence of an effective hamiltonian @xmath10 where @xmath11 gev , and @xmath12 run over possible gamma matrix structures .
the dimensionless coefficients @xmath13 are real in a basis where the standard model @xmath14 effective hamiltonian has a real coefficient . in the case
that the dominant term is @xmath15 , @xmath16 the two basic issues of cp violation are * what is the underlying physics which leads to @xmath17 ? is it a very small effect originating at the weak scale , as suggested by the form @xmath18 , or is it a larger effect generated by physics at higher energies ? * how can the magnitude @xmath19 to @xmath20 be understood ?
in the standard model all information about flavor and cp violation originates from the yukawa coupling matrices .
after electroweak symmetry breaking , this is manifested in the cabibbo - kobayashi - maskawa ( ckm ) matrix of the charged current interactions of the @xmath21 boson @xcite . a one loop box diagram with internal top quarks
gives the dominant contribution to @xmath17 via @xmath22 \label{eq : csm}\ ] ] where @xmath23 is the result of the loop integration , and @xmath24 is the su(2 ) gauge coupling constant .
for a suitable choice of the ckm matrix elements , @xmath25 , the standard model can provide a description of the observed cp violation .
the fundamental reason for the size of the cp violation observed in nature remains a mystery , however , and must await a theory of flavor which can explain the values of @xmath26 and the ckm phase . if the ckm matrix contained no small parameters one would expect @xmath27 to be of order @xmath28 to @xmath29 rather than the observed value of order @xmath30 to @xmath20 . of course , measurements of cp conserving observables have shown that @xmath31 are small for @xmath32 , and , given the measured values of @xmath33 and @xmath34 , it is convenient to use the wolfenstein parameterization@xcite of the ckm matrix , in which case ( [ eq : csm ] ) becomes @xmath35 if we _ assume _ that the ckm matrix does not have any other small parameters , the standard model yields a value of @xmath4 of the observed order of magnitude .
while this is not a prediction , it is an important success of the standard model , and has made the ckm theory the leading candidate for cp violation . to our knowledge
, there is no similar success in any published alternative to the ckm theory of cp violation , since in these theories the order of magnitude of @xmath36 can only be fixed by fitting to the measured value of @xmath4 . in this letter
we present such an alternative theory .
two further measurements of @xmath31 , with @xmath32 , would determine both @xmath37 and @xmath38 allowing a prediction of @xmath27 and @xmath4 .
a fit to the two observables @xmath39 and @xmath40 , but _ not _
@xmath4 , is shown in figure 1 . for all numerical work
, we use the data and parameters listed in table 1 for a discussion of these , and references , see @xcite .
unfortunately the large uncertainties make this a very weak prediction : @xmath41 is allowed even at the 68% confidence level .
hence , from this one can not claim strong evidence for ckm cp violation .
recent observations at lep have improved the limit on @xmath42 mixing , so that @xmath43 at 95% confidence level @xcite .
the result of a @xmath44 fit in the standard model to @xmath37 and @xmath38 using the three observables @xmath39 , @xmath40 and @xmath0 , but _ not _
@xmath4 , is shown in figure 2 . for @xmath45 mixing the amplitude method is used @xcite . comparing figures 1 and 2 ,
it is clear that the @xmath0 limit is now very significant . at 68% confidence level
the standard model is able to predict the value of @xmath4 to within a factor of 2 ; however , at 90% confidence level @xmath41 is allowed , so that at this level there is no prediction , only an upper bound . while this is an important success of the ckm theory , it is still worth pursuing credible alternative theories of @xmath3 violation .
.values of observables and parameters [ cols="^,^",options="header " , ]
a superweak theory @xcite is one in which the ckm matrix is real , so @xmath46 , and @xmath17 of eq .
( 1 ) originates from physics outside the standard model .
we define a _ pure _ superweak theory to be one where all flavor changing phenomena ( other than @xmath4 ) are accurately described by the real ckm matrix . comparing figures 1 and 2 at low @xmath38
, one sees that the new limit on @xmath45 mixing has excluded superweak theories with negative @xmath37 .
this has important phenomenological consequences for pure superweak theories .
we have computed @xmath47 in pure superweak theories , using as input the three observables @xmath39 , @xmath40 and @xmath0 .
we find that all negative values of @xmath37 are excluded at greater than 99% confidence level . at positive @xmath37
only the two observables @xmath39 and @xmath40 , are relevant , and we find the most probable value of @xmath37 to be + 0.27
. however , even this value of @xmath37 corresponds to the pure superweak theory being excluded at 92% confidence level . since the uncertainties are dominated by the theory of @xmath48 , we take the view that this does not exclude purely superweak theories . in such theories
positive values of @xmath37 are 40 times more probable than negative values , and hence large values for @xmath49 mev and small values for @xmath50 are predicted .
a pure superweak description of @xmath3 violation implies @xmath51 an important consequence of the new limit on @xmath45 mixing is the strong preference for positive @xmath37 and the resulting small values for @xmath52 .
this is numerically significant : without the @xmath45 mixing result the superweak theory can also have negative values of @xmath37 which give @xmath53 about a factor of two larger than the positive @xmath37 case . with the @xmath45 result , a pure superweak theory must have @xmath53 at the lower end of the standard model range . thus in a pure superweak theory , @xmath54
is predicted to be @xmath55 by comparison , in the standard model @xmath56 at 68% ( 95% ) confidence level . in the standard model , the branching ratio @xmath57 is given by @xcite @xmath58 where @xmath59 and @xmath60 .
in pure superweak theories , since @xmath37 is positive and @xmath41 , the branching ratio is lowered to @xmath61 relative to the standard model prediction of @xmath62 .
mixing increases @xmath37 even in the standard model .
] the recent observation of a candidate event for this decay @xcite is not sufficient to exclude pure superweak theories , but further data from this experiment could provide evidence against such theories .
pure superweak theories are artificial : they do not possess a symmetry which allows @xmath17 of eq . ( 1 ) , while forbidding similar @xmath63 operators .
if @xmath4 is generated by new physics , why does this new physics not contribute to @xmath64 mixing ?
in general it would be expected to also contribute to @xmath65 and @xmath66 processes . in the absence of a fundamental theory of flavor ,
the relative sizes of the various flavor changing operators can be estimated only by introducing arguments based on approximate flavor symmetries .
we assume that the underlying theory of flavor possesses a flavor symmetry group , @xmath67 , and a mass scale @xmath68 .
the breaking of @xmath67 , whether explicit or spontaneous , is described in the low energy effective theory by a set of dimensionless parameters , \{@xmath69 } , each with a well defined @xmath67 transformation .
the low energy effective theory of flavor is taken to be the most general operator expansion in powers of @xmath70 allowed by @xmath67 and \{@xmath69}. in the case that the ckm matrix can be made real , we call these _ general _ superweak theories . the phenomenology of such theories depends on @xmath67 , @xmath68 and \{@xmath69 } and will typically not coincide with the pure superweak phenomenology . the @xmath63 operators may lead to exotic cp violation in neutral @xmath71 meson decays and may contribute to @xmath40 , allowing large values of @xmath53 invalidating ( [ eq : bssw ] ) . similarly the @xmath14 operators may invalidate ( [ eq : kpisw ] ) , and may give an observable contribution to @xmath72 .
the dominant flavor changing neutral current ( fcnc ) interactions of the down sector of the standard model result from the `` 3 mechanism '' : small flavor breaking parameters which mix the light quarks with the heavy third generation quarks , together with a large , order unity , breaking of the flavor symmetry that distinguishes the third generation from the first two .
hence , beneath the weak scale , the standard model yields an effective hamiltonian with dominant fcnc operators which contain a factor @xmath73 for each flavor changing current @xmath74 , and a factor @xmath75 from the loop integration .
the relevant diagrams are all 1 loop , giving the @xmath76 factor , and involve the large gim violation of the top quark mass ; since there is no small flavor violating parameter , the rest of the loop integral has an order of magnitude given by dimensional analysis as @xmath77 .
now consider physics beyond the standard model where the entire flavor structure of the theory beneath @xmath68 is controlled by @xmath67 and \{@xmath69 } both the yukawa matrices of the standard model , @xmath78 , and the non - standard model operators in @xmath79 .
since the dominant down sector , fcnc effects from @xmath80 are known to arise from the `` 3 mechanism '' , we assume that @xmath67 and \{@xmath69 } are chosen so that the dominant such effects from @xmath81 are also from the `` 3 mechanism '' .
the most general parameterization of the `` 3 mechanism '' in the down sector involves four complex parameters : @xmath82 and @xmath83 , @xmath84 , which describe the mixing of @xmath85 and @xmath86 with @xmath87 and @xmath88 . assuming all phases to be of order unity
, we can describe the `` 3 mechanism '' in terms of just four real small parameters @xmath89 and @xmath90 .
we make the additional simplifying assumption that @xmath91 , yielding the non - standard model interactions @xmath92 \label{eq : h3}\end{aligned}\ ] ] where @xmath93 are complex coefficients of order unity , and @xmath94 is a lepton field . a sum on possible gamma matrix structures
is understood for each operator . since the flavor changing interactions from both the standard model and the new physics
are governed by the same symmetry , we can choose @xmath95 and @xmath96 .
such interactions can arise from many choices of @xmath67 and \{@xmath69 } ; the particular choice is unimportant , however , as the phenomenology rests only on three assumptions * there is an underlying theory of flavor based on symmetry @xmath67 and breaking parameters \{@xmath69}. * the dominant non - standard model fcnc operators of the down sector arise from the `` 3 mechanism '' . * the symmetry breaking parameters of the down sector are left - right symmetric , and have phases of order unity . in the standard model ,
the dominant fcnc of the down sector arises from the `` 3 mechanism '' , so that it is useful to describe the effective theory beneath the weak scale by eq .
( [ eq : h3 ] ) with @xmath97 and @xmath93 real .
this special case of the `` 3 mechanism '' has a restricted set of gamma structures due to the left - handed nature of the weak interaction .
we have argued that pure superweak theories are artificial , and we now study superweak theories where fcnc interactions are generated by the `` 3 mechanism '' and yield @xmath98 of ( [ eq : h3 ] ) .
why should such theories have @xmath25 real when @xmath93 are complex ?
one possibility is that @xmath67 forces the yukawa matrices @xmath78 to have a sufficiently simple form that they can be made real by field redefinitions .
another possibility will be discussed later .
since @xmath98 will be the origin of all @xmath3 violation , one may wonder if it could also account for all of @xmath99 .
this is not possible charged current measurements , together with the unitarity of @xmath100 , imply @xmath53 and @xmath101 are sufficiently large that @xmath21 exchange contributes a significant fraction of @xmath99 . given that the fcnc of both the standard model and exotic interactions have the form of ( [ eq : h3 ] ) , it would appear that the exotic interactions must give a large fraction of @xmath99 since they are responsible for all of @xmath4 .
this is not the case ; in the standard model the @xmath102 and @xmath63 operators have chirality @xmath103 , whereas for a generic `` 3 mechanism '' they will have all chiral structures .
it is known that the @xmath104 , @xmath8 , operator has a matrix element which is enhanced by about an order of magnitude relative to that of the @xmath103 operator @xcite , and that there is no similar enhancement in the @xmath63 case .
furthermore , the @xmath104 operator is enhanced by qcd radiative corrections in the infrared @xcite ; with the enhancement at 1 gev about a factor of 3 larger than at 5 gev . hence we conclude _ in a generic superweak theory , we expect that _
@xmath98 _ leads to @xmath105 contributions to @xmath99 .
_ there is considerable uncertainty in this percentage because of the uncertainty in the overall enhancement of the @xmath8 and @xmath63 contributions from the @xmath104 operator , and because of the unknown order unity @xmath93 coefficients . given this result
, we must evaluate how well these generic superweak theories can account for the data , and to what extent they lead to predictions .
let @xmath106 and @xmath107 be the standard model and new physics contributions to @xmath108 first we consider a perturbation around the pure superweak case , where the fractional contributions from new physics @xmath109 are small . the central value of @xmath37 , from @xmath40 alone , changes by @xmath110 for very small @xmath111 ( @xmath112 for @xmath113 ) . for positive @xmath111 ,
this improves the fit of general superweak theories to @xmath40 and @xmath39 .
for example , @xmath114 gives a central value of @xmath115 with @xmath116 , which corresponds to 68% c. l. since the allowed range of @xmath37 is little changed from eq .
( 5 ) , the prediction of small @xmath53 persists in these general superweak theories , so that the prediction of eq .
( 7 ) for low values of @xmath57 applies .
similarly , since @xmath37 is little altered , the prediction for @xmath45 mixing is @xmath117 , where the pure superweak prediction @xmath118 is given in eq .
( 6 ) . in this case
the general superweak theory also predicts large values of @xmath119 , although for negative @xmath120 , it is not quite so large as @xmath118 .
there is a second class of general superweak theories which is not a perturbation about the parameters of the pure superweak theories . in general superweak theories ,
the limit @xmath121 ps@xmath122 can be expressed as @xmath123 . for negative @xmath111 and positive @xmath120 , the negative @xmath37 region could become allowed .
for example , @xmath124 gives a theory in which @xmath37 has a probability 25% ( 9% ) of being negative .
this class of superweak theories _ requires _ values of @xmath125 which are larger than our expectation , and appear somewhat improbable .
they have @xmath53 and @xmath57 at the upper end of the standard model range . in these theories
@xmath126 is likely to be low , although it depends on @xmath127 .
in general , the alternative theory of @xmath3 violation of @xmath98 from the `` 3 mechanism '' is not a strong competitor to the ckm theory of @xmath3 violation .
the ckm theory , with two small measured parameters , @xmath33 and @xmath34 , yields the correct order of magnitude for @xmath4 , while superweak theories with the `` 3 mechanism '' apparently require a new scale @xmath128 tev . however , there is the interesting possibility that the new physics generates fcnc operators only at 1 loop , as in the standard model .
this would give @xmath129 , with the mass of the new quanta close to the weak scale at @xmath130 tev .
we therefore take the view that the `` 3 mechanism '' generating fcnc operators at 1 loop at the weak scale is a credible alternative to the ckm theory of cp violation . while not as minimal as the ckm theory , it correctly accounts for the order of magnitude of @xmath4 .
let @xmath94 represent @xmath131 or @xmath132 , left or right handed .
new interactions of the form @xmath133 , where @xmath134 is some new heavy field , will generate fcnc at tree level , whereas @xmath135 generates them at 1 loop .
thus the exotic new heavy particles at the weak scale should possess a parity so that they appear only in pairs .
weak scale supersymmetry allows a symmetry description of the weak scale , and leads to a successful prediction for the weak mixing angle .
furthermore , it incorporates the economical higgs description of flavor of the standard model .
@xmath136 parity ensures that superpartners appear pairwise in interactions , so that the dominant supersymmetric contributions to fcnc processes occur only at one loop .
supersymmetric theories have several new generation mixing matrices in particular @xmath137 at the gluino interaction @xmath138 . a flavor symmetry , @xmath67 ,
can ensure that the largest contribution from superpartner exchange to fcnc occurs via the `` 3 mechanism''@xcite .
if the small symmetry breaking parameters are left - right symmetric and real , this gives @xmath139 of ( [ eq : h3 ] ) with @xmath140 where @xmath141 is the average mass of the colored superpartners in the loop .
as the superpartners are at the weak scale , @xmath142 , and comparing with ( [ eq : h3sm ] ) one finds that , with weak scale supersymmetry , it may well be that @xmath4 receives comparable standard model and supersymmetric contributions .
operator relative to the @xmath103 , and given the further order of magnitude enhancement of @xmath143 relative to @xmath144 from qcd scaling , one generically expects the supersymmetric contribution to be larger .
however , these factors may be outweighed by colored superpartner masses somewhat larger than @xmath145 , some degree of degeneracy between the third generation scalars and those of the lighter generations , and by @xmath146 somewhat less than @xmath25 .
we note that the qcd enhancement of @xmath143 for the @xmath8 operator@xcite was not included in @xcite ] here we stress that weak scale supersymmetry can provide an important example of the general superweak theories discussed in this letter .
the absence of ckm @xmath3 violation would be guaranteed if @xmath3 violation were _ soft _ restricted to operators of dimension two and three .
the yukawa matrices would then be real , so that there would be no @xmath3 violation from diagrams with internal quarks , but the scalar mass matrices would contain phases , so that @xmath3 violation would arise from diagrams with internal squarks . and @xmath21 matrices such that even the supersymmetric contribution to @xmath4 involves a phase originating from the yukawa couplings . ]
soft @xmath3 violation in supersymmetric theories , with fcnc operators arising from the `` 3 mechanism '' , represents a well - motivated and credible alternative to ckm @xmath3 violation , and will be explored in detail elsewhere .
fits of the ckm matrix to @xmath39 , @xmath40 and @xmath0 show that at 68% c.l .
the standard model correctly predicts @xmath4 to better than a factor of two , while at 90% c.l .
not even the order of magnitude can be predicted .
on one hand the standard model is highly successful ; on the other , there is still room for an alternative theory of cp violation .
the recent improvement on the limit on @xmath126@xcite implies that pure superweak theories with negative @xmath37 are excluded , while at positive @xmath37 they are somewhat disfavored .
pure superweak theories allow @xmath147 at 95% c.l . , and predict high values for @xmath126 and @xmath2 and low values for @xmath39 , @xmath57 and @xmath72 .
we have argued that _ pure _ superweak theories are artificial , and have introduced _ general _ superweak theories , in which all fcnc are governed by an approximate flavor symmetry and the `` 3 mechanism . '' in this case the new physics induces other flavor changing operators in addition to the @xmath8 operator responsible for @xmath4 ; in particular , @xmath148 contributions to @xmath149 mixing are expected .
there are two important classes of general superweak theories , one with positive @xmath37 and the other with negative @xmath37 .
the first can be viewed as a perturbation about the superweak case , with an improved fit to data , while retaining the characteristic predictions mentioned above .
the negative @xmath37 possibility appears less likely , and arises only if the new physics contributes more than 10% of @xmath99 . in this case
future data should show a high value for @xmath57 and low values for @xmath126 , @xmath2 , @xmath39 , and @xmath72 .
all these superweak theories predict low values for the @xmath3 asymmetries in @xmath71 meson decays .
weak scale supersymmetric theories with softly broken @xmath3 can provide an important example of general superweak theories . as in the ckm theory , assuming phases of order unity yields a correct prediction for the order of magnitude of @xmath4 .
in addition they have @xmath150 at tree level , and it is interesting to seek a flavor symmetry which would sufficiently protect @xmath151 from radiative corrections to solve the strong @xmath3 problem .
this work was supported in part by the director , office of energy research , office of high energy and nuclear physics , division of high energy physics of the u.s .
department of energy under contract de - ac03 - 76sf00098 and in part by the national science foundation under grant phy-95 - 14797 .
c. bernard et al .
( milc collaboration ) b mixing on lattice : @xmath152 , @xmath153 and related quantities
hep - ph/9709328 ; v. gimnez and g. martinelli ,
b @xmath154 mixing in hqet , rome prep .
96/1153 , ftuv 96/26-ific 96/30 , hep - lat/9610024v2 ; results presented in hamburg by c. sachradja at the electron - photon 1997 conference ( hamburg ) .
|
recent improvements to the limit of @xmath0 imply that pure superweak theories , while not excluded , no longer provide a good fit to the data . a class of general superweak theories is introduced in which all flavor changing interactions are governed by an approximate flavor symmetry which gives a `` 3 mechanism '' .
these theories are in good agreement with data , and predict low values for @xmath1 and cp asymmetries in b decays , and high values for @xmath0 and @xmath2 .
an important example of such a theory is provided by weak scale supersymmetric theories with soft @xmath3 violation .
the @xmath3 violation originates in the squark mass matrix , and , with phases of order unity , flavor symmetries can yield a correct prediction for the order of magnitude of @xmath4 . plus 1000pt minus 1000pt # 1 # 1= to # 1#2#1#2 # 1#2#3 ^ 2 # 1#2 # 3 # 1#1| # 1| # 1 # 1#1 # 1/ # 1| # 1| # 1#1^ lbnl-41104 + ucb - pth-97/57 + sns / ph/1997 - 8 + lal 97 - 91 .25 in * alternative theories of cp violation * 0.3 in riccardo barbieri@xmath5 , lawrence hall@xmath6 , achille stocchi@xmath7 , and neal weiner@xmath6 0.1 in @xmath5 _ scuola normale superiore , pisa , italy ; + and infn , sezione di pisa , italy + _ 0.1 in @xmath6 _ department of physics and lawrence berkeley national laboratory + university of california , berkeley , california 94720 + _ 0.1 in @xmath7 _ laboratoire de lacclrateur linaire + in2p3-cnrs et universit de paris - sud , f-91405 orsay _ .3 in
|
expreimental studies of coherent backscattering ( cbs ) of light from disordered media have been completed by many groups during the last decade.@xmath1 cbs measurements are of great interest because they have allowed to test the physics of weak localization , which was first uncovered in electronic transport,@xcite when applied to light waves and also to determine the light mean - free - path , @xmath2 in random media . in cbs measurements
the normalized reflected light intensity ( also called the albedo , @xmath3 ) is monitored versus the deviation angle , @xmath4 from the backscattering direction of the incident laser beam . in the multiple scattering regime interference effects are averaged out except in a small angular range around @xmath5 , where constructive interference originating from reciprocal light paths enhance the diffusive reflection from the sample . as a result a coherent scattering intensity cone is formed in the albedo , @xmath6 for small @xmath7 . from the full width , @xmath8 of the cbs cone at quarter maximum , @xmath2 is obtained using the relation @xmath9 , where @xmath10 is the impinging laser wavelengh .
the first cbs measurements were performed on suspensions of polystyrene spheres.@xcite since the suspension particles were very small in diameter , their brownian motion provided a natural ensemble averaging . for solid materials
, however cbs usually shows a noisy but repeatable mesoscopic pattern.@xcite under these conditions the cbs cone around @xmath5 can be obtained by applying an elaborate ensemble - averaging technique in order to suppress the dominant mesoscopic fluctuations . the averaging techniques employed to get the cone include spot variation,@xcite tilt angle variation,@xcite and sample spin - rotation.@xcite all the above experimental studies@xmath1 were performed on completely disordered , isotropic and homogeneous media .
photonic crystals have recently attracted a lot of attention@xcite because of potential optical device applications .
these applications pertain to photonic crystals with _
complete _ band gaps _
i.e. _ with no propagating solutions of the maxwell equations within a certain frequency range
. however , most of the available three dimensional photonic crystals nowadays have _
incomplete _ gaps in their photonic band structure , with light propagation forbidden only along certain directions . along these directions and for frequencies within the gap the light intensity decays with a certain decrement @xmath11-that is the bragg attenuation length . in the ideal case ( when disorder is negligibly small ) the length @xmath11 can be extracted from spectroscopic reflectivity measurements .
since @xmath11 is directly related to the gap width , then its value can be deduced from the angular ( or spectral ) width of the bragg peak in reflectivity . however , in the more realistic situation disorder is present in the sample and this broadens the bragg band .
we also note that even if @xmath12 broadening can be substantial due to the sensitivity of the bragg diffraction to the properties of the surface boundary . under these conditions
a separate determination of @xmath11 and @xmath2 is a challenge . in general , despite the conceptual elegance of the cbs technique ,
the main quantitative information it has provided@xmath13 so far is @xmath2 value in _ macroscopically homogeneous _ and isotropic disordered media .
we note that @xmath2 value in such media can be also determined from more conventional measurements of the transmission versus thickness@xcite that do not rely on the interference of clockwise and anticlockwise scattering paths . in the present work
we show that for optically _ periodic _ structures _ with disorder _ the cbs measurements provide a unique opportunity to separate the two effects , namely bragg attenuation and light diffusion lengths .
this reveals the tremendous potential of the cbs technique as compared to the more conventional optical spectroscopies .
for our studies of periodic structures we have chosen the synthetic @xmath14 opals that are actually three - dimensional photonic crystals.@xcite the synthetic opal single crystals were fabricated from a colloidal suspension of mono - dispersed silica ( @xmath14 ) spheres with a mean diameter , @xmath15=@xmath16 and dispersion in @xmath15 of about @xmath17 .
in contrast to the natural gem opals that exist in the market the space between @xmath14 spheres in artificial opals is filled with air .
a sintering process leads to the formation of inter - sphere necks that mechanically connect neighboring silica spheres in a face - centered - cubic ( fcc ) lattice .
the opals with air - filled voids are chalk - like diffuse light scatterers that show some iridescence due to bragg diffraction from their crystal planes .
the iridescence is dramatically enhanced , however if the voids are filled with an index matching liquid .
in opals there exist photonic gaps , @xmath18 , in the frequency spectrum around each @xmath19 , where @xmath20 are the miller indices of the crystal planes .
the central frequencies , @xmath19 , are determined by the bragg condition : @xmath21 , where @xmath22 is the inter - planar spacing , @xmath23 is the opal effective index of refraction ( @xmath24 for air - filled opals ) and @xmath25 is the speed of light . in the absence of disorder the stop band width ,
@xmath26 is related to the bragg attenuation length , @xmath11 , by the relation @xmath27 ( with implied @xmath28 indices ) , where @xmath29 . as mentioned above , however disorder in opals
dramatically broadens @xmath26 so that this relation is no longer valid . for our cbs measurements
, we cut a centimeter - size opal single crystal into planes orthogonal to several @xmath0 $ ] directions including the fast growing @xmath30 $ ] direction and a prominent @xmath31 $ ] direction .
we note that the @xmath32 planes in opals are rather disordered .
we used a standard experimental set - up@xcite with angular resolution in @xmath7 of less than @xmath33 .
the coherent light beam was directed from continuous wave lasers such as @xmath34 and @xmath35 , as well as various well - collimated semiconductor lasers .
the laser beam was linearly polarized with transverse electric field ( @xmath36 ) polarization .
the reflected intensity from the opal @xmath37 surface with @xmath36 polarization was measured versus @xmath7 using a silicon detector and phase - sensitive techniques .
the incident beam ( @xmath38 in diameter ) was directed at various angles , @xmath39 between @xmath40 to @xmath41 relative to the surface normal , and @xmath7 was varied within @xmath42 . to obtain specle - free cbs cones we performed configuration averaging using three different methods .
these include different illuminated areas of the opal surface , different inclination angles @xmath43 of the illuminated surface with respect to the laser beam , and sample rotation about the normal direction to the crystal surface . in fig .
1 we show typical cbs cones measured at various incident angles @xmath39 with respect to @xmath30 $ ] that we obtained using an @xmath34 laser at @xmath44 .
the cbs cones are apparent in all cases and their width , @xmath8 at quarter maximum versus @xmath39 is summarized in fig .
2 . except for a small @xmath39 interval , @xmath45 ( about @xmath46 ) around @xmath47 , @xmath8 monotonously increases with @xmath39 . within the interval
@xmath45 around @xmath47 , however there is an anomalous increase in @xmath48 of about a factor of two ( fig .
insight into the origin of the observed anomaly in @xmath8 ( @xmath39 ) may be obtained from fig . 3 in which the angle - dependent reflectivity spectra , @xmath49 , from the @xmath50 opal surface are shown .
it is seen that for @xmath51 the angle @xmath52 , at which the anomaly in the cbs cones occurs , in fact corresponds to the bragg conditions .
another evidence that the cbs anomaly is related to the bragg conditions can be observed from the obtained cbs cones at @xmath39 around @xmath53 ( fig .
these cbs cones were measured using a @xmath35 laser at @xmath54 that is close to @xmath55 at @xmath56 ( fig .
3 , inset ) . fig .
4 inset shows that @xmath8 for cbs cones measured at normal incidence stays approximately constant at about @xmath57 for laser wavelengths outside the bragg conditions .
however , for cbs measured at @xmath58 , @xmath8 in the cbs cone increases by a factor of @xmath59 . also the incident angle interval ( @xmath60 , fig .
4 ) in which the cbs anomalies are observed with laser illumination at @xmath61 is much larger than that obtained at @xmath44 ( @xmath62 , fig .
2 ) . both observations , namely the anomaly in @xmath8 ( @xmath39 ) at @xmath63 and strengthening of the anomaly at @xmath64 require an explanation .
a possible model is given in the next section .
the bragg - related anomalies of the cbs cones show that light transport and localization is substantially affected by the @xmath30 $ ] photonic band gap in the opal , where light becomes more localized . to understand these results we first recall the most transparent derivation of the cbs cones in the literature and apply it to laser frequencies outside the photonic gap.@xcite
we then modify the theory to interpret the anomalous cbs results at intra - gap laser frequencies at @xmath65 .
the coherent scattering albedo , @xmath66 , at an angle @xmath67 from a semi - infinite slab at @xmath68 due to an incident laser beam at an angle @xmath39 is determined by the first and last light scattering events occurring at the spatial points @xmath69 and @xmath70 , respectively , and by the light diffusion process that occurs between these events .
if @xmath71 and @xmath72 are the projections of @xmath73 and @xmath74 on the @xmath71-axis , and @xmath75 is the projection of @xmath76 on the sample interface plane , then the coherent part of albedo is given by the expression @xmath77\right\ } \pi(\rho , z , z^{\prime}){\cal p}_{\phi^{\prime}}(z^{\prime}).\ ] ] here @xmath78 and @xmath79 , where the angles @xmath80 and @xmath81 are , respectively , the propagation directions of the incoming and outgoing light _ inside _ the slab . these angles are related to @xmath39 and @xmath67 through the snell law @xmath82 . in eq .
( [ albedo ] ) @xmath83 is the probability density for the incident light to propagate up to the depth @xmath71 _ before _ the occurence of the _ first _ scattering event .
analogously , @xmath84 is the surviving probability density _ after _ the occurence of the _ last _ scattering event .
@xmath84 has the same form as @xmath83 , where @xmath80 , @xmath71 are replaced by @xmath81 , @xmath72 respectively . since @xmath83 satisfies the poisson statistics , it is given by @xmath85 .
the propagator @xmath86 in eq .
( [ albedo ] ) describes the diffusive motion of light between the first and last scattering events . to take into account the fact that diffusion occurs in the presence of the boundary @xmath87 ,
it is customary@xcite to choose the propagator @xmath88 in the form @xmath89 where @xmath90^{-1}$ ] is the diffusion propagator in an unbounded medium ; @xmath91 is the light diffusion coefficient , and @xmath92 is the mirror image of @xmath72 with respect to the plane @xmath93 , where @xmath94.@xcite the choice ( [ pi ] ) of the propagator @xmath88 insures that the boundary conditions @xmath95 are obeyed .
the evaluation of integrals in eq .
( [ albedo ] ) for small @xmath96 yields@xcite @xmath97 where @xmath7 enters into the r.h.s . of eq .
( [ alpha ] ) through @xmath98 and @xmath39 and @xmath80 are related through snell s law .
( [ alpha ] ) contains only a single free parameter , namely @xmath2 .
we used eq .
( [ alpha ] ) to fit the cbs cones in @xmath99 at various @xmath39 outside the anomaly angle interval @xmath45 around @xmath100 .
the excellent fits shown in fig .
1 for @xmath101 and @xmath102 were obtained with a single @xmath103 . from this value for the optical mean - free - path length and eq .
( [ alpha ] ) , we calculated the width of the albedo for the cbs cone at quarter maximum , @xmath8 , versus @xmath39 , as shown in fig .
2 . except near @xmath104 the overall agreement between the calculated and the experimentally determined @xmath8 is excellent showing that light transport is quite ordinary for frequencies outside the photonic gap . for describing the cbs process at @xmath65 and
correspondingly at @xmath105 we note that the diffusive propagation of light between the first and last scattering events that is given by @xmath86 in eq .
( [ albedo ] ) is affected very little by the bragg diffraction .
the reason for this is the small angle interval , @xmath106 , around @xmath104 where bragg condition is satisfied .
an upper bound estimate for @xmath106 can be obtained by relating it to the spectral width of the bragg gap @xmath107 , where @xmath108 .
if we choose for @xmath109 the measured value of the reflectivity peak @xmath110 , then for @xmath111 and @xmath112 we calculate @xmath113 to be @xmath114 radian .
this forms a stereo - angle of about @xmath115 stereo - radian around the bragg direction that is only about @xmath116 of the complete @xmath117 stereo - angle .
therefore , even if the light beam is incident at @xmath100 , it is diverted after the first scattering event and is then followed by an almost regular diffusion ; this is true since the probability for light to propagate close to @xmath104 again is very small .
this permits the generalization of the cbs theory to photonic crystals with relatively small stop bands , simply by only modifying the probability densities @xmath83 and @xmath84 in eq .
( [ albedo ] ) . for @xmath65 ,
light intensity decays as @xmath118 due to the bragg diffraction from the photonic crystal planes .
therefore @xmath83 can be written at @xmath100 as a product of the surviving probabilities : @xmath119 and the bragg attenuation , @xmath118 .
equivalently , we may represent @xmath120 at @xmath100 as a single exponent , @xmath121 , where @xmath122 .
this expression shows that eq .
( [ alpha ] ) describing the cbs cone is still valid at @xmath63 , but with a smaller effective mean - free - path , @xmath123 . a smaller @xmath123 leads to a broader cbs cone at @xmath63 , as indeed seen in fig . 2 for @xmath124 .
if @xmath39 and @xmath67 are close , but not equal to @xmath104 , then the bragg attenuation becomes weaker .
it can be shown that this effect is taken into account by the following modification of @xmath125 @xmath126^{1/2}.\ ] ] since the argument under the square root in eq .
( [ eff ] ) must be positive , we use this constraint to estimate the angle interval @xmath45 around @xmath104 where the cbs anomalies due to the photonic crystal are observed .
this leads to the expression : @xmath127 .
2 shows that an excellent fit is obtained for both @xmath8 enhancement and @xmath45 , using eqs .
( [ alpha ] ) and ( [ eff ] ) with @xmath128 and the previously obtained @xmath103 . using these values of @xmath2 and @xmath11 along
@xmath30 $ ] we evaluated @xmath129 , which agrees well with the data of fig .
we also calculated the `` pure '' bragg band @xmath107 to be @xmath130 .
we note , that this value is 3 times smaller than @xmath131 extracted from the bragg reflectivity band in fig .
3 . this illustrates the ambiguity in determining the photonic band structure form the more conventional reflectivity spectroscopy . for @xmath132
_ i.e. _ close to normal incidence and @xmath133 , eq .
( [ eff ] ) is no longer valid , since @xmath134 . in this case
it can be shown that the expression for the effective light mean - free - path is given by @xmath135^{1/2}.\ ] ] again the constraint that the argument under the square root is positive now yields @xmath136 .
this value is much larger than the obtained @xmath129 for @xmath44 .
however this much larger value is consistent with our measurements at @xmath137 ( fig .
4 ) . moreover , from eq .
( [ eff1 ] ) we calculated theoretical curve @xmath8 @xmath138 shown in fig .
4 . this curve fits very well the width enhancement of the cbs cone at @xmath53 , when @xmath39 is changed for a fixed @xmath10 ( fig . 4 for @xmath139 ) .
( [ eff1 ] ) is also consistent with the obtained results when @xmath10 is changed for a fixed @xmath140 ( fig
. 4 inset ) .
the agreements obtained validate both our modified cbs model and the extracted optical parameters for the opal photonic crystal .
we again emphasize that consistent values of @xmath11 were inferred from both the magnitude @xmath48 @xmath141 and the width @xmath142 of the cbs anomaly .
the widths of cbs cones from @xmath143 planes measured with @xmath144 are shown in fig .
5 , where @xmath8 is plotted versus the incident angle @xmath39 measured from @xmath31 $ ] direction . for @xmath144 the bragg cbs anomaly in @xmath8 occurs at @xmath145 , and is clearly seen in fig .
5 . fitting this cbs anomaly in a similar way as that at @xmath146 in fig .
2 , we deduced a bragg length , @xmath147 for the photonic gap in @xmath31 $ ] direction . in our fitting
we used the value of mean - free - path , @xmath148 that was extracted from the widths of the cbs cones for incident light beams close to the normal direction to the @xmath143 planes .
we note that , in addition to @xmath11 along @xmath31 $ ] , @xmath2 is also different from that extracted before from measurements along @xmath30 $ ] .
apart from the bragg - related anomalies we note that three _ qualitative _ features of the data presented above indicate that the description of cbs from opal photonic crystal within the conventional model of a medium with randomly positioned point - like scatterers is not completely adequate .
these features are : ( _ i _ ) the rayleigh scattering law @xmath149 that is characterisitc for point - like scatterers does not hold in opals .
this fact becomes evident from fig .
4 inset ; except for a narrow @xmath10 interval near @xmath150 , the cbs cone width @xmath8 remains approximately unchanged with @xmath10 for @xmath10 interval between @xmath151 to @xmath152 .
( _ ii _ ) the value of @xmath153 for @xmath30 $ ] direction appreciably exceeds @xmath154 for @xmath31 $ ] direction . ( _ iii _ )
the cone width @xmath8 increases with @xmath39 for cbs along @xmath30 $ ] ( fig .
2 ) but shows no dependence on @xmath39 for cbs along @xmath31 $ ] ( fig . 5 ) .
the question that we would like to address here is whether these cbs features can help in pinpointing the size , shape , and orientation of the generic defect scatterers in opal .
firstly , independence of @xmath2 on @xmath10 suggests that the size of the defects is larger than @xmath10 . on the other hand , defects with size @xmath155 scatter most of the light in the forward direction .
more precisely , the scattered light intensity is restricted to a cone with aperture @xmath156 . as a direct consequence of this almost forward scattering
, we should have @xmath157 , where @xmath158 is the _ transport _ mean - free - path that determines the diffusion coefficient ( @xmath159 ) in this case .
the conclusion that @xmath158 is much bigger than @xmath2 in opal has some dramatic consequencies . as can be seen from the general expression for the coherent albedo [ eq .
( [ albedo ] ) ] , the dependence @xmath99 is determined by two factors : the surviving probabilities @xmath160 , @xmath161 and the propagator @xmath88 . for point - like
scatterers both ingredients change within the same characteristic spatial scale , namely @xmath2 . however if @xmath157 , then the scale of the propagator that describes the diffusion process is @xmath158 , whereas the scale of @xmath160 , @xmath161 is still @xmath2 .
this would make the cone width @xmath162 rather than @xmath163 as in the usual case .
consequently @xmath8 would be _ independent _ of the incident angle @xmath39 , and , correspondigly insensitive to the bragg anomaly , contrary to the data .
a possible resolution of this apparent contradiction with the experiment that we can suggest here is to assume that the scatterers are _ highly anisotropic _
, such as disk - like with radius @xmath164 and thickness @xmath165
. moreover , in order to insure that cbs is different from the @xmath50 and @xmath143 opal surfaces as found experimentally , we have to assume that the disks are not oriented randomly , but rather their orientation is correlated .
although we did not extend the quantitative theory of cbs to this situation , intuitively it seems likely that the disks normals be parallel to the @xmath30 $ ] direction .
for such an orientation the light would diffuse easier _ along _ the @xmath50 planes . under these conditions of strongly anisotropic diffusion , longer paths do not necessarily lead to narrower cones as in the case of isotropic scatterers . since cbs is essentially the `` structure factor '' of the propagation _ along the boundary_,@xcite then the easier diffusion along @xmath50 would lead to narrower cones , _
i.e. _ to @xmath166 } < \delta\theta_{\left[220\right]}$ ] , in agreement with the experimental results ( figs . 2 and 5 ) .
a possible realization of the geometry and arrangement of scatterers as conjectured above can be provided by stalking faults defects .
such defects can be easily created along the @xmath30 $ ] opal fastest growing direction . also , this type of defects has been recently shown to dominate disorder in colloidal crystals,@xcite and we note that synthetic opals grow from silica colloids .
in conclusion , we have demonstrated that cbs is a powerful technique for investigating photonic crystals with incomplete photonic band gaps .
these measurements show that the remarkable universality of cbs characteristics for diverse disordered materials is not valid for photonic crystals at near bragg conditions . using the major effects on light transport near bragg conditions and their influence on cbs cones , we deduce the values of several fundamental parameters of an opal photonic crystal .
remarkably , despite the existence of substantial scattering due to disorder , nevertheless we showed in this work that the bragg attenuation length and , thus the photonic band gap width of an `` ideal '' crystal along @xmath30 $ ] , as well as light mean - free - path can be readily extracted from the cbs measurements , even if they are comparable in magnitude ( unlike the situation in ref . ) .
the reason why for @xmath167 disorder is unable to completely mask the contribution of bragg scattering to the cbs inside the photonic crystal , is that at bragg conditions the cbs involves @xmath168 light propagation _ before the first _ and _ after the last _ scattering events , when the existing disorder is _ not yet relevant_. this makes the cbs measurements a much more sensitive tool for characterizing the inherent band gap features for disordered photonic crystals as compared to the more conventional measurements of optical reflectivity . the only other study of cbs from photonic crystals that we know of was recently reported in ref . .
the cbs measurements in that work were performed on photonic crystals based on polystyrene spheres and also on crystals composed of air spheres in a @xmath169 matrix .
in contrast to the present study that has been performed on a _ single _ opal sample where the stop - band was `` swept '' by changing the angle of incindence , @xmath39 ; in ref .
, on the contrary the incident angle was kept close to normal . under this condition , in order to explore the cbs within the stop - band frequency range in the polystyrene opals , the authors measured cbs on a _ set _ of samples with different sphere diameters using three different wavelenghts for the incident light beam . also the band edge frequencies were determined from _ reflectivity _ measurements , rather than by cbs as in our work . even under these assumptions , the authors failed to measure cbs cones _ inside _ the stop - band.@xcite for air - sphere @xmath169 crystal , however , the authors@xcite used a setup allowing continuous tuning of @xmath10 .
they observed that between the red and blue edges of the stop - band , @xmath8 _ decreases _ by a factor of @xmath170 .
this observation , which actually contradicts our own study , was explained on the basis of a previous theoretical work.@xcite it was claimed that for light frequencies above the lower band - edge the electromagnetic wave , in course of the diffusion process , can occasionally `` hit '' the boundary interface at the bragg angle @xmath63 ; thus rather than exiting the sample , it experiences instead the bragg diffraction to the interior . as a result of such events
the diffusive trajectories become longer ( on average ) leading to the narrowing of the cbs cone . in light of the cbs model that was outlined in section iii
, the above explanation corresponds to the modification of the propagator @xmath86 in eq .
( [ albedo ] ) .
in contrast , our analysis suggests that @xmath88 is not appreciably affected by the bragg conditions , whereas the prime effect of the photonic band structure is the modification of the surviving probability densities @xmath83 and @xmath171 in eq .
( [ albedo ] ) . the apparent contradiction between our results and those of ref .
can be probably accounted for by the difference in the `` strenghts''of the bragg diffraction process , or @xmath11 in synthetic @xmath14 opals and air - sphere crystals in @xmath169 .
we note , however that the relative width of the bragg reflectivity peaks in both samples is comparable @xmath172 . , ed .
by c. m. soukoulis ( plenum , new york , 1993 ) ; j. d. joannopoulos , r. d. meade , and j. n. winn , _ photonic crystals : molding the flow of light _ ( princeton university press , princeton , new jersey , 1995 ) .
|
we studied coherent backscattering ( cbs ) of light from opal photonic crystals in air at different incident inclination angles , wavelengths and along various @xmath0 $ ] directions inside the opals .
similar to previously obtained cbs cones from various random media , we found that when bragg condition with the incident light beam is not met then the cbs cones from opals show a triangular line shape in excellent agreement with light diffusion theory . at bragg condition , however , we observed a dramatic broadening of the opal cbs cones that depends on the incident angle and @xmath0 $ ] direction .
this broadening is explained as due to the light intensity decay in course of propagation along the bragg direction _ before the first _ and _ after the last _ scattering events .
we modified the cbs theory to incorporate the attenuation that results from the photonic band structure of the medium .
using the modified theory we extract from our cbs data the light mean free path and bragg attenuation length at different @xmath0 $ ] .
our study shows that cbs measurements are a unique experimental technique to explore photonic crystals with disorder , when other spectroscopical methods become ambiguous due to disorder - induced broadening .
|
the interest in millimeter wave communications has been tremendously increased as a viable technology for fifth generation wireless cellular systems .
this is due to the fact that millimeter wave communications support the very high data rates necessary for broadband and multimedia communications thanks to the availability of large bandwidth at the high frequencies .
however , communications at these high frequencies suffer from two main drawbacks .
the first is that the millimeter wave signal suffers from severe pathloss . to overcome this
, there is an active research going on designing beamforming techniques in order to extend the signal range and enable communication between the targeted transmitter and receiver @xcite-@xcite .
the second drawback is that the millimeter wave signal gets severely attenuated in the case of non line of sight ( nlos ) @xcite,@xcite . to improve communications in case of nlos , the use of intermediate relays that have los ( or in general a better ) signal with the source , the destination or among each others
is suggested .
hence constructing a path between the source and destination using those relays improves the source - destination communication . in this paper , we focus on the second challenge and attempt to design a multihop relaying technique for a millimeter wave networks .
relaying in general is a well studied topic in wireless communications , and there is a vast literature covering multihop relaying .
however , multihop relaying in the context of millimeter wave communications is yet a new topic , and there has been still few works that deal with this issue .
the work in @xcite selects for a given source - destination pair employing millimeter wave rf the best relay within the beamwidth of the source to assist in the transmission in case of non line of sight ( nlos ) between the source and the destination .
also based on this relay selection mechanism , a scheduling algorithm is presented for the case when multiple source - destination pairs are present . in @xcite a centralized algorithm for multihop relaying routing that takes into account the characteristics of the millimeter wave transmissions is presented . in particular ,
the presence of multiple source - destination pairs , where each source is interested in video streaming to its destination is assumed .
further , the performance is measured in terms of a differentiated quality function of each flow .
the algorithm then finds a feasible route of relays for each source - destination pair , and the objective is to maximize the sum of differentiated quality functions for all flows .
our work considers the multihop routing problem for multiple source - destination pairs employing millimeter wave rf as in @xcite . however , our approach is different because we formulate the multihop relaying problem as a _ network formation game_. network formation games have been recently used for multihop relaying in wireless networks ( see @xcite ) but not yet in particular for millimeter wave networks . in @xcite , an algorithm based on network formation game
is presented that constructs a uplink multicast tree of relays , to which the mobiles can connect to in order to communicate with the base station . in this game , the relays are the players , and their objective is to connect to the tree in such a way that maximizes their utilities , where the utility is measured in terms of per hop delay and bit error rate . in @xcite ,
a network formation game is formulated for multihop relaying in cognitive radio networks .
the game has also a stackelberg approach in which the primary source - destination pairs are considered as the leaders , and the secondary users are considered as the followers , and the objective is to construct a path of secondary users ( that act as relays ) between each source - destination pair so as to improve its transmission and to eventually give the secondary users chance of channel access . in both problems ,
the network formation games are non cooperative i.e. the player moves are based on maximizing their individual utilities .
in contrast , our approach is based on _ cooperative _ network formation games and in particular a coalitional graph game in which the path between each source destination pair is constructed in a distributed fashion i.e. each relay decides on joining the path of a certain source - destination pair .
each group of relays along the same path forms a coalition .
but as opposed to all previously mentioned works , our algorithm achieves _ proportional fairness _ in order to maintain an acceptable quality for every source - destination pair and to ensure fairness in relay assignment among the different source destination pairs .
we consider a set @xmath0 of source - destination pairs @xmath1 , ( @xmath2 , @xmath3 ) where each source @xmath4 has a file of @xmath5 bits to deliver to destination @xmath6 , and a set @xmath7 of deployed relays .
it is assumed that the nodes employ millimeter wave rf .
then , the received power @xmath8 is given by @xmath9 where @xmath10 and @xmath11 are the antenna gains at the transmitting and receiving nodes respectively , @xmath12 and @xmath13 are the pathloss coefficient and exponent , @xmath14 is the transmitted power , and @xmath15 is the distance between the transmitting and receiving nodes .
millimeter wave signals get severely attenuated with distance .
hence , we assume that all nodes employ directional beamforming , and that each pair of communicating nodes engage in a beamstearing algorithm in order to achieve the maximum directivity gains .
it is further assumed that the beamwidth is very small as transmitting at very high frequencies permits very narrow beamforming .
( some current products such as @xcite demonstrate that the beamwidth can be as small as 2 degrees and that interference can be eliminated even with nodes along the same path . )
this makes it very unlikely for two pairs of nodes to interfere with each other and therefore interference is neglected .
additive white gaussian noise with zero mean and power spectral density @xmath16 is assumed to be present at each node .
hence , the received signal to noise ratio ( snr ) is given by @xmath17 and we assume that the achieved rate @xmath18 is related to the snr through shanon s capacity formula i.e. @xmath19 where @xmath20 is the available bandwidth .
further , millimeter wave signals can get severely attenuated with blockage , and thus the signal can get considerably degraded in the case of non line of sight ( nlos ) .
hence , the channel quality between any pair of nodes is dependent whether a line of sight ( los ) signal exists or not . in particular , measurements ( such as in @xcite ) have shown that different pathloss models exist for the los and the nlos cases .
we define @xmath21 and @xmath22 to be the pathloss coefficient and exponent respectively for the nlos case and @xmath23 and @xmath24 to be the pathloss coefficient and exponent for the los case .
since the direct channel between each source - destination may be nlos , the achieved rate using direct transmission may be low and incur significant delay to deliver file from the source and the destination .
the objective is then to devise multihop relaying i.e. to try to find a path between each source - destination pair using relays ( as some relays might have los signal with the source and destination and among each other ) so as to improve the communication between each source - destination .
the performance is assessed by computing the multihop delay i.e. the time spent through the path to deliver the file from the source to its destination . in order to find a path for each source - destination pair , we design an algorithm using a cooperative network formation game or more specifically a coalition graph game in which each relay chooses to connect to one source - destination pair .
our algorithm also ensures proportional fairness among the source - destination pairs .
we formulate our problem as a cooperative network formation game or more specifically a coalition graph game , where the players are the relays .
each relay chooses to assist one source - destination pair by connecting itself along the path between the chosen source and destination in a way that achieves the best performance possible .
thus , a group of relays assisting the same source - destination pair is considered as a coalition .
this leads us to the following definitions : a path @xmath25 between source @xmath4 and destination @xmath6 is a sequence @xmath26 , @xmath27, ... ,@xmath28,@xmath29 , where @xmath30 is the number of relays in the path , @xmath31 and @xmath32 and @xmath27, ... ,@xmath28 are the relays along @xmath25 . in other words , it is the set of edges given by @xmath33 . each relay @xmath34
( which is presently either unused or assisting a source - destination pair @xmath35 ) can decide to perform action @xmath36 and assist source - destination @xmath37 by inserting itself between two consecutive nodes along the path of ( @xmath38 ) in a way that achieves the minimum possible multihop delay .
in other words , if @xmath39 is the current path between @xmath40 and @xmath41 , @xmath42 is the path formed by inserting relay @xmath18 between nodes @xmath43 and @xmath44 along the path : @xmath45 relay @xmath34 inserts itself between nodes @xmath46 and @xmath47 such that @xmath48 where @xmath49 is the multihop delay along path @xmath42 .
we denote by @xmath50 the resulting path . also , we define action @xmath51 where the relay decides not to assist any pair .
the action set for each relay is then @xmath52 .
we are interested in allocating the relays to the source - destination pairs in a fair way so as to avoid situations in which all relays would be allocated to a few source - destination pairs ( that have better channels with the relays ) enjoying very enhanced performance , while other source - destination pairs would not be adequately assisted by the relays and get poor performance .
one approach in cooperative game theory is the nash bargaining solution ( nbs ) @xcite . in the nbs ,
the objective is to choose the strategies of the players that maximize the following objective function , commonly known as the nash product : @xmath53 , where @xmath54 is the number of players and @xmath55 the utility for player @xmath56 when all players take actions represented by vector @xmath57 and vector @xmath58 is known as the disagreement point , where each @xmath59 corresponds to the utility value of player @xmath56 when no agreement is reached .
more often , the value of the @xmath6s are assumed to be zero .
using this assumption and by taking the logarithm of the objective function we get our objective function : @xmath60 which corresponds to the proportional fairness @xcite .
the value ( or utility ) of the coalition of each source - destination pair ( @xmath61 ) is expressed in terms of the multihop delay , i.e. the time required to deliver the file from @xmath4 to @xmath6 , which we denote @xmath62 , with @xmath25 the path corresponding to the relays in the coalition @xmath63 .
hence , in order to minimize the proportional fair sum of @xmath64s , we define the value of coalition @xmath63 to be @xmath65 then , maximizing the sum of the values of the coalitions amounts to maximizing the proportional fairness of the utilities of the source - destination users with their utilities being proportional to their transfer rates @xmath66 .
we consider the multihop delay as being the sum of delays of all edges from @xmath4 to @xmath6 .
hence , it is given by the following expression@xmath67 where @xmath30 is the number of relays along the path , @xmath5 is the size of the file to transfer , @xmath68 is the rate achieved between node @xmath69 and node @xmath70 .
again , we assume that node @xmath26 is the source @xmath4 and node @xmath29 is the destination @xmath6 .
any other node @xmath69 ( @xmath71 ) is the @xmath72 relay along the path .
note that in the delay expression of equation [ mhdelay ] , it is assumed that each relay decodes the whole file before transmitting it to the next relay along the path .
the multihop delay can be improved in the case where the file is divided into packets and transmissions occur packet by packet .
however , the expression is more complicated to handle and relies on assumptions on the packet based system ( see , e.g. @xcite and @xcite , in which packets arrive at each source at a certain rate , and the average delay at each hop is computed based on modeling the packet service system as @xmath73 queue ) .
hence , our expression constitutes a simple upper bound on the delay for packet based transmissions .
the following theorem shows how the optimal actions of the relays maximize the proportional fairness sum . if each relay @xmath34 chooses to connect to the source - destination pair @xmath74 that maximizes its marginal contribution ( i.e. that has the maximum @xmath75 if it is positive and to chooses not to assist any pair - i.e. to remain unused - otherwise ) , the corresponding equilibria are the maximizers of the proportional fair sum .
we consider the game of transferable utility in which the utility ( welfare ) of each relay in a coalition is proportional to the collective contribution of all relays in the coalition i.e. the coalition value .
hence , when relay @xmath34 is assisting source - destination @xmath74 , the coalition value is divided equally among the @xmath30 relays in the coalition : @xmath76 .
we also set the utility of any unused relay to be zero ( as if they were belonging to a dummy path with null coalition value ) .
now , we define the repercussion utility of relay @xmath34 in coalition @xmath63 as @xmath77 by substituting the values of the utilities @xmath78 , @xmath79 , and @xmath80 into @xmath81 , we get : @xmath82 recall that we assume that the utility of each unused relay is zero .
the importance of this assumption is to prevent the relay to join a path that it would harm .
this happens in the case when all its repercussion utilities for all paths are negative .
due to our assumption that the network is interference free , the value of each coalition is not dependent on the other coalitions .
it has been proven in @xcite that a coalition game that satisfies this property and where repercussion utilities are used is an exact potential game with the sum of the original utilities as the potential function .
hence , our game is a potential game where the potential function is the negative of the proportional fair sum of delays of all source - destination pairs .
the result in theorem 1 follows since an exact potential game has the property that ( at least ) one pure nash equilibrium exists and that the nash equilibria are the local maximizers of the potential function .
we present in algorithm [ alg1 ] a distributed algorithm for our cooperative network formation game in which the relays select to connect to a particular source - destination @xmath74 based on proportional fairness .
we assume that the relays have full knowledge of the network topology , and that they store the current value of the multihop delay as well as the current path of each source - destination pair .
we assume the size of the broadcasted messages is small ( i.e. do not require high bit rates ) and thus omnidirectional transmission is used during this phase and that a round - robin algorithm is chosen to select each relay @xmath34 periodically .
note that we introduce some randomness as we allow each relay to take some non - optimal decision .
indeed , each relay joins the path that yields the maximum repercussion utility with probability @xmath83 , where @xmath84 is commonly known as the mutation probability @xcite .
otherwise , the relay will randomly join the path of any other source - destination pair .
due to the mutation probability , the evolution of paths of all source - destination pairs forms a markov chain which is irreducible and aperiodic .
hence , it has a unique stationary distribution .
it is shown in @xcite that as @xmath84 tends to zero , the process converges to a unique limiting distribution . also , since our game is a finite exact potential game , it admits one or several pure nash equilibria that are the local maximizers of the nash product .
hence , as @xmath84 tends to zero , the algorithm converges to a deterministic nash equilibrium . in order to reach the global maximum of the nash product , it is useful to incorporate gibbs sampling techniques ( such as the algorithm in @xcite ) .
the drawback of gibbs sampling is that the convergence time might be unacceptably large for some scenarios .
in order to simulate our multihop relaying algorithm , we consider @xmath85 source - destination pairs and @xmath86 deployed relays .
the coordinates of all nodes are generated randomly according to a uniform distribution on a @xmath87 meters rectangular grid .
we set all direct channels between source - destination pairs to be nlos .
all other links are chosen los or nlos randomly .
the probability of los , @xmath88 is taken according to the statistical blockage model of @xcite : @xmath89 , where @xmath15 is the distance between the nodes and @xmath90 is a parameter known as the average los range of the network and is related to the density and average blockage sizes and set to @xmath91 meters . as for the pathloss models for both cases of los and nlos , we use the the values obtained from the measurements in @xcite , i.e. @xmath92 and @xmath93 for the pathloss exponents , @xmath94 for the pathloss coefficients and @xmath95 for all antennas gains . the transmission power for all nodes
is set to be @xmath96 watts and the awgn variance is set to @xmath97 dbm .
the bandwidth is set to @xmath98 ghz and the size of all files is @xmath99 gb . for the algorithm , we choose the mutation probability @xmath100 .
[ multihop1],width=8,height=226 ] , width=8,height=226 ] in order to investigate the potential benefits of proportional fairness , we compare its results to a modified version of the algorithm in which each relay joins the path of the source - destination pair that has the minimum delay .
this modified version can be interpreted as a greedy approach whose convergence points are the nash equilibria of the system
. figures [ multihop1 ] and [ multihop2 ] show the paths formed between each source destination pair by using our multihop relaying algorithm and the modified minimum delay algorithm respectively .
the red , green , and blue circles represent source nodes 1,2 , and 3 respectively .
the red , green , and blue squares represent destination nodes 1,2 , and 3 respectively , and the black circles represent the relay nodes . in this run ,
most of the formed edges are los .
table [ delay1 ] shows the delay values computed for the cases of direct transmission , the proportional fair multihop relaying algorithm , and the modified minimum delay algorithm respectively .
first , it is easy to see that the delay values have significantly decreased when multihop relaying ( for both cases of proportional fairness and minimum delay ) is employed , which confirms the benefits of the multihop relaying algorithm in improving the transmission of the different source destination pairs . also by comparing the delay values obtained from the two algorithms
, we find that when using the minimum delay algorithm , the delay of @xmath101 has slightly dropped from 0.2505 sec ( for the proportional fairness algorithm case ) to 0.1058 sec while the delay of @xmath102 has increased from 0.1428 sec ( for the proportional fairness case ) to 1.2451 sec .
further , we compute the variances of the delays of all paths for both the proportional fairness algorithm and the minimum delay algorithm .
we find out the the variance is 0.282 for the minimum delay algorithm while it is 0.0028 for the proportional fairness algorithm .
this shows that proportional fairness can provide a better distribution of the relays among the source destination pairs .
.delay values ( in seconds ) [ cols="<,^,^,^",options="header " , ] further , we run both both algorithms for a thousand times . in each simulation
, we randomly generate the coordinates of the nodes and choose which links are los .
after each run , we record the sum of delays of all paths for both algorithms . then , we compute the average sum of delays for each algorithm out of the 1000 runs . we find that the average is 6.8 sec for our algorithm while it is 9.2 sec for the minimum delay algorithm and 86 sec when no multihoping is used ( i.e. through the direct path ) .
these values demonstrate the power of proportional fairness compared to the minimum delay approach : while the minimum delay approach can slightly benefit to some users , it does so at the cost of a decreased global performance .
both techniques exhibit excellent global performance compared to the direct transmission .
further , although it converges to a local optima of the nash product , we observe very good performance metrics . hence , while gibbs sampling techniques could ensure convergence to global optima , it is anticipated that the cost of the convergence time will not be compensated by significant performance improvement .
we have considered a cooperative network formation algorithm in order to construct a multihop path through relays to improve the transmission of source - destination pairs employing millimeter wave rf . also , we have considered proportional fairness in our algorithm . due to the assumption of negligible interference in millimeter wave networks
, we could show that our network formation game can be turned into a potential game , whose nash equilibria maximize the proportional fair sum of the transmission rates .
we further proposed a distributed algorithm and assessed its performance by numerical simulations .
the results show the considerable performance improvement brought by multihop relaying especially in the case where the sources and destinations are nlos .
they also confirm the benefits of proportional fairness in achieving balanced allocations among the source - destination pairs while maintaining a good overall performance of the system .
w. roh , j. seol , j. park , b. lee , j. lee , y. kim , j. cho , k. cheun , and f. aryanfar , `` millimeter - wave beamforming as an enabling technology for 5 g cellular communications : theoretical feasibility and prototype results '' , _ ieee communications magazine _ , vol . 52 , no . 2 , pp .
106 - 113 , 2014 .
s. hur , t. kim , d.j .
love , j.v .
krogmeier , t.a .
thomas , and a. ghosh , `` millimeter wave beamforming for wireless backhaul and access in small cell networks '' , _ ieee transactions on communications _ , vol . 61 , no . 10 , pp .
4391 - 4403 , 2013 .
w. saad , z. han , t. basar , m. debbah , and a. hjrungnes , network formation games among relay stations in next generation wireless networks ,
_ ieee trans . on comm .
59 , no . 9 , pp . 2528 - 2542 , 2011 . f.p .
kelly , a.k .
maulloo and d.k.h .
tan , rate control for communication networks : shadow prices , proportional fairness and stability
, _ journal of the operational research society _ vol .
49 , pp . 237252 , 1998 .
rappaport , f. gutierrez , e. ben - dor , j.n .
murdock , y. qiao , and j.i .
tamir , broadband millimeter - wave propagation measurements and models using adaptive - beam antennas for outdoor urban cellular communications ,
_ ieee trans . on antennas and propagations _ , vol .
61 , no . 4 , pp .
1850- 1859 , 2013 .
|
millimeter wave channels suffer from considerable degradation in the channel quality when the signal is non line of sight ( nlos ) between the source and the destination .
multihop relaying is thus anticipated to improve the communication between a source and its destination .
this is achieved by transmitting the signal to a sequence of relays in which a line of sight ( los ) signal exists between two nodes along the path , or more generally when the signal is better than the transmitted signal directly from the source to the destination . in this paper
, we consider a millimeter wave network composed of multiple source - destination pairs and a set of deployed relays .
we formulate the problem of multihop relaying as a cooperative network formation game in which each relay chooses which source - destination pair to assist in order to improve the end - to - end performance , that is , the multihop delay between the source and the destination .
further , we present an algorithm based on the nash bargaining solution to ensure fairness among the different source - destination pairs and assess its efficiency on numerical simulations .
|
the existence of dark matter ( dm ) is confirmed by a lot of astrophysical observations . however , there is no candidate in the standard model ( sm ) particle content .
one of properties required for dm is a long lifetime which should be longer than the age of our universe . to realize this stability
, an additional symmetry should be exist .
for example , when the dm particle has odd parity and others do even parity under a @xmath4 symmetry , the dm particle can be stable .
recent analyses of the observation of x - ray from the perseus galaxy clusters and the andromeda galaxy by the xmm - newton observatory reported an unknown x - ray line spectrum around @xmath1 @xcite .
the origin of this line can not be explained by astrophysical phenomena so far .
therefore , we try to explain the x - ray line spectrum in the particle physics . an interesting explanation is given by decay of a light dm particle .
after these announcements , a lot of articles @xcite-@xcite in which dm decays into a photon have been researched .
one attractive candidate for a decaying dm is the right - handed neutrino , @xmath5 , with mass around @xmath6 . in this scenario ,
the experiments show the mixing angle between left- and right - handed neutrinos , @xmath7 , to be @xmath8 .
on the other hand , it is well known that the right - handed neutrinos can generate the tiny neutrino masses through the type - i seesaw mechanism @xcite .
the lagrangian for the seesaw mechanism is described as @xmath9 with @xmath10 .
@xmath11 and @xmath12 are lepton and higgs doublets , respectively . here
, @xmath13 is the majorana mass matrix of the right - handed neutrinos . @xmath14 and @xmath15 represent the flavor of left - handed leptons and right - handed neutrinos , respectively .
if the right - handed neutrinos have a large mass hierarchy , it can explain several issues such as the tiny neutrino masses , dm , and baryon asymmetry of the universe simultaneously ( _ e.g. _ see @xcite ) .
such a hierarchical structure among the majorana masses can be realized by _
e.g. , _ the froggatt - nielsen mechanism @xcite , the split seesaw mechanism @xcite , a lepton flavor symmetry @xcite , an extra - dimensional extension @xcite and an extended b - l structure @xcite , and so on .
however , once the right - handed neutrino with the mass of order kev gives the solar or atmospheric neutrino mass scale through the type - i seesaw mechanism , the size of left - right mixing angle can not be explained .
after breaking the electroweak symmetry , we have the dirac mass term , @xmath16 with @xmath17 , and the majorana mass term .
the seesaw mechanism can explain the tiny neutrino masses as @xmath18 with @xmath19 even under hierarchical mass spectra of majorana masses .
the mixing angle between the left- and right - handed neutrinos is given by @xmath20 .
when the lightest majorana mass is of order @xmath21 , the seesaw relation induces the left - right mixing angle as @xmath22 it is worth noting that the suitable magnitude of the mixing angle , @xmath23 , never achieves unless giving up the seesaw relation .
it is because a tiny neutrino yukawa couplings only for the first generation with the left - right mixing angle , @xmath23 induces the mass scale of @xmath24 , which is much smaller than the atmospheric neutrino mass scale , @xmath19 . in this work , we suggest a new possibility to realize the suitable left - right mixing due to the deviation from the seesaw relation by an extending sm .
we introduce a dm - philic higgs field , @xmath25 , which constitutes a yukawa interaction of dm with ordinary sm fields .
the results from x - ray observation experiments indicate the existence of an unknown particle which has a mass around @xmath6 .
if it is a right - handed neutrino , it should not contribute to the active neutrino mass scale through the seesaw relation due to its tiny mass and the suitable mixing angle as discussed above .
such property could imply the right - handed neutrino dm ( rh@xmath26 dm ) has a different interaction from other particles .
we introduce an extra field : a rh@xmath26dm - philic higgs boson , @xmath25 . by introducing @xmath25 ,
the relevant lagrangian is given by @xmath27 where @xmath28 and we suppose that @xmath29 at this moment .
therefore , we can realize a small mixing of @xmath30 by taking small vacuum expectation value ( vev ) , @xmath31 . under this simple setup
, the mixing angle can be represented as @xmath32 where the third and fourth equalities are given by eq .
( [ eq : fn ] ) .
one can see that when @xmath33 is @xmath34 , the observed x - ray flux can be explained . therefore , @xmath31 is needed to be @xmath35 .
instinctively , the deviation from the seesaw formula is realized by the small value of @xmath31 rather than a fine - tuning of coupling constants .
this is our main idea and this possibility is really simple . here
, we show an example of realistic model .
the charge assignment of each field is summarized in tab .
[ tab : assign ] .
.the charge assignments for each field .
@xmath36 is an arbitrary natural number . [
cols="^,^,^,^,^,^,^",options="header " , ] we also introduce gauged @xmath37 which is responsible not only for the generation of the majorana masses of rh@xmath26s but also the dm production .
two global symmetries @xmath38 and @xmath4 provide the sufficient mass hierarchy and couplings between @xmath30 and others . @xmath37 and @xmath38 are spontaneously broken by a b@xmath39l higgs boson , @xmath40 and another higgs boson , @xmath41 , respectively .
those breaking scales are denoted as @xmath42 and @xmath43 , respectively .
the lagrangian under these symmetries are given as @xmath44 where @xmath45 contains the terms under the sm gauge symmetries , @xmath46 and @xmath47 are dimension - less coupling constants , and @xmath48 represents a cutoff scale .
next , we consider the higgs potential . for the realistic model ,
the small magnitude of @xmath31 is essential .
we show the validity of this point below .
the corresponding higgs potential of this model is @xmath49 \,,\label{v}\end{aligned}\ ] ] where @xmath50 terms break @xmath4 symmetry softly .
thus , the lightest rh@xmath26 becomes dm which can decay .
a stationary condition @xmath51 gives @xmath52 where we assume @xmath53 .
note that a tiny value of @xmath31 can be realized when @xmath54 in eq .
( [ eq : vdm ] ) . after
each scalar field takes its vev , the dirac and majorana mass matrices can be written as @xmath55 the hierarchical structure of the majorana masses as @xmath29 and the different coupling property in the dirac mass term are realized .
the seesaw mechanism provides the mass matrix for the active neutrinos , @xmath56 , as @xmath57 where @xmath58 are obtained by the diagonalization of eq .
( [ eq : majorana ] ) .
@xmath56 can be diagonalized by the so - called mns matrix , @xmath59 , as @xmath60 . roughly speaking
, the order of @xmath61 can be estimated in the normal mass ordering case as @xmath62 where @xmath63 is of @xmath64 which corresponds to the atmospheric neutrino mass scale and the squared difference between @xmath65 and @xmath66 explains the solar mass - squared difference .
thus , the @xmath37 breaking scale can be determined as @xmath67 .
note that we can reproduce the observed mixing angles in the mns matrix by taking proper values of neutrino yukawa couplings , @xmath68 .
we are at the position to investigate suitable model parameters by cosmological observations .
we assume that all of the dimension - less couping constants are unity just for simplicity and the cutoff scale is the reduced planck scale , @xmath69 .
firstly , we estimate the breaking scale of the extra symmetries by the lightest right - handed neutrino mass , @xmath70 , which is given by eq.([eq : majorana ] ) as @xmath71 the left - right mixing angle of the neutrinos is estimated as @xmath72 from the observations , each value is constrained as @xcite @xmath73 and here , we fix these as @xmath74 and @xmath75 which are the best fit values of the experiments .
we first mention to the scale of @xmath31 .
one can rewrite eq .
( [ eq : mixing ] ) by the seesaw neutrino mass , @xmath76 , as @xmath77 this relation is free from the scale of @xmath43 and the charge @xmath36 . and , eq .
( [ eq : mixing ] ) determines @xmath31 as @xmath78 this value can be obtained when @xmath79 in eq .
( [ eq : vdm ] ) . meanwhile , the remaining ambiguities of this model are the values of @xmath43 and @xmath36 .
these values are also determined by the constraint of the dm mass .
the correlation between @xmath43 and @xmath36 is shown in fig .
[ fig : n - vfn ] . for example , when @xmath80 , the breaking scale of @xmath38 should be @xmath81 to explain dm properties .
on the other hand , when @xmath36 becomes large , the ordinary froggatt - nielsen mechanism @xcite can work for generating the fermion mass hierarchy without any fine - tuning of the coupling constants .
we would like to give a brief comment on the production mechanism of dm .
one simple production mechanism is the so - called dodelson - widrow mechanism @xcite in which rh@xmath26 dm is produced via the mixing after the big bang nucleosynthesis .
however , the required magnitude of the mixing angle for this mechanism is much larger than the experimental bound .
thus , we need to consider other possibilities .
one possibility is the production via the b - l gauge boson s - channel exchange @xcite . in this case , the relic abundance is evaluated as @xmath82 where @xmath83 is the dimensionless hubble parameter , and @xmath84 denotes the reheating temperature of the universe .
when we adopt the thermal production , the data from the lyman-@xmath85 forest @xcite-@xcite should be taken into account .
however , this bound is relaxed by the late time entropy dilution @xcite .
the total amount of thermal abundance seems to be smaller than eq .
( [ eq : dm_abun ] ) , but we can adjust the abundance by controlling the reheating temperature . although the sufficient amount of dm can also be explained in our setup , we skip the detailed estimation here . moreover , in this setup
, we can also explain the baryon asymmetry of the universe via the so - called leptogenesis scenario @xcite by non - thermal decay of the second lightest right - handed neutrino , @xmath86 .
the lightest one is a candidate for dm as discussed above and the other right - handed neutrinos can not decay into this lightest right - handed neutrino due to the difference of the @xmath4 charge shown in tab .
[ tab : assign ] .
such setup is the same as the split seesaw case @xcite .
lastly , we would like to add a comment on the detectability of rh@xmath26dm - philic higgs boson , @xmath25 . @xmath25
pair production through the decay of @xmath12 is possible when the mass of @xmath25 is less than the half of the sm higgs boson mass . in this case
, the invisible decay of the higgs boson might be seen in the collider experiment depending on the strength of the portal couplings @xcite . on the other hand , @xmath25 is hardly produced because of its odd parity under @xmath4 symmetry with @xmath87 .
detailed quantitative estimation is the future works at the moment .
we have suggested a new model in which the right - handed neutrino dark matter is naturally included .
it can explain the recently observed @xmath1 x - ray signal by the xmm - newton observatory .
one of the most impressive features of our model is the existence of the dark matter - philic higgs field , @xmath25 .
the smallness of the left - right mixing angle of the lightest neutrinos is ensured by the smallness of the vacuum expectation value of this dark matter - philic higgs field . on the other hand ,
the smallness of the mass of the dark matter can be clearly understood by the froggatt - nielsen mechanism .
we have shown that the cosmological observations fix the scale of @xmath31 as @xmath2 and the intriguing point is that the uncertainty of this model to realize the dm mass is not essential to determine this scale .
the authors thank to alexander merle and kei yagyu for giving us valuable comments .
this work is partially supported by scientific grant by ministry of education and science , no .
24540272 . the work of r.t .
is supported by research fellowships of the japan society for the promotion of science for young scientists .
100 e. bulbul , m. markevitch , a. foster , r. k. smith , m. loewenstein and s. w. randall , astrophys .
j. * 789 * ( 2014 ) 13 [ arxiv:1402.2301 [ astro-ph.co ] ] .
a. boyarsky , o. ruchayskiy , d. iakubovskyi and j. franse , arxiv:1402.4119 [ astro-ph.co ] .
h. ishida , k. s. jeong and f. takahashi , phys .
b * 732 * ( 2014 ) 196 [ arxiv:1402.5837 [ hep - ph ] ] .
d. p. finkbeiner and n. weiner , arxiv:1402.6671 [ hep - ph ] .
t. higaki , k. s. jeong and f. takahashi , phys .
b * 733 * ( 2014 ) 25 [ arxiv:1402.6965 [ hep - ph ] ] .
j. jaeckel , j. redondo and a. ringwald , phys .
d * 89 * ( 2014 ) 103511 [ arxiv:1402.7335 [ hep - ph ] ] .
h. m. lee , s. c. park and w. -i .
park , arxiv:1403.0865 [ astro-ph.co ] . j. -c .
park , s. c. park and k. kong , phys .
b * 733 * ( 2014 ) 217 [ arxiv:1403.1536 [ hep - ph ] ] .
m. t. frandsen , f. sannino , i. m. shoemaker and o. svendsen , jcap * 1405 * ( 2014 ) 033 [ arxiv:1403.1570 [ hep - ph ] ] .
s. baek and h. okada , arxiv:1403.1710 [ hep - ph ]
. k. -y .
choi and o. seto , phys .
b * 735 * ( 2014 ) 92 [ arxiv:1403.1782 [ hep - ph ] ] .
m. cicoli , j. p. conlon , m. c. d. marsh and m. rummel , phys .
d * 90 * ( 2014 ) 023540 [ arxiv:1403.2370 [ hep - ph ] ] .
t. tsuyuki , phys .
d * 90 * ( 2014 ) 013007 [ arxiv:1403.5053 [ hep - ph ] ] .
r. allahverdi , b. dutta and y. gao , phys .
d * 89 * ( 2014 ) 127305 [ arxiv:1403.5717 [ hep - ph ] ] . c. kolda and j. unwin , phys .
d * 90 * ( 2014 ) 023535 [ arxiv:1403.5580 [ hep - ph ] ] .
s. p. liew , jcap * 1405 * ( 2014 ) 044 [ arxiv:1403.6621 [ hep - ph ] , arxiv:1403.6621 ] .
p. ko , z. kang , t. li and y. liu , arxiv:1403.7742 [ hep - ph ] .
s. v. demidov and d. s. gorbunov , arxiv:1404.1339 [ hep - ph ] .
f. s. queiroz and k. sinha , phys . lett .
b * 735 * ( 2014 ) 69 [ arxiv:1404.1400 [ hep - ph ] ] .
e. dudas , l. heurtier and y. mambrini , arxiv:1404.1927 [ hep - ph ] .
k. s. babu and r. n. mohapatra , phys .
d * 89 * ( 2014 ) 115011 [ arxiv:1404.2220 [ hep - ph ] ] .
modak , arxiv:1404.3676 [ hep - ph ] .
j. m. cline , y. farzan , z. liu , g. d. moore and w. xue , phys .
d * 89 * ( 2014 ) 121302 [ arxiv:1404.3729 [ hep - ph ] ] .
h. okada and t. toma , arxiv:1404.4795 [ hep - ph ] .
j. p. conlon and f. v. day , arxiv:1404.7741 [ hep - ph ] .
d. j. robinson and y. tsai , arxiv:1404.7118 [ hep - ph ] .
s. chakraborty , d. k. ghosh and s. roy , arxiv:1405.6967 [ hep - ph ] . n. chen , z. liu and p. nath , arxiv:1406.0687 [ hep - ph ] .
h. ishida and h. okada , arxiv:1406.5808 [ hep - ph ] .
b. dutta , i. gogoladze , r. khalid and q. shafi , arxiv:1407.0863 [ hep - ph ] .
a. merle , int .
j. mod .
d * 22 * ( 2013 ) 1330020 [ arxiv:1302.2625 [ hep - ph ] ] .
p. minkowski , phys .
b * 67 * , 421 ( 1977 ) , t. yanagida , in proceedings of the _ `` workshop on the unified theory and the baryon number in the universe '' _ , tsukuba , japan , feb .
13 - 14 , 1979 , edited by o. sawada and a. sugamoto , kek report kek-79 - 18 , p. 95 , and _ `` horizontal symmetry and masses of neutrinos '' _ , prog .
* 64 * ( 1980 ) 1103 ; m. gell - mann , p. ramond and r. slansky , in
_ `` supergravity '' _ ( north - holland , amsterdam , 1979 ) _
eds_. d. z. freedom and p. van nieuwenhuizen , print-80 - 0576 ( cern ) .
t. asaka , s. blanchet and m. shaposhnikov , phys .
b * 631 * ( 2005 ) 151 [ hep - ph/0503065 ] .
a. kusenko , f. takahashi and t. t. yanagida , phys .
b * 693 * , 144 ( 2010 ) [ arxiv:1006.1731 [ hep - ph ] ] .
j. barry , w. rodejohann and h. zhang , jcap * 1201 * ( 2012 ) 052 [ arxiv:1110.6382 [ hep - ph ] ] . c. -s .
chen and r. takahashi , eur .
j. c * 72 * ( 2012 ) 2089 [ arxiv:1112.2102 [ hep - ph ] ] . c. d. froggatt and h. b. nielsen , nucl .
b * 147 * , 277 ( 1979 ) .
a. merle and v. niro , jcap * 1107 * ( 2011 ) 023 [ arxiv:1105.5136 [ hep - ph ] ] .
m. lindner , a. merle and v. niro , jcap * 1101 * ( 2011 ) 034 [ erratum - ibid .
* 1407 * ( 2014 ) e01 ] [ arxiv:1011.4950 [ hep - ph ] ] .
r. takahashi , ptep * 2013 * ( 2013 ) 6 , 063b04 [ arxiv:1303.0108 [ hep - ph ] ] .
h. ishida , k. s. jeong and f. takahashi , phys .
b * 731 * ( 2014 ) 242 [ arxiv:1309.3069 [ hep - ph ] ] . c. d. froggatt and h. b. nielsen , nucl .
b * 147 * ( 1979 ) 277 .
s. dodelson and l. m. widrow , phys .
* 72 * ( 1994 ) 17 [ hep - ph/9303287 ] .
s. khalil and o. seto , jcap * 0810 * ( 2008 ) 024 [ arxiv:0804.0336 [ hep - ph ] ] .
m. viel , j. lesgourgues , m. g. haehnelt , s. matarrese and a. riotto , phys .
* 97 * ( 2006 ) 071301 [ astro - ph/0605706 ] .
u. seljak , a. makarov , p. mcdonald and h. trac , phys .
* 97 * ( 2006 ) 191303 [ astro - ph/0602430 ] .
a. boyarsky , j. lesgourgues , o. ruchayskiy and m. viel , jcap * 0905 * ( 2009 ) 012 [ arxiv:0812.0010 [ astro - ph ] ] .
a. boyarsky , j. lesgourgues , o. ruchayskiy and m. viel , phys .
* 102 * ( 2009 ) 201304 [ arxiv:0812.3256 [ hep - ph ] ] . f. bezrukov , h. hettmansperger and m. lindner , phys .
d * 81 * ( 2010 ) 085032 [ arxiv:0912.4415 [ hep - ph ] ] .
m. nemevsek , g. senjanovic and y. zhang , jcap * 1207 * ( 2012 ) 006 [ arxiv:1205.0844 [ hep - ph ] ] . m. raidal and a. strumia , phys .
b * 553 * ( 2003 ) 72 [ hep - ph/0210021 ] .
n. haba and k. tsumura , jhep * 1106 * ( 2011 ) 068 [ arxiv:1105.1409 [ hep - ph ] ] .
|
we suggest a new model of @xmath0 kev right - handed neutrino dark matter inspired by a recent observation of @xmath1 x - ray line signal in the xmm - newton observatory .
it is difficult to derive the tiny masses with a suitable left - right mixing of the neutrino in a framework of ordinary simple type - i seesaw mechanism .
we introduce a new higgs boson , a dark matter - philic higgs boson , in which the smallness of its vacuum expectation value can be achieved .
we investigate suitable parameter regions where the observed dark matter properties are satisfied .
we find that the vacuum expectation value of dark matter - philic higgs boson should be about @xmath2
. 0.6 cm su - het-09 - 2014 * @xmath3 dark matter - philic higgs for 3.5 kev x - ray signal * 1.2 cm naoyuki haba , hiroyuki ishida and ryo takahashi 0.4 cm _ graduate school of science and engineering , shimane university , matsue , 690 - 8504 japan _ 0.2 cm .5 in
|
the egret _ ( energetic gamma ray experiment telescope ) _ detector on board of the _ compton gamma - ray observatory _ discovered strong mev @xmath0-ray emission from 66 blazars , mainly from flat spectrum radio quasars and unidentified flat spectrum radio sources @xcite .
ground - based cherenkov telescopes discovered tev @xmath0-ray emission from 6 blazars , 4 of which have not been detected by egret .
all the established tev detections belong to the class of bl lac objects , blazars with relatively low luminosity but with spectral energy distributions ( seds ) that peak at extremely high energies ( see table [ detect ] ) .
+ .tev blazars with a significant detection by at least 2 experiments ( as of july 2003 ) . [ cols="^,^,^",options="header " , ] the large detection area of cherenkov telescopes of @xmath110@xmath2 m@xmath3 makes it possible to sample the @xmath0-ray lightcurves with a time resolution of several minutes . large amplitude flux variability on 30 min time scales implies that the tev emission originates from a small region very near the supermassive black hole @xcite .
the relativistic doppler factor of the emitting plasma is defined as @xmath4 , where @xmath5 is the bulk lorentz factor of the emitting plasma , @xmath6 is its bulk velocity in units of the speed of light , and @xmath7 is the angle between jet axis and the line of sight as measured in the observer frame .
the requirement that the emitting volume is optically thin for the tev @xmath0-radiation , together with assumptions about the optical to uv energy spectrum of co - spatially emitted synchrotron emission gives a lower limit of @xmath8 9 for the emitting plasma @xcite . the lorentz boost of the observed jet luminosity by the factor @xmath9 makes the rather weak jet emission readily detectable .
+ the main scientific drivers for broadband observations of tev blazars are : ( i ) the study of the matter , energy content , and structure of jets from active galactic nuclei ( agns ) , ( ii ) the time resolved investigation of particle acceleration processes , ( iii ) the research on the connection between the accreting supermassive black hole system and the jet parameters , and ( iv ) the measurement of the total infrared / optical luminosity of the universe .
+ in this paper we give an update on the status of the observations ( sect .
[ obs ] ) and the four major scientific topics ( sects . [ t1][t4 ] ) .
we conclude with an outlook in sect .
we will focus on observations and modeling of tev blazars ( blazars detected by tev instruments ) .
reviews on observations and models of mev / gev blazars ( detected by egret ) have been given in @xcite ; recent overviews of tev astronomy have been given in @xcite .
tev @xmath0-ray observations have achieved a state of maturity with good agreement in fluxes and reasonable agreement in energy spectra between different experiments .
the measurements of the steady flux from the crab nebula with the cat , hegra , and whipple experiments lie within 18.5% from the common mean comparable to the systematic error of about 22% quoted by the groups @xcite .
the spectral indices agree to within 0.2 from the common mean , a difference that is larger than the systematic errors of between 0.04 and 0.06 estimated by the 3 groups .
systematic error on flux and spectral changes from one day to the next are estimated to be about 10% and 0.05 , respectively . +
high - quality tev @xmath0-ray energy spectra have now been measured for mrk 421 , mrk 501 , h 1426 + 428 , and 1es 1959 + 650 . fig .
[ seds ] compares the x - ray and tev @xmath0-ray seds of these 4 sources .
compared to the other 3 blazars , the low - energy ( synchrotron ) component of mrk 421 peaks at lower frequencies .
comparison of the high - energy tev @xmath0-ray energy spectra is not straight forward as it depends on the uncertain extent of extragalactic absorption ( see sect . [ t4 ] ) .
spectral variability has been established for the two strongest ones , mrk 421 @xcite and mrk 501 @xcite .
the significance of measurements of spectral changes can not be overestimated .
spectral changes ( like flux changes ) are free of the still substantially uncertain extent of extragalactic @xmath0-ray absorption .
modeling of simultaneously taken x - ray and @xmath0-ray data with well measured spectral changes in both bands has not yet been performed but has the potential of breaking basically all model degeneracies .
+ several extensive multiwavelength campaigns have been performed .
the correlation of x - ray and tev @xmath0-ray fluxes had first been seen in mrk 421 data @xcite . in the meantime ,
correlations with high statistical significances have been observed from mrk 501 @xcite and from mrk 421 @xcite .
a more recent campaign on the tev blazar 1es 1959 + 650 shows an orphan @xmath0-ray flare without an x - ray or optical counterpart @xcite .
the orphan flare shows that the simplest models ( 1-zone synchrotron self - compton models with a randomly oriented magnetic fields and isotropic electron distributions ) are not able to account for the data .
+ simultaneous measurements of blazar energy spectra with imaging cherenkov telescopes and solar array cherenkov detectors have the potential of producing highly interesting results .
first lightcurves from stacee @xcite and a first energy spectrum from celeste @xcite have been reported recently .
* synchrotron - compton models : * in synchrotron compton ( sc ) models of tev blazars the radio to x - ray emission is produced as synchrotron radiation from a population of non - thermal electrons ( and possibly positrons ) .
the same electron population emits @xmath0-rays through inverse compton processes by electrons scattering synchrotron photons in so - called synchrotron self - compton ( ssc ) models or with `` external '' photons that originate outside the jet in so - called external compton ( ec ) models ( see @xcite and references therein ) . in the case of bl lac objects ,
the lack of strong emission lines is commonly taken as evidence that ambient photon fields are not important , and that ssc models rather than ec models are more likely to explain the data . + a method to distinguish between ssc and ec models is to measure the time lag @xmath10 between x - ray and @xmath0-ray flux variability @xcite . in ssc models
one expects a time - lag of the order of @xmath11 @xmath12 @xmath13 with @xmath11 the radius of the emission volume and @xmath14 the speed of light .
the time lag originates from the fact that a population of freshly accelerated electrons immediately radiates synchrotron emission ; in contrast , the inverse compton component peaks only after the new synchrotron seed photons had time to propagate through the source .
no such time lag has yet been observed with high statistical confidence . + tev blazar data have been modeled intensively with a variety of codes .
the more simpler ones assume 1-zone homogeneous emission volumes and use phenomenological electron energy spectra to fit the broadband data ( e.g. @xcite ) . modeling with time - dependent self - consistently evolved codes ( e.g. @xcite )
assures that the energy spectra are physically realizable from initial acceleration energy spectra .
time dependent modeling has to be used whenever the flare duration is comparable or shorter than the longest time scale of the microprocesses that modify the electron energy spectra ( e.g. acceleration energy gains , radiative energy gains and losses ) .
+ several conclusions can be drawn from the ssc and ec modeling performed by various authors : ( i ) relatively simple 1-component and 2-component ssc models are able to explain most multiwavelength data ; ( ii ) even for the nearest blazars with @xmath15 0.03 , extragalactic absorption has to be taken into account and substantially modifies the regions in the parameter space that correspond to acceptable solutions ( e.g. @xcite ) ; ( iii ) the data prefer high jet doppler factor of between 50 and 100 @xcite ; ( iv ) in the jet frame the energy density of non - thermal particles is much higher than the energy density of the magnetic field ( the jet plasma is a low @xmath16-plasma ) ( e.g. @xcite ) ; ( v ) only a small fraction ( @xmath17 ) of the energy transported by the jet is converted into radiation ( e.g. @xcite ) .
+ the last two conclusions constrain the structure of the jet at its base .
while electromagnetic models of jet formation produce high-@xmath16 jets with large electromagnetic field energy to particle energy densities , ssc models indicate low-@xmath16 jets .
the low radiative efficiency indicates that the sub - parsec jet can smoothly evolve into the parsec - scale jet , without the need for plasma re - collimation or re - acceleration that would have to follow a more efficient conversion of bulk plasma energy into random particle energy and radiation ( see also @xcite ) . + * hadronic models : * in hadronic models the continuum emission is explained by hadronic interactions of a highly relativistic baryonic outflow which sweeps up ambient matter @xcite , by interactions of high - energy protons with gas clouds moving across the jet @xcite , or by interactions of ultra high energy protons with ambient photons @xcite , with the jet magnetic field @xcite , or with both @xcite . although attractive since they involve ultra high energy cosmic ray acceleration , hadronic models have difficulties to explain the good x - ray / tev @xmath0-ray flux correlations found for mrk 421 and mrk 501 .
furthermore , little work has yet been done to see if proton models can explain more than single snapshots of data .
among the many particle acceleration mechanisms that are discussed , fermi acceleration at strong magnetohydrodynamic ( mhd ) shocks has received most attention . indeed , the data are largely consistent with an electron acceleration energy spectrum @xmath18 @xmath19 with @xmath20 between 2 and 2.23 , expected for non - relativistic and ultra - relativistic shocks , respectively ( see @xcite and references therein ) . based on a time - dependent sc code kirk & mastichiadis ( 1999 )
have shown that fermi energy gains and radiative energy losses can result in characteristic hardness
intensity correlations during individual flares @xcite .
x - ray and recently also tev @xmath0-ray data do show the predicted signatures : during flares the sources seem to go through clockwise and anti - clockwise loops in the x - ray or @xmath0-ray hardness intensity planes ( e.g. @xcite ) .
however , the same source shows a wide range of different signatures ( e.g. @xcite ) , indicating that the relative length of the characteristic time scales of particle acceleration and radiative cooling change from flare to flare .
the lack of a prevailing signature has cast doubts on whether `` cooling loops '' and `` particle acceleration loops '' have really been observed .
+ most sc models of tev blazars invoke a minimum lorentz factor @xmath21 of accelerated particles on the order of @xmath22 to account for the data ( e.g. @xcite and references therein ) .
such high @xmath21-values could either result from a yet unknown pre - acceleration mechanism , or , from an upstream bulk lorentz factor on the order of @xmath22 .
a low - entropy , extremely relativistic upstream plasma could be detectable by its inverse compton emission if exposed to a suitable external radiation field .
from the theoretical side , the connection between black hole and jet properties is far from being understood .
purely electromagnetic @xcite and gr - electromagnetic models @xcite of jet formation face the well known @xmath16-problem , namely that they predict poynting flux dominated energy transport while sc models indicate strongly particle energy dominated jets .
hydrodynamic , mhd , and gr - mhd scenarios of jet formation ( e.g. @xcite ) face the difficulty that they predict jets with bulk lorentz factors @xmath23 while sc models favor @xmath24 .
+ from the observational side , first studies of the black hole jet connection are just becoming possible based on high - quality black hole mass estimates from galactic stellar velocity dispersion measurements @xcite .
the studies performed so far do not indicate any correlation of the black hole mass and the peak location or peak luminosity of the low - energy or high - energy seds , the time scale of flux variability ( see fig . [ bm ] ) , or the flare duty cycle .
the correlations may be masked by variation of parameters like jet viewing angle , jet magnetic field , or intensity of the ambient photon field @xcite .
alternatively , the black hole mass estimators may not work in the case of blazars , as discussed in @xcite .
extragalactic absorption of tev @xmath0-rays by the cosmic infrared background ( cib ) and the cosmic optical background ( cob ) in @xmath25 @xmath26 pairproduction processes @xcite substantially complicates the study of the astrophysics of blazar jets .
however , the absorption effect allows us to use tev @xmath0-ray observations to measure the energy spectrum of the cib and the cob .
these backgrounds are extremely interesting owing to their close relation to the total electromagnetic luminosity of the universe since decoupling of matter and radiation @xmath1300,000 years after the big bang .
+ the tev energy spectra of extragalactic sources carry the imprint of extragalactic absorption in the form of `` characteristic high - energy cutoffs '' in the 30 gev to 10 tev energy range .
the measurements of these cutoffs makes it possible to constrain the energy spectrum of the infrared to optical background radiation .
these constraints are very valuable , as direct measurements are plagued by large systematic errors , owing to strong foreground emission from our solar system and the milky way ( e.g. @xcite ) . while the spectral measurement of the nearby blazars mrk 421 and mrk 501 ( @xmath15 0.03 )
have been used to set upper limits on the cib intensity in the 1 - 20 @xmath27 m wavelength range @xcite , the recent measurement of the energy spectrum from the more distant blazar h 1426 + 428 ( @xmath28 0.129 ) has given the first demonstration of the full potential of tev @xmath0-ray observations to decide between different background models @xcite .
+ further progress may best be accomplished with a two - pronged approach : ( i ) detailed source modeling makes it possible to disentangle , at least partially , source inherent high - energy cutoffs from those caused by extragalactic extinction and thus to improve on the constraints from each single source @xcite ; ( ii ) since the spectral cutoffs from extragalactic absorption are expected to occur at energies which depend only on source redshift , a statistical analysis of the constraints from many sources can be used to test and strengthen the constraints from individual sources .
[ ol ] over the last several years , the field of tev blazar studies has benefited greatly from detections of new sources and from very intensive multiwavelength campaigns .
further progress may derive from extending the time scales of the broadband observations .
high - accuracy measurement of gamma - ray burst ( grb ) afterglows is boosting the theoretical progress in the case of grbs . in the same way , detection of increased flux levels in the radio to uv bands following major blazar flaring phases may prove crucial for pinning down the jet properties .
for this purpose regular broadband monitoring is required over time scales longer than the typical 10 days of current multiwavelength campaigns .
as the quality of the data improves , it now becomes apparent that simple one - zone sc model fit snapshots of the data , but are not able to give us deep insights into the jet structure and the nature of the particle acceleration processes .
most present theoretical work focuses on the emission mechanism and largely neglects the flare origin and the development of the plasma that supports the non - thermal particles during the flare .
a notable exception is the recent work concerning the internal shock model @xcite . in future work , combining 3-dimensional mhd simulations of flares with kinetic simulations of the non - thermal particle population would make it possible to explore more complex observational signatures than possible with present codes .
+ in view of the next - generation cherenkov telescopes cangaroo iii , h.e.s.s . ,
magic , and veritas presently under construction , the upcoming launches of the x - ray observatories swift and astro - e2 , and the @xmath0-ray observatories agile and glast to be launched in 2005 and 2006 , respectively , we can expect a wealth of exciting blazar results in the next 5 years .
+ _ acknowledgements : _ the author gratefully acknowledges support by nasa through the grant nasa nag5 - 12974
. + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
|
since the first tev blazar markarian ( mrk ) 421 was detected in 1992 , the number of established tev @xmath0-ray emitting bl lac objects has grown to 6 , with redshifts ranging from 0.031 ( mrk 421 ) to 0.129 ( h 1426 + 428 ) .
the intensive study of these sources has had a major impact on our understanding of the blazar phenomenon .
the most notable observational results have been extremely fast large amplitude flux and spectral variability on hour time scales , and a pronounced x - ray - tev @xmath0-ray flux correlation . in this paper
we discuss recent observational results and report on progress in their theoretical interpretation .
galaxies : bl lacertae objects galaxies : jets gamma rays : observations gamma rays : theory
|
observations by fermi and the various tev experiments , and advances in numerical techniques have led to a progress in our understanding of relativistic jets : i ) in m87 , combined vlba and tev data ( acciari et al .
2009 ) seem to indicate that the tev emission is produced on horizon scales by either some magnetospheric process , or at the base of the vlba jet .
the observational constraints raise interesting questions about the structure of the bh magnetosphere and the jet formation mechanism , that appear to be relevant also to other tev agns .
ii ) the lat spectrum of blazars provide now a better than before probe of opacity sources on sub - parsec scales .
the data reveal relationship between source power and spectral features that can be interpreted in terms of blr properties .
iii ) combined fermi and tev observations of vhe blazars may be used to probe intergalactic magnetic fields ( e.g. , neronov et al .
2010 ; tavecchio et al .
attempts to derive constrains on the igmf have been published , and although inconclusive yet , the results demonstrate the potential in pursuing further these efforts .
iv ) fermi lat detections of several grbs indicate very high lorentz factors in outflows having opening angles larger than the causality scale .
those detections triggered recent theoretical studies and numerical simulations of magnetic outflows .
the conclusion emerging from these studies is that processes beyond ideal mhd are crucial .
hydrodynamical outflows , perhaps driven by neutrino annihilation , may provide an alternative .
the nature of the gev emission in grbs is yet an open issue .
v ) in several bursts a prominent , quasi - thermal spectral component has been detected , challenging the `` standard model '' .
it opens up the issue of dissipation and emission during the prompt phase .
in particular , part of the emission is likely produced behind relativistic radiation mediated shocks . in what follows ,
these issues are discussed in some greater detail .
over 700 agns , most of which are blazars , are listed in the first fermi lat catalogue ( abdo , et al .
2010c ) . around 3 dozens agns have been detected by various tev experiments at energies above 100 gev , with a spectrum extending up to 10 tev in some cases .
at least 50 percents of the tev sources have gev counterparts .
the lat spectrum of many bright fermi blazars exhibits curvature or breaks at gev energies ( abdo et al .
2010b ) , conceivably associated with attenuation by radiation emitted from highly ionized clouds in the broad line region ( blr ) .
eleven non - blazar agns , - 7 fri radio galaxies and 4 frii radio sources - exhibit vhe ( @xmath1 mev ) emission ( abdo , et al .
2010a ) , and some of them , e.g. , m87 ( wagner et al .
2009 ) , cen a ( aharonian , et al . 2009 ) , are also tev sources . according to the unified model , those radio galaxies are misaligned blazars , and the question then arises if the vhe photons seen represent side emission from the jet , or have a different origin . in the former case ,
the high luminosities measured in some of the objects may imply surprisingly small doppler factors , or extremely high jet power .
it is conceivable that the misaligned blazars exhibit spectral components that are difficult to detect in blazars because they are overwhelmed by the beamed emission from jet .
emission from a starved magnetosphere is one possibility ( levinson 2000 ) .
recent vlba and tev observations of m87 ( acciari et al .
2009 ) strongly motivates reconsideration of this scenario .
below we propose that the variable tev emission observed in m87 could be a manifestation of the jet formation process .
combined vlba and tev observations of m87 reveal a rapidly varying tev emission that appears to be associated with the m.a.s vlba jet .
the rapid flaring activity of the tev source , with timescales @xmath2 day as low as 1 - 2 days , implies a source size of @xmath3 for a black hole mass @xmath4 solar masses .
this , and the fact that the tev emission appears to be correlated with the vlba jet but not with emission from larger scales ( and in particular hst-1 ) , motivates the consideration that the observed tev rays originate from the black hole magnetosphere .
a plausible magnetospheric process discussed in the literature is curvature and/or ic emission by particles , either hadrons or leptons , accelerating in a vacuum gap of a starved magnetosphere .
an alternative explanation is emission from small regions located at larger radii , @xmath5 , as , e.g. , in the misaligned minijets model of gainnios et al .
( 2010 ) .
schematic representation of the magnetosphere structure : a vacuum gap of height @xmath6 accelerates particles ( electrons or positrons ) to high lorentz factors .
the gap is exposed to soft radiation emitted from a source of size @xmath7 . curvature emission and inverse compton scattering of ambient radiation produce vhe photons with a spectrum extending up to @xmath8 tev .
photons having energies below a few tev can escape freely to infinity .
interactions of ic photons having energies well above 10 tev with the ambient radiation initiate pair cascades just above the gap , leading to a large multiplicity .
a force free outflow is established just above the pair formation front , and appears as the vlba jet .
intermittencies of the cascade process give rise to the observed variability of the tev emission , and the resultant force - free outflow , as indicated by the morphological changes of the vlba jet.,width=377,height=302 ] the presence of the vlba jet implies that a force - free ( or ideal mhd ) flow is established on scales @xmath9 , so that the magnetosphere is anticipated to be screened in the sense that the invariant @xmath10 nearly vanishes everywhere .
however , as in pulsar theory , there must be a plasma source that compensates for the loss of particles which escape the system ( both , to infinity and across the horizon ) along the open magnetic field lines in the polar region .
the nature of this plasma source is poorly understood at present .
the injection of charges into the magnetosphere may be associated with the accretion process .
direct feeding seems unlikely , as charged particles would have to cross magnetic field lines on timescale shorter than the accretion time in order to reach the polar outflow .
free neutrons , if exist , decay over length scale much smaller than the gravitational radius of the black hole .
annihilation of mev photons produced in the radiative inefficient flow , may provide the required charge density , depending on accretion rate and other details .
a possible plasma source is cascade formation in starved magnetospheric regions .
the size of the gap then depends on the conditions in the magnetosphere and the pair production opacity . as shown elsewhere
( levinson 2000 ) , vacuum breakdown by back - reaction is unlikely , as it requires magnetic field strength in excess of a few times @xmath11 g , higher than the equipartition value for eddington accretion .
pair production via absorption of tev photons by the ambient radiation field is more likely .
however , the spectrum of the vhe photons observed by hess extends up to 10 tev , and the assumption that these photons originate from the magnetosphere ( or even the vlba jet ) implies that the pair production opacity at these energies must not exceed unity .
this raises the question whether pair cascades in the magnetosphere can at all account for the multiplicity required to establish a force - free flow . in an upcoming paper ( levinson & rieger , 2010 , in preparation )
it is shown that inverse compton scattering ( ic ) of ambient photons by electrons accelerated in the gap can lead to a large multiplicity , in excess of @xmath12 , while still allowing photons at energies of up to a few tev to freely escape the system .
the seed electrons are generated by the mev emission from the riaf .
this requires that the ambient radiation source contributing the opacity will have a characteristic size @xmath13 .
preliminary results indicate that the electromagnetic cascade is initiated by ic photons having much higher energies , @xmath14 tev , for which the @xmath15 optical depth is much larger .
it is found that the gap width is not much smaller than @xmath16 .
the luminosity of the vhe photons produced in the gap , roughly a fraction @xmath17 of the maximum bz power , can easily account for the tev luminosity observed by hess and fermi .
any intermittencies of the cascade formation process would naturally lead to variability of both , the magnetospheric tev emission and the resultant force - free flow , as observed .
a schematic illustration of the model is presented in figure [ fig : gap ] .
this model may be applicable also to some of the other misaligned vhe blazars .
the nature of opacity sources in blazars is an issue of considerable interest , particularly in regards to the location of the vhe emission zones .
much efforts have been devoted over the years to identify different opacity sources , and detailed calculations of @xmath15 attenuation by these sources are presented in the literature .
recent fermi observations have led to reconsideration of the effect of the broad line region on the gev spectrum of blazars .
strong departure from a single power law appears to be a common feature in fsrqs ( abdo et al .
the break energies , typically lying in the energy range 2 - 10 gev , are too low to be due to absorption by ly@xmath18 radiation .
recently , it has been pointed out ( putanen & stern 2010 ) that the lat spectrum of fsrqs can be well reproduced by a double absorber model .
detailed numerical calculations of the blr spectrum for different ionization parameters indicates that the observed gev breaks can be accounted for by @xmath0-ray absorption through pair production on he ii and h i recombination continua .
the exact location of the break energy depends on the ionization parameter , as described in putanen & stern ( 2010 ) .
the fact that gev breaks are seen mainly in the brightest blazars is naturally explaned by the scaling of the blr size with luminosity , @xmath19 , obtained from reverberation mapping techniques ( kaspi et al .
2007 ) , that implies @xmath20 .
over a dozen grbs , both long and short , have been detected thus far by lat onboard fermi .
the origin of the gev emission is yet an open issue .
the flux level of the lat emission and its long lasting light curve , suggest that this component is produced during the afterglow phase , although other explanations have been offered . for sources that show sub - second variations of the gev flux , e.g. , 080916c ,
opacity arguments yield high lorentz factors , @xmath21 . while various factors render those estimates uncertain ( e.g. , yuan - chuan et al .
2010 ) , it seems difficult to avoid the conclusion that the outflow reaches high lorentz factors .
such high lorentz factors posses a great challenge to jet models , as discussed in the following .
another interesting result is that in some gev bursts , e.g. , grb 090902b ( peer et al . 2010 ) , a prominent quasi - thermal spectral component has been detected , whereas in others , e.g. , 080916c , a fit by a band spectrum is claimed to be satisfactory , and there appears to be no indication for thermal emission .
this , presumably , reflects the conditions in the dissipation region .
below it is argued that in bursts that exhibit a thermal component a substantial fraction of the outflow bulk energy dissipates below the photosphere , in relativistic radiation mediated shocks , whereas in other bursts most of the energy dissipates above the photosphere , behind collisionless shocks .
extraction of rotational energy from a rapidly spinning black hole by magnetic fields is a widely discussed jet production mechanism ( e.g. , levinson & eichler 1993 ; lyutikov & blandford 2003 ; giannios & spruit 2005 ) .
up to 30 percents of the black hole mass can be extracted from a maximally rotating hole , which is more than sufficient to account for the energetics of essentially all grbs .
the rate at which the energy is extracted depends on the strength of the magnetic field in the vicinity of the horizon .
recent numerical simulations confirm that a blandford - znajek process does operate in kerr spacetime , and that for reasonable assumptions about the progenitor star , sufficiently large power can be extracted in the form of an outflow .
moreover , a very low baryonic content is anticipated along horizon threading field lines , allowing , in principle , high asymptotic lorentz factors of the polar outflow
. pressure support by the external medium may give rise to collimation of the central outflow ( bromberg & levinson 2007 ; lyubarsky 2009 ) .
magnetic outflows have been considered also in the context of the unexpected paucity of optical flashes seen in grbs ( e.g. , zhang & kobayashi 2005 ; mimica et al .
2009 ) , and it has been shown that even moderate magnetization ( @xmath22 ) would strongly suppress the reverse shock .
the question remains as to how the internal shocks that produce the prompt emission form under these conditions .
alternative explanations for the paucity of optical flashes , e.g. , an early onset of a r - t instability ( levinson , 2009 , 2010a ) , have been offered recently .
an key issue in the theory of magnetically dominated outflows is magnetic energy dissipation .
the process by which magnetic energy is converted into kinetic energy has not been identified yet .
stationary magnetic outflows allow , in general , only partial conversion of magnetic - to - kinetic energy . for a split monopole field acceleration ceases at an asymptotic lorentz factor @xmath23 , and magnetization @xmath24 , where @xmath25 is the initial magnetization of the expanding shell .
a better conversion can be achieved if the outflow is collimated into a small opening angle , @xmath26 ( komissarov et al .
such small opening angles are problematic , particularly for the gev bursts , for which @xmath27 is inferred .
corking from a star that , within the framework of the collapsar model , may be relevant to long bursts can alleviate the latter condition ( tchekhovskoy et al .
, 2010 ; komissarov et al .
however , even then @xmath28 can not be much smaller than unity ( lyubarski 2010 ) .
furthermore , there is evidence for extreme lorentz factors also in short bursts , where strong deconfinement is not anticipated .
recently it has been shown ( granot et al . , 2010 ; hereafter gks10 ; lyutikov 2010a , b ) that time - dependent effects may play a crucial role in the acceleration of a magnetized flow . unlike a stationary flow , for which acceleration ceases at @xmath23 , an impulsive spherical shell expelled by a central source continues accelerating , even after loosing causal contact with the engine , until reaching nearly complete conversion of magnetic energy into bulk kinetic energy .
the terminal lorentz factor of a shell expanding in vacuum is @xmath29 . during the acceleration phase
the major fraction of the shell energy is contained in a layer of width @xmath30 , where @xmath31 is the initial shell s width , bounded between the front of a rarefaction wave reflected from the central source and the head of the shell .
the average lorentz factor of the shell , roughly equals the lorentz factor of the fluid at the rarefaction front , evolves as @xmath32 .
once the shell enters the coasting phase its width starts growing and its magnetization continues to drop , until nearly full conversion is accomplished .
however , for sufficiently high values of the initial magnetization @xmath25 the evolution of the system is significantly altered by the ambient medium well before the shell reaches its coasting phase ( levinson 2010b ) .
the maximum lorentz factor of the shell is limited to values well below @xmath25 . to be concrete , for a shell of initial energy @xmath33 erg and size @xmath34 cm expelled into a medium having a uniform density @xmath35 the lorentz factor is limited to @xmath36 in the high sigma limit .
the reverse shock and any internal shocks that might form if the source is fluctuating are shown to be very weak .
the restriction on the lorentz factor is more severe for shells propagating in a stellar wind .
intermittent ejection of multiple , thin shells does not seem to help , as even in vacuum unlikely small duty cycle is required in order for shells to collide after reaching the coasting phase .
such episodes are expected to produce a smooth , relatively fast rising slowly decaying ( power law ) light curve , even in a multi - shell scenario .
events like grb080916c and grb090510 ( abdo et al . 2009a ,
b ) are not easily accounted for by the impulsive high - sigma shell model .
optical depth at the shells collsion radius as a function of lorentz factor , for different values of @xmath37 - the time interval between ejections of the two colliding shells .
the upper curves correspond to the dynamical time of a 3 solar mass black hole , @xmath38 s. the lower curves are limits from observations , with @xmath39 s for grb080916c , and @xmath40 s for grb090902b .
the range of @xmath41 in each source represents various estimates taken from the literature ( bromberg et al .
2010).,width=377,height=302 ] shocks that form by overtaking collisions can dissipate energy at radii @xmath42 , where @xmath41 is the lorentz factor of the slow shell and @xmath43 is time interval between ejections of the two consecutive shells . in blazars and microquasars with @xmath44 dissipation by internal shocks
is expected close to the bh , consistent with ( but not necessarily implied by ) the short durations of strong flares observed in these objects , particularly in tev blazars . in grbs @xmath45 or
so for the lorentz factors envisaged .
a rough condition for internal shocks to form above the photosphere is @xmath46 , where @xmath47 is the burst luminosity in units of @xmath48 erg / s , and @xmath49 .
results of more exact calculations ( bromberg et al .
2010 ) are exhibited in figure [ fig : a ] , where the thompson optical depth at the radius of shock formation , @xmath50 , is plotted agains the flow lorentz factor @xmath41 .
the two colored areas correspond to a range of @xmath37 and @xmath41 for grb 080916c ( blue ) and grb 090902b ( green ) , as explained in the figure caption .
as shown , in case of grb 080916c internal shocks are expected to form above the photosphere for shells accelerated to @xmath51 .
such shocks are probably collisionless , and can fermi accelerate particles to nonthermal energies .
slower shells ejected over time intervals of the order of the dynamical time will collide below the photosphere , at a moderate optical depth @xmath52 . the parameter space explored in figure [ fig : a ] indicates that in this burst the majority of the explosion energy dissipates above the photosphere , behind collisionless shocks , so that a thermal component may be very weak or absent . in grb 090902b , on the other hand , the major fraction of the energy is likely to dissipate below the photosphere , albeit in a region of moderate optical depth , @xmath53 , as seen in figure [ fig : a ] .
only sufficiently slow shells that have large separations , such @xmath54 is much larger than the anticipated duty cycle of the engine , will collide above the photosphere to form collisionless shocks .
the shocks that form below the photosphere , where the thomson depth exceeds unity , are mediated by compton scattering ( bromberg & levinson 2008 ; katz et al .
2010 ) . under conditions anticipated in grbs , such relativistic radiation mediated shocks ( rrms ) convect enough radiation upstream to render photon production in the shock transition negligible ( bromberg , et al . , 2010 ) , unlike the case of shock breakout in supernovae and hypernovae ( katz et al . 2010 ) .
bulk comptonization then produces a relatively low thermal peak , followed by a broad , nonthermal component in the immediate downstream that extends up to the kn limit in the shock frame ( or up to @xmath55 in the observer frame ) , and perhaps even beyond ( budnik et al .
2010 ) , depending on details .
the radiation produced downstream is trapped for times much longer than the shock crossing time , and diffuses out after the shock breaks out of the photosphere and becomes collisionless . at what depth downstream equilibrium
is established is yet an open issue .
since the enthalpy downstream is dominated by radiation , a full equilibrium is not expected for the moderate optical depths found above .
we contend that a thermal component plus a hard tail , as seen in grb 090902b , can be emitted from the downstream of a rrms .
a full treatment requires detailed transfer calculations , or monte - carlo simulations .
abdo , a. a. , et al .
2009a , _ science _
, 323 , 1688 abdo , a. a. , et al .
2009b , _ arxiv0908.1832 _
abdo a. a. et al . , 2010a _ apj _ , 720 , 912 abdo a. a. et al . ,
2010b _ apj _ , 710 , 1271 abdo a. a. et al . ,
2010c _ apjs _ , 188 , 405 acciari , v. a. et al . 2009 , _ science _
, 325 , 444 aharonian , f. , et al .
( h.e.s.s .
collaboration ) 2009 , _ apj _ , 695 , l40 bromberg , o. & levinson , a. 2007 , _ apj _ , 671 , 678 bromberg , o. , mikolitzky , z. & levinson , a. 2010 , ( to be submitted ) budnik , r. et al .
2010 , _ arxiv:1005.0141 _
giannios d. & spruit h. , 2005 , _ a&a _ , 430 , 1 giannios , d. , uzdensky , d. a. , & begelman , m. c. , 2010 , _ mnras _ , 402 , 1649 granot , j. , komissarov s. & spitkovsy , a. , 2010 , _
kaspi , s. et al .
2007 , _ apj _ , 659 , 997 katz , b. budnik , r. & waxman , e. 2010 , _ apj _ , 716 , 781 levinson , a. 2000 , _ phys . rev . lett _ , 85 , 912 levinson , a. 2009 , _ apj _ , 705 , 213 levinson , a. 2010 , _ gapfd _ , 104 , 84 levinson , a. 2010 , _ apj _ , 720 , 1490 levinson a. & eichler d. , 1993 , _ apj _ , 418 , 386 levinson , a. & bromberg , o. _ phys .
_ , * 100 * ( 2008 ) 131101 lyubarsky , y. 2009 , _ apj _ , 698 , 1570 lyubarsky , y. 2010 , _ mnras _ , 402 , 353 lyutikov m. & blandford r. d. 2003 , _ arxiv : astro - ph/0312347 _ lyutikov m. 2010a , _ arxiv:1004.2428
_ lyutikov m. 2010b,_arxiv:1004.2429 _ mimica , p. , giannios , d. , & aloy , m. a. , 2009 , a&a , 494 , 879 neronov , a. et al .
2010 , _ apj _ , 719 , l130 peer , a. et al .
2010 , arxiv:1007.2228 poutanen , j. & stern , b. 2010 , _ apj _ , 717 , l118 tavecchio , f. et al .
2010 , _ mnras _ , 406 , l70 tchekhovskoy , a. , narayan , r. , & mckinney , j. c. 2010 , _ new astron .
_ wagner , r. m. et al .
2009 , _ proc .
icrc , arxiv:0907.1465 _ yuan - chuan , z. , yi - zhong , f. & piran , t. 2010 _ arxiv:1008.2253 _ zhang , b. & kobayashi , s. 2005 , _ apj _ , 628 , 315
|
a recent progress in the study of @xmath0-ray jets is reviewed , with a focus on some theoretical interpretations of the vhe emission from m87 , and possibly other misaligned blazars ; the connection between the gev breaks exhibited by bright lat blazars and opacity sources in the broad line region ; the consequences of the detection of gev emission from grbs to models of magnetic outflows ; and the implications of the thermal emission observed is some grbs to dissipation of the outflow bulk energy .
|
the pion scattering length is an important observable characterizing dynamical effects of the strong interaction and its ab initio calculation on the lattice is an important nonperturbative test of qcd . in the literature several attempts to compute the scattering length with wilson fermions @xcite and staggered fermions @xcite can be found .
however , calculations with a chiral dirac operator which allows for a better control of the small mass region are still missing . in full qcd the scattering length is a quantity which vanishes in the chiral limit , while it is power divergent in the quenched theory @xcite and eventually a study in the full theory is desirable . in this contribution
, we present results of the scattering length calculated using the p - fp action .
( calculations with chirally improved fermions is also underway . )
the chiral and scaling properties of the light hadron spectrum using p - fp and chirally improved actions were reported by the collaboration in an earlier publication @xcite .
the @xmath2 scattering length is extracted using the standard finite volume technique of lscher @xcite .
the energy shift of the two pion state , @xmath3 , in a finite volume is related to the scattering length , @xmath4 , in the infinite volume limit through the relation , @xmath5 where @xmath6 and @xmath7 . the two pion state is created using a gaussian smeared source for the quark propagator on a single timeslice .
a second smeared source which was displaced in time by one lattice unit was used on small lattices to check for the systematics and the often claimed complications due to fierz mixing .
point sink operators were used on a single timeslice regardless of the position of the sources .
cutoff effects were studied in a small volume @xmath8 using three different lattice spacings ( @xmath9 ) and in an intermediate volume of @xmath10 with @xmath11 and @xmath12 fm .
a wide range of quark masses were studied , with the smallest quark mass resulting in @xmath13 . in this quenched study , we keep those masses with @xmath14 .
= 2.5 in = 2.5 in
we extract the energy shift of the two pion state in a finite volume by taking ratios of two pion correlation functions and single pion correlation functions .
we follow the procedure and notation of ref .
@xcite and construct the `` direct '' and `` crossed '' contractions of @xmath15 where @xmath16 annihilates a @xmath17 state and @xmath18 are the gaussian sources for the pions and divide by the two single pion propagators .
the large time behaviour of the ratio is given by @xmath19 in the full theory .
periodic boundary conditions were used in the time direction so that we must take into account a backwards propagating pion .
the fitting form of the correlation function was @xmath20 the pion mass was fitted from the pion propagator alone within the same bootstrap sample .
the momentum spacings of the pions in the largest volume is quite small which resulted in significant excited state contamination ( higher momenta states ) . in this case
, we fit to a sum of exponentials ( assuming this is the case ) to take this into account .
only the amplitude of the excited state was left as an extra fitting parameter and we vary the energy of this state over a wide range to monitor the effect of the choice of this energy .
the results are shown in fig .
[ excited ] .
it can be seen in this example that the ground state energy is nearly independent of the value of the energy of the excited state , however , with much larger errors than with a single exponential fit .
= 2.5 in the effect of displacing the second source by one timeslice is shown in fig .
[ fierz ] .
the asymptotic value agrees within errors but there is less excited state contamination when the two pions are not created on the same timeslice .
the effect of the second source is being investigated on the largest lattice where this contamination is the source of the large errorbars as was mentioned in the previous section .
the cutoff effects of @xmath21 in the small volume , @xmath8 , is shown in fig .
[ cutoffv ] .
we observe no significant lattice artifacts for pseudo - scalars lighter than @xmath22 gev .
a similar analysis in the large volume for @xmath23 ( fig . [ cutoffv ] ) confirms that the lattice artifacts are well under control and that the results from @xmath24 and @xmath25 are very close to the continuum and infinite volume limit . in fig .
[ summary ] , we plot @xmath26 , against @xmath27 in physical units .
we also include in this plot some recent results from other groups as well as quenched chiral perturbation theory @xcite and chiral perturbation theory predictions @xcite .
the mass range of the perturbative calculations shown in this figure were chosen for illustration and does not suggest convergence of the series .
the effect of the @xmath28 term in the quenched theory is still not entirely visible at the present quark masses .
the results also suggest that pseudo - scalars with masses less than @xmath29 may be needed for a reliable extrapolation to the physical pion mass .
it remains a challenge to perform a full qcd simulation with such light quarks .
= 2.5 in = 3.1in=3.0 in
k.j . would like to thank g. colangelo and j. gasser for providing details of their calculations and insightful discussions .
the simulations were done on the hitachi sr8000 at the leibniz rechenzentrum in munich and at the swiss center for scientific computing in manno .
this work was partially supported by the european community s human potential programme under contract hprn - ct-2000 - 00145 .
9 r. gupta _ et al .
_ , nucl .
* b383 * ( 1992 ) 309 .
m. fukugita _ et al .
_ , phys .
* d52 * ( 1995 ) 3003 .
c. liu _ et al .
_ , nucl .
* b624 * ( 2002 ) 360 .
jlqcd collaboration , phys .
* d66 * ( 2002 ) 077501 .
cp - pacs collaboration , phys .
* d67 * ( 2003 ) 014502 . c. bernard and m. golterman , phys .
* d53 * ( 1996 ) 476 ; g. colangelo and e. pallante , nucl .
* b520 * ( 1998 ) 433 . c. gattringer _
et al . _ , hep - lat/0307013 .
m. lscher , commun . math
* 104 * ( 1986 ) 177 ; * 105 * ( 1986 ) 153 ; nucl .
* b354 * ( 1991 ) 531 .
g. colangelo _ et al .
_ , nucl .
* b603 * ( 2001 ) 125 .
|
we report on the @xmath0 scattering length in the @xmath1 channel using the parametrized fixed point ( p - fp ) action .
pion masses of 320 mev were reached in this quenched calculation of the scattering length .
|
human action detection at real - time has become a topic of increasing interest due to its wide practical use .
applications like human - machine interaction , surveillance and gaming , all require accurate and low - latency action detection .
action detection on raw videos is difficult because it is first needed to localize a person in a scene full of objects and clutter , then try to recognize the type of action being performed . on the other hand , the recent low - cost depth sensors , like microsoft kinect , provided a more convenient way for data capture .
the 3d positions of body joints can be estimated from depth maps at low - latency and with acceptable accuracy .
filtering out background clutter , it is now more adequate to perform action detection based on skeleton data .
recently , skeleton - based approaches to action recognition and detection have been widely adopted .
while action recognition focuses on identifying the action label of pre - segmented video sequences , action detection tackles the more challenging problem of temporally localizing the action in an unsegmented stream of frames .
the main contribution of this paper is a novel approach for action detection from skeleton data , that we refer to as efficient linear search ( els ) .
we show that a combination of simple components and specializing them towards skeleton - based action detection can achieve state - of - the - art results and overcome the limitations of similar approaches .
the proposed approach is flexible : it can be used with a wide class of classifier functions and with different types of action local descriptors . as a byproduct contribution
, we propose a simple skeleton - based local descriptor that , when used in a simple bag - of - features model , produces state - of - the - art results on different datasets .
the proposed framework works online and is suitable for real - time applications .
moreover , it can be used for real - time video segmentation , since it specifies both the start and end frames of the action .
the rest of this paper is organized as follows : section [ sec : relwork ] gives an overview about recent related work in the literature .
we show the used action representation and our proposed descriptor in section [ sec : mainidea ] .
we , then , explain our efficient linear search approach in section [ sec : els ] .
experimental evaluation is presented in [ sec : experiments ] . and
finally , we conclude in [ sec : conclusion ] .
forming suitable skeletal - based descriptors for action recognition has been the focus of many recent research works @xcite .
the objective is to facilitate the recognition task via a discriminative descriptor .
some of these descriptors capture both the pose of the skeleton and the kinematics at the same time on the frame level .
for example , nowozin _ et al . _
@xcite proposed a local descriptor that uses 35 angles between triplets of joints , @xmath0 , along with angular velocity , @xmath1 , to encode joints kinematics .
joints angles are a powerful cue to skeleton pose .
moreover , they are invariant to body translation and rotation .
later , zanfir _ et al . _
@xcite proposed the moving pose descriptor , which captures both the body pose at one frame , as well as the speed and acceleration of body joints within a short time window centered around the current frame .
another class of descriptors is focused on computing a fixed length descriptor for the whole action sequence , like @xcite .
et al . _
@xcite used a 2d trajectory descriptor , called `` histogram of oriented displacements '' ( hod ) , where each displacement in the trajectory casts a vote , weighted by its length , in a histogram of orientation angles .
et al . _
@xcite modeled human actions as curves in a lie group , since 3d rigid body motions are members of the special euclidean group .
et al . _
@xcite used the 3d joints positions to construct a descriptor of relative positions between joints .
however , descriptors on the whole sequence suffer from much higher dimensionality over those that were designed for the frame level .
this higher dimensionality led sometimes to the need for feature selection as done in @xcite . while most of the focus on skeletal data was about action recognition , fewer works focused on the online problem .
the trade - off between latency and accuracy was addressed in recent works @xcite . in
@xcite the notion of action points was first introduced and the detection problem was cast as a classification problem for every overlapping 35-frames intervals .
the same notion of action points was utilized in the work of @xcite , but they could handle different scales .
other works , such as @xcite used the standard sliding window protocol for online action detection .
et al . _
@xcite proposed a feature extraction method , called `` structured streaming skeleton '' ( sss ) , which constructs a feature vector for each frame using a dynamic matching approach .
the sss feature vectors are then used for detecting the start and end of actions .
et al . _
@xciteused a modified knn classifier to detect the start and end of actions
. however , both @xcite can not handle multi - scale actions , where the same action can be performed at different speeds .
the concept of local features first appeared in object detection in images @xcite .
the main idea is that each object has a set of discriminative local features that , if appeared together , signify the existence of the object .
same concept applies to actions . for a specific action
, we can identify a set of key frames that best capture the discriminative poses of this action .
however , skeleton poses alone can not distinguish between some actions , e.g. standing up vs. sitting down , since other information like the direction and speed of motion play an important role in identifying the action .
so , differential quantities that describe joints kinematics must be included in an action s local features .
we then define a _
gesturelet _ to be any general local feature that , for any frame , captures both the skeleton pose and kinematic information of body joints at this point in time . in the following ,
we first introduce our action representation as a bag of gesturelets .
then , we explain our local features descriptor for representing gesturelets . based on extracted gesturelets from action sequences ,
we make use of a bag - of - gesturelets ( bog ) representation of human actions .
we first extract features at each frame .
resulting descriptors are then clustered to produce a @xmath2-entry codebook .
we then represent any action sequence or sub - sequence by its cluster histogram , where the histogram counts how many local features from each cluster index have occurred . in order to relax the assignment of each gesturelet to its representative cluster in the codebook
, we apply ; letting each gesturelet cast a vote to its @xmath3 nearest clusters .
the vote will be weighted according to the distance between the gesturelet and the corresponding cluster , so that closer clusters get higher weight .
implementation details for soft binning are presented in section [ sec : implementationdetails ] .
the bog representation is necessary for our approach for action detection , as we show in section [ sec : els ] .
however , as shown in section [ sec : exp_recognition ] , it is also very effective for action recognition .
this bog representation is independent of the choice of the local descriptor used to represent a gesturelet .
possible descriptors that capture both pose information and joints kinematics are @xcite .
the type of local descriptor has a direct impact on the recognition performance .
we experimented with different descriptors , and as a byproduct contribution , we achieved best results with a proposed local descriptor that is a weighted concatenation of the angles descriptor @xcite and a slight modification of the moving pose descriptor @xcite .
the angles descriptor uses 35 angles between triplets of joints , @xmath0 , along with angular velocity , @xmath1 . on the other hand ,
the moving pose descriptor relies on joints positions , @xmath4 , relative to a reference joint , namely the hip center . and
to capture kinematic information , it includes the first and second order derivatives ; @xmath5 and @xmath6 . the final form of our descriptor is @xmath7 $ ] @xmath8 @xmath9 $ ] , where @xmath10 and @xmath11 are parameters defined in @xcite , and @xmath12 is a weighting parameter to the relative importance of the two concatenated descriptors . in our experiments in section [ sec : experiments ] , we show that using our simple descriptor and with basic dictionary learning , we can achieve state - of - the - art results over different datasets .
in this section , we first explain our approach on offline action detection .
then , we show how it can be easily extended to work for online detection as well .
figure [ fig : pipeline ] gives an overview of the approach .
although many successful skeleton - based action recognition systems exist , most of them havent been extended to action detection .
sliding window approaches can be used for this task , evaluating the classifier function over all candidate sub - intervals .
so , for a video sequence s , it will identify the sub - interval , @xmath13 $ ] , for which the classifier function produces the maximum score .
@xmath14 this identifies only one occurrence of the target action in the sequence .
if multiple occurrences are to be found , then we can simply remove sub - intervals corresponding to previously identified actions , and repeat the search for the next @xmath15 .
however , for a sequence of @xmath16 frames , we have @xmath17 candidate sub - intervals , which incurs significant computational complexity .
thanks to lampert _
et al . _
@xcite , an efficient branch - and - bound method was proposed to search for the optimal bounding box of an object in an image .
a limitation to branch - and - bound approaches is that they typically work offline , requiring the whole search space beforehand , as in @xcite . in @xcite ,
yuan _ et al .
_ proposed a direct extension of @xcite on action detection from rgb videos .
they used a bag - of - features model based on spatio - temporal features , and proposed an offline action detection system .
another limitation to @xcite is that , optimizing equation [ eq : maxsubwindow ] may not always produce the desired behavior for a detection procedure since the optimal interval may contain multiple consecutive instances of an action .
this problem is less likely to happen in the case of 2d images , for which the original branch - and - bound approach was first introduced @xcite . in the following , we present a specialization of the branch - and - bound approach to the case of skeleton - based action detection
this turns out to be easily cast as one of the well - known dynamic programming problems that can be solved in linear time .
next , we show that a greedy approximation can effectively address the offline limitation of the branch - and - bound approach , as well as the problem of combining consecutive actions .
we assume two conditions : ( 1 ) a bag - of - gesturelets representation of action sequences , and ( 2 ) a linear binary classifier with good recognition accuracy trained for a specific target action .
while the linear classifier constraint is not necessary , as we later show , we will use this assumption for simplicity of explanation .
0.43 0.43 0.43 0.43 for any linear classifier , the corresponding scoring and decision functions take the form of : @xmath18 where @xmath19 is a test sequence or sub - sequence , @xmath20 is the feature vector of @xmath19 , @xmath21 is the weight vector learned by the classifier , and @xmath22 is a constant bias . with the linearity of the dot product , @xmath23 , and the fact that @xmath20 is a histogram that counts the occurrence of each cluster index ,
the scoring function [ eqn : linearclassifier ] can be rewritten as : @xmath24 where , @xmath25 is the cluster index to which gesturelet @xmath26 belongs , and @xmath27 is the total number of gesturelets extracted from sequence @xmath19
. we can then evaluate the classifier function over any sub - sequence @xmath28 by summing the weights , @xmath29 , of gesturelets @xmath26 that only belong to the sub - sequence @xmath30 .
since we want to identify the sub - sequence @xmath28 that maximizes equation ( [ eq : svmform2 ] ) , we can safely drop the bias term , @xmath22 . the offline detection procedure will then be as follows : construct an empty 1d score array , with length equal to the number of frames .
this array will represent the per - point contribution of all extracted features from test sequence @xmath19 .
for each feature @xmath26 , identify its cluster index @xmath25 , then add up its weight , @xmath29 , to the frame index in the score array . finding the start and end frames of the action , @xmath31 $ ] , that satisfy equation [ eq : maxsubwindow ] can now be mapped to finding the maximum subarray sum in the score array .
thanks to kadane s algorithm @xcite , this can be done in linear time in the number of frames , using dynamic programming . if @xmath32 , where @xmath33 is a learned threshold for action class @xmath34 , then we specify that action @xmath34 has occurred , and we return the start and end frames of the sub - sequence @xmath15 . to learn the action threshold for each action class , we compute the score of all training sequences , and search for the score threshold , @xmath33 , that minimizes the binary classification error on the training data for class @xmath34 . to detect different action classes
, we will construct a one - versus - all binary classification model for each action class .
what differs from one action class to another is the classification model weights assigned to each entry in the codebook , along with the learned threshold @xmath33 .
the detection procedure for each of the classes can then be run concurrently on the action sequence with different per - point feature weights for each action class .
it is noted that , while the linearity of the classifier is a sufficient condition , it is not necessary .
any type of classification method that can output a classification score at the frame level can be used with our approach , since we later learn the score threshold , @xmath33 , explicitly .
this gives flexibility to our approach with a wider range of classifiers . in this section
we extend our approach to handle online action detection . to find the maximum subarray sum
, kadane s algorithm scans through the array . at each position ( frame ) , it identifies the subinterval ending at this position with the maximum sum .
since we only trigger the action if the maximum sum is larger than a learned threshold @xmath33 , then at any time @xmath35 , if the maximum sum ending at frame @xmath36 exceeds @xmath33 , then we know that action @xmath34 is being performed , and our task , then , is to find at which frame the action will end .
we expect that after the action ends , we will encounter a contiguous sequence of frames with negative scores .
so , for the next frames , as long as the maximum sum grows , the action is still on - going . once the subarray sum starts to decrease
, we can assume that the action has ended .
there will be a compromise between latency and confidence .
if we specify the end of action at the first negative point score , this will be very sensitive to noise but it will achieve low latency .
if we wait till we encounter two consecutive points with negative scores , then this will achieve higher confidence at the expense of increased latency , and so on . in our experiments
, we found that triggering the action at the first frame with negative score , after exceeding the threshold @xmath33 , causes a slight decrease of accuracy , while achieving very low latency .
figure [ fig : detection ] shows an example of how online detection works . as we hinted before , a problem to the offline detection procedure
is that it may combine consecutive repetitions of the same action in one detection .
if an action is repeated twice , with a short pause between them , we expect that each repetition will yield a high positive response , while the pause between them will have a negative response , but not as high .
so , optimizing equation [ eq : maxsubwindow ] would combine the two repetitions together . to separate the two repetitions ,
the separating negative score must be high .
this problem is solved in online detection with the greedy approach to terminating the action , which only requires consecutive frames with negative scores , without further restrictions on their weight .
section [ sec : offlinevsonline ] compares offline and online detection to highlight this offline detection limitation .
detecting multiple classes in the online case is similar to the offline case , except that if one action class is detected at time @xmath37 , then the detection procedure for all other classes is reset , and the search for a new detection starts from time @xmath38 .
in this section , we first describe the two datasets used in our experiments .
we then expand on our implementation details .
then , we compare the performance of our approach to the state of the art on both datasets .
we start with the performance on the more classical action recognition task , then , we move to online action detection .
next , we compare offline and online detections . after that , we show how the performance of our approach is affected by changing its main parameters . finally , we demonstrate its real - time performance . * * msrc-12:**the microsoft research cambridge-12 dataset @xcite is a large dataset designed for action detection .
it contains more than @xmath39 frames in 594 unsegmented sequences , encompassing @xmath40 gesture instances , recorded for 30 subjects performing 12 different gestures .
the samples consist of the 3d positions of 20 joints of the body skeleton captured using the microsoft kinect sensor at 30 fps .
the msrc-12 dataset is annotated using the notion of an action point , which is a pose within the gesture that clearly identifies its completion . + * * msr - action3d:**msr - action3d dataset @xcite is a standard dataset for action recognition .
it consists of @xmath41 pre - segmented sequences , with more than @xmath42 frames .
there are 10 subjects performing 20 different action gestures .
similar to msrc-12 , the 3d positions of 20 joints are captured using the microsoft kinect sensor . for the linear classifier through our experiments
, we use an svm @xcite with a linear kernel .
the local descriptor has three parameters , @xmath43 , and @xmath12 .
while @xmath43 are inherited from the moving pose descriptor @xcite , we introduced @xmath12 to weight the relative importance of the two concatenated descriptors .
coarse - grain values for the parameters were learned from the training set , using different combinations of the 3 parameters in a brute - force manner .
trial - and - error was then used to fine - tune the parameters values . for msr - action3d , we split the training set , persons @xmath44 , into training and validation sets , using @xmath45 of the training data ( persons 7 and 9 ) as a validation set . learned parameters ( @xmath46 ) were then fixed for all test experiments on msr - action3d . for msrc-12 , the parameters were learned over one modality ( @xmath47 of the dataset ) , namely the video modality .
learned values ( @xmath48 ) were then used in test experiments over all modalities . for soft - binning , we set the number of neighbors , @xmath3 , to 3 .
the vote to the @xmath49 nearest cluster is weighted by @xmath50 . a control experiment for choosing @xmath3
is shown in section [ sec : sensitivity ] .
the unsegmented sequences for the msrc-12 dataset contain pauses between consecutive action instances , in which the actor often stands still in a neutral pose .
such a neutral pose also occurs at the beginning and ending of most action instances .
therefore , the neutral pose does not discriminate between different action classes , and hence , the classifier may be tempted to give it a neutral weight , possibly positive in sign .
this would cause a problem for our detection procedure , which relies on having negative scores right before and after an action instance . to overcome this problem , we add hard negatives to the negative training samples used to train our binary classifiers .
each of these hard negatives consists of one positive instance followed by a pause between two consecutive instances . in this way
, the classifier is forced to give strong negative scores to the neutral poses in order to discriminate between those hard negatives and the positive samples , which solves the issue for the detection procedure .
a similar problem occurs in the detection on unsegmented sequences from the msr - action3d dataset .
actions start and end with neutral poses , which could again cause issues in localizing the beginning and ending of action instances . in this case , we include concatenations of two action instances as hard negatives in the binary classifier training .
a weighted moving - average on the frames scores was applied , where the anchor frame had weight equal to number of its neighbors . for msrc-12
, we used a window of 5 frames , anchored at the middle frame ( just as in the local descriptor ) . for msr - action , we used a window of 3 frames , since msr - action3d sequences are much smaller than those of msrc-12 .
.comparative recognition results on msr - action3d . [
cols="^,^",options="header " , ] the goal of our framework is not only to perform online action detection , but also to do this in real - time .
there are 3 main factors affecting the running time of our approach : 1 ) size of the codebook ; 2 ) size of the local descriptor .
3 ) number of neighbors , @xmath3 , to vote for , in soft binning . for this experiment
, we used a large codebook of size 3000 .
the dimensionality of our descriptor is 250 , and we used @xmath51 .
the average running - time per frame of our matlab implementation was measured to be @xmath52 .
so , our framework can process approximately @xmath53 frames per second .
the running - time was measured on a machine with 2.2 ghz intel quad - core core - i7 processor and 12 gb ram .
we have proposed both a simple skeleton - based descriptor and a novel approach for action detection .
the proposed approach maximizes a binary classifier score over all possible sub - sequences , typically in linear time .
it can be used in conjunction with a large class of classifiers and with any local descriptor .
our proposed approach works online and at real - time with low latency .
it detects a gesture by specifying its start and end frames in an unsegmented video sequence , which makes it suitable for real - time video temporal segmentation .
while the proposed method relies on simple components , we showed that a specialization for skeleton - based action detection can be established which , not only outperforms the state - of - the - art , but also overcomes the limitations of similar approaches .
the authors would like to thank smartci research center for supporting this research .
|
sliding window is one direct way to extend a successful recognition system to handle the more challenging detection problem . while action recognition decides only whether or not an action is present in a pre - segmented video sequence , action detection identifies the time interval where the action occurred in an unsegmented video stream .
sliding window approaches can however be slow as they maximize a classifier score over all possible sub - intervals .
even though new schemes utilize dynamic programming to speed up the search for the optimal sub - interval , they require offline processing on the whole video sequence . in this paper
, we propose a novel approach for online action detection based on 3d skeleton sequences extracted from depth data .
it identifies the sub - interval with the maximum classifier score in linear time .
furthermore , it is suitable for real - time applications with low latency .
|
in recent years , several experiments under well - controlled laboratory conditions were carried out to explore pedestrian characteristics in bottlenecks .
most experiments focused on the relationship between bottleneck width and pedestrian flow @xcite .
the flow was once announced to grow in a stepwise manner along with lane formation @xcite .
however , it was shown afterwards by more detailed experiments to be linearly dependent on bottleneck width .
the slope of the linear function is approximately 1.9 ( m@xmath0s)@xmath1 with bottleneck width ranging from 0.7 m to 5.0 m @xcite . when bottleneck width is smaller than 0.7 m , the slope of the linear function decreases with increasing bottleneck width @xcite .
other bottleneck geometrical factors were also studied .
shorter bottlenecks provide higher flow than longer ones @xcite .
wider width of the passage in front of bottleneck leads to higher flow @xcite .
the flow also increases with increasing distance between bottleneck and holding area @xcite .
usually bottleneck geometry was made up of boards higher than 2.0 m to prevent pedestrians bodies overlapping the boundaries . but nagai et al .
@xcite used desks with the height of approximately 0.8 m , which was pointed out to actually provide wider bottleneck widths @xcite . helbing et al .
@xcite and yanagisawa et al .
@xcite studied the influence of an obstacle in front of the bottleneck , and found that the existence of the obstacle leads to higher flow especially when shifted from the center .
furthermore , the influences of non - geometrical factors were studied .
nagai et al .
@xcite changed the initial density of the participants in the holding area , and found that the flow increases with increasing initial density but the rate of the increase decreases .
daamen et al .
@xcite changed the composition of the participants , and found that the experiment with mainly children has the highest flow and that with disabled pedestrians has lower flow .
most experiments were conducted under normal situations , in which the participants were asked to walk through the bottleneck with normal speed .
only a few experiments were conducted under competitive @xcite , hurried @xcite , pushing @xcite or stressful @xcite situations .
unfortunately , no coincident result is made considering the influence of different situations .
it is worth noting that pedestrian movement includes transient state and steady state . in pedestrian dynamics experiments ,
transient state depends strongly on initial conditions while steady state is a good indicator of the independency of initial conditions , especially when the duration of experiments is short .
thus steady state is a distinguished significance for the interpretation of the results , and transient state should be excluded when combining different experiments to get universal conclusions .
however , few research about bottleneck experiments considered the difference between transient state and steady state .
cepolina @xcite showed the time series of pedestrian flow .
the flow changes significantly in transient state which is at the beginning and at the end , but keeps constant in steady state .
yet she did not further analyse the flow in steady state .
seyfried et al .
@xcite used regression analysis to calculate the stationary values for the density and speed , but the flow in steady state was not considered .
rupprecht et al .
@xcite found that the flows with all participants and with specified participants have different trends , but the specified participants were selected arbitrarily not considering steady state .
liao et al .
@xcite compared the flows in all states and in steady state , and found that they are both linearly dependent on bottleneck width but with different slopes .
nevertheless , the duration of steady state was determined according to `` small fluctuations '' in density and speed time series of the experiments , but the demarcation of the `` small fluctuations '' is ambiguous . moreover
, the specific relationship between steady state and the flow is unknown . up to now , no generally accepted method was used to detect steady state in bottleneck experiments . manually selected steady state can be used , but the result varies with different steady states from different researchers . for automatical method ,
krausz et al .
@xcite did similar work by using a cumulative sum control chart ( cusum ) algorithm to detect transitions from the optical flow computations at the loveparade 2010 in duisburg . in this work , a feasible method to detect steady state
is proposed and the steady state of pedestrian flow in bottleneck experiments is studied .
the remainder of the paper is organized as follows . in section 2 a cusum algorithm
is proposed and improved to automatically detect steady state , and the threshold of the detection parameter in the algorithm is calibrated using an autoregressive model .
section 3 applies the modified cusum algorithm to three groups of bottleneck experiments , and compares the flows in all states and in steady state .
finally , the conclusions are made in section 4 .
in pedestrian dynamics , time series of the variables density @xmath2 , speed _ v _ and flow _ j _ and their steady states are used to characterize the transport of pedestrian streams .
the classic definition of flow is the number of the pedestrians passing a line per unit time : @xmath3 where @xmath4 is the time gap between the crossing of two following pedestrians , and _
t _ is the time for all the pedestrians _ n _ passing the measurement line .
a reasonable resolution is necessary to identify steady state , but microscopic measurements of the flow by time gaps or mean values over small intervals could lead to strong fluctuations .
these flow fluctuations are caused by the relationship between the size of the objects ( here pedestrians ) and the size of the facilities in combination with ordering phenomena like the zipper effect ( see figure 5 and figure 7 in @xcite ) . to deal with this problem
, we refer to the flow equation from fluid dynamics : @xmath5 where _ w _ indicates the width of the facility . here
@xmath2 and _ v _ is the mean density and mean speed in the measurement area , respectively .
following this equation , the steady states based on density and speed are a good indicator for the steady state of flow . in this paper , we aim to detect steady state from the flow in equation ( 1 ) with relatively short time series .
the series lengths are too short to obtain precise flow estimations , so we use the density and speed variables in equation ( 2 ) to determinate if the system is in steady state .
the aforementioned density and speed are calculated using the voronoi method to obtain estimations with low fluctuations ( see equation ( 8) and ( 9 ) in @xcite ) .
this approach allows precise detection of the steady state at the scale of only few seconds . since the detection of steady state is a sort of transition detection , the cusum algorithm
is proposed in this work .
it is a sequential analysis technique which is initially used for monitoring transition detection @xcite .
the precondition of the cusum algorithm is a reference process with observations in normal situation .
let @xmath6 denote a sequence of observations in density or speed .
the manually selected steady state can be regarded as the reference process @xmath7 , where @xmath8 and @xmath9 ( see the interval between the two green dash - dotted lines in figure 1 ) .
q(@xmath10 ) _ and _ q(1-@xmath10 ) _ represent the upper and lower @xmath10-percentile of the distribution in the reference process @xmath7 , respectively .
the cusum algorithm continuously accumulates the deviations of the observations @xmath6 from the @xmath10-percentiles @xcite : @xmath11 the statistics @xmath12 and @xmath13 show the degree of the fluctuations in @xmath6
. if the distribution of @xmath6 is different from the distribution of @xmath7 , then the statistics increase .
therefore , the cusum algorithm is able to detect the transitions of the observations from the given reference process .
the specific way is combining @xmath12 and @xmath13 with a threshold of the detection parameter @xmath14 , then the interval with both @xmath12 and @xmath13 under @xmath14 is the steady state
. 0.49 in ( a ) density and ( b ) speed .
the interval between the two green dash - dotted lines is the manually selected steady state , which is regarded as the reference process @xmath7 , where @xmath8 and @xmath9 .
the interval between the two red dashed lines is the final steady state , which is obtained using the detected steady state minus the corresponding reaction time.,title="fig : " ] 0.49 in ( a ) density and ( b ) speed .
the interval between the two green dash - dotted lines is the manually selected steady state , which is regarded as the reference process @xmath7 , where @xmath8 and @xmath9 .
the interval between the two red dashed lines is the final steady state , which is obtained using the detected steady state minus the corresponding reaction time.,title="fig : " ] the preliminary cusum algorithm can detect the steady state , but there exist problems during the detection procedure .
first , since the statistics @xmath12 and @xmath13 are combined to detect the steady state , it is redundant to calculate them separately . therefore , new statistics @xmath15 are introduced ( equation ( 5 ) ) .
second , the response of the statistics to the fluctuations in @xmath6 is not sensitive enough .
the increase rate and decrease rate of the statistics are not the same .
to solve this problem , a step function is introduced to enhance the sensitivity of the response ( equation ( 6 ) ) .
third , the criterion in equation ( 6 ) is not the @xmath10-percentiles of @xmath7 any more .
the replacement is @xmath16 , which is the upper @xmath10-percentile in a standard normal distribution @xmath17 .
accordingly , the standard score transformation of the observations @xmath6 is used ( equation ( 7 ) ) .
fourth , the maximum value of the statistics is very large that it needs a long time to get back to steady state again . as an improvement
, a boundary @xmath18 is introduced to limit the increase of the statistics .
last but not least , since it is unknown whether at the beginning the statistics are in steady state or not , the value of @xmath19 should be equal to the boundary @xmath18 . from the above ,
the modified cusum algorithm is as follows : @xmath20 where @xmath21 is the mean of @xmath7 , and @xmath22 is the standard deviation of @xmath7 . as shown in figure 2 , the statistics @xmath15 calculated by the modified cusum algorithm have sensitive response to the fluctuations , and the trend of the statistics is clear .
0.49 in figure 1 ( a ) and ( b ) , respectively .
@xmath23 shows the degree of the fluctuations in the observations @xmath6 .
the interval with @xmath23 under @xmath14 is the detected steady state.,title="fig : " ] 0.49 in figure 1 ( a ) and ( b ) , respectively .
@xmath23 shows the degree of the fluctuations in the observations @xmath6 .
the interval with @xmath23 under @xmath14 is the detected steady state.,title="fig : " ] we should note that the detected transitions ( the interactions in figure 2 ) are not the real transitions in the observations @xmath6 .
there is a reaction time caused by the detection procedure itself .
when reaching steady state and leaving steady state , the reaction time is calculated as follows , respectively : @xmath24 where @xmath25 is the frame number per second in the observations . the final steady state is using the detected transitions minus the corresponding reaction times ( see the interval between the two red dashed lines in figure 1 ) . the accurate detection of steady state depends on the detection parameter s threshold @xmath14 to be calibrated .
for a neyman - pearson statistical test , @xmath26 is the upper @xmath27-percentile of the statistics @xmath28 for the reference process @xmath7 . in this way ,
we are able to control the probability of detecting a false transition , which is the risk level 1-@xmath27 . different methods to calibrate @xmath14 are shown in figure 3 .
the simplest way is the direct calibration with the reference process @xmath7
( @xmath8 and @xmath9 ) .
this method requires that the number of the selected observations in the reference process should be large enough for calibration .
if not , a bootstrap method is used to expand the reference process @xcite .
the requirement is that the observations in @xmath7 should be independent of each other , or that the number of the observations in @xmath7 should be large enough ( moving block bootstrap ) . .
] in this work , the observations do not meet any requirements of the two empirical methods in figure 3 .
therefore , an autoregressive process @xmath29 is proposed to model the standard score of the reference process @xmath30 : @xmath31 where @xmath32 is the first autocorrelation of @xmath30 , and @xmath33 are independent normal random variables .
the autoregressive process @xmath29 is stationary with normal distribution , and @xmath34 for all @xmath35 . using the autoregressive model ,
the threshold @xmath26 can be obtained by simulations or analytically by using mathematical tools of stochastic processes . if by simulations , the simulation time @xmath36 should be sufficiently long enough to exclude fluctuations and precisely estimate the distribution of @xmath28 .
moreover , several runs must be done to control the precision of the estimation . when @xmath37 , the computational effort to get good estimations is important for standard computers .
for instance if @xmath38 , the process has to be simulated @xmath39 steps to obtain precise results .
the threshold @xmath26 with the autoregressive process can be faster obtained analytically .
the couple @xmath40 is a markov chain with stationary distribution @xmath41 ( @xmath42 , @xmath43 ) such that ( global balance equation ) : @xmath44 for all @xmath45 , and at the borders @xmath46 if @xmath47 , and @xmath48 if @xmath49 . here
@xmath50 .
unfortunately , equations ( 10 - 12 ) do not have an explicit solution for @xmath51 .
yet they can be approximated by using the rectangular numerical scheme @xmath52 with @xmath53 , @xmath54 and @xmath55 .
the numerical approximation yields in the linear equation @xmath56 where @xmath57 and @xmath58 with @xmath59 & & b_1&&b_2\\ & & & b_1&b_1{\end{matrix}\right]},\quad b_1 = \!{\left[\begin{matrix}}b_{0,0}\ldots\ b_{0,k}\\[2 mm ] b_{i_a,0}\ldots\ b_{i_a , k}\\[2 mm ] \\[-0 mm ] b_{i_b,0}\ldots\ b_{i_b , k}\\[2 mm ] b_{k,0}\ldots\ b_{k , k } { \end{matrix}\right]},\quad b_2 = \!{\left[\begin{matrix}}\\[2 mm ] b_{i_a+1,0}\ldots\ b_{i_a+1,k}\\[2 mm ] b_{i_b-1,0}\ldots\ b_{i_b-1,k}\\[2 mm ] \ { \end{matrix}\right ] } , \tag{15 } \end{aligned}\ ] ] @xmath60 and @xmath61 .
the stationary distribution of the couple @xmath62 is approximated by solving a linear system with @xmath63 equations .
the complexity of the resolution is in @xmath64 by using the thomas - algorithm ( see in appendix ) , which makes in general the numerical approximation of the analytical solution faster than the simulations .
it is difficult to estimate the error of the numerical approximation .
tests show that @xmath65 gives good results .
the error can be reduced by using trapezoidal or polynomial schemes instead of rectangular ones . for a series of observations @xmath6 , the steady state detected by the modified cusum algorithm
depends on the combination of the statistics @xmath15 and the threshold of the detection parameter @xmath14 .
scrutinizing equations ( 5 - 7 ) and ( 9 ) , six key parameters determining the values of @xmath15 and @xmath14 are summarized in table 1 .
@xmath10 , @xmath27 and @xmath18 are independent of the observations so they should be calibrated independently .
the value of @xmath10 should be close to 1 to diminish outliers in @xmath7 as much as possible .
the value of @xmath27 should be close to 1 to diminish false detections as much as possible .
the boundary @xmath18 of the statistics @xmath15 should be approximately two times of @xmath14 . in this work , the values of the three calibrated parameters are fixed , and @xmath10 = 0.99 , @xmath27 = 0.99 , @xmath18 = 100 . on the contrary ,
@xmath21 , @xmath22 and @xmath32 vary with different observations because they are measured from the original reference process @xmath7 . @xmath21 and @xmath22 affect the statistics @xmath15 ( equation ( 7 ) ) , and @xmath32 affects the threshold of the detection parameter @xmath14 ( equation ( 9 ) ) . in the situation that @xmath15 and @xmath14 both change with different reference processes
, the detected steady states might be different .
.key parameters in the modified cusum algorithm .
[ cols="^,^ , < " , ] the modified cusum algorithm is applied to detect the steady states of the three groups of bottleneck experiments . for the experiments
eg _ and _ ao _ the measurement area of the detected variables , density and speed , is at the beginning of the bottleneck with size _ b _ ( see the region slashed in green in figure 5 ) . for the experiment _ uo _
the measurement area is in front of the bottleneck with size _ b _ ( see the region back - shaded in blue in figure 5 ) .
the measurement line of pedestrian flow is always at the beginning of the bottleneck .
the relationship between flow and bottleneck width is shown in figure 6 .
all the flows are calculated according to equation ( 2 ) .
the error bars represent the standard errors of the flows in steady states .
the calculation of the standard error is by dividing the standard deviation by the square root of the number of the measurements that make up the mean .
since the observation ( flow ) in each frame is dependent with the observations nearby , they can not be used as the measurements directly .
thus the observations in steady state are divided into groups that the number of the groups _
p _ is the number of the measurements to calculate the standard error : @xmath66 where _ f _ is the number of the observations in steady state , and _
p _ is the number of the observations in each group . here _
p _ should be large enough to make the groups independent of each other , but also small enough to ensure the sufficient number of the measurements to diminish fluctuations when making up the mean . in this work , @xmath67 and if @xmath68 the bias of an estimator is used .
as shown in figure 6 , the flow is linearly dependent on bottleneck width but the slope of the linear function varies in different experiments .
this reconfirms the result that the flow per unit width is constant as bottleneck width changes @xcite . for each run in each experiment ,
the flows in all states and in steady state are compared and the difference between them is analysed .
as shown in figure 6 , the difference changes significantly in the last few runs of the experiments _ eg _ and _ ao _ but slightly in the experiment _ uo_. to scrutinize what exactly impacts the difference , the parameters in table 2 are studied .
bottleneck length _
l _ , passage width _ w _ , the distance between bottleneck and holding area _
, pedestrian number _
n _ and initial density @xmath69 are same in the experiments _ eg _ and _ ao_. since the difference changes significantly in part runs of these two experiments , the above parameters have no direct influence on the difference
. bottleneck width _
b _ is positively correlative with the difference in the experiments _ eg _ and _ ao _ , but the relationship is ambiguous in the experiment _
uo_. in this situation , we can not judge that bottleneck width _
b _ is directly proportional to the difference .
afterwards , different combinations of the parameters in table 2 are studied . the ratio of pedestrian number to bottleneck width _
n / b _ is found to have a direct influence on the aforementioned difference , which is represented by _
z _ in figure 7 .
_ z _ is calculated by using the flow in all states minus the flow in steady state and taking the absolute value . as shown in figure 7 ,
_ z _ decreases with the increase of the ratio _
n / b_. the critical value of the ratio _
n / b _ is approximately 115 persons / m , which determines if the difference between the flows in all states and in steady state is acceptable . considering the practical significance of _ n / b _ , it describes whether the number of participants is large enough for the corresponding bottleneck width to reach a steady state . when the value of _ n / b _ is higher than the critical value , the steady state is quickly reached and the duration of steady state is long .
conversely , it takes a long time to reach steady state and the duration of steady state is short when the value of _ n / b _ is lower than the critical value . in the case of extremely small values of _ n / b _ , the steady state even can not be reached .
from the above , the ratio of pedestrian number to bottleneck width _
n / b _ is the main influence factor of the difference between the flows in all states and in steady state .
other factors , such as the composition of participants and the motivation of the experiments , also impact the difference aforementioned , but they are beyond the consideration of this work .
in this work , a feasible cusum algorithm is proposed to automatically detect steady state from density and speed time series of bottleneck experiments . to improve the algorithm , a step function
is introduced to calculate the statistics for enhancing the sensitivity of the response to the fluctuations .
in addition , a boundary is added to limit the increase of the statistics .
the threshold of the detection parameter in the algorithm is calibrated using an autoregressive model . comparing the detected steady states with previous manually selected ones ,
the modified cusum algorithm gives more reproducible results .
for the applications , three groups of bottleneck experiments are analysed by the modified cusum algorithm .
the steady states of density and speed in each run are detected separately .
then the interval which is included both in the steady states of density and speed is regarded as the steady state of the flow .
the flows in all states and in steady state are measured and compared .
the results reconfirm that the flow per unit width is a constant as bottleneck width changes .
furthermore , the difference between the flows in all states and in steady state mainly depends on the ratio of pedestrian number to bottleneck width .
the critical value of the ratio is approximately 115 persons / m . when the value of the ratio is higher than the critical value , the steady state
is quickly reached and lasts longer .
in this situation , the flow in all states is almost identical with the flow in steady state . considering the ratio , we have more possibilities to compare the flows from different experiments , especially when the detection of steady state is difficult . in future studies , the flow in steady state
should be used when combing the flows from different bottleneck experiments . for the experiments in which it is difficult or impossible to detect steady state
, the ratio of pedestrian number to bottleneck width should be calculated to estimate the difference between the flows in all states and in steady state .
the matrix @xmath70 in equation ( 13 ) has a size of @xmath71 .
solving @xmath70 directly with gaussian elimination takes @xmath72 operations , which is not feasible for great @xmath18 .
since @xmath70 is of block - tridiagonal structure , a block thomas - algorithm @xcite is used to solve the system efficiently in @xmath73 .
essentially it is a block matrix variant of the thomas - algorithm , which is a simplified version of the gaussian elimination on tridiagonal matrices . where @xmath75 are matrices of size @xmath76 and @xmath77 are vectors of size @xmath78 .
since the zero blocks on the off diagonals are not necessary , using a sparse storage scheme reduces the memory amount for the matrix from @xmath79 to @xmath80 .
j. liddle , a. seyfried , w. klingsch , t. rupprecht , a. schadschneider , and a. winkens .
an experimental study of pedestrian congestions : influence of bottleneck width and length . in traffic and granular flow 2009 , 2009 .
a. seyfried , m. boltes , j. khler , w. klingsch , a. portz , t. rupprecht , a. schadschneider , b. steffen , and a. winkens .
enhanced empirical data for the fundamental diagram and the flow through bottlenecks . in w. w. f. klingsch , c. rogsch ,
a. schadschneider , and m. schreckenberg , editors , pedestrian and evacuation dynamics 2008 , 145 - 156 , 2010 .
t. rupprecht , w. klingsch , and a. seyfried
. influence of geometry parameters on pedestrian flow through bottleneck . in r. d. peacock , e. d. kuligowski , and j. d. averill , editors ,
pedestrian and evacuation dynamics 2010 , 71 - 80 , 2011 .
w. song , j. zhang , and a. seyfried .
experimental study of pedestrian flow in the channel through bottleneck . in r. d. peacock , e. d. kuligowski , and j. d. averill , editors ,
pedestrian and evacuation dynamics 2010 , 875 - 879 , 2011 .
w. song , w. lv , and z. fang .
experiment and modeling of microscopic movement characteristic of pedestrians .
procedia engineering , 62(0 ) : 56 - 70 , 2013 .
9th asia - oceania symposium on fire science and technology .
w. liao , a. seyfried , j. zhang , m. boltes , x. zheng , and y. zhao .
experimental study on pedestrian flow through wide bottleneck .
transportation research procedia , 2(0 ) : 26 - 33 , 2014 .
the conference on pedestrian and evacuation dynamics 2014 ( ped 2014 ) , 22 - 24 october 2014 , delft , the netherlands .
w. liao , m. chraibi , a. seyfried , j. zhang , x. zheng , and y. zhao .
validation of fds+evac for pedestrian simulations in wide bottlenecks .
in intelligent transportation systems ( itsc ) on 2014 ieee 17th international conference , 554 - 559 , october 2014 .
d. yanagisawa , a. kimura , a. tomoeda , n. ryosuke , y. suma , k. ohtsuka , and k. nishinari .
introduction of frictional and turning function for pedestrian outflow with an obstacle .
physical review e , 80(3 ) : 036110 , 2009 .
h. c. muir , d. m. bottomley , and c. marrison .
effects of motivation and cabin configuration on emergency aircraft evacuation behavior and rates of egress . the international journal of aviation psychology , 6 : 57 - 77 , 1996 .
j. zhang , w. klingsch , a. schadschneider , and a. seyfried .
transitions in pedestrian fundamental diagrams of straight corridors and t - junctions .
journal of statistical mechanics : theory and experiment , 6 : 06004 , 2011 .
|
experiments with pedestrians could depend strongly on initial conditions
. comparisons of the results of such experiments require to distinguish carefully between transient state and steady state . in this work , a feasible algorithm - cumulative sum control chart
- is proposed and improved to automatically detect steady states from density and speed time series of bottleneck experiments .
the threshold of the detection parameter in the algorithm is calibrated using an autoregressive model . comparing the detected steady states with previous manually selected ones
, the modified algorithm gives more reproducible results .
for the applications , three groups of bottleneck experiments are analysed and the steady states are detected .
the study about pedestrian flow shows that the difference between the flows in all states and in steady state mainly depends on the ratio of pedestrian number to bottleneck width .
when the ratio is higher than a critical value ( approximately 115 persons / m ) , the flow in all states is almost identical with the flow in steady state .
thus we have more possibilities to compare the flows from different experiments , especially when the detection of steady states is difficult .
keywords : bottleneck , experiments , flow , steady state , cusum
|
in einstein - hilbert gravity coupled to various matter fields , the exterior geometry of a 3 + 1 dimensional asymptotically flat black hole ( bh ) spacetime typically approaches a stationary solution at late times .
non - stationary perturbations decay both by falling across the horizon and dispersing to infinity , as described by the ringdown of quasi - normal modes followed by power - law tails .
our purpose here is to explore a corresponding late - time limit of the associated black hole interiors .
we will argue that as far as the observations of late - infalling physical observers are considered , the result is well - described by a simple effective geometry which contains the part of the corresponding stationary bh solution to the past of the inner horizon .
however , the regular inner horizon is replaced by singular components of two different types : ( i ) the _ ingoing _ section of the inner horizon the cauchy horizon ( ch)is replaced by a null , weak , curvature singularity , and ( ii ) the _ outgoing _ section of the inner horizon is replaced by an outgoing shock - wave singularity .
the presence of a null , weak , curvature singularity at the ch is a well - known phenomenon since the pioneering works of hiscock on the reissner - nordstrom - vaidya solution @xcite and of poisson and israel on the mass - inflation model @xcite ( see also @xcite ) .
it is the second singular component
the outgoing shock - wave singularity which will be our main concern in this paper . our study is motivated in part by the picture of extreme black holes at late times suggested in @xcite and explored further in @xcite .
this picture agrees with the extreme limit of our results below .
the starting point for our analysis is the large body of literature studying perturbations of the reissner - nordstrm ( rn ) and kerr interiors . with the assumption of spherical symmetry
, these works establish that perturbations transform at least the initial part of the ingoing inner horizon ( the ch ) of rn into a null curvature singularity often called the mass - inflation singularity .
this singularity is weak in the sense that the metric remains continuous at the singularity , though it is not differentiable . in particular , the area - radius @xmath0 of the spheres is well - defined at this singularity , taking the value @xmath1 near the point marked @xmath2 on the conformal diagram shown in figure [ summary ] ( left ) and shrinking toward the future as described by the raychaudhuri equation , eventually reaching @xmath3 . in the spherical case , at least when the matter content includes a minimally - coupled massless scalar field , it was numerically established @xcite that when @xmath0 shrinks to zero the weak null singularity meets a strong spacelike singularity these numerical simulations showed that a transition from a null to a spacelike singularity occurs in a region which ( when mapped to a collapsing - shell spacetime ) would correspond to being _ outside _ the shell ( i.e. where the electric field is non - zero ) .
however , when the initial scalar perturbations are sufficiently weak , the full focussing of the ch to @xmath3 will only occur inside the shell .
we strongly expect the formation of a spacelike singularity in this case as well , and have drawn this scenario in figure [ summary ] .
] , along which @xmath3 .
this situation is depicted in the left panel of figure [ summary ] .
note that curvature scalars diverge at both the above spacelike and null singularities .
while the establishment of a spacelike singularity is less certain for other forms of matter ( e.g. perfect fluids ) , it is nevertheless expected that all future - directed timelike and null curves inside the black hole will be incomplete due to reaching a curvature singularity of some form .
investigations of nonlinearly - perturbed spinning bhs reveal a similar scenario .
perturbative analyses @xcite again indicate the formation of a null weak scalar - curvature singularity at the ch ( though this time the singularity is generically oscillatory @xcite , as opposed to the monotonic mass - inflation singularity in the spherical case ) .
the presence of such a null singularity is also supported by an asymptotic local analysis of the einstein equations @xcite , as well as by exact analytical constructions of locally - generic classes of null weak singularity @xcite . yet , to the best of our knowledge , no numerical verification of this scenario has yet been carried out in the spinning case . *
left : * a reissner - nordstrm black hole formed by collapse of a thin spherical charged shell and subject to spherical perturbations by a massless scalar field .
two types of singularities form : a null weak singularity at the ch ( the diagonal dashed line ) and a spacelike @xmath3 singularity ( the solid horizontal line ; though see footnote [ insideshell ] ) .
depending on the strength of the initial perturbation , the spacelike piece of the singularity might also form mcuh earlier .
it may intersect the worldline describing the shell or even entirely remove the would - be outgoing inner horizon . *
right : * our proposed effective geometry of the perturbed ( but spherically symmetric ) collapsing shell spacetime as seen by late - time observers .
it consists precisely of the region of the unperturbed eternal reissner - nordstrm black hole with @xmath4 , together with two types of singularities on its future boundary ( at @xmath5 ) : the diagonal dashed line at the upper right represents the mass - inflation singularity , a null weak singularity located at the ch .
the diagonal solid line at the upper left represents a null shock - wave singularity , where the metric tensor is effectively discontinuous .
, title="fig : " ] * left : * a reissner - nordstrm black hole formed by collapse of a thin spherical charged shell and subject to spherical perturbations by a massless scalar field .
two types of singularities form : a null weak singularity at the ch ( the diagonal dashed line ) and a spacelike @xmath3 singularity ( the solid horizontal line ; though see footnote [ insideshell ] ) . depending on the strength of the initial perturbation , the spacelike piece of the singularity might also form mcuh earlier .
it may intersect the worldline describing the shell or even entirely remove the would - be outgoing inner horizon . *
right : * our proposed effective geometry of the perturbed ( but spherically symmetric ) collapsing shell spacetime as seen by late - time observers .
it consists precisely of the region of the unperturbed eternal reissner - nordstrm black hole with @xmath4 , together with two types of singularities on its future boundary ( at @xmath5 ) : the diagonal dashed line at the upper right represents the mass - inflation singularity , a null weak singularity located at the ch .
the diagonal solid line at the upper left represents a null shock - wave singularity , where the metric tensor is effectively discontinuous .
, title="fig : " ] the results above provide a good ( if not yet fully complete ) understanding of the internal structure of generic ( charged , rotating , non - stationary ) asymptotically flat , isolated ( i.e. non - accreting @xcite ) black holes in classical general relativity .
however , the detailed experiences of any given observer inside such a black hole will in general depend on the process by which the black hole was formed and on the particular perturbations generated .
in contrast , we argue below that the spacetime effectively simplifies from the point of view of observers who enter the black hole at late times , which we call late infall - time ( or just late - infall ) observers .
the simplified spacetime may be described by the simple , stationary bh solution up to the inner horizon and an effective outgoing shock wave at the outgoing portion of the latter ( plus a null weak singularity at the ch ) .
this structure is depicted in the right panel of figure [ summary ] . due to the key role it plays in our analysis ,
it is useful to describe the notion of _ late - infall observers _ in more detail .
we start with a simple demonstration ( though not necessarily the most precise or most general one ) of this concept .
recall that perturbations outside the bh decay at late time , where of course late means relative to the formation of the black hole and the onset of any significant new perturbations .
therefore , for observers who fall into the bh at sufficiently late time , the exterior geometry will be well approximated by the stationary ( and axially - symmetric ) bh metric .
this allows one to associate specific values of energy @xmath6 ( @xmath7 ) and angular momentum @xmath8 ( @xmath9 ) to the late - time geodesics . more importantly ,
owing to the approximate time - translation of the external geometry , from any seed " infalling geodesic @xmath10 , we may construct a one - parameter family @xmath11 of similar geodesics , obtained from @xmath10 by time translation to the future .
( we emphasize that in the present construction the members of @xmath11 are _ exactly _ geodesics , all related to @xmath10 by the approximate time translation .
may be constructed as follows .
let @xmath12 be a set of coordinates for the bh exterior , such that at late time the associated metric functions are approximately independent of @xmath13 .
we pick a certain point @xmath14 on the seed " geodesic @xmath10 outside the bh , and time - translate it to the future by a certain amount @xmath15 . at the new point , which we denote @xmath16 , we set the four - velocity components @xmath17 to be numerically the same as those of @xmath10 at @xmath14 .
the geodesic @xmath18 which emanates from @xmath16 with those initial four - velocity components @xmath17 now becomes a member of the set @xmath11 . ] ) note in particular that all geodesics in @xmath11 share ( approximately ) the same values of @xmath6 and @xmath8 .
now , each member of @xmath11 is characterized by the parameter @xmath19 , namely the value of the advanced time @xmath20 ( eddington s advanced null coordinate ) at which the geodesic crosses the event horizon .
for any given seed geodesic @xmath10 , the _ late - infall observers _ are those members of @xmath11 characterized by sufficiently large values of @xmath19 .
our main objective in this paper is to characterize the experience of such late infall observers who move toward the outgoing section of the inner horizon in a generically - perturbed charged ( or spinning ) bh .
we shall see that such observers experience abrupt changes in the amplitude of various perturbing fields as well as the metric itself while crossing the ( would - be ) inner horizon .
these changes occur within a short proper - time interval whose magnitude decreases exponentially with the infall time @xmath19 . for an observer with fixed resolution and sufficiently large @xmath19 this proper - time interval
is so tiny that he experiences the perturbation as an effective shock wave .
the above - mentioned concept of late - infall observers may be generalized , and reformulated in a somewhat more precise manner ( though we will not attempt a fully precise definition here ) .
consider a continuous one - parameter family of inextendible causal curves labeled by the advanced time @xmath19 at which they cross the event horizon , with @xmath19 taking values in some range @xmath21 so that the family includes curves that enter the black hole at arbitrarily late times . note that since perturbations outside the black hole decay , the advanced time @xmath20 can indeed be used as a coordinate along the horizon at sufficiently late times .
we require that , in the limit @xmath22 , the part of these curves to the past of the event horizon approaches a _
stationary _ family of such curves in some stationary spacetime ; i.e. for which curves with different values of @xmath19 are related by the corresponding time - translation .
we further require that , for large @xmath19 , the parts of our curves to the future of the event horizon all have the same proper accelerations when expressed using a reference frame parallel - propagated along the world line in terms of the proper time along the worldline after crossing the event horizon .
one may think of this as the assumption that all observers in a given family are equipped with identically pre - programmed rocket ships .
of course we insist that these reference frames at the event horizon are also related by an approximate time translation at large @xmath19 . for our rather qualitative purposes below it
will not be necessary to specify the precise rate at which these limiting behaviors are approached , though some such specification will certainly be needed to derive more precise results . after a brief review of charged spherical black holes in section [ preliminaries ] , we study the experiences of late - time observers in stationary spacetimes subject to linear perturbations in section [ perturbations ] .
such observers experience no perturbation at all until they would expect to encounter an inner horizon .
however , they effectively encounter a shockwave at the outgoing inner horizon .
while we include a discussion of linearly perturbed kerr black holes , our treatment mainly focuses on the simpler spherically symmetric case .
non - linear perturbations are addressed in sections [ non - linear ] and [ latet ] .
here we consider only spherical black holes .
section [ non - linear ] addresses the model of a charged bh perturbed by a self - gravitating scalar field .
we show that the experiences of freely - falling late - time observers again agree with those in unperturbed reissner - nordstrm up to the point where they would expect to encounter the ( outgoing section of the ) inner horizon .
however , instead of finding a smooth null surface at that point , they effectively encounter a gravitational shockwave at which the metric is discontinuous .
section [ latet ] then gives a heuristic argument that the experiences of more general late - time observers are similar , and in particular that they are described by the effective spacetime shown in the right panel of figure [ summary ] .
the final discussion in section [ disc ] describes possible generalizations to rotating black holes and to black holes in any dimensions and the implications for finite - time observers who fall into astrophysical black holes .
we also discuss several aspects of the gravitational shock - wave phenomenon which takes place at the inner horizon .
the rn solution is the unique spherically - symmetric electrovac geometry . in schwarzschild
coordinates @xmath23 it takes the form @xmath24 where @xmath25 , and @xmath26 is the unit two - sphere . throughout this paper
we shall consider the non - extreme black - hole ( bh ) case , namely @xmath27 . in this case
@xmath28 vanishes at two @xmath0 values , @xmath29 .
the larger root @xmath30 corresponds to the event horizon , and the smaller one @xmath31 to the inner horizon .
figure [ rnfig ] ( left ) depicts a part of the penrose diagram of the eternal , analytically - extended , rn geometry .
note that the inner horizon has two separate portions the two intersecting null lines denoted @xmath32 " .
* left : * the eternal reissner - nordstrm black hole .
the vertical lines are singularities at @xmath3 , while the diagonal lines describe copies of asymptotically flat future or past null infinities ( @xmath33 and @xmath34 ) , outer horizons ( @xmath30 ) , and inner horizons ( @xmath31 ) .
the full analytically extended solution consists of a periodic vertical array of copies of the regions shown . *
right : * a conformal diagram for the spacetime of a collapsing ( but otherwise unperturbed ) spherically symmetric massive charged thin shell .
the @xmath3 curve is a regular origin in the flat region inside the shell .
the cauchy horizon is also indicated.,title="fig : " ] * left : * the eternal reissner - nordstrm black hole .
the vertical lines are singularities at @xmath3 , while the diagonal lines describe copies of asymptotically flat future or past null infinities ( @xmath33 and @xmath34 ) , outer horizons ( @xmath30 ) , and inner horizons ( @xmath31 ) .
the full analytically extended solution consists of a periodic vertical array of copies of the regions shown . *
right : * a conformal diagram for the spacetime of a collapsing ( but otherwise unperturbed ) spherically symmetric massive charged thin shell .
the @xmath3 curve is a regular origin in the flat region inside the shell .
the cauchy horizon is also indicated.,title="fig : " ] later we shall also consider the spherically - symmetric spacetime of a charged collapsing thin shell .
the geometry outside the shell is described by the reissner - nordstrom ( rn ) metric ( [ rn ] ) . inside the shell
the geometry is flat , i.e. minkowski .
figure [ rnfig ] ( right ) displays this hybrid spacetime . in this
non - eternal bh spacetime the ingoing portion of the inner horizon is a cauchy horizon ( ch ) .
the diagram only displays the globally - hyperbolic piece of the spacetime
namely , the region up to the cauchy horizon ( the extension beyond the ch will not concern us in this paper ) .
let us examine the free - fall orbits of observers who jump into the bh . throughout this paper we assume , as usual ( and without loss of generality ) that the motion is confined to the equatorial plane .
these geodesic orbits are characterized by two constants of motion : the energy " @xmath35 , and the angular momentum @xmath36 .
they satisfy the radial equation , @xmath37 , @xmath38 flips its sign ( a phenomenon known as gravitational bounce " ) . in this paper , however , we focus on the behavior of orbits up to their first approach to @xmath5 , hence @xmath39 . ]
@xmath40 of particular importance will be the behavior of these worldlines in the neighborhood of @xmath5 , where the last result becomes @xmath41 in the analysis below it will be useful to express the rn metric in double - null coordinates .
a particularly useful form is given by the eddington coordinates @xmath42 where @xmath43 is the tortoise coordinate defined by @xmath44 the line element then becomes @xmath45 where @xmath0 is to be regarded as a function of @xmath46 and @xmath20 , determined ( implicitly ) by setting @xmath47 in eq .
( [ tortoise ] ) . notice that in the region which will mostly concern us here the domain @xmath48 @xmath46 is _ past - directed _
( this choice of sign simplifies many of the expressions below ) .
note that @xmath49 at @xmath5 , implying that either @xmath46 or @xmath20 must diverge there .
it follows that at the ingoing section of the inner horizon @xmath50 ( as @xmath46 is regular ) , whereas at the outgoing section @xmath51 ( and @xmath20 is regular ) . since @xmath52 vanishes at the inner horizon , in its neighborhood we may approximate @xmath53 , where @xmath54 ( note that @xmath55 ) .
it then follows ) defines @xmath43 up to an integration constant .
we use this freedom and choose the convenient pre - factor @xmath56 in the right - hand side of eq .
( [ delta_r ] ) , which fixes this arbitrary constant . ] from eq .
( [ tortoise ] ) that near the inner horizon @xmath57 that is , @xmath43 diverges logarithmically ( in @xmath58 ) at the inner horizon .
the metric ( [ eddington ] ) is singular at the inner horizon ( where @xmath59 vanishes ) . to remove this coordinate singularity
we define the inner - horizon s kruskal - like coordinates @xmath60 , @xmath61 . with this choice of signs , both @xmath62 and @xmath63 are future - directed .
note that @xmath63 ( @xmath62 ) vanishes at the ingoing ( outgoing ) section of the inner horizon .
the line element now reads @xmath64 .
we shall not need here the specific form of the metric functions @xmath65 and @xmath66 .
we shall just note that both functions are regular .
furthermore , @xmath67 turns out to be a smooth function of @xmath0 which ( unlike @xmath52 ) is nonvanishing at @xmath31 . as a consequence ,
the kruskal metric is regular at the inner horizon .
in this section we consider several types of linear perturbations , and explore how these perturbations are experienced by late - time infalling observers .
we shall start our analysis by addressing the simplest type of perturbation , namely a purely - outgoing , spherically - symmetric , test scalar field .
then we shall proceed to consider more realistic types of linear perturbations , deferring non - linear perturbations to the following sections .
consider a free , massless , minimally - coupled , test scalar field @xmath68 on the rn background . the general behavior of a scalar field of this type inside the bh will be discussed in the next subsection . here
we focus on a spherically - symmetric scalar field , restricting our attention to the very neighborhood of the cauchy horizon .
as it turns out , in this region the field becomes purely outgoing , namely ) ] .
furthermore , the @xmath20-dependent component of @xmath68 decays as @xmath69 at large @xmath20 , leaving us with only the @xmath46-dependent component . ]
@xmath70 . in our first example
we shall further assume , for simplicity , that @xmath71 vanishes until a certain value @xmath72 ; then it grows monotonically up to @xmath73 , where it reaches its final value @xmath74 ( and remains constant afterward ) .
while this type of function @xmath71 is certainly oversimplified , it transparently demonstrates the mechanism responsible for the shock - wave formation . needs not be monotonic ( see discussion below ) .
here we picked a monotonic function @xmath71 only for the sake of conceptual simplicity . ] both null lines @xmath75 are assumed to reside in the internal range @xmath48 .
we shall now explore the behavior of @xmath68 along the worldline of the infalling observer , as a function of proper time @xmath76focussing our attention on late - time observers .
a free - falling observer in a rn spacetime crosses the inner horizon @xmath5 through its _ outgoing _ section ( where @xmath20 is finite ) .
let us denote the values of @xmath20 as the observer crosses the event and inner horizons by @xmath19 and @xmath77 , respectively .
since we are interested in late - time observers we shall assume @xmath78 .
is sufficiently large that the orbit is confined to the shell s exterior throughout the range @xmath79 .
also , in the case @xmath80 we further demand @xmath81 , such that @xmath82 as @xmath83 is approached . ] along a timelike worldline @xmath20 increases monotonically , hence in the range between the two horizons @xmath84 . obviously , @xmath19 and @xmath77 depend on the infalling geodesic .
however , for fixed parameters @xmath6 and @xmath8 , the difference @xmath85 will be independent of the infall time , owing to the time - translation symmetry of the rn metric .
@xmath86 is typically of order @xmath56 ( we assume here that @xmath6 is of order unity , i.e. not too large or too close to zero , and @xmath87 is @xmath88 or smaller ; and similarly the bh is not too close to schwarzschild or to extremality . ) next we evaluate the proper times @xmath89 at the two events where the worldline intersects the null lines @xmath75 . for convenience we set @xmath90 at the worldline s intersection with the inner horizon . from the above assumptions ( in particular @xmath78 )
it immediately follows that @xmath82 and hence @xmath91 throughout the portion @xmath92 of the worldline .
we can therefore use eqs .
( [ delta_tau ] ) and ( [ delta_r ] ) in this domain to obtain @xmath93 substituting @xmath47 we find for @xmath94 and @xmath95 @xmath96 where @xmath97 respectively denote the value of @xmath20 at the worldline s intersections with @xmath75 . since @xmath98
, we readily find @xmath99 hence the difference @xmath100 is bounded by @xmath101 the last inequality tells us at once that the late - time infalling observers will see a scalar - field profile @xmath102 which rises from zero ( at @xmath72 ) to its maximal value @xmath103 ( at @xmath104 ) within an arbitrarily short proper - time interval , @xmath105 .
for a sufficiently large @xmath19 , this proper - time interval will presumably be unresolved by an observer with fixed sophistication .
the large-@xmath19 observers will thus experience the scalar perturbation as a sharp shock - wave of finite amplitude .
note that in terms of the proper time of these observers , the shock wave is detected effectively at @xmath90 [ the limit @xmath106 of eq .
( [ bound])]namely , just at the crossing time of the ( outgoing portion of the ) inner horizon .
it should be emphasized that the two outgoing null lines @xmath107 and @xmath73 need _ not _ be close to the ( outgoing section of the ) inner horizon in any sense . to further clarify this point , consider the intersection of these two outgoing rays with an ingoing null geodesic @xmath108 located just after the collapse .
( to be more specific , within the shell - collapse scenario we may choose @xmath109 such that it passes through the intersection point of the collapsing shell with @xmath73 . )
let us denote the @xmath0 values at the two intersection points of @xmath110 with @xmath72 and @xmath104 by @xmath111 and @xmath112 respectively .
our point here is that @xmath111 and @xmath112 need _ not _ be close to @xmath31 ; rather , they can take values anywhere in the range @xmath113 . still , since @xmath47 , at a sufficiently large @xmath20 both these @xmath114 lines will necessarily attain @xmath82 values , and hence @xmath0 values arbitrarily close to @xmath31 .
it may be instructive to complement eq .
( [ difbound ] ) by an actual estimate of ( rather than just a bound on ) @xmath115 , for late infalling observers . based on eq .
( [ tau12 ] ) we express @xmath115 as @xmath116 } - e^{\kappa[u_{2}+(v_{ih}-v_{2 } ) ] } \right ) e^{-\kappa v_{ih } } .
\label{deltatau12}\ ] ] consider now the quantity @xmath117 .
for fixed parameters
@xmath6 and @xmath8 , @xmath118 is a well - defined function of @xmath0 ( independent of @xmath19 ) . in the late - infall limit , @xmath119 when the orbit reaches @xmath72 .
the quantity @xmath118 at @xmath72 similarly approaches its inner - horizon value , which ( for @xmath120 ) turns out to be a definite finite number , @xmath121 .
in addition , from eq .
( [ bound ] ) it follows that @xmath122 in the late - infall limit , and thus that @xmath123 .
obviously exactly the same argument applies to @xmath124 as well . using @xmath125 and @xmath126 in eq .
( [ deltatau12 ] ) , we thus obtain @xmath127 e^{-\kappa v_{eh}}. \label{difestimate}\ ] ] ( note that the pre - factor in squared brackets is independent of the infall time @xmath19 . ) the typical behavior of @xmath68 as a function of the observer s proper - time is depicted in fig .
[ simple ] for several values of infall time @xmath19 .
the typical behavior of @xmath68 as a function of the observer s proper time is depicted for several values of @xmath19 . ] summarizing , the late - time infalling observers will experience the scalar perturbation as a sharp shock wave of finite amplitude ( @xmath128 ) , with effectively vanishing rising time ( @xmath129)located just at the outgoing section of the inner horizon .
so far we have considered the case of a ( near - ch ) outgoing field @xmath70 which varies monotonically from @xmath130 at @xmath83 until @xmath73 where it saturates at @xmath131 .
the extension of the argument to a more general function @xmath71 ( not necessarily monotonic ; and not necessarily one with well - defined final saturation value ) is straightforward : it is sufficient to assume that in the near - ch region , @xmath71 varies over a certain range @xmath92 , from @xmath132 at @xmath72 to some @xmath133 at @xmath104 .
then , the above analysis yields that for a late - infall observer the finite jump @xmath134 in the value of @xmath68 will take place within an extremely small proper - time interval , which decreases as @xmath135 . for an observer of given sophistication ,
this exponentially small proper time will become unresolvable at sufficiently large @xmath19 .
hence , a late - infall observer will again experience a kind of effective shock - wave phenomenon : a finite change in the field , within an effectively - vanishing proper - time interval .
we turn now to consider more realistic perturbations of a spherical charged bh .
we still assume that in the region of interest the perturbations are small and can therefore be treated linearly .
again , we would like to explore how these perturbations will be experienced by late - time infalling observers .
the linearized perturbations both outside and inside the bh may conveniently be toy - modeled by a minimally - coupled massless scalar field @xmath68 , satisfying @xmath136 @xcite .
the perturbation field may be decomposed in spherical harmonics @xmath137 in the usual way , @xmath138 the individual perturbation modes @xmath139 all satisfy a hyperbolic partial differential equation of the form @xmath140 with @xmath141 , and with an @xmath142-dependent effective potential given by @xmath143 .
\label{scalarpotential}\ ] ] one then finds that outside the bh all modes @xmath144 decay to zero at late time @xcite .
this decay typically proceeds in two stages : first is the stage of quasi - normal ringing " ( i.e. exponentially - damped oscillations ) .
subsequently , after the ringing has damped , the late - time perturbations are dominated by inverse - power tails .
all modes @xmath144 decay ( along worldlines of constant @xmath0 ) as @xmath145 , with @xmath146 ( or @xmath147 for initially - static multipoles ) @xcite the same inverse - power form as in the schawarzschild case @xcite . throughout the rest of the paper
we shall focus our attention on the inverse - power tails ( rather than the quasi - normal ringing ) , because it is this component which eventually dominates at late time .
investigations of the dynamics of linearized perturbations inside the bh reveal a behavior which parallels the external dynamics in many respects , though there also are some important differences .
the infalling power - law tails lead to a similar inverse - power decay inside the bh .
thus , along lines of constant @xmath0 ( between the event and inner horizons ) , the perturbations still decay as @xmath145 @xcite .
note , however , that this time @xmath13 is a spacelike coordinate , so what we face here is a _ spatial _ rather than temporal decay .
( for a discussion of this issue see ref .
@xcite . ) being proportional to @xmath52 , the effective potential @xmath148 vanishes at @xmath31 ( exponentially in @xmath43 , as it also does at @xmath30 ) . as a consequence , in the neighborhood of the inner horizon the perturbations
take the simple form of a superposition of outgoing and ingoing modes , that is , @xmath149 hereafter we omit the sub - indices @xmath150 from the @xmath151-coefficients for brevity . in the asymptotic region @xmath152 , both functions @xmath153 and
@xmath154 admit simple inverse - power forms : @xmath155 again with @xmath147 or @xmath146 ( depending on the presence or absence of initial static multipole ) .
the coefficients @xmath156 are determined by the scattering problem , and are given in ref .
@xcite ( see also @xcite ) .
we consider now a late - infall observer , and examine how this observer will record the scalar perturbations , focusing on a particular multipole @xmath144 .
we restrict our attention to the orbit s section between two hypersurfaces @xmath83 and @xmath73 ( both reside in the domain between the event and inner horizons , and satisfy @xmath157 ) .
we focus on the quantity @xmath158 , where as before , @xmath89 respectively denote the proper times when the observer crosses the two hypersurfaces @xmath75 . for a sufficiently late infall time
, the observer will reach @xmath83 only when the orbit is already in the near - ch region , where eq .
( [ nearih ] ) applies .
@xmath159+[p_v(v_2)-p_v(v_1 ) ] , \ ] ] where @xmath160 . furthermore , since @xmath161 at large @xmath20 ( and @xmath162 ) , it follows that the second term in squared brackets vanishes at the late - infall limit .
thus , for a sufficiently large @xmath19 the expression for @xmath163 simplifies to @xmath164 for generic choice of @xmath165 the rhs is non - vanishing .
( this is most explicitly verified in the case where @xmath166 are both in the early domain , @xmath167 , where @xmath168 .
should be achieved when @xmath104 is not @xmath169 ( and @xmath72 is not too close to @xmath104 ) . ] ) yet , the proper - time difference @xmath170 is still given by eq .
( [ difestimate ] ) , namely , @xmath171 .
thus , for sufficiently late infall , the observer will watch a finite jump @xmath172 in the field component @xmath144 , during an effectively vanishing ( i.e. physically non - resolvable ) proper - time interval . in the rn spacetime , owing to the presence of electric field , the ( electrovac ) gravitational and electromagnetic perturbations are mutually coupled already at the linear level .
still , one can write decoupled field equations for a pair of combined electromagnetic / gravitational variables ( i.e. two specific linear combinations of the electromagnetic and gravitational variables ) @xcite . in particular , based on a formalism developed by moncrief @xcite , gursel et al .
@xcite constructed such a pair of electromagnetic / gravitational field variables @xmath173 ( for the various angular modes @xmath150 ) .
these variables satisfy the decoupled equations @xmath174 with the effective potential @xmath175 , \label{emgpotential}\ ] ] where @xmath176^{1/2 } $ ] . in turn
, the gravitational and electromagnetic perturbations may be recovered from the fields @xmath173 ( by certain linear combinations of the latter fields and their derivatives ) @xcite .
the potentials @xmath177 are again @xmath178 and therefore vanish at the two horizons , hence near the ch each of the fields @xmath173 is decomposed into free ingoing and outgoing components , i.e. @xmath179 , as in eq .
( [ nearih ] ) .
furthermore , the leading - order behavior of these two components in the early - ch domain @xmath152 was found @xcite to be of the same form as in the scalar case , i.e. eq . ( [ tails ] ) .
as before , our objective is the proper - time variation of the perturbation fields , as recorded by a late - infall observer .
the analysis of the previous subsections apply here with no modifications , implying that the variables @xmath173 undergo finite variations within proper - time intervals @xmath129 .
the gravitational and electromagnetic perturbations ( constructed from the variables @xmath173 and their derivatives ) are likely to yield a similar structure of an effective shock wave . in this subsection
we very briefly address the case of a linearly - perturbed kerr bh .
the latter s internal structure is known to be similar in many respects to that of a rn bh .
in particular there are two horizons , an event horizon located at @xmath180 and an inner horizon at @xmath5 , where @xmath181 . not surprisingly , we find that the effective shock - wave phenomenon takes place in the kerr case as well .
we focus here on the _ gravitational _ perturbation , which is apparently the perturbation field of greatest physical relevance here .
( note that the kerr background unlike the electrovac rn background admits pure vacuum gravitational perturbations . )
the behavior of late - time gravitational perturbations inside a kerr bh has been analyzed using two different formulations : ( i ) analysis of metric perturbations ( mp ) @xmath182 @xcite , ( ii ) analysis @xcite of the evolution of the teukolsky variables @xcite @xmath183 and @xmath184 .
both analyses examined the late - time gravitational perturbations , employing the so - called _ late - time expansion _
they both focussed on the near - ch behavior , and led to similar ( and mutually - consistent ) results . for the sake of the present analysis
the key result may be summarized as follows : near the ch ( that is , @xmath185 ) , the linear mp decouple into a superposition of outgoing and ingoing components , namely and @xmath186 , with @xmath43 now defined through @xmath187 ( where @xmath13 and @xmath0 are the boyer - lindquist time and radial coordinates ) . ]
@xmath188 furthermore , @xmath189 decays at @xmath190 as an inverse power of @xmath20 , hence for a late - infall observer @xmath191 near the ch . in the kerr background ( unlike the rn case ) , infalling timelike geodesics may intersect the inner horizon @xmath5 either at its ingoing or outgoing section . here
we shall focus on those orbits intersecting the _ outgoing _ section ( these include , for example , all infalling geodesics with positive @xmath6 and @xmath192 ) .
the effective shock - wave phenomenon will only occur for this class of orbits .
consider now an observer which falls into the kerr bh ( and heads towards the outgoing section of the inner horizon ) , at the late - infall limit .
as before , we focus on the observer s history while moving between @xmath83 and @xmath73 .
just like in the rn case , both proper - times @xmath89 scale at the late - infall limit as @xmath129 and so does their difference @xmath115 . here
@xmath193 is a certain positive constant , the inner - horizon s surface gravity .
again , we find a finite jump @xmath194 in the metric , within an effectively - vanishing proper time @xmath129 namely , an effective gravitational shock wave . on the other hand , geodesics with @xmath195 will intersect the _ ingoing _ section of @xmath5 . in the late - infall limit , these observers will hit the ch at its past boundary .
we shall provide here a simple interpretation of the effective shock - wave phenomenon derived above . for concreteness and simplicitly we present the explicit argument only for the rn case , but it applies to the kerr case as well .
the line element ( [ eddington ] ) preserves its form under a coordinate transformation of the form @xmath196 where @xmath197 is any constant ( this invariance reflects the time - translation symmetry of rn )
. we shall refer to different sets of eddington coordinates corresponding to different choices of @xmath197as different _ eddington frames _
( this terminology is borrowed from the analogous notion of _ lorentz frames _ in minkowski spacetime ) .
note that all tensors constructed from the metric are unaffected by this transformation .
in addition , @xmath43 preserves its functional form , @xmath198 .
consider two infalling observers , which move along two identical worldlines related to each other by a time translation .
these observers cross the event horizon ( eh ) at eddington times @xmath199 and @xmath200 , respectively ( with @xmath201 ) . owing to this difference in @xmath19
, the two observers will not share the same function @xmath202 ( or @xmath203 ) . to bridge this difference , we equip the second observer with its own eddington frame @xmath204 , setting @xmath205 in eq .
( [ edshift ] ) . since now @xmath206 , it is not difficult to show that @xmath207 of the second observer will be the same function as @xmath202 of the first observer and the same relation will apply between @xmath208 and @xmath203 .
consider now some linear perturbation field @xmath209 on the background spacetime ( [ eddington ] ) . like all other tensorial quantities , @xmath210 is invariant under shifts in the eddington frame .
we assume that near the ch @xmath210 decouples to ingoing and outgoing components ( like all linear fields considered above ) ; and we shall be concerned here with the field s outgoing component , which we denote @xmath211 .
pick two @xmath46 values @xmath166 , with the only requirement that @xmath212 , where @xmath213 .
we shall now examine how the second observer will experience this variation in @xmath214 from @xmath215 to @xmath216 , as a function of its own proper time .
to this end we re - formulate the problem in terms of @xmath217 rather than @xmath46 .
the change from @xmath215 to @xmath216 thus occurs while the second observer moves from @xmath218 to @xmath219 , where @xmath220+v_{eh}^{(2 ) } .\ ] ] let us now fix @xmath221 , and yet consider the late - infall limit for the second observer : @xmath222 . evidently , in this limit both @xmath218 and @xmath219
are pushed toward @xmath223 .
the corresponding ( second - observer ) proper times @xmath224 will thus be pushed to the same ( finite ) limiting value @xmath225 , that is , the moment of inner - horizon crossing . in particular , the proper - time difference @xmath226 will vanish in this late - infall limit .
we conclude that at the late - infall limit , the finite variation @xmath227 in the perturbation field @xmath210 ( which takes place between a certain pair of @xmath46 values @xmath166 ) occurs within a vanishing proper - time interval at a moment which ( at the limit ) coincides with that of inner - horizon crossing .
thus we recover the effective shockwave phenomenon for late - infall observers .
in this section we shift our focus from linear perturbations on a rn ( or kerr ) background , to nonlinearly - perturbed bhs .
the main new ingredient is that now the infall orbit is disturbed by the mp , which in turn may influence the observer s experience of the perturbation . for simplicity
, we shall restrict attention here to a spherically - symmetric model of a nonlinearly - perturbed charged bh .
we shall first present the model and describe the perturbed bh geometry , and then analyze the experience of late - infall observers in such a spacetime demonstrating that the effective shock - wave phenomenon occurs in nonlinearly - perturbed bhs as well .
let us consider a spherical charged bh perturbed by a spherically - symmetric self - gravitating scalar field .
this model was investigated by several authors , primarily numerically @xcite but also analytically @xcite ( assisted by insights gained from earlier analytical investigations of the mass - inflation model @xcite ) .
the model consists of a massless , minimally - coupled , scalar field @xmath68 , satisfying the covariant wave equation @xmath136 on the ( self consistently - perturbed ) metric of a spherically - symmetric charged bh .
the scalar - field energy - momentum tensor @xmath228 acts as a source term in the einstein equations ( in addition to the electromagnetic contribution to energy - momentum ) , yielding a system of nonlinear field equations for the metric functions [ e.g. @xmath229 and @xmath230 in eq .
( [ metric ] ) below ] . as for initial conditions
, we consider here initial configurations wherein @xmath68 is initially compactly supported outside the bh , as in @xcite .
( alternatively , one may prescribe the initial data for @xmath68 , including its presumed inverse - power tails , directly on the eh , as done in @xcite . ) evolving the initial data one then finds not surprisingly that at late times perturbations die out , and the bh settles down asymptotically to a member of the rn family , with charge @xmath231 and a certain final mass @xmath56 .
the scalar perturbations decay as inverse - power tails .
in particular , along the eh , @xmath232 ( typically with @xmath233 ) at late times @xcite .
these radiation tails fall into the bh and perturb its internal geometry .
the perturbed metric in the bh interior is conveniently expressed in double - null coordinates . in particular , in eddington - like coordinates may naturally be defined in the perturbed spacetime by using characteristic initial - value formulation , and setting @xmath234 and @xmath235 along the outgoing initial ray ( beyond the end of the compact initial support of @xmath68 ) to be the same functions as in the unperturbed rn spacetime with the eddington metric ( [ eddington ] ) ( setting @xmath236 therein ) .
the key property of @xmath20 is that it diverges at future null infinity , with @xmath237 on approaching the latter .
an analogous procedure may in principle be applied to define @xmath46 , yielding an ingoing null coordinate which diverges at the eh .
] we write the line element in the form @xmath238 the infalling scalar - field tail triggers the formation of a curvature singularity at the ch .
this is a direct consequence of the infinite blue - shift that takes place at the inner horizon @xcite , which leads to ( almost ) exponential divergence of the gradient of @xmath68and of curvature .
it turns out , however , that this is actually a _ weak _ @xcite curvature singularity , located exactly at the ch ( @xmath239 ) .
the metric tensor ( in appropriate coordinates ) approaches a finite , non - singular , limit as @xmath239 .
for example , we may use the kruskal - like coordinates @xmath240 with the line element @xmath241 ( with @xmath242 ; note that @xmath62 and @xmath63 are both future - directed , and correspondingly @xmath243 ) .
the ch is located at @xmath244 .
both @xmath0 and @xmath245 have _ finite , nonvanishing _ values at the ch .
yet , @xmath246 diverges at @xmath244 , implying the presence of a null curvature singularity there ( though a weak one ) .
the scalar field @xmath68 behaves in a manner similar to @xmath0 : it is finite at @xmath247 , yet @xmath248 diverges at that limit .
perturbation theory predicts @xcite ( and numerical simulations @xcite confirm ) that in the early portion ( i.e. @xmath167 ) of the ch , the metric functions deviate only slightly from their respective values in the unperturbed rn solution .
the domain @xmath152 is amendable to perturbative treatment .
correspondingly we express @xmath0 and @xmath245 as @xmath249 where the suffix rn " denotes the corresponding function in the unperturbed rn spacetime .
the perturbations @xmath250 vanish in the limit @xmath251 .
( this limit corresponds to the past boundary of the ch , but also to @xmath252 along spatial lines of constant @xmath253 . ) in the domain @xmath152 , the scalar field is dominated by its linear perturbation term , and the mp @xmath254 by the second - order perturbation , as described below ( see appendix ) . as was mentioned above , @xmath0 ( like @xmath245 and @xmath68 ) is finite along the ch .
it initially starts at @xmath255 with @xmath256 , but subsequently shrinks steadily with time ( @xmath257 ) , due to the focussing induced by the outflux of scalar - field energy - momentum across the ch . at some stage @xmath0 shrinks to zero at which point the null weak
ch singularity terminates , and connects to a strong , spacelike , @xmath3 singularity @xcite .
we turn now to investigate the experience of late - infall observers in this spacetime of a nonlinearly - perturbed spherical charged bh .
we consider ( equatorial ) infalling geodesics which are not necessarily radial .
the angular - momentum parameter @xmath8 is conserved in these geodesics , though @xmath6 is no longer conserved . for concreteness we shall focus here on the behavior of the metric function @xmath0 , which the infalling observer probes as a function of his proper time @xmath76 .
physically , a rapid change in @xmath0 will imply ( for a finite - size observer ) a rapid deformation in the tangential directions @xmath258 .
again , we choose two @xmath114 hypersurfaces , denoted @xmath75 ( with @xmath259 ) , requiring that both hypersurfaces intersect the ch ( _ before _
@xmath0 shrinks to zero ) . while the observer progresses from @xmath260 to @xmath261 , @xmath0 changes by the amount @xmath262 .
hereafter a quantity with a sub - index 1 " or 2 " will denote the value of this quantity as the worldline crosses the hypersurface @xmath72 or @xmath104 , respectively .
since along the ch @xmath0 is steadily shrinking , one finds that @xmath263 .
let us now evaluate the proper - time interval @xmath264 , using @xmath265^{1/2 } du \ , .\ ] ] note that @xmath245 ( like @xmath0 ) is bounded throughout the domain @xmath92 , so we can easily bound @xmath266 by @xmath267 where @xmath268 denotes the maximal value of @xmath269 in the worldline s section between @xmath72 and @xmath104 .
the integral on the rhs is nothing but the proper - time length of a timelike curve connecting the two points @xmath270 and @xmath271 , in a fiducial two - dimensional spacetime with the flat metric @xmath272 .
it is bounded above by the ( timelike ) geodesic connecting these edge points , whose length is @xmath273 , where @xmath274 and @xmath275 . clearly , @xmath276 , therefore @xmath277 consider now the late - infall limit , which is the limit of large @xmath19 . in this limit @xmath278 . therefore @xmath268 approaches @xmath279 , which is the maximal value of @xmath269 along the section @xmath92 of the ch .
we obtain our bound on @xmath266 ( for late - infall observers ) in its final form : @xmath280 where @xmath281 is a parameter which depends on @xmath72 and @xmath104 but not on the orbit s infall time .
we conclude that late - infall observers will measure a non - vanishing variation @xmath282 in the metric function @xmath0 , within a short proper - time difference @xmath266 which shrinks exponentially in the infall time @xmath19which is again an effective shock - wave phenomenon . our analysis
so far did not make use of the perturbative nature of the metric field ( at the early portion of the ch ) .
we merely assumed that the ch singularity is null and weak and more specifically , that @xmath245 admits a finite limiting value along the ch .
correspondingly , there was no need to restrict @xmath72 and @xmath104 to the perturbative domain ( @xmath167 ) : we only required that the surfaces @xmath83 and @xmath283 intersect the ch , rather than the spacelike singularity .
however , there still was one hidden assumption : we implicitly assumed that the late - infall observers will make it all the way from @xmath72 to @xmath104 _ without intersecting the ch _ ( that is , with finite @xmath20 ) .
once an observer intersects the ch , we can not make any concrete statement about his subsequent experience , because the ch is by definition the boundary of the domain of unique prediction ( for e.g. the metric functions ) .
we therefore still need to complete this missing piece of the analysis .
we shall show that as long as @xmath104 is located in the weakly - perturbed domain of the ch ( that is , @xmath104 is large compared to @xmath56 ) , the late - infall orbits indeed arrive at @xmath73 with finite @xmath20 .
the control on the growth of @xmath20 will be achieved by monitoring the evolution of the geodesic s energy " parameter @xmath284 note that @xmath6 ( unlike @xmath8 ) is no longer conserved , because the perturbations destroy the exact @xmath13-translation invariance of the rn background . yet , following the evolution of @xmath6 will enable us to control @xmath285 , and thereby the evolution of @xmath20 along the orbit .
the lower - index covariant geodesic equation , applied to the eddington - like metric ( [ metric ] ) , reads @xmath286 therefore @xmath287 to get rid of the term @xmath288 in the rhs , we use the normalization condition @xmath289 , which for the metric ( [ metric ] ) yields @xmath290 the quantity @xmath291 then becomes @xmath292.\ ] ] noting that the last term in squared brackets is equal to @xmath293 , we re - express @xmath294 in terms of the kruskal - like metric function @xmath245 : @xmath295 -\frac{l^{2}}{r^{3 } } ( r_{,u}+r_{,v } ) .
\label{eq : evolve}\ ] ] note that in the background rn metric , both @xmath245 and @xmath0 are functions of @xmath253 , hence @xmath6 is conserved .
evolution of @xmath6 will thus only result from the mp @xmath296 and @xmath297 .
so far all equations were exact . to proceed beyond this point
, we restrict attention to the early portion of the ch , where perturbations are presumably small , and carry out the analysis at the leading order in the mp . obtained from the metric functions etc . ) to first order in the mp .
however , we do _ not _ linearize the worldline - related quantities , like @xmath6 , @xmath294 , @xmath285 , @xmath298 , etc . ] also , since we are dealing with the orbit s evolution very close to the ch , we may replace the background s functions @xmath299 and @xmath300 by their corresponding inner - horizon values @xmath31 and @xmath301 .
we obtain @xmath302 where @xmath303 and @xmath304 are constants , @xmath305 recall that @xmath306 ( like @xmath31 ) is a finite , non - vanishing constant .
we re - write eq .
( [ eq : edot ] ) as @xmath307 in which we view @xmath298 as a ( yet unknown ) function of the parameter @xmath46 along the geodesic
. one might choose to approximate this function @xmath308 by the corresponding function for geodesics in the unperturbed rn geometry .
we shall not proceed here in this way , because we do not want to assume a - priori that the accumulating perturbations in the orbital parameters must be small .
instead , we shall proceed by expressing @xmath309 in terms of @xmath6 . to this end
we use the contravariant version of eq .
( [ edefined ] ) , namely @xmath310 this , together with eq .
( [ eq : normalization ] ) , constitutes a closed algebraic system for the two unknowns @xmath311 .
one can of course write down the exact solution of this algebraic system . however , it will be more instructive to employ here the approximate solution , associated with the smallness of @xmath312 : we are dealing here with the near-@xmath32 region , where ( valid in the perturbative domain considered here ) also implies that @xmath313 ( the perturbation in @xmath312 ) satisfies @xmath314 ; that is , @xmath315 . ]
@xmath316 . the algebraic system
thus yields the simple approximate solution ( to leading order in the small parameter @xmath312 ) and @xmath317 are interchanged .
however , since our late - infall observers enter the @xmath318 region with large @xmath309 , it is the solution ( [ eq : uupvalues ] ) which actually takes place . ]
@xmath319 substituting this expression for @xmath298 in eq .
( [ eq : dedu ] ) we find @xmath320 f . \label{eq : de2}\ ] ] setting @xmath321 ( where , recall , @xmath322 ) , we obtain @xmath323 e^{\kappa ( u - v ) } , \label{eq : de2f}\ ] ] where we have absorbed the above @xmath324 ( in @xmath52 ) in @xmath325 and @xmath326 .
note that the rhs in this equation is a prescribed function of @xmath46 and @xmath20 ( with no reference whatsoever to four - velocity ) . at this point
it will be useful to refer to the concrete form of the mp @xmath296 and @xmath297 .
we focus here on the early portion of the ch ( i.e. @xmath167 ) , where the mp are small and decay as inverse powers of @xmath46 and/or @xmath20 .
we denote by @xmath327 the modification in @xmath6 acquired in the near - ch region , up to some @xmath328 , due to the presence of mp .
based on the inverse - power form of the mp , in the appendix we derive the bound @xmath329 where @xmath330 is a certain parameter ( independent of @xmath19 ) . in particular , we find that for late - infall orbits @xmath331 .
we proceed now to analyze the evolution of @xmath20 , showing that it remains finite throughout @xmath332 ( for any @xmath333 . from eq .
( [ eq : uupvalues ] ) we have @xmath334 now for late - infall orbits we already found that @xmath331 and hence we may regard @xmath6 as constant ( essentially the entrance value of @xmath6 ) . also we may set @xmath91 , and take the near - ch form of @xmath312 , namely @xmath335 .
we obtain @xmath336 where @xmath337 is some positive constant .
re - writing this as @xmath338 , we obtain @xmath339 [ notice that the last two equations are the same as those describing late - infall geodesics in the _ unperturbed _ rn geometry . ]
this expression is bounded above ( by the @xmath324 in the rhs ) .
thus , @xmath20 remains finite throughout the range @xmath167meaning that the orbit can not cross the ch ( located at @xmath50 ) in that domain .
we found that the late - infall observers can not cross the ch in the regime @xmath167 .
the analysis in section [ nonlinear_orbits ] then shows that the proper - time for these observers to move from @xmath83 to @xmath73 , for any @xmath340 , decreases exponentially in the infall time @xmath19 .
since @xmath0 varies during this range by a finite amount @xmath341 , we inevitably face here the phenomenon of effective _ gravitational shock - wave _ : a discontinuity in the metric tensor , which propagates along a null hypersurface ( the outgoing section of the inner horizon in our case ) .
physically , this means that any extended object will undergo a sudden deformation , by a certain amount , within an effectively - vanishing proper time .
the amplitude of such a gravitational shock - wave may naturally be characterized by the ( dimensionless ) magnitude of the object s deformation . specifically in the spherical model studied here ,
the shock s amplitude may be taken to be the dimensionless quantity @xmath342 .
two additional remarks are in order here : 1 .
since the discussion in section [ nocrossing ] assumed @xmath328 in order to treat perturbations perturbatively , one might mistakenly conclude that the amplitude of the gravitational shock wave can be weak ( i.e. @xmath343 ) .
but this is not the case .
it is clear from the above analysis that late - infall geodesics can not fall across the ch until after the perturbations grow to be of order 1 .
we therefore see from section [ nonlinear_orbits ] that late freely - falling observers must face a strong gravitational shock wave whose amplitude is at least of order 1
. 2 . in order to allow a simple discussion of the quantity @xmath344 , which is conserved along geodesics in exact rn ,
we have so far assumed freely - falling worldlines . but let us now consider an accelerated late infall - time observer .
we choose the acceleration as a function of proper time to be bounded and such that , in the unperturbed rn geometry , the worldline would reach the outgoing inner horizon before crossing the ingoing inner horizon .
note that in the limit of large @xmath19 all of these accelerations occur before reaching @xmath83 . as a result ,
in the unperturbed spacetime @xmath6 becomes some constant @xmath345 for all @xmath346 .
corresponding late infall - time observers in the perturbed spacetime may thus be analyzed just as for the freely - falling observers discussed above but with the entrance value of @xmath6 replaced by @xmath345 .
we conclude that any accelerated late infall - time observer who would have reached the outgoing inner horizon in unperturbed rn " also experiences a shockwave in the perturbed spacetime .
we have seen above that , in the limit of late infall times , observers who enter a perturbed reissner - nordstrm black hole experience an effectively unperturbed reissner - nordstrm geometry up to the point where they would expect to encounter an inner horizon at @xmath5 . at this point
, observers who would have reached the outgoing inner horizon in exact rn then encounter a shockwave across which the metric changes discontinuously .
describing the detailed nature of this discontinuity requires further investigation .
while we will not attempt a precise treatment here , it is natural to expect that the above shockwave in fact contains a curvature singularity , as we know that our observers will eventually reach such a singularity and we expect that , since they are already `` nearly null '' in the region described in section [ non - linear ] , all proper times along their worldlines will be compressed to zero in the large @xmath19 limit note that this argument is equally valid if , for some kinds of matter , the spacetimes contain no spacelike singularity . in such cases
the ingoing worldline will reach the weak null singularity , which is also a curvature singularity . ] .
in addition , it is clear that some accelerated observers will reach the ingoing weak null singularity shown in figure [ summary ] ( left ) . for observers who enter the black hole very late , in the region where perturbations are very small
, this may be accomplished using roughly the same set of accelerations ( as defined in their own reference frame ) as would be required to reach the ingoing part of the inner horizon in an unperturbed rn black hole .
furthermore , in the limit of late infall times , such observers arrive at the portion of the weak null singularity close to @xmath2 in figure [ summary ] where the singularity is extremely weak , so that again such observers will measure no noticeable departure from unperturbed rn until @xmath0 is exceedingly close to @xmath31 . to compliment the precise calculations of sections [ linear ] and [ non - linear ] , we find it useful to give a brief heuristic argument that reinforces the above statements . to do so , note that taking our observer to enter the black hole at late times is equivalent to considering some fixed infall time and replacing the given spacetime with one in which the black hole formed ( along with its full set of perturbations ) at a much earlier time .
as a particular example , we might consider a scenario which starts with empty minkowski space in the interior and where both the matter that forms the black hole and our observer are dropped in from some large value of @xmath0 at finite advanced times @xmath347 and @xmath348 respectively .
for example , the observer might be ejected from a static space station .
the matter could be dropped from a dense network of such space stations , or simply assumed to cross the @xmath0-coordiante associated with the observer s space station near the advanced time @xmath347 .
the observer is also given some particular instructions for executing accelerations as a function of proper time along his worldline and we take the matter to be released as some particular function of advanced time peaked around some @xmath347 . allowing @xmath347 to vary leads to a one - parameter family of ( isometric ) spacetimes .
we wish to study the limit @xmath349 holding @xmath348 fixed .
it is useful to assume that the perturbed spacetime preserves spherical symmetry .
the advanced time @xmath20 is thus a well - defined null coordinate everywhere in the globally hyperbolic part of the spacetime ( shown at left in figure [ summary ] ) and the retarded time @xmath46 is a similarly well - defined null coordinate in the region outside the event horizon .
furthermore , it is clear that two spacetimes differing by a shift of @xmath347 can be described by essentially the same metric functions , albeit again with a corresponding shift of @xmath350 .
for example , introducing @xmath351 and @xmath352 , we may take the collapsing spacetimes to be described by the metric @xmath353 for some fixed functions @xmath354 , @xmath355 which do not explicitly depend on @xmath347 .
similarly , the center of spherical symmetry is described by some fixed curve @xmath356 and the event horizon is the curve @xmath357 .
the fact that for any fixed @xmath347 the spacetime approaches rn as @xmath358 implies that in the formal @xmath349 limit with @xmath350 fixed we have @xmath359 and @xmath360 , where @xmath28 and @xmath229 are the functions defined in section [ preliminaries ] for exact rn .
we would like to find a similar construction for the region to the future of the event horizon . in that region the retarded time @xmath46
can not be defined simply by tracing outgoing radial null rays to @xmath33 . rather than attempt to find a physically preferred definition of @xmath46 ,
let us therefore consider any null coordinate @xmath46 which is smooth inside the event horizon , has past - directed gradient @xmath361 , and for which the metric again takes the form ( [ uvform ] ) on each member of our family of spacetimes where the functions @xmath362 again approach @xmath363 and @xmath229 as defined in section [ preliminaries ] in the limit of large @xmath350 with fixed @xmath347 .
note that inside the event horizon @xmath217 will decrease to some @xmath347-independent minimum value @xmath364 at the point where the center of symmetry hits the singularity .
we choose this @xmath364 to be finite .
* left : * some lines of constant @xmath350 drawn on the spherical spacetime describing a collapsing shell perturbed by a massless scalar field . the long - dashed line ( top right boundary )
is the mass - inflation singularity .
in contrast , the short dashes mark coordinate lines lying outside the physical spacetime ( since @xmath365 ) .
the vector field @xmath366 is tangent to any surface @xmath367 or @xmath368 .
the arrows indicate the direction of this vector field along such surfaces . the point marked @xmath2
is fixed under the action of this vector field .
* right : * the effective geometry for late - time observers obtained by flowing the interior of our spacetime _
backward _ along the vector field @xmath366 , while similarly deforming the boundaries . as for figure 1 ( right )
, it consists of the region of the unperturbed eternal reissner - nordstrm black hole with @xmath4 , together with certain shockwave singularities .
the shock along the lower left boundary descirbes a jump from @xmath180 to @xmath3 ( corresponding to the regular center of spherical symmetry shown in the left diagram ) . though it arises only in the formal late - infall limit , this shock is not actually accessible to any late - infall observer . in this sense
the shock is fictitious . in contrast , the shock along the upper left boundary is accessible to late - time observers .
it contains the final piece of the shell worldline and describes a jump from @xmath5 to @xmath3 ( corresponding to the spacelike singularity shown in the left diagram ) .
, title="fig : " ] * left : * some lines of constant @xmath350 drawn on the spherical spacetime describing a collapsing shell perturbed by a massless scalar field . the long - dashed line ( top right boundary )
is the mass - inflation singularity .
in contrast , the short dashes mark coordinate lines lying outside the physical spacetime ( since @xmath365 ) .
the vector field @xmath366 is tangent to any surface @xmath367 or @xmath368 .
the arrows indicate the direction of this vector field along such surfaces . the point marked @xmath2
is fixed under the action of this vector field .
* right : * the effective geometry for late - time observers obtained by flowing the interior of our spacetime _
backward _ along the vector field @xmath366 , while similarly deforming the boundaries . as for figure 1 ( right ) , it consists of the region of the unperturbed eternal reissner - nordstrm black hole with @xmath4 , together with certain shockwave singularities .
the shock along the lower left boundary descirbes a jump from @xmath180 to @xmath3 ( corresponding to the regular center of spherical symmetry shown in the left diagram ) . though it arises only in the formal late - infall limit , this shock is not actually accessible to any late - infall observer . in this sense
the shock is fictitious .
in contrast , the shock along the upper left boundary is accessible to late - time observers .
it contains the final piece of the shell worldline and describes a jump from @xmath5 to @xmath3 ( corresponding to the spacelike singularity shown in the left diagram ) .
, title="fig : " ] we expect that late - time observers experience an effective spacetime described by the formal @xmath349 limit ( with @xmath350 held fixed ) of ( [ uvform ] ) for _ some _ definition of the internal @xmath46 coordinate above will be determined by analytic extension of the action on the exterior metric , and this interior action can be used to define an interior null coordinate @xmath46 that simply shifts with @xmath347 .
it is clear that the resulting @xmath46 will have the properties required above near @xmath2 . ] .
moreover , the essential features of this limit are independent of the particular definition of @xmath46 .
indeed , by construction this limit is described by @xmath369 and @xmath370 on the domain @xmath371 .
all of the additional structure has been sent off to @xmath372 , which of course nevertheless lies at finite affine parameter along future - directed geodesics .
thus , as shown in figure [ pertt ] , this limiting spacetime is just the region of unperturbed rn to the future of its past horizon and to the past of its outgoing inner horizon .
however , rather than being smooth , the outgoing null surface @xmath372 is a shockwave into which all of the time - dependent structure of the spacetime has been compressed .
we should also recall that , although it is no longer apparent from the limiting form of the metric , the entire ingoing surface @xmath373 to the future of @xmath2 is a curvature singularity ( i.e. the mass - inflation singularity ) for any finite value of @xmath347 .
our work above examined the experiences of observers who enter classical spherically - symmetric charged asymptotically flat black holes subject to both linear and non - linear perturbations .
we also considered linear perturbations of kerr black holes .
our emphasis was on the limit of late infall times , @xmath22 . in this limit
, our observers observations agree precisely with those of similar observers in the unperturbed ( stationary ) black hole spacetime up to the point where the latter observers would reach the inner horizon . at that point , however , those late infall - time observers who would have reached the ( smooth ) outgoing inner horizon in unperturbed rn instead encounter an effective gravitational shock - wave in the perturbed spacetime .
the shock s width ( expressed in terms of the proper time of infalling observers ) decreases with infall time as @xmath135 , hence it effectively vanishes for late - infall observers .
this may be thought of fundamentally a time - dilation effect .
recall that the killing field @xmath374 of a stationary black hole acts like a minkowski - space boost in the region near where the ingoing inner horizon meets the outgoing inner horizon .
this is precisely the region in which late - infall observers encounter any structure present in the black hole interior . as a result ,
observers who enter at later and later times arrive here in more and more highly boosted reference frames .
thus they approach at nearly the speed of light and transit through in vanishing proper time , experiencing any structure as a shock - wave . in such a gravitational shock - wave ,
the metric undergoes a discontinuity and changes by a finite amount in vanishing time .
an extended physical object hitting this shockwave will thus undergo a sudden deformation ( i.e. shear and/or contraction ) , by a finite amount , within an effectively - vanishing proper time .
owing to their short time scale , the gravitational tidal forces entailed in the shock will presumably dominate over all internal interactions .
thus , each single nucleon ( say ) will be deformed by a certain amount ( typically of order 1 ) while traversing such a shock - wave .
we also gave a heuristic argument that the experiences of late - time observers are described by the simple effective spacetime shown at right in figure [ pertt ] , which in particular reproduces the above exact results .
on the basis of this argument one expects the outgoing shockwave to contain a strong curvature singularity . ] , with the area - radius @xmath0 shrinking to zero across the shockwave for the case described in figure [ pertt ] .
however , these final details remain to be verified by other techniques .
while the explicit non - linear analysis in section [ non - linear ] was restricted to the spherically - symmetric case , we strongly expect that the same phenomenon of shockwave formation will take place in nonspherical black holes as well , and particularly in perturbed spinning black holes .
this expectation is based on the combination of several pieces of evidence : the experience of late - infall observers inside a linearly - perturbed kerr bh ( discussed in section [ linear ] ) , the sub - dominance of nonlinear perturbations in generically - perturbed spinning black holes ( particularly near the early portion of the ch ) , and more generally on the profound similarity between the inner structures of charged and spinning black holes
. it may be interesting to compare the properties of the two different types of null singularity that develop at the inner horizon : the shock - wave singularity at the latter s outgoing section , and the weak curvature singularity at the ch .
we argue that from the physical point of view the shock - wave singularity is the more violent one particularly for very late - infalling observers . to see this , consider two representative orbits in the rn geometry : orbit ( a ) hits the ingoing section of @xmath5 , whereas orbit ( b ) hits the latter s outgoing section .
let us now consider the experiences of these two observers on their approach to @xmath5 , in case the rn spacetime is perturbed focussing our attention on the late - infall orbits belonging to the two families ( a , b ) .
observers of type ( a ) will hit a true curvature singularity at the ch .
this singularity is weak , however , and any observer will only experience a finite integrated deformation up to the ch .
furthermore , this deformation will decrease with increasing @xmath19 ( as an inverse power thereof ) , and will vanish in the late - infall limit . on the other hand , observers of type ( b ) will experience , on approaching @xmath5 , a certain amount of deformation which does _ not _ decrease with increasing @xmath19 .
this is because the shock s amplitude approaches a certain limiting value at large @xmath19 ( it is only the shock s width which evolves at late time , it decreases exponentially in @xmath19 ) .
thus , at least as far as the overall tidal deformation is considered , at the late - infall limit observers of type ( a ) will effectively feel no singularity at the ch , whereas late - infall observers of type ( b ) will feel a violent gravitational shock - wave singularity , whose magnitude is typically of order unity ( at least ) . while our considerations were completely classical , our conclusions also suggest properties of quantum black hole microstates . in particular , the second law of thermodynamics implies that the late - time limit of any black hole spacetime should approximate the typical quantum state of the relevant ensemble .
it is thus natural to conjecture that the typical state in an ensemble of charged or rotating black holes corresponds to the geometry displayed in figure [ pertt ] ( right ) together with quantum corrections , as opposed to say that shown in [ pertt ] ( left ) .
some of our results carry over directly to generic ( stable ) black holes in any dimension .
indeed , in retrospect our linear analysis depended only on the general form of a ( future - directed ) time - translation @xmath374 between the inner and outer horizons of the unperturbed black hole , and in particular on the pattern of surfaces invariant under its action .
the essential ingredients are just that : i ) the ( smooth ) inner and outer horizons are invariant under @xmath374 , ii ) the point @xmath375 in figure [ rnfig ] ( left ) is an attractive fixed point of @xmath374 while @xmath376 is a repulsive fixed point , and iii ) perturbations in the immediate vicinity of the outgoing inner - horizon decouple into independent @xmath46 and @xmath20 components and taking the limit @xmath377 to zoom in on the region near @xmath244 .
corresponding results may also hold for higher spin fields . ] .
whenever these features arise , late infall - time observers who reach the outgoing inner horizon will experience any given linear perturbation as a shockwave .
note that , despite our use of eddington coordinates @xmath350 in the main text , this more geometric form of the argument is manifestly coordinate invariant .
at the non - linear level , the assumption that the perturbed spacetime approaches a stationary geometry at @xmath375 which is an attractive fixed point of @xmath374 again implies that late infall - time observers experience an essentially unperturbed solution up to the point where they would expect to reach an inner horizon .
one is tempted to draw a conformal diagram similar to that in figure [ pertt ] ( right ) and again conjecture that infalling observers experience an effective outgoing shockwave . however , investigating this conjecture in detail will require more sophisticated methods or , perhaps , numerical simulations . throughout this work
we have considered what one may call ` test observers , ' which experience perturbations of the spacetime but do not source further perturbations .
in contrast , any physical observer who falls into the black hole may be expected to create additional perturbations which now typically fall into the black hole at an advanced time comparable to @xmath19 ( and thus which can not generally be ignored simply by taking the limit of large @xmath19 ) .
it would be interesting to understand what this implies for the experiences of such physical observers .
for example , consider an observer who reaches the would - be ingoing inner horizon ( the ch ) . due to the decay of other perturbations
as reviewed in section [ non - linear_metric ] , for sufficiently large @xmath19 the perturbations at the relevant part of the ch will generically be dominated by those sourced by our observer himself .
one would expect him to encounter a null weak singularity whose strength ( typically tiny , of order the mass ratio between the bh and observer ) is determined by perturbations of his own creation .
in contrast , even if our observer instead reaches the would - be outgoing inner horizon , we see no reason for perturbations sourced by the observer to destroy the outgoing shock wave set up by earlier perturbations or to significantly change the experience described above . in our work
above , we assumed that the spacetime approaches some non - extreme stationary black hole geometry near @xmath2 . to simplify the analysis
, we assumed `` generic '' parameter values for this limiting bh , and in particular that it is not too close to extremality . while extending these results to nearly - extreme black holes
is not difficult ( see in particular footnote [ extend ] in appendix a ) , it remains to understand the precisely extreme case .
taking the formal extreme limit of our late - time results ( i.e. , first taking @xmath22 with @xmath378 and then taking @xmath379 ) agrees with the picture suggested in @xcite ( and supported by @xcite ) in which late - time observers experience a singularity which effectively resides at the final event horizon .
this suggests that such a picture should hold at least for horizons that approach extremality sufficiently slowly relative to the production of perturbations ( and relative to the time scale set by the surface gravity ) .
this in particular should apply for large black holes when the approach to extremality is due to quantum processes in cases where the extreme bh is quantum mechanically stable ( e.g. , with enough supersymmetry ) .
however , a complete analysis is again left for future work . to simplify the discussion
, our analysis has focussed on very late time observers who enter asymptotically flat black holes . but
let us now comment on the implications for observers who enter astrophysical black holes at finite times .
so long as the current accretion rate and any effects from the expansion of the universe are small ( e.g , at the 1% level ) , and so long as a few light - crossing times have passed since the black hole experienced a large perturbation ( including the initial formation of the black hole ) , a finite infall - time observer who reaches the would - be outgoing inner horizon should also experience our effective shock wave .
this can be seen from the following points : ( i ) by causality , nothing that enters the black hole more than a few light - crossing times to the future of our observer can affect his experience .
thus the experience of our observer can be described by modeling the astrophysical black hole as an asymptotically flat black hole with a small accretion rate .
( ii ) although in various parts of this paper we have focused on the internal structure of the bh at very late time ( after the decay of radiative tails ) , one should bear in mind that the process of shock formation is in fact rather quick : owing to the exponential decrease of the shock width , a @xmath19 value of only a few light - crossing times is required for the very narrow shock configuration to build up . during such a @xmath19 interval of ,
say , 10 - 20 times @xmath56 , a weak accretion will not have a chance to cause a significant modification to the process of shock formation . for observers who fall into such an accreting bh with a much larger @xmath19
, the accumulating effects of continuing accretion may possibly be more significant ( see e.g. @xcite ) .
nevertheless , in the limit of weak accretion , any effects of this accretion can cause only small changes from the scenario described above . while this might give our effective shock a finite width ( independent of @xmath19 ) or perhaps make the effective shock timelike or spacelike instead of null , in the weak accretion limit observers with fixed resolution would still describe their experience as an encounter with an effective shock .
thus our scenario provides a good first approximation to the experiences of observers who enter weakly - accreting astrophysical black holes at finite times . while it would be interesting to compute these corrections quantitatively ( e.g. by comparing with @xcite ) , the fact that in many astrophysical situations the mass accretion rate is far , far less than @xmath56 per light crossing time seconds .
even if such a black hole accreted a solar mass per year , we would have @xmath380 . ]
suggests that this approximation is quite good indeed .
dm thanks patrick brady , steve giddings , gary horowitz , harvey reall , and eric poisson for numerous discussions of black hole interiors over many years .
he in particular thanks steve giddings for emphasizing the importance of the late time limit .
ao would like to thank the ucsb physics department for kind hospitality when part of this work was carried out .
d.m . was supported in part by the national science foundation under grant no phy08 - 55415 , and by funds from the university of california .
was supported in part by the israel science foundation ( grant no .
1346/07 ) .
in this appendix we derive a bound on @xmath327 , based on eq .
( [ eq : de2f ] ) and on the inverse - power form of the mp ( at the early portion of the ch ) . for a _ generic _ situation of a perturbed charged ( or spinning ) bh
, the mp are dominated by the linear term , which is usually a superposition of a term @xmath381 and a term @xmath382 , typically with @xmath146 or possibly @xmath147 , where @xmath142 is the multipolar number of the mp mode under consideration .
however in our case the situation is different because we are only considering here _ spherically - symmetric _ perturbations , and there are no matter - free spherical mp . therefore it is only the scalar field which is excited at the linear level .
the mp are in turn sourced by the scalar - field energy - momentum tensor ( [ tscalar ] ) , which is quadratic in ( derivatives of ) @xmath68 ; hence the mp will first appear as ( and will be dominated by ) _
second - order _ perturbations .
thus , whereas in our case @xmath68 is dominated by terms @xmath381 and @xmath382 ( with @xmath383 ) , the mp will be dominated by inverse - power terms of overall power @xmath384 rather than @xmath385 . restated in other words ,
the mp will be dominated by a superposition of terms of the form @xmath386 , with non - negative integers @xmath387 satisfying @xmath388 , where @xmath389 are certain constants
. decays as @xmath69 along the eh , then perturbation analysis reveals @xcite that in the neighborhood of the early portion of the ch @xmath296 is dominated by a term @xmath390 plus another term @xmath391 , and @xmath297 by a term @xmath392 ( plus , possibly , additional terms @xmath393 and @xmath394 , though these two would be more sensitive to the choice of gauge ) .
note that as a consequence many of the coefficients @xmath395 ( to be defined shortly ) actually vanish , though some of them ( for example @xmath396 ) do not . ]
correspondingly , the term in squared brackets in the rhs of eq .
( [ eq : de2f ] ) will be dominated by superposition of terms @xmath397 with non - negative integers @xmath398 satisfying @xmath399 ( with some new constants @xmath395 ) .
let us now restrict our attention to the contribution to @xmath400 coming from a single such term @xmath397 : @xmath401}/du \equiv c'_{j'k ' } v^{-j ' } u^{-k ' } e^{\kappa ( u - v ) } .
\label{dejk}\ ] ] we can bound this contribution by @xmath402}/du | < \left [ |c'_{j'k'}| ( v_{min})^{-j'}e^{-\kappa \ , v_{min } } \right ] u^{-k ' } e^{\kappa u } , \label{dejkbound}\ ] ] where @xmath403 is the minimal value of @xmath20 in the orbit s section under consideration .
( @xmath403 is always @xmath404 ; later we shall take @xmath403 to be the value of @xmath20 when the orbit approaches the neighborhood of the ch . )
the term in the rhs of eq .
( [ dejkbound ] ) is an explicit function of @xmath46 .
consider now its integral between a certain pair of @xmath46 values @xmath405 satisfying @xmath406 .
we can now set @xmath403 to be @xmath407 , namely , the value of @xmath20 when the orbit approaches @xmath408 . also , since @xmath167 throughout the relevant domain , over the basic exponential time scale @xmath409 the factor @xmath410 does not vary in an appreciable manner , therefore it may be pulled out of the integral .
the latter is thus well approximated by just the integrand namely the rhs of eq .
( [ dejkbound]) divided by @xmath411 ( the error being smaller typically by a factor @xmath412 ) .
thus , denoting the @xmath413 contribution to the variation in @xmath414 ( accumulated between @xmath415 and @xmath416 ) by @xmath417}$ ] , we find @xmath418 } \right|
< \left [ \kappa^{-1 } |c'_{j'k'}| ( v_{i})^{-j'}e^{-\kappa v_{i } } \right ] \left [ u^{-k ' } e^{\kappa u } \right]_{u_f } ^{u_i } .\ ] ] clearly this may be further bounded by the contribution at the upper limit ( @xmath408 ) , yielding the bound @xmath418 } \right| < \kappa^{-1 } |c'_{j'k'}| ( v_{i})^{-j ' } ( u_i)^{-k'}e^{\kappa ( u_i - v_i ) } .
\label{deltabound}\ ] ] the situation which concerns us is the motion of late - infall observers between two @xmath114 hypersurfaces , @xmath72 to @xmath104 ( as described in sections [ perturbations],[non - linear ] ) .
our goal here is to show that the late - infall orbits will not cross the ch before approaching @xmath283 ( for any @xmath328 ) .
we therefore need to bound the variation in @xmath419 from the stage where the orbit approaches the neighborhood of the ch , up to @xmath73 .
correspondingly , in the above bounds on @xmath420 we should in principle set @xmath421 ( though @xmath416 actually drops out from our final bound ) , and choose @xmath415 to be the value of @xmath46 at which the orbit approaches the ch neighborhood .
this notion of approaching the ch neighborhood " is most naturally formulated by means of a certain value of @xmath43 , which we denote @xmath422 .
that is , a particular orbit is said to having arrived at the ch neighborhood at the point where @xmath423 , and we set our parameters @xmath415 and @xmath407 to be the @xmath424 values of that point ( this in particular implies @xmath425 ) . from eq .
( [ delta_r ] ) , @xmath422 is ( roughly speaking ) the value of @xmath43 at which @xmath426 becomes @xmath427 . , but it does not affect our final results [ e.g. eqs .
( [ deltaboundf ] ) and ( [ deltaebound ] ) below ] , which are independent of @xmath422 . note also that in the analysis here we do not attempt to include modifications in @xmath6 caused by mp outside the bh , or even inside the bh but not in the ch neighborhood .
these non - ch " contributions do not concern us here , they can be shown by simpler arguments to decrease as an inverse power of infall time @xmath19 . ] substituting @xmath428 in eq .
( [ deltabound ] ) , the exponent at the rhs reads @xmath429 .
also , since we are concerned here with the late - infall limit , in evaluating the factor @xmath430 we can safely ignore @xmath422 compared to @xmath407 ( which is @xmath431 ) , after which the rhs of eq .
( [ deltabound ] ) becomes @xmath432 note , however , that the difference between @xmath407 and @xmath19 is , approximately , the quantity @xmath86 ( defined earlier in sec . [ monotonic ] ) , which is @xmath433 ( and independent of infall time ) , and hence is @xmath434 in the late - infall limit .
therefore we can further replace @xmath435 by @xmath436 .
also , since @xmath437 ( it is in fact @xmath427 ) , we are allowed to entirely drop it from our bound . recalling that @xmath438
, we obtain the bound on @xmath417}$ ] in a much simpler form : @xmath418 } \right| < \kappa^{-1 } \ , |c'_{j'k'}| \
, ( v_{eh})^{-(2n+1 ) } .
\label{deltabound1}\ ] ] it remains to sum over the relevant pairs of integers @xmath439 ( recall , these are certain non - negative integers satisfying @xmath438 ) .
this summation is trivial and we get the bound on @xmath440 in its final form : @xmath441 where @xmath442 ( setting @xmath443 ) .
obviously the constant @xmath444 is independent of @xmath19 .
we have thus established that at the late - infall limit , @xmath445 inevitably becomes @xmath446 .
equation ( [ deltaboundf ] ) thus also implies that @xmath327 ( namely the modification in @xmath6 accumulated in the near - ch region up to some @xmath328 ) is bounded for late - infall orbits by for simplicity , the above derivation assumed typical values of the black hole parameters such that @xmath447 ; i.e. , the black hole is not too close to extremity .
we note in passing that a similar argument yields a bound of the same form for any @xmath448 . ]
@xmath449 where @xmath450 .
a. ori , `` inner structure of a charged black hole : an exact mass - inflation solution , '' phys .
lett . * 67 * , 789 ( 1991 ) .
d. marolf , `` the dangers of extremes , '' gen .
grav . * 42 * , 2337 - 2343 ( 2010 ) .
[ arxiv:1005.2999 [ gr - qc ] ] .
l. m. burko , `` structure of the black hole s cauchy horizon singularity , '' phys .
* 79 * , 4958 ( 1997 ) [ gr - qc/9710112 ] .
a. ori , `` structure of the singularity inside a realistic rotating black hole , '' phys .
* 68 * , 2117 ( 1992 ) .
a. ori , `` evolution of linear gravitational and electromagnetic perturbations inside a kerr black hole , '' phys .
d * 61 * , 024001 ( 2000 ) .
a. ori , `` oscillatory null singularity inside realistic spinning black holes , '' phys
. rev .
lett . * 83 * , 5423 ( 1999 ) [ gr - qc/0103012 ] .
p. r. brady , s. droz and s. m. morsink , `` the late time singularity inside nonspherical black holes , '' phys .
d * 58 * , 084034 ( 1998 ) [ gr - qc/9805008 ] .
a. ori and e. e. flanagan ,
`` how generic are null space - time singularities ?
, '' phys .
d * 53 * , 1754 ( 1996 ) [ gr - qc/9508066 ] ; a. ori , `` null weak singularities in plane symmetric space - times , '' phys .
d * 57 * , 4745 ( 1998 ) [ gr - qc/9801086 ] . for accreting black holes ,
see a. j. s. hamilton and p. p. avelino , `` the physics of the relativistic counter - streaming instability that drives mass inflation inside black holes , '' phys .
rept . * 495 * , 1 ( 2010 ) [ arxiv:0811.1926 [ gr - qc ] ] ; a. j. s. hamilton and g. polhemus , `` the interior structure of rotating black holes 1 .
summary , '' arxiv:1010.1269 [ gr - qc ] ; a. j. s. hamilton , `` the interior structure of rotating black holes 2 .
uncharged black holes , '' arxiv:1010.1271 [ gr - qc ] ; a. j. s. hamilton , `` the interior structure of rotating black holes 3 . charged black holes , '' arxiv:1010.1272 [ gr - qc ] ; a. j. s. hamilton , `` towards a general description of the interior structure of rotating black holes , '' arxiv:1108.3512 [ gr - qc ] . for a closely related scalar - field based model
see also l. m. burko , `` black hole singularities : a new critical phenomenon , '' phys .
lett . * 90 * , 121101 ( 2003 ) [ erratum - ibid . *
90 * , 249902 ( 2003 ) ] [ gr - qc/0209084 ] ; r. h. price , `` nonspherical perturbations of relativistic gravitational collapse .
scalar and gravitational perturbations , '' phys .
d * 5 * , 2419 ( 1972 ) .
j. bicak , gravitational collapse with charge and small asymmetries : i. scalar perturbations , " gen .
gravitation 3 , 331 ( 1972 ) .
y. gursel , v. d. sandberg , i. d. novikov and a. a. starobinsky , `` evolution of scalar perturbations near the cauchy horizon of a charged black hole , '' phys .
d * 19 * , 413 ( 1979 ) .
a. ori , perturbative approach to the inner structure of a rotating black hole , " gen .
29 , 881 ( 1997 ) .
a. ori , `` transmission and reflection coefficients for a scalar field inside a charged black hole , '' phys .
d * 57 * , 2621 ( 1998 ) [ gr - qc/9711034 ] .
s. chandrasekhar , the mathematical theory of black holes ( oxford university press , new york , 1983 ) . v. moncrief , `` odd - parity stability of a reissner - nordstrom black hole , '' phys .
d * 9 * , 2707 ( 1974 ) .
y. gursel , i. d. novikov , v. d. sandberg and a. a. starobinsky , `` final state of the evolution of the interior of a charged black hole , '' phys .
d * 20 * , 1260 ( 1979 ) .
s. a. teukolsky , `` rotating black holes - separable wave equations for gravitational and electromagnetic perturbations , '' phys .
* 29 * , 1114 ( 1972 ) .
a. ori , `` evolution of scalar field perturbations inside a kerr black hole , '' phys .
d * 58 * , 084016 ( 1998 ) .
a. ori ( unpublished ) l. m. burko and a. ori , `` analytic study of the null singularity inside spherical charged black holes , '' phys .
d * 57 * , 7084 ( 1998 ) [ gr - qc/9711032 ] .
r. penrose , structure of space - time , " in 1968 _ battelles rencontres _ , edited by b.s .
dewitt and j.a .
( benjamin , new york , 1968 ) .
|
we investigate the interiors of 3 + 1 dimensional asymptotically flat charged and rotating black holes as described by observers who fall into the black holes at late times , long after any perturbations of the exterior region have decayed . in the strict limit of late infall times , the initial experiences of such observers are precisely described by the region of the limiting stationary geometry to the past of its inner horizon . however , we argue that late infall - time observers encounter a null shockwave at the location of the would - be outgoing inner horizon .
in particular , for spherically symmetric black hole spacetimes we demonstrate that freely - falling observers experience a metric discontinuity across this shock , that is , a _ gravitational shock - wave_. furthermore , the magnitude of this shock is at least of order unity .
a similar phenomenon of metric discontinuity appears to take place at the inner horizon of a generically - perturbed spinning black hole .
we compare the properties of this null shockwave singularity with those of the null weak singularity that forms at the cauchy horizon .
|
when investigating the equilibrium properties of a complex polyatomic system , it is customary to identify a suitable _ reaction coordinate _ @xmath2 that maps atomic positions @xmath3 onto the points of some manifold @xmath4 , and then to study its equilibrium probability density : @xmath5\big>,\ ; \xi\in\mathbb{q}\ ] ] ( angular brackets denote an ensemble average ) .
the density @xmath6 provides information about the relative stability of states corresponding to different values of @xmath7 along with useful insights into the transitional kinetics between various stable states . in practice , the landau free energy@xcite @xmath8 is typically preferred over @xmath6 , because it tends to be more intuitive .
either @xmath6 or @xmath9 is said to provide a coarse - grained description of the system in terms of @xmath7 alone with the rest of the degrees of freedom of the original system integrated out .
quite naturally , the reaction coordinate ( often also referred to as _ collective variable _ or _ order parameter _ ) is typically chosen to represent the slowest degrees of freedom of the original system , although this is not formally required . in the past few years
, several methods targeting the computation of @xmath9 using non - equilibrium dynamics have become popular .
first methods that introduced a time evolving potential to bias the original potential energy were the the ` l`ocal ` e`levation ` m`ethod ( ` lem`)@xcite , by huber , torda and van gunsteren in the ` md ` context and the wang - landau approach in ` mc ` one@xcite .
more recent approaches include the adaptive - force bias method@xcite , and the non - equilibrium metadynamics@xcite method .
these methods all estimate the free energy of the reaction coordinate from an `` evolving '' ensemble of realizations@xcite , and use that estimate to bias the system dynamics , so as to flatten the effective free energy surface .
collectively , they can all be considered as umbrella sampling methods , with an evolving potential . in the long time
limit , the biasing force is expected to compensate for the free energy gradient , so that the biasing potential eventually reproduces the free energy surface . in this work
, we present an ` a`daptively ` b`iased ` m`olecular ` d`ynamics ( ` abmd ` ) method whose implementation is particularly efficient and suited for free energy calculations .
the method has an @xmath0 scaling with molecular dynamics time t and is characterized by only two control parameters .
in addition , the method allows for extensions based on _
multiple walkers _ and _ replica exchange _ for both temperature and/or the collective variables .
the ` abmd ` method has been implemented in the ` amber ` software package@xcite , and is to be distributed freely . before discussing ` abmd ` , it is helpful to review the salient features of the metadynamics ( ` mtd ` ) method .
essentially , the ` mtd ` method is built upon the ` lem ` method by exploiting car - parrinello dynamics : the phase space of the system is extended to include additional dynamical degrees of freedom harmonically coupled to the collective variable .
these additional degrees of freedom are assumed to have masses associated with them , and evolve in time according to newton s laws .
the masses are supposed to be large enough , so that the dynamics of these extra - variables is driven by the free energy gradient .
their trajectory is then used to construct a history - dependent biasing potential by means of placing many small gaussians along the trajectory . when combined with car - parrinello _
ab - initio _ dynamics , ` mtd ` has been successfully used to explore complex reaction pathways involving several energy barriers . while ` mtd ` continues to be used successfully ,
there are several known limitations associated with the initial implementation of the method , which provided the motivation for the development of the ` abmd ` method .
first , in order to calculate reliable free energies with a _ controllable accuracy _ , long runs are needed , especially for the `` corrective '' follow - up at equilibrium@xcite .
this is especially true for biomolecular systems , which typically are characterized by many degrees of freedom and non - negligible entropy contributions to the free energies .
long runs , however , may be precluded by the ` mtd ` method , because of its unfavorable scaling with molecular dynamics ` md ` time @xmath1 . while one can readily speed up the original ` mtd ` method using such tricks as truncated gaussians and kd - trees@xcite , the bottleneck there is the explicit calculation of the history - dependent potential .
since at every ` md ` step , gaussians from all previous time steps need to be added , the number of gaussians grows linearly with @xmath1 .
the numerical cost of ` mtd ` therefore grows as @xmath10 which , in some cases , may prove itself to be prohibitively expensive , especially when long runs are needed .
another undesirable feature is that the ` mtd ` method ( at least in its original implementation ) is characterized by a relatively large number of parameters ( _ e.g. _ , the masses and spring constants associated with the collective variable , the characteristics of the gaussians to be added , multiple control parameters , etc ) , all of which influence the dynamics in an entangled and non - transparent way .
a successful ` mtd ` run often requires a careful balancing of these parameters , which is especially nontrivial for multidimensional collective variables .
more recent implementations of ` mtd ` @xcite have reduced the number of parameters . as will be discussed , the ` abmd ` method is characterized by only _ two _ control parameters and scales as @xmath0 with simulation time .
the ` abmd ` method is formulated in terms of the following set of equations : @xmath11,\ ] ] @xmath12,\ ] ] where the first set represents newton s equations that govern ordinary ` md ` ( temperature and pressure regulation terms are not shown ) augmented with the additional force coming from the time - dependent biasing potential @xmath13 ( with @xmath14 ) , whose time evolution is given by the second equation . in the following , we refer to @xmath15 as _ flooding timescale _ and to @xmath16 as _ kernel _ ( in analogy to the _ kernel density estimator _ widely used in statistics@xcite ) .
the kernel is supposed to be positive definite ( @xmath17 ) and symmetric ( @xmath18 ) .
it can be perceived as a smoothed dirac delta function . for large enough @xmath15 and small enough width of the kernel
, the biasing potential @xmath13 converges towards @xmath19 as @xmath20@xcite .
our numerical implementation of the ` abmd ` method involves the following .
we stick with @xmath21 where @xmath22 is either @xmath23\in\mathbb{r}^1 $ ] or a one - dimensional torus , and use cubic b - splines ( or products of thereof for @xmath24 ) to discretize @xmath13 in @xmath4 : @xmath25 @xmath26 we use the biweight kernel@xcite for @xmath16 : @xmath27 and an euler - like discretization scheme for the time evolution of the biasing potential : @xmath28,\ ] ] where @xmath29 is at time @xmath1 .
note that this time discretization may be readily improved .
this , however , is not really important here , since the goal is not to recover the solution of the ` abmd ` equations _ per se _ , but rather to flatten @xmath30 in the @xmath20 limit .
note also , that the numerical cost of evaluation of the time - dependent potential is constant over time , and so ` abmd ` scales trivially as @xmath0 , which is computationally quite favorable .
the storage requirements of the ` abmd ` are also quite reasonable , especially if sparse arrays are used for @xmath31 . in addition , it is characterized by only _ two _ control parameters : the flooding timescale @xmath15 and the kernel width @xmath32 . `
abmd ` admits two important extensions .
the first is identical in spirit to the _ _ multiple walkers metadynamics__@xcite .
it amounts to carrying out several different ` md ` simulations biased by the same @xmath13 , which evolves via : @xmath33\;,\ ] ] where @xmath34 labels different ` md ` trajectories .
a second extension is to gather several different ` md ` trajectories , each bearing its own biasing potential and , if desired , its own distinct collective variable , into a generalized ensemble for `` replica exchange '' with modified `` exchange '' rules@xcite .
both extensions are advantageous and lead to a more uniform flattening of @xmath30 in @xmath4 .
this enhanced convergence to @xmath9 is due to the improved sampling of the `` evolving '' canonical distribution .
the peptide in a @xmath35-hairpin conformation ( sketch ) . ] we have implemented the ` abmd ` method in the ` amber ` package@xcite , with support for both replica exchange and multiple - walkers . in pure `` parallel tempering '' replica exchange ( same collective variable in all replicas )
, @xmath36 replicas are simulated at different temperatures @xmath37 , @xmath38 .
each replica has its own biasing potential @xmath39 , @xmath38 , that evolves according to its dynamical equation .
exchanges between neighboring replicas are attempted at a prescribed rate , with an exchange probability given by@xcite : @xmath40 @xmath41\\ \nonumber & - & \frac{1}{k_bt_n}\big[u^n(\xi^n ) - u^n(\xi^m)\big],\end{aligned}\ ] ] where @xmath42 denotes the atomic potential energy .
the biasing potentials are temperature - bound and converge in the @xmath20 limit to the free energies at their respective temperatures .
we have also implemented a more _ general _ replica exchange scheme , where different replicas can have different collective variables and/or temperatures , and can experience either an evolving or a static biasing potential ( the latter obviously includes the case of @xmath43 ) .
exchanges between random pairs of replicas are then tried at a prescribed rate .
this method is simply a generalization @xcite of the `` hamiltonian replica exchange '' method described in ref . , and reduces to it when all biasing potentials are static .
the big advantage here is that , by using replicas with different collective variables , it is possible to obtain several _
one- _ or _
_ two-__dimensional projections of the free energy surface for the corresponding variables .
this is very useful because it not only increases the amount of information that can be gathered from a given simulation , but it also allows for previously obtained information for a collective variable to be used to compute the free energy associated with different variables . for instance , suppose that in the course of a simulation it becomes apparent that one wishes to address additional questions involving different collective variables . instead of starting from `` scratch '' , one can re - use the already obtained biasing potentials and thereby greatly accelerate the free energy calculation for the new variables .
it is also worth noting that for replicas running at the same temperature , the exchange probability does not depend on the atomic potential energies ( eqs .
( [ eq : exchange1])-([eq : exchange2 ] ) above ) .
this implies that the number of replicas needed to maintain acceptable exchange rates can be made independent of the solvent degrees of freedom , provided that one is interested in the properties of the solute only ( so that the collective variables do not depend explicitly on the atomic coordinates of the solvent ) and that the structure is adequately solvated .
this can be exploited to sample a solute with a minimum amount of solvent , and to accelerate the averaging over the solvent degrees of freedom .
these last two applications of the _ general _ replica exchange method are illustrated in the next section .
rms error of the free energy over @xmath44 at @xmath45 for the ` abmd ` ( blue ) and reference ` mtd ` ( red ) simulations . ]
to illustrate the method , we have simulated the hydrophobic peptide ( sketched in fig.[fig:1 ] ) in the gas phase and in a solvated environment , using cyclohexane as the explicit solvent .
the free energy of this peptide at @xmath46 in gas phase has previously been investigated with the ` mtd ` method@xcite , and is characterized by a `` double - well '' structure ( see fig.[fig:8 ] ) , with the wells corresponding to the peptide in a `` globular '' ( left minimum in the fig.[fig:8 ] ) and a @xmath35-hairpin ( right minimum in the fig.[fig:8 ] ) folded conformation , respectively . while simple enough , the molecule possesses all the typical features of larger peptide systems usually studied with biomolecular simulations .
simulation parameters are as in a previous study@xcite : the atoms are described by the 1999 version of the cornell _ et al .
_ force field@xcite , with no cutoff for the non - bonded interactions .
the berendsen thermostat is chosen with @xmath47 for temperature control .
the ` md ` time step ( @xmath48 ) is 1 fs for the parallel tempering simulations , and 2 fs otherwise .
rms error of the free energy over @xmath44 for ` abmd ` simulations at @xmath45 with @xmath49 ( 1 ) , @xmath50 ( 2 ) , @xmath51 ( 3 ) and @xmath52 ( 4 ) . ]
the radius of gyration of the heavy atoms was chosen to be the collective variable : @xmath53 here , @xmath54 is the center of mass , with @xmath55 , and the sum runs over all atoms except hydrogen .
the initial configuration is the fully unfolded peptide . a reference free energy profile , whose error@xcite in the region of interest is less than 0.15 kcal / mol , was computed for benchmarking purposes ( see appendix [ ap : reference ] for details ) . as a measure of the rms free energy error ,
the following construction was used : @xmath56 where @xmath57 accounts for the arbitrary additive constants in the free energies @xmath58 . here
@xmath59 and @xmath60 , which correspond to the physical region of interest ( see fig.[fig:8 ] ) .
rms error of the free energy over [email protected] for multiple walkers ` abmd ` simulations with @xmath62 using one ( 1 ) , two ( 2 ) , four ( 4 ) and eight ( 8) trajectories at @xmath63k . ]
figure [ fig:2 ] presents the time dependence of the rms free energy error for the ` abmd ` and reference ` mtd ` run@xcite .
both simulations have exactly the same kernel width @xmath64 , and flooding timescale @xmath65 that corresponds to the _ a posteriori _ hills acceptance rate reported in ref .. it is evident that the ` ambd ` run is more accurate than the corresponding ` mtd ` run .
the ` abmd ` method owes its better convergence to the smoother time evolution of the biasing potential and accurate discretization in @xmath4 .
the amount of memory used by the ` abmd ` simulation to store @xmath31 values was only @xmath66 bytes ( considering double precision ) for roughly @xmath67 tiny `` hills '' that were accumulated by the end of the run .
the reference ` mtd ` simulation with merely @xmath68 gaussians required roughly 25 times more memory for the biasing potential ( with only the positions of the gaussians stored explicitly ) .
one can expect , that ` abmd ` will be even more economical when it comes to dealing with multidimensional collective variables , provided that sparse arrays are used for @xmath31 with only non - zero elements being stored explicitly .
although an _ a priori _ error estimate for this type of non - equilibrium simulation is really not feasible , it is expected that the error should decrease for increased @xmath15 .
this point is illustrated in fig.[fig:3 ] , which shows the error for increasing values of @xmath15 . in order to decrease the simulation time required for accurate free energy estimates
even further , the multiple - walker variation of ` abmd ` proves to be useful . for a moderate number of walkers ,
the speedup is nearly linear , with an additional increase in accuracy coming from the better sampling of the `` evolving '' canonical distribution ( see fig.[fig:4 ] ) .
parallel tempering improves both the speed and the accuracy even more . to this end
, we first ran ` abmd ` with @xmath69 using 2 , 4 , 6 and 8 replicas at @xmath46 , @xmath70 , @xmath71 , @xmath72 , @xmath73 , @xmath74 , @xmath75 and @xmath76 ( during equilibrium ` md ` runs , the peptide configuration jumps between the two minima on a picosecond timescale at @xmath77 ) . in all cases ,
the @xmath78 was found to be @xmath79 kcal / mol , or less as @xmath20 ( data not shown ) .
again , the improvement in accuracy stems from the better sampling of the `` evolving '' canonical distribution .
then , we ran 8 replicas with smaller values @xmath15 and were surprised that the accuracy does not degrade , even for @xmath80 as shown in fig.[fig:5 ] .
rms error of the free energy over @xmath44 at @xmath45 for parallel tempering ` abmd ` simulations using eight replicas running at @xmath46 , @xmath70 , @xmath71 , @xmath72 , @xmath73 , @xmath74 , @xmath75 and @xmath76 with @xmath69 ( 1 ) , @xmath81 ( 2 ) , @xmath82 ( 3 ) and @xmath83 ( 4 ) . ]
finally , we turn to aspects related to the _ general _ replica exchange method and illustrate its potential . as already noted , by using replicas with different collective variables and swapping these at prescribed rates , it is possible to obtain projections of the free energy surface for the corresponding variables .
it is also possible to use previously obtained information with respect to one collective variable to compute the free energy associated with a different variable .
for example , suppose that instead of the _ one - dimensional _ free energy profile as as a function of @xmath84 already discussed , one realizes that what is actually needed is a _ two - dimensional _ profile that includes information with respect to the number of o - h bonds along the backbone .
the two - dimensional free energy map is computationally quite expensive , but the calculation can be greatly accelerated with the help of the general replica exchange method
. we therefore simulated @xmath85 replicas .
the eight replicas were run at the previously stated temperatures , with each replica biased by a _ static _ ( not evolving ) biasing potential corresponding to the negated free energy associated with the radius of gyration @xmath84 at the corresponding temperature , as shown in fig.[fig:8 ]
. the additional ninth replica was run at @xmath46 with ` abmd ` flooding in the two collective variables , _
@xmath84 and the number of o - h bonds along the backbone as given by @xmath86 where @xmath87 is the distance between a pair of hydrogen and oxygen atoms and @xmath88 .
the sum runs over the unique o - h pairs ( _ i.e. _ , each o - h pair is counted only once ) , with o and h separated by one or more amino - bases along the backbone ( 27 pairs in total ) . in other words
, we `` re - use '' the previously computed free energies for @xmath84 to get the free energy in the two - dimensional space @xmath89 : the eight first replicas serve as a `` sampling enhancement device '' for the ninth replica .
the calculation is carried out in two stages : a `` coarse '' stage ( @xmath90 with @xmath91 and @xmath92 , @xmath93 ) followed by a `` fine '' stage ( @xmath94 with @xmath95 and @xmath96 , @xmath97 ) . in both runs exchanges between four randomly chosen pairs of replicas
were attempted every @xmath98 .
the final free energy map is shown in fig.[fig:6 ] .
it is clear that this two - dimensional free energy landscape conveys additional information not contained in the one - dimensional free energy plots already discussed . in particular
, it allows for a better characterization of the `` globular '' states of the : specifically , it is apparent from the fig.[fig:6 ] that there are at least two such states with different values of @xmath99 ( both correspond to the left minimum in fig.[fig:8 ] ) .
of course , one could have re - used the information in the one - dimensional @xmath84 profiles to include other collective variables , in addition to @xmath99 .
free energy map for peptide in the gas phase as a function of the collective variables @xmath84 and @xmath99 .
( see ref . for details regarding the algorithm used to make this plot ) . ]
the general replica exchange ` abmd ` may also be advantageous for explicit solvent simulations , which are often notoriously lengthy . specifically , if one is interested in the solute and the collective variables
do not depend on the solvent degrees of freedom , then the number of replicas required to maintain an adequate exchange rate depends only very weakly on the amount of solvent ( which must of course be sufficient as to adequately solvate the structure ) , provided that all the replicas are simulated at the same temperature .
this is because the exchange probability does not explicitly depend on the potential energy difference when the temperature of the replicas is the same . while not every choice of collective variables for different replicas will lead to decent exchange rates , one can nevertheless take advantage of this property and use general replica exchange to enhance the sampling in a solvated environment . in order to demonstrate the method in this regime
, we simulated the peptide at @xmath46 solvated by 171 cyclohexane ( @xmath100 ) molecules ( the total number of atoms was 3139 ) under periodic boundary conditions using the general@xcite ` amber ` force - field ( ` gaff ` ) for the solvent .
we used a _ truncated octahedron _ cell of fixed size ( constant volume ) that corresponds to the equilibrium density at @xmath46 ( the equilibrium density value was obtained from a @xmath101 simulation under constant pressure at @xmath46 ) .
the particle - mesh ewald ( ` pme`@xcite ) method was used for the electrostatic forces , with a @xmath102 ` fft ` grid and an @xmath103 cutoff for the direct sum ( same cutoff was used for van der waals interactions ) .
first , we ran 10 replicas in the `` flooding '' mode ( _ i.e. _ , under evolving biasing potentials ) using as collective variables the distances @xmath104 between the backbone carbons separated by at least 2 amino - acids ( there are 10 such distances for ) .
we ran for @xmath105 with @xmath106 and @xmath107 , and then for another @xmath105 with @xmath108 and @xmath109 , attempting exchanges between five randomly selected pairs every @xmath110 . as expected , at the start of the simulation , the exchange rate was nearly 100% decreasing later as the biasing potentials were built up . by the end of the simulation , when all possible values of the distances had been covered , the exchange rate was very disparate between different pairs of replicas . however , for every replica there was at least one other replica such that the exchange rate between them was reasonable ( _ i.e. , _ the whole simulation did not degenerate into ten different non - interacting trajectories ) .
we then set @xmath111 in these 10 replicas , and added an eleventh replica whose collective variable was chosen as the radius of gyration of the heavy atoms .
in other words , as before , we use the ten replicas as a `` sampling enhancement device '' for the last one .
we then ran a two - stage flooding scheme : a @xmath105 coarser stage , with @xmath112 and @xmath92 ; followed by a @xmath101 finer stage , with @xmath113 and @xmath114 .
as before , the exchange attempts between 5 randomly selected pairs of replicas were performed every @xmath110 .
the `` raw '' ` abmd`-computed free energy associated with @xmath84 after that stage is shown in fig.[fig:7 ] . in a next step
, we ran 64 biased simulations ( each comprising of 11 replicas ) for @xmath115 ( first @xmath116 for equilibration followed by @xmath105 of `` production '' runs ) starting from different initial configurations .
we set @xmath111 in all replicas ( static biasing potentials ) and recorded the values of @xmath84 in the eleventh replica every 10 picoseconds .
we then used the log - spline algorithm of stone@xcite _ at .
_ to estimate the ( biased ) log - density of the @xmath84 values at _
equilibrium_. this led us to the final shape of the free energy curve shown in fig.[fig:7 ] .
compared to the gas phase , the folded @xmath35-turn in the cyclohexane solvated peptide is clearly favored over the globular structure . free energy for peptide solvated in cyclohexane at @xmath46 as obtained via : coarse ( non - equilibrium ) replica - exchange ` abmd ` ( 1 ) ; finer ( non - equilibrium ) replica - exchange ` abmd ` ( 2 ) ; including the correction coming from equilibrium biased replica exchange ( 3 ) . ]
in summary , we have presented an ` abmd ` method that computes the free energy surface of a reaction coordinate using non - equilibrium dynamics .
the method belongs to the general category of umbrella sampling methods with an evolving potential , and is characterized by only two control parameters ( the flooding timescale and the kernel width ) and a favorable @xmath0 scaling with molecular dynamics time t. this scaling can be very important for large - scale classical ` md ` biomolecular simulations when long simulation times are required ( see , for example ref . , and references therein ) .
` abmd ` has also been extended to support multiple walkers and replica exchange .
both variations improve speed and accuracy of the method due to the better sampling of the `` evolving '' canonical distribution .
the replica exchange ` abmd ` has been generalized to include different temperatures and/or collective variables , that move under either an evolving or a static biasing potential . aside from enhancing the sampling , this swapping of replicas has several important practical advantages .
most importantly , it enables one to obtain projections of the free energy surface for _ any number of collective variables _ one might wish to investigate .
in addition , one can re - use previously obtained results in order to enhance the sampling of new collective variables .
it is also possible to exploit the fact that exchange rates at the same temperature are independent of the potential energy to enhance sampling of a solute in a minimum amount of solvent ( for collective variables independent of solvent atom coordinates ) .
we have implemented the ` abmd ` method in the ` amber ` package@xcite , and plan to distribute it freely . here
, we have demonstrated the workings of the ` abmd ` method with a study of the folding of the peptide , the application of ` abmd ` to more complicated biomolecular systems is reserved for future publications .
this research was partly supported by nsf under grants itr-0121361 and career dmr-0348039 .
in addition we thank nc state hpc for computational resources .
here , we provide simulation details with regards to the reference free energy curve .
we first ran short ( 5 ns , eight walkers with @xmath117 and @xmath118 ) multiple - walkers ` abmd ` at @xmath119 to reconstruct the global well .
this was followed by parallel tempering ` abmd ` runs , using the biasing potential obtained from the multiple - walkers simulation as the zero - time value for the biasing potentials at different temperatures .
we used eight replicas at @xmath46 , @xmath70 , @xmath71 , @xmath72 , @xmath73 , @xmath74 , @xmath75 and @xmath76 and attempted exchanges every 100 ` md ` steps ( @xmath120 ) .
the simulation started with @xmath117 and @xmath118 , and ran for @xmath121 exchanges .
we then set @xmath15 to @xmath122 , @xmath32 to @xmath123 and ran for @xmath124 more exchanges .
finally , this was followed up with @xmath125 more exchanges with @xmath126 and @xmath127 .
we then ran a very long ( @xmath128 exchanges , @xmath120 between exchanges ) biased parallel tempering simulation in the spirit of ref .
, in order to get an _ a posteriori _ error estimate .
from the resulting histogram it follows that the error does not exceed @xmath129 for [email protected] .
the rms error is probably much smaller , since 0.15 corresponds to the absolute non - uniformity of the histogram , _
i.e. _ , the maximum error , over [email protected] . the accurate free energy curves as a function of temperature are shown in fig.[fig:8 ] .
|
we present an ` a`daptively ` b`iased ` m`olecular ` d`ynamics ( ` abmd ` ) method for the computation of the free energy surface of a reaction coordinate using non - equilibrium dynamics .
the ` abmd ` method belongs to the general category of umbrella sampling methods with an evolving biasing potential , and is inspired by the metadynamics method .
the ` abmd ` method has several useful features , including a small number of control parameters , and an @xmath0 numerical cost with molecular dynamics time @xmath1 .
the ` abmd ` method naturally allows for extensions based on _ multiple walkers _ and _ replica exchange _ , where different replicas can have different temperatures and/or collective variables . this is beneficial not only in terms of the speed and accuracy of a calculation , but also in terms of the amount of useful information that may be obtained from a given simulation .
the workings of the ` abmd ` method are illustrated via a study of the folding of the peptide in a gaseous and solvated environment .
|
the population of @xmath1-ray binaries is a subclass of the high mass x - ray binary ( hmxb ) population ; they are binary systems comprised of a high mass ( @xmath10 m@xmath11 ) star and a compact object whose spectral energy distribution ( sed ) peaks above 1 mev .
indeed , they exhibit emission at high energies ( he ; 0.1100 gev ) and/or very high - energies ( vhe ; @xmath0100 gev ) .
the non - thermal high energy radiation from these objects indicates that they are sites of natural particle acceleration . at present only a handful of confirmed @xmath1-ray binaries are known : * ls i + 61@xmath2303 @xcite * ls5039 @xcite * psr b1259@xmath1263 * hess j0632 + 057 @xcite * 1fgl j1018.6@xmath125856 @xcite cyg x@xmath123 @xcite and cyg x@xmath121 @xcite show @xmath1-ray flaring activity , but their seds are peaked in x - rays .
so they do not belong to the population of the @xmath1-ray binaries .
ls i + 61@xmath2303 , ls5039 and psr b1259@xmath1263 have all been detected at both gev and tev energies ; 1fgl j1018.6@xmath12585 was discovered at gev energies solely with data from the _ fermi gamma - ray space telescope _ large area telescope ( _ fermi_-lat ) and has recently been reported to have a possible tev counterpart , hess j1018@xmath12589 @xcite .
hess j0632 + 057 was the first @xmath1-ray binary to be discovered at vhe prior to its detection in other wavebands ; the h.e.s.s .
collaboration reported an unidentified point - like vhe source lying close to the galactic plane @xcite .
it was noted that a potential optical counterpart to the source was a massive emission - line star , mwc 148 , of spectral type b0pe .
these facts led to the hypothesis that the source may be a new @xmath1-ray binary , a nature that was confirmed by the discovery of an orbital period of @xmath13 days detected in the x - ray counterpart by @xcite using data from a long - term _ swift_-xrt monitoring campaign .
@xcite determined the ephemeris and the orbital parameters of the system using the x - ray orbital period and through optical spectroscopy of mwc 148 .
they found the binary to be in a highly eccentric orbit ( @xmath14 ) . following
the discovery of @xcite observations made by the veritas collaboration reported upper limits at the 99% confidence level that are @xmath152.4 times lower than the flux found by the h.e.s.s .
collaboration , implying a variable flux behaviour of the source at vhe @xcite .
hess j0632 + 057 was detected again at vhe by magic and veritas in february 2011 when tev observations were simultaneously taken with an outburst in x - rays @xcite . in the x - ray regime @xcite
identified the x - ray counterpart of hess j0632 + 057 xmmu j063259.3 + 054801 from _
xmm_-newton observations .
the x - ray counterpart exhibits a hard spectrum with spectral index @xmath16 that is observed to soften from 1.2 to 1.6 with rising flux @xcite .
the source exhibits flux variability on timescales of hours .
similar x - ray characteristics together with a softer spectrum at tev energies are also seen in other @xmath1-ray binaries e.g. ls i + 61@xmath2303 and ls5039 @xcite .
@xcite explain the observed emission and variability as synchrotron and inverse compton emission from a population of relativistic electrons in a region within a few au of the star .
similar to ls i + 61@xmath2303 , the peak x - ray emission occurs @xmath30.3 orbital phases after the periastron passage . in the radio regime
@xcite detected a point - like , variable radio source at the position of mcw 148 with the giant metrewave radio telescope ( gmrt ) and the very large array ( vla ) at 1.28 and 5ghz , respectively .
@xcite observed hess j0632 + 057 with the european vlbi network ( evn ) at 1.6ghz in two epochs : during the 2011 january / february x - ray outburst and 30 days later .
in the first epoch the source appears point - like , whereas in the second it appears extended with a projected size of @xmath375au assuming a distance to the system of 1.5kpc .
the brightness temperature of 2@xmath1710@xmath18k at 1.6ghz together with the average spectral index of @xmath3 is indicative of non - thermal synchrotron radiation producing the radio emission @xcite .
the projected displacement of the peak of the radio emission in the 30 day gap between observations is 21au .
the radio morphology and the size of hess j0632 + 057 are similar to those observed in the well - established @xmath1-ray binaries psr b1259@xmath1263 , ls5039 and ls i + 61@xmath2303 . within the small class of known @xmath1-ray binaries the nature of the compact source is known only for one system : psr b1259@xmath1263 contains a 48ms radio pulsar @xcite .
the nature of the compact objects in ls5039 , ls i + 61@xmath2303 1fgl j1018.6@xmath125856 and hess j0632 + 057 is still unknown @xcite . @xcite
conducted a _
chandra _ observing campaign during the high state of x - ray and tev activity of hess j0632 + 057 in 2011 february .
their timing analysis did not find pulsations in the x - ray data and they derived 3@xmath19 upper limits on the x - ray pulsed fraction of @xmath330% , similar to those derived for ls i + 61@xmath2303 and ls5039 @xcite . presented here are the results of a detailed analysis of @xmath343 months of _ fermi _ lat survey observations of hess j0632 + 057 .
we describe the procedures implemented to analyse this complex field and the results of searching for persistent , orbitally variable and transient high - energy emission from hess j0632 + 057 . a search for radio pulsations using the 100-m green bank telescope ( gbt ) is also reported .
the large area telescope ( lat ) is the primary instrument onboard _ fermi _ ; it is a pair conversion telescope sensitive to photons with energies from @xmath320mev to more than 300gev .
the direction of an incident photon is derived by tracking the electron - positron pair in a high - resolution converter tracker , and the energy of the pair is measured with a hodoscopic csi(tl ) crystal calorimeter .
the _ fermi_-lat has a peak on - axis effective area of 8000 @xmath20 , a 2.4sr field of view , and an angular resolution of @xmath30.6@xmath2 at 1gev ( for events converting in the front section of the tracker ) .
furthermore , an anti - coincidence detector vetoes the background of charged particles @xcite .
the observatory operates principally in survey mode ; the telescope is rocked north and south on alternate orbits to provide more uniform coverage so that every part of the sky is observed for @xmath330 minutes every 3 hours .
the analysis presented here makes use of all data taken from 2008 august 4 until 2012 march 3 .
the _ fermi _ science tools v09r28p00 package was used to analyse the ` source ' event class data .
all events within a circular region of interest ( roi ) of 15@xmath2 radius centred on the known location of the massive star mwc 148 associated with hess j0632 + 057 ( r.a.(j2000 ) = 98247 , dec(j2000 ) = 5800 ) and in the energy range 0.1100 gev were extracted .
the good time intervals are defined such that the roi does not cross the @xmath1-ray - bright earth limb ( defined at 100@xmath2 from the zenith angle ) , and that the source is always inside the lat field of view , namely within a cone angle of 66@xmath2 .
the ` p7source_v6 ' instrument response functions ( irfs ) are used throughout the analysis .
when constructing spectral - spatial models of the roi the models used for the galactic diffuse emission ( gal_2yearp7v6_v0.fits ) and isotropic backgrounds ( iso_p7v6source.txt ) were those currently recommended by the lat team and were used by the _
fermi_-lat collaboration to build the second _ fermi_-lat source catalog ( hereafter 2fgl , * ? ? ?
hess j0632 + 057 lies at the galactic latitude @xmath21 , in the active region of the monoceros loop .
an image of this region in the @xmath1-ray band is shown in figure [ cmap ] .
this is a background subtracted count map of the @xmath22 region around hess j0632 + 057 .
it is generated by selecting events with energies greater than 1 gev , and subtracting the counts due to the galactic diffuse emission and the isotropic emission as calculated using the respective models with the science tool gtmodel . in figure [ cmap ]
it is evident that the region is dominated by the @xmath1-ray emission of psr j0633 + 0632 ( 2fgl j0633.7 + 0633 ) , which is only 078 away from hess j0632 + 057 .
sky region centered on hess j0632 + 057 .
the pixel size is 01@xmath1701 and a gaussian smoothing with 3 pixel kernel radius is applied .
the 2fgl sources are labeled with green crosses , hess j0632 + 057 with a diamond and psr j0633 + 0632 with the circle.,scaledwidth=50.0% ] psr j0633 + 0632 is a bright , radio - quiet @xmath1-ray pulsar discovered in the first six months of the _ fermi _ mission @xcite .
it has a flux ( @xmath0100 mev ) of @xmath23 ph @xmath6 s@xmath7 , and the pulse profile has the characteristic two narrow peaks separated by @xmath24 in phase @xcite .
the pulse profile obtained with the dataset analysed in this paper is shown in figure [ psrj0633 ] . to coherently assign a phase to each photon over such a long observation period ( @xmath25 years )
requires an up - to - date ephemeris for the pulsar . as psr j0633 + 0632 is radio - quiet , the new ephemeris has been calculated using the _ fermi_-lat data with the method described by @xcite .
a phase was associated to each event using the _ fermi _ plug - in of the tempo2 software package @xcite .
a weight was also assigned to each event , corresponding to the probability that the gamma ray is emitted by the pulsar , rather than by a nearby source or by the diffuse emission @xcite ; weights are calculated using the science tool gtsrcprob .
the spectral models of the pulsar and the nearby sources adopted to calculate the weights are those from the 2fgl catalogue . finally , the pulse profile in figure [ psrj0633 ] was obtained by summing the weights in each phase bin for the events within 2 of the pulsar position . to minimise contamination from the strong emission of psr j0633 + 0632 in the analysis of hess j0632 + 057 the pulsar emission was ` gated ' by only using events which occurred outside the pulse peaks , during phases 0.660.98 and 0.210.49 ( hereafter out - of - peak phases ) .
the off - pulse and intra - peak phases of psr j0633 + 0632 were defined by applying a bayesian block algorithm adapted for weighted - counts light curves , described in the next section .
years of _ fermi_-lat data . the off - pulse , and the intra - peak phases ( shaded regions )
are defined by means of the bayesian block algorithm : ( 0.66 - 0.98 ) and ( 0.21 - 0.49 ) , respectively.,scaledwidth=45.0% ] we found that an efficient way to define the off - pulse phase range is by subdividing the pulse profile into ` bayesian blocks ' , and choosing the block with the lowest count rate .
the intra - peak phase range is instead the lowest block between the two peaks .
the bayesian blocks algorithm was proposed by @xcite : the algorithm defines the phase ranges of constant rate , called bayesian blocks , on the basis of the likelihood ratio test applied to a poissonian distribution .
this is perfect for count light curves , but not for weighted pulse profiles , as in this case the weights are not integer values , and they do not follow a poissonian distribution . furthermore , the pulse profile is periodic , meaning that it does not have a well defined start and end . in order to manage weighted counts , in our method the bayesian blocks
are defined using the @xmath26 ratio test .
specifically , given a light curve of weighted counts , the null hypothesis is that the light curve is flat , and it is fitted by a single block .
the associated @xmath27 value for a single block is defined as : @xmath28 where @xmath29 are the weighted source counts in the @xmath30-th bin , and @xmath31 are the average weighted counts all along the light curve .
@xmath32 is the total number of bins .
the test hypothesis is that two blocks fit better than a single block .
the @xmath27 of the two blocks hypothesis is @xmath33 \label{chi2}\ ] ] where @xmath34 , @xmath35 and @xmath36 , @xmath37 are the average weighted counts , and the weighted counts in the @xmath30-th and @xmath38-th bin of the first and second block , respectively .
finally , the significance of the improvement given by two blocks rather than a single block is calculated by means of the f - test .
if two blocks fit the data significantly better than a single block then the division point between the blocks is recorded and the first of the new blocks ( on the left ) is tested for further subdivision .
the process continues until further subdivision is not significant or the block length reaches a predefined minimum length . in this way , the shortest , furthest - left significant block is defined .
the subdivision restarts considering the remaining light curve on the right .
the issue relating to the periodicity of the pulse profile is solved by selecting the starting ( and ending ) point of the light curve to be the phase of the highest peak .
this is the choice that least affects the definition of the off - pulse region .
the algorithm has three parameters : the number of bins of the pulse profile ( 100 ) , the minimum possible length of a block ( 3 bins ) , and the threshold on the two blocks significance ( 1.5@xmath19 ) .
the values in parenthesis are the settings used for psr j0633 + 0632 .
we tested this algorithm on all the @xmath1-ray pulsars detected so far with the _
fermi_-lat , finding that these settings give good results for all the bright pulsars .
having restricted the events to those selected from within the out - of - peak phases of psr j0633 + 0632 ( shaded phase ranges in figure [ psrj0633 ] ) , the analysis of the region around hess j0632 + 057 was performed by means of the binned maximum - likelihood method @xcite .
two different tools were used to perform the spatial and spectral analysis : science tool gtlike and pointlike .
gtlike is the official tool supplied with the publicly available science tools to fit spectral models to lat data .
pointlike is an alternative binned likelihood implementation , optimised to localise the source and to characterise its extension .
it operates quickly on long datasets making it better suited for iteratively constructing source models of a region .
results from pointlike have been extensively compared with gtlike and found to be consistent ( see e.g. , * ? ? ?
* ; * ? ? ?
the likelihood - ratio test statistic ( ts ) is employed to evaluate the significance of the @xmath1-ray flux of a test source .
the ts value is defined to be : @xmath39 where @xmath40 is the likelihood function maximised by adjusting all parameters of the model with and without the test source ( test and null hypothesis , respectively ) .
a ts = 25 corresponds to a significance of just over 4@xmath19 when the testing source is modelled with a simple power law @xcite .
figure [ tsmaps ] shows two ts maps of the @xmath41 region centred on hess j0632 + 057 .
they are generated by testing for an additional point - like source located at the centre of each pixel ( of size @xmath42 ) . in the left panel of figure [ tsmaps ]
is plotted a diffuse subtracted ts map , meaning that the spectral - spatial model for the null hypothesis includes only the galactic and isotropic emission models .
in contrast , the right panel shows the residual ts map when the spectral - spatial model also includes the 2fgl sources within a radius of 20@xmath2 from the roi centre , in addition to the diffuse models .
the 2fgl sources within 5@xmath2 of hess j0632 + 057 ( 11 sources ) are modelled with their flux normalizations allowed to vary , while the other 45 sources had all their parameters fixed to the values given in the 2fgl catalog .
the spectral model of the isotropic diffuse emission had free normalization and the spectral model of the galactic diffuse emission is scaled by a simple power - law with normalisation and index free to account for any local variations in the diffuse emission . in the final fit of the data
the fitted value for the power - law index is consistent with 0 within the fit error , indicating that no adjustment to the local galactic diffuse spectral shape was required .
hess j0632 + 057 lies in a region of high galactic diffuse emission . since at low energies
the diffuse emission is stronger , as well as its uncertainties , the following analysis was restricted to using data at @xmath43 mev .
the ts maps in figure [ tsmaps ] show that the region around hess j0632 + 057 is not well modelled solely by the sources from the 2fgl catalogue .
this is not unexpected considering that the 2fgl catalog was constructed with 2 years of _
fermi_-lat data , while in this paper the dataset is almost doubled ( @xmath44 years ) . searching for emission from hess j0632
+ 057 requires an accurate modeling of the entire region , and hence an improved source model of the region is required .
the roi source model was iteratively improved by adding a new point source at the location of the highest residual ts and then re - calculating the residual ts map ( see sec .
initially the 3@xmath45 region around hess j0632 + 057 was fine tuned before extending the radius to 5@xmath45 .
a detailed description of the process and results are given in the following sections .
[ cols="<,^,^,^,^,^",options="header " , ] @xmath46 range given due to the transient / flaring nature of the source .
+ @xmath47@xcite ; @xmath48@xcite ; @xmath49@xcite ; @xmath50@xcite ; @xmath51@xcite ; @xmath52@xcite ; @xmath53@xcite ; @xmath54@xcite ; [ binary_pop ]
a rigorous search for emission in the gev energy band from the @xmath1-ray binary hess j0632 + 057 using @xmath33.5 years of _
fermi_-lat data has not yielded a significant detection .
there is no significant persistent , flaring or orbital phase - dependent emission .
this makes hess j0632 + 057 unique among the handful of known @xmath1-ray binaries as the only source eluding detection by the lat .
the 95% lat flux upper limit for the 0.1100 gev energy band of @xmath55 is consistent with the flux predicted by the one - zone ssc model shown in figure 4 of @xcite and @xcite , which assume a magnetic field of b = 70 mg and a radiation density of u@xmath56= 1 erg @xmath57 .
however , it is clear from the lat data that there must be a turnover in the vhe spectrum of hess j0632 + 057 below @xmath3136 gev in order for the _
fermi_-lat data to be compatible with the data obtained at higher energies by magic and h.e.s.s .
the `` striped pulsar wind '' model invoked by @xcite to explain the emission from psr b1259@xmath1263 could explain what is observed in hess j0632 + 057 , however a pulsar nature for the compact object in this system has not been confirmed and hence other models may offer feasible explanations . numerical simulations were performed to calculate the radio flux fraction expected to be observed along the orbit by a pulsar with its radio beam pointing toward us , while absorbed by free - free interaction with the strong wind ( polar and disk ) from the companion .
the results of the simulations lead us to exclude the truncation of the disk in favor of an indefinitely extended disk .
alternatively we could exclude the hypothesis that the radio beam of the plausible pulsar is pointing toward the earth .
the _ fermi _ lat collaboration acknowledges generous ongoing support from a number of agencies and institutes that have supported both the development and the operation of the lat as well as scientific data analysis .
these include the national aeronautics and space administration and the department of energy in the united states , the commissariat lenergie atomique and the centre national de la recherche scientifique / institut national de physique nuclaire et de physique des particules in france , the agenzia spaziale italiana and the istituto nazionale di fisica nucleare in italy , the ministry of education , culture , sports , science and technology ( mext ) , high energy accelerator research organization ( kek ) and japan aerospace exploration agency ( jaxa ) in japan , and the k. a. wallenberg foundation , the swedish research council and the swedish national space board in sweden .
additional support for science analysis during the operations phase is gratefully acknowledged from the istituto nazionale di astrofisica in italy and the centre national dtudes spatiales in france .
hill , a. b. , dubois , r. , torres , d. f. , on behalf of the fermi - lat collaboration 2010 , proceedings of the 1st session of the sant cugat forum of astrophysics .
nanda rea & diego f. torres ( editors ) , 2011 , springer , issn : 1570 - 6591 .
isbn 978 - 3 - 642 - 17250 - 2
|
the very high energy ( vhe ; @xmath0100 gev ) source hess j0632 + 057 has been recently confirmed as a @xmath1-ray binary , a subclass of the high mass x - ray binary ( hmxb ) population , through the detection of an orbital period of 321 days .
we performed a deep search for the emission of hess j0632 + 057 in the gev energy range using data from the _ fermi _ large area telescope ( lat ) .
the analysis was challenging due to the source being located in close proximity to the bright @xmath1-ray pulsar psr j0633 + 0632 and lying in a crowded region of the galactic plane where there is prominent diffuse emission .
we formulated a bayesian block algorithm adapted to work with weighted photon counts , in order to define the off - pulse phases of psr j0633 + 0632 .
a detailed spectral - spatial model of a 5@xmath2 circular region centred on the known location of hess j0632 + 057 was generated to accurately model the lat data .
no significant emission from the location of hess j0632 + 057 was detected in the 0.1100 gev energy range integrating over @xmath33.5 years of data ; with a 95% flux upper limit of f@xmath4 @xmath5 ph @xmath6 s@xmath7 . a search for emission over different phases of
the orbit also yielded no significant detection . a search for source emission on shorter timescales ( days months ) did not yield any significant detections .
we also report the results of a search for radio pulsations using the 100-m green bank telescope ( gbt ) .
no periodic signals or individual dispersed bursts of a likely astronomical origin were detected .
we estimated the flux density limit of @xmath8jy at 2/9ghz .
the lat flux upper limits combined with the detection of hess j0632 + 057 in the 136400 tev energy band by the magic collaboration imply that the vhe spectrum must turn over at energies @xmath9136 gev placing constraints on any theoretical models invoked to explain the @xmath1-ray emission .
[ firstpage ] binaries : individual : hess j0632 + 057 gamma - rays : observations binaries : close stars : variables : other x - rays : binaries .
|
in recent years several mechanisms generating harmonics of electromagnetic radiation have been discovered . among these harmonics from gases @xcite , harmonics from a laser pulse propagating through underdense plasma @xcite , and harmonics production from the plasma - vacuum boundary when a laser pulse impinges on a solid target @xcite are the most prominent ones .
especially the high - order gas - harmonics , exhibiting a `` plateau '' instead of a rapid decrease with the harmonic order , seem to be a promising source for xuv `` water - window''-radiation . apart from the effort to make progress towards shorter wavelengths , another goal is to achieve shorter pulse durations because the temporal resolution in pump - probe experiments clearly depends on the pulse length .
one scheme proposed to generate attosecond pulses is based on phase - matching pulse trains which are produced by a laser pulse focused into a jet of rare gases @xcite .
another method makes use of the fact that the efficiency of gas - harmonics generation is sensitive to the ellipticity of the incident laser light @xcite .
the method to generate an ultra - short _ low order _ harmonic laser pulse as proposed in this paper is based on the time - dependent electron density of the target material in laser pulse - solid interaction .
the mechanism is thus entirely different from those mentioned above which are based on phase - matching of nonlinear single atom - responses @xcite , relativistic ( and thus nonlinear ) electron trajectories under the influence of the surrounding plasma @xcite or the oscillating vacuum - target - interface owing to the @xmath0-nonlinearity in the lorentz force .
however , the key idea of the method proposed in this paper is similar to the one in ref .
@xcite , namely to control the time duration of effective harmonics production by the incident pulse itself . in ref .
@xcite it is the ellipticity which is the relevant parameter governing harmonics creation while in the present paper it is the ionization rate which determines the rise in the free electron density of the target .
the harmonic pulse duration is of the order of this risetime and thus can be tuned appropriately by varying the intensity of the incident pulse .
it turned out that although our method is capable of generating very short pulses of third and fifth harmonic light ( about 2 fundamental laser cycles in length ) the mechanism is not appropriate for generating particularly _ high order _ harmonics efficiently . in a recently published conference proceeding @xcite similar studies of `` ionization harmonics ''
are presented , but the authors did not focus on the generation of ultra - short pulses there . in section [ modeling ]
we introduce our simple one - dimensional analytical model to study how the laser pulse propagation is influenced by the ionizing thin foil . in section [ numresults ]
we discuss our particle - in - cell simulation results .
finally , we conclude in section [ concl ] .
we assume a linearly polarized laser pulse impinging perpendicularly on a thin foil target . in the following analytical and numerical calculations
the whole setup is treated one dimensionally in space , i.e. , the laser pulse propagates along @xmath1 and the electric field is in @xmath2-direction .
the foil will be ionized by the pulse . to calculate the pulse propagation through a medium with varying free electron density one has to solve the inhomogeneous wave equation e(x , t ) - e(x , t ) = j(x , t ) .
[ inh_wave ] the green s function of this equation is @xmath3 / 2 $ ] where @xmath4 is the step - function , i.e. , @xmath5 for @xmath6 and 0 otherwise . the solution of can be written as the sum of the incident field @xmath7 and the radiation field produced by the current @xmath8 , i.e. , @xmath9 , with e_r(x ,
t)=-dt dx j(x,t ) .
[ radi ] in order to model thin foils we now assume a delta - like current in space @xcite .
if the thin foil is located at @xmath10 the current is @xmath11 \ell \delta(x)$ ] , with @xmath12 and @xmath13 the electron and ion density , and velocity , respectively , and @xmath14 the ion s charge state . integrating @xmath8 over
@xmath1 one finds that the current per unit area equals that of a `` real '' physical thin foil of thickness @xmath15 ( as long as there is no strong electron density gradient across the foil ) . inserting the current @xmath8 into and performing the spatial integration lead to e_r(x , t)= - dt j_h(0,t ) [ er_mediumstep ] where @xmath16 .
if we assume that the pulse hits the target at @xmath17 we finally get e(x , t)=e_0(x , t)- j_h(0 , ) [ eq_for_field ] for the electric field ( @xmath18 is the retarded time ) .
the current @xmath19 itself depends on the electric field . neglecting the ionic contribution to the current
, we have j_h(0,)= n(0 , ) _ 0^ dt e(0,t ) where @xmath20 . here
it has been assumed that all newly created electrons are born with the appropriate fluid element velocity and that collisional as well as relativistic effects are negligible . besides , we neglect in our analytical treatment energy subtraction from the pulse due to the finite ionization energy of the target material . how this energy loss as well as momentum transfer due to the velocity distribution of the ionization produced electrons can be incorporated in a fluid description
is studied in @xcite .
all pulse intensities considered in this article do not cause relativistic electron motion . supposing an ionization rate @xmath21 applicable for pulse intensities under consideration has been chosen , the electron density @xmath22 in the foil is given by @xmath23\right)\right]$ ] . when the target is fully ionized the electron density is @xmath24
we finally end up with the following integral equation for the electric field @xmath25 , e(x , t)=e_0(x , t)-_0^ dt e(0,t ) , [ integr_eq ] where = =( ) ^2 _ 1 .
@xmath26 is the plasma frequency of the fully ionized target , @xmath27 , and @xmath28 and @xmath29 are the incident em wave s frequency and length , respectively .
the dimensionless parameter @xmath30 determines how strong the propagation of the incident pulse is affected by the foil . for @xmath31
the foil is optically `` thin '' .
if the foil is not pre - ionized or ionization is not completed already during the very early part of the pulse , remains nonlinear due to the electron - density shape - factor which depends on the electric field through the rate @xmath32 $ ] .
therefore one expects harmonics in the transmitted and reflected light . in what follows
we will restrict ourselves to study in first order in @xmath30 ( i.e. , we assume a `` thin '' foil and iterate once ) . at the position of the foil
then e(0,t)=e_0(0,t)-_0^t dt e_0(0,t ) [ integr_eq_first ] holds . here , the difficulty is to calculate @xmath33)$ ] .
the ionization rate @xmath34 depends on the _ absolute value _ of the electric field , i.e. , the rate has two maxima per fundamental laser cycle . supposed that the pulse envelope @xmath35 is sufficiently adiabatic the rate may be expanded in a fourier - series with even multiples of the fundamental frequency only , and a slowly time - dependent envelope @xmath36 , = \ { a_0 + _ n=1^ ( a_2n 2n_1 t + b_2n 2n_1 t ) } .[fourier_expans ] since the rate @xmath34 is a complicated functional of the field , in general all terms in the expansion are present .
however , if we assume the incident pulse ( divided by its amplitude ) to be an even function in time , @xmath37 , all coefficients @xmath38 in vanish .
this finally leads to @xmath39 here , @xmath40 , @xmath41 , and @xmath42 is the modified bessel - function .
note that @xmath43 are slowly time - dependent due to their dependence on @xmath36 . from we
can deduce that by ionization in first order of @xmath30 only _ odd harmonics _ will be produced : in the term in curly brackets is composed of even harmonics , but multiplied with @xmath44 odd harmonics are created .
harmonics production is enhanced when @xmath30 is increased .
therefore one expects that increasing the density or the thickness of the foil acts in favor of the creation of harmonics .
however , one has to bear in mind that a perturbative treatment with respect to @xmath30 preceded here .
furthermore , increasing the density @xmath24 or the thickness @xmath15 simply makes the foil less transparent .
a 1d2v particle - in - cell ( pic ) code was used to simulate laser pulse - solid interaction . in order to incorporate ionization
the rate equations governing the ionization state of the target were solved during each `` pic - cycle '' @xcite . for simplicity only one ionization state with the ionization energy of hydrogen ( 13.6 ev ) was assumed .
landau s tunneling - rate @xcite was used which is a reasonable choice for the field strengths and frequencies under consideration .
the short risetime of the electron density forces a tiny time step .
usually one fundamental laser cycle ( wavelength ) was sampled by 1000 temporal ( spatial ) gridpoints . about @xmath45 computer particles , sampling the physical charge densities of the thin foil
were found to be sufficient .
the ions were mobile ( although this is unimportant for the effect under consideration ) and 1836 times heavier than the electrons ( hydrogen ) . in fig .
[ figure_one ] numerically computed spectra of the transmitted light are shown for 5 different peak field strengths @xmath35 , corresponding to intensities @xmath46 and @xmath47 w/@xmath48 .
all other parameters were held constant : wavelength @xmath49 nm , foil thickness @xmath50 , incident @xmath51-shaped laser pulse of duration @xmath52 fs , and the density was the critical one with respect to the fundamental frequency , i.e. , @xmath53 @xmath54 .
the higher the field strength , the broader are the harmonics peaks in the spectrum .
the pulse length of the harmonics radiation is closely related to the risetime of the electron density in the foil since as soon as the density remains constant harmonics production will stop . in fig .
[ figure_two ] the normalized electron density is plotted vs time for the 5 field strengths of fig .
[ figure_one ] .
a risetime covering @xmath55 fundamental periods for the weakest pulse and only one cycle for the strongest pulse can be inferred from the plot ( each stair in the density corresponds to one half cycle ) .
the density risetime is very sensitive to field strength and pulse shape .
increasing the field strength leads to a decreasing density risetime and hence to a shorter harmonics pulse length .
however , the conversion efficiency decreases when the field strength of the incident pulse is increased since the nonlinearity switches off too soon .
furthermore the `` harmonics '' peaks are shifted and asymmetrically broadened if complete ionization occurs within only one fundamental cycle ( or even less ) . in the limit of a step - like behavior of the electron density
the spectrum resembles the fourier - transform of the @xmath56-function with no @xmath28-harmonics structure at all . in fig .
[ figure_one ] the pulse length of the harmonic radiation can be estimated by fitting the peaks in the spectrum to a fourier transformed `` test envelope '' @xmath57 .
@xmath58 is the pulse duration of the @xmath22th harmonic .
for the 5 cases of fig.[figure_one ] one finds for the pulse length of the 3rd harmonic @xmath59 times the fundamental period @xmath60 .
a lower limit for @xmath61 certainly is @xmath62 itself because a shorter risetime of the electron density leads to a vanishing @xmath28-structure in the spectrum .
the power in the 3rd harmonic is about @xmath63 of the fundamental .
this conversion efficiency is similar to the one in ref .
@xcite while all other methods mentioned in the introduction are superior as far as power transferred to ( short pulse ) harmonic radiation is concerned .
one may object that the incident pulse intensity we took in our numerical simulations was already small ( at least for `` up - to - date '' short pulse laser systems ) so that the 3rd - harmonic pulse with only a millionth of its intensity is not acceptable at all .
however , the incident pulse might be a stronger but defocused pulse so that the 3rd harmonic output , when focused , becomes considerable .
besides , using a shorter fundamental wavelength ( and correspondingly a thinner and/or denser foil ) would require a higher field strength to fully ionize the target within the same number of cycles . by examining the dimensionless parameters @xmath30 and @xmath64 one can estimate the `` experimental parameters '' @xmath15 , @xmath24 and @xmath65 in order to meet the desired harmonic pulse duration @xmath61 .
the practical limit for @xmath61 found in the numerical simulations is about @xmath66 .
if the incident light has already a rather short wavelength ( e.g. , if light , produced by one of the high - harmonics mechanisms described above are used ) then the @xmath66-limit can be shifted towards the attosecond - domain .
however , in that case it would be certainly a challenging task to find the optimal parameters @xmath15 and @xmath24 for a manufacturable thin foil . in recent experiments with thin foil targets @xcite the thickness @xmath15 was about @xmath67@xmath68 nm and the electron density @xmath24 ( when fully ionized ) was of the order of @xmath69 @xmath54 , i.e. , by a factor 100 denser than assumed in our numerical simulations presented so far .
one may wonder if it is really essential to take a _
thin _ foil .
when a sufficiently strong laser pulse impinges on a thick overdense target only a thin plasma layer is created anyway and our analytical treatment in section [ modeling ] should therefore apply as well .
we thus expect also from thick targets short harmonic radiation , at least in reflection . in figure [ spek_thick]a spectra calculated from timerows of the transverse electric field @xmath25 in front ( reflected light ) , in the middle , and at the rear edge ( transmitted light ) of a thicker target are presented .
the target was one wave length thick , @xmath70 , and ten times overcritical , @xmath71 @xmath54 .
the incident pulse had an intensity @xmath72 w/@xmath48 .
the other parameters were @xmath49 nm , @xmath73 fs , square - sine in shape again .
the spectra have a rather complicated structure owing to the complex spatial and temporal dependence of the free electron density .
this can be seen in figure [ spek_thick]b where the target density is shown as a contour plot vs time and space . during the pulse the target gets fully ionized only in a thin layer at the front . the electron density quickly rolls off and at the rear side not even the critical density is reached . in the spectra we clearly observe the well - known effect that higher frequency radiation penetrates easier the target than light of the fundamental frequency does since for third harmonic light
the target density is only slightly overcritical while it is even undercritical for the 5th harmonic .
this effect was utilized as an experimental method to determine the plasma density @xcite .
we observe also frequency shifts since the peaks are not located any longer at the odd harmonics positions .
obviously , thick targets are not as easily accessible by means of simple physical reasoning as it is the case for thin foils where the @xmath74-current model described in section [ modeling ] works well .
our experiences from several pic runs with different target densities and thicknesses can be summarized as follows : as soon as there is a density gradient across the foil the spectra of the reflected and transmitted light get distorted ( compared to the thin foil - spectra in fig .
[ figure_one ] ) . to avoid this density gradient
the optimal @xmath75-product is about @xmath76 which is , with present day technology , not easy to achieve .
using a higher incident pulse intensity in order to ionize the target more homogeneously reduces the density risetime and thus the conversion efficiency .
apart from these `` technical problems '' one may argue that during the plasma formation process electron - ion - collisions might be important , especially for the relatively low field strengths about @xmath77 w/@xmath48 .
since it is during the plasma formation where the harmonics are produced there might be serious distortions in the spectrum of the transmitted light . in order to take electron - ion - collisions into account we introduced a collision frequency @xmath78 into our 1d2v pic - code .
this leads to dissipation of energy due to friction of the oscillating charge - sheets ( note that in a 1d pic - code each `` computer - particle '' represents an actual charge sheet @xcite ) .
the dissipated energy is used to determine a `` sheet - temperature '' which , in turn , enters into @xmath78 .
we found collisions causing mainly distortions at high frequencies but the 3rd and 5th harmonic - peaks were almost unaltered . for stronger incident pulses
the effect of collisions is even less .
with our pic - code we also examined the effects of energy subtraction due to the ionization energy of the target material ( according to the model in @xcite ) .
the most prominent effect , as far as harmonics generation is concerned , is that ionization gets slowed down slightly ( which can be compensated by choosing a higher incident pulse intensity ) .
it is worth mentioning that the observed effect of harmonics production due to the rise of the electron density in a thin foil may be used to _ measure _ the ionization time of the foil instead of presupposing an ionization rate .
this would offer an opportunity to check the validity of ionization models experimentally .
in summary , we have studied the spectrum of a perpendicularly incident laser pulse when transmitted through a rapidly ionizing foil .
a simple analytical method was utilized to show what the underlying mechanism for the generation of short pulse odd harmonic radiation is .
the pulse duration of the harmonic radiation is only a few cycles with respect to the frequency of the incident laser light .
the pulse length is governed by the risetime of the electron density in the target and therefore it can be easily tuned through adjusting the peak field strength of the incident pulse .
this might be a promising way towards the generation of attosecond pulses .
this work was supported by the european commission through the tmr network silasi ( super intense laser pulse - solid interaction ) , no.erbfmrx-ct96-0043 .
one of the authors ( d.b . ) would like to thank mr.g .
baier for stimulating discussions .
99 n. sarukura , k. hata , t. adachi , r. nodomi , m. watanabe and s. watanabe , phys .
a * 43 * , 1669 ( 1991 ) ; + g. farkas and c. toth , phys .
a * 168 * , 447 ( 1992 ) ; + j. l. krause , k. j. schafer and k. c. kulander , phys .
lett . * 68 * , 3535 ( 1992 ) ; + a. lhuillier , ph .
balcou , s. candel , k. j. schafer and k. c. kulander , phys . rev .
a * 46 * , 2778 ( 1992 ) ; + reviews by several authors about recent advances in the field of gas harmonics can be found in _
multiphoton processes 1996 _ , institute of physics conference series no .
154 , ( iop publishing , bristol , 1997 ) phillip sprangle and eric esaray , _ phys.fluids b _ * 4 * , 2241 ( 1992 ) ; + j. m. rax and n. j. fisch , _ phys . rev .
lett . _ * 69 * , 772 ( 1992 ) r. l. carman , d. w. forslund , and j. m. kindel , phys
. rev .
lett . * 46 * , 29 ( 1981 ) ; + r. l. carman , c. k. rhodes , and r. f. benjamin , phys .
a * 24 * , 2649 ( 1981 ) ; + paul gibbon , phys .
lett . * 76 * , 50 ( 1996 ) ; + r. lichters , j. meyer - ter - vehn , and a. pukhov , phys .
plasmas * 3 * , 3425 ( 1996 ) ; + h. ruhl , r. a. cairns , phys .
plasmas * 4 * , 2246 ( 1997 ) philippe antoine , anne lhuillier , and maciej lewenstein , phys .
lett . * 77 * , 1234 ( 1996 ) ; + philippe antoine , dejan b. miloevi , anne lhuillier , mette b. gaarde , pascal salires , and maciej lewenstein , _
phys.rev . a _ * 56 * , 4960 ( 1997 ) ; p. b. corkum , n. h. burnett , and m. y. ivanov , opt .
* 19 * , 1870 ( 1994 ) ; + m. yu ivanov , p. b. corkum , t. zuo , and a. bandrauk , phys .
lett . * 74 * , 2933 ( 1995 ) enrique conejero jarque and luis plaja , in : _ superstrong fields in plasmas _ , ed . by m. lontano , g. mourou , f.pegoraro , and e. sindoni , ( aip proceedings 426 , new york , 1998 ) , p. 360
v. vshivkov , n. naumova , f. pegoraro , and s. bulanov , submitted to phys .
plasmas p. mulser , f. cornolti , and d. bauer , accepted for publication in phys .
plasmas c. k. birdsall and a. b. langdon , _ plasma physics via computer simulation _ , ( iop publishing , bristol , 1991 ) l. d. landau and e. m. lifshitz , _ quantum mechanics _ , ( pergamon press , oxford , 1977 ) p. 295
w. theobald , r. hner , c. wlker , and r.sauerbrey , _ phys .
lett . _ * 77 * , 298 ( 1996 ) d. giulietti , l. a. gizzi , a. giulietti , a.macchi , d. teychenn , p. chessa , a. rousse , g. cheriaux , j.p .
chambaret , and g. darpentigny , _ phys .
_ * 79 * , 3194 ( 1997 )
|
a thin and dense plasma layer is created when a sufficiently strong laser pulse impinges on a solid target .
the nonlinearity introduced by the time - dependent electron density leads to the generation of harmonics .
the pulse duration of the harmonic radiation is related to the risetime of the electron density and thus can be affected by the shape of the incident pulse and its peak field strength .
results are presented from numerical particle - in - cell - simulations of an intense laser pulse interacting with a thin foil target . an analytical model which shows how the harmonics are created is introduced .
the proposed scheme might be a promising way towards the generation of attosecond pulses .
+ pacs number(s ) : 52.40.nk , 52.50.jm , 52.65.rr # 1#2 _ null_0 # 1 # 1#2 # 1#2 # 1([#1 ] )
|
consider a system of 4-dimensional spins with basis states @xmath191 .
suppose the hamiltonian is written as @xmath192 where @xmath193 contains only @xmath194 , while @xmath195 contains only @xmath196 .
both @xmath193 and @xmath195 are gapped with a minimal gap @xmath13 , but the ground state energy of @xmath193 is zero for @xmath197 and @xmath13 for @xmath198 , whereas that of @xmath195 is @xmath13 for @xmath197 and zero for @xmath198 . the ground state is then spanned by @xmath194 for @xmath197 and by @xmath196 for @xmath198 . in this case , the overlap between @xmath199 and @xmath200 vanishes for any @xmath201 .
consider a mapping from @xmath15 to @xmath14 , as in fig . 1 of the main text .
as the hamiltonian @xmath9 differs from @xmath38 only by a local term @xmath202 , it is reasonable to approximate @xmath14 with @xmath79 by finding an optimal unitary operator @xmath203 supported on @xmath81 so that the approximation gets better as @xmath83 is increased .
let us define a function @xmath204 apparently , @xmath205 is a non - increasing function of @xmath83 and @xmath206 .
we can also define @xmath207 which is also non - increasing .
one can think of @xmath208 as a characteristic function of the system that quantifies how far the boundary effect penetrates into the bulk .
this is also intimately related to the existence of a thermodynamic limit .
for instance , if @xmath209 for @xmath210 with @xmath211 being a certain constant , @xmath212 becomes identical to @xmath213 for @xmath210 ( see fig . 1 of the main text ) .
consequently , if one takes any region sufficiently away from the boundary , the corresponding reduced density matrix becomes independent of the system size and hence all the local quantities of the bulk become intensive quantities , i.e. , there exists a thermodynamic limit of the system at the zero temperature .
ordinarily , @xmath208 would be a certain _ decreasing _ function , whose functional form characterizes how fast the system converges to the thermodynamic limit .
note that the condition ( 1 ) of the main text is violated if and only if @xmath214 for all @xmath83 ( the system in the previous section is an example ) . for gapped systems satisfying the condition ( 1 ) , i.e. , @xmath215 for some @xmath201
, one can show that @xmath208 asymptotically decreases at least exponentially , and hence the thermodynamic limit should exist ( this will be discussed elsewhere ) . throughout the proof , we omit superscript @xmath216 for brevity .
let @xmath217 with eigenvalues @xmath218 and eigenstates @xmath219 .
without loss of generality , @xmath220 and @xmath221 for @xmath222 . in the same manner ,
let @xmath223 with @xmath224 and @xmath225 for @xmath222 .
@xmath226 is readily diagonalized with eigenstates @xmath227 and @xmath228 with eigenvalues @xmath229 and @xmath230 , respectively . for brevity ,
let us write the state as @xmath231 and similarly for others .
[ fig4](a ) depicts the spectrum of @xmath226 .
the two lowest energy levels @xmath232 and @xmath233 cross only at @xmath124 .
consequently , by adding an additional term that couples @xmath232 and @xmath233 , one can open a gap at @xmath124 , making the hamiltonian gapped for all @xmath128 .
@xmath234 can do this as @xmath235 for @xmath124 .
if @xmath236 is close to one , it would be easy to find @xmath120 making @xmath237 gapped .
if @xmath236 is very small , however , it is not necessarily obvious because @xmath232 and @xmath233 mostly couple to higher energy levels , whereas their mutual coupling is very small . without loss of generality , choose the phase so that @xmath238 is real
. let @xmath239 a part of the hamiltonian @xmath240 has eigenvalues @xmath241 with @xmath242 .
denote by @xmath243 the corresponding eigenstate .
let @xmath244 it is easy to see that for any state @xmath245 with @xmath246 , @xmath247 .
hereafter , we omit @xmath248 for brevity when the meaning is clear from the context ( e.g. , instead of @xmath249 , we simply write @xmath250 ) .
we can write the hamiltonian as @xmath251 let @xmath252 and @xmath253 denote , respectively , the first and the second line of this hamiltonian .
@xmath252 has a gap @xmath254 ( see fig .
[ fig4](b ) ) .
note that @xmath255 .
let @xmath256 with eigenvalues @xmath257 and eigenstates @xmath258 for @xmath259 . in what follows , we prove that @xmath260 for all @xmath248 if @xmath261 we regard @xmath253 as a perturbation to @xmath252 . as @xmath264 , the first - order perturbation to @xmath265 vanishes .
moreover , as @xmath266 , the second - order perturbation to @xmath265 is at most @xmath267 , which can be much smaller than the unperturbed gap @xmath268 if @xmath269 is sufficiently small . in order to obtain a rigorous bound , let us write the perturbed eigenstate of @xmath270 as @xmath271 , where @xmath272 , @xmath273 , and the perturbed eigenvalue @xmath274 . we then have @xmath275 which leads to @xmath276 from this set of equations , we find @xmath277 note that the left hand side is larger than @xmath278 and that @xmath279 .
note also that when @xmath280 , the left hand side is larger than the right hand side . consequently , the lowest - energy solution @xmath281 exists when @xmath282 and the second lowest solution @xmath283 is positive .
we thus find @xmath284 note that @xmath286 let us write @xmath287 with real @xmath288 , where @xmath289 .
it follows that @xmath290\\ \ge&(1-c_0 ^ 2)\omega_-+c_0 ^ 2\omega_+-2c_0\sqrt{1-c_0 ^ 2}f\delta .
\end{split}\ ] ] these two inequalities lead to @xmath291 similarly , write @xmath292 with real @xmath293 and @xmath294 . for @xmath295 to vanish , we require @xmath296 which results in @xmath297 .
it thus follows that @xmath298 consequently , we end up with @xmath299 it is easy to see that this bound is larger than @xmath300 for @xmath285 .
33ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * ,
( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( )
|
we consider general locally - interacting arbitrary - dimensional lattice spin systems that are gapped for any system size .
we show under reasonable conditions that nondegenerate ground states of such systems obey the entanglement area law . in so doing ,
we offer an intuitive picture on how a spectral gap restricts the correlations that a ground state can accommodate and leads to such a special feature . over the last decade or so
, quantum information theory has emerged as an indispensable tool in studying strongly - correlated many - body systems .
for example , entanglement is essential in classifying quantum phases of matter @xcite , especially topological quantum phases @xcite , the density matrix renormalization group ( dmrg ) method provides the best way to numerically simulate low energy physics of one - dimensional spin systems @xcite , and various universal features of general many - body systems have been explored with new tools and insights @xcite . while there are many different avenues in such quantum information approaches to many - body physics , they are essentially built upon the grounds of common theoretical foundations .
the entanglement area law ( or simply the area law ) is one of the prominent @xcite . for a many - body pure state @xmath0 ,
the bipartite entanglement between a subregion ( @xmath1 ) and the rest ( @xmath2 ) is quantified by the entanglement entropy @xmath3 , the von neumann entropy of the reduced density matrix @xmath4 .
when @xmath3 has an upper bound proportional to the surface area of @xmath1 , we say @xmath0 obeys the area law @xcite .
it turns out that ground states of local hamiltonians typically obey the area law @xcite , possibly with a multiplicative logarithmic correction @xcite , although one can deliberately construct a counterexample @xcite .
the area law is indeed a very special feature because in a large hilbert space , almost all states , in the sense of the haar measure , exhibit a volume - law scaling of the entanglement entropy ; the states obeying the area law actually belong to a measure - zero set @xcite .
this anomaly leads to diverse and profound implications across various fields , e.g. , in classical simulations of quantum systems @xcite , topological quantum phases @xcite , and hamiltonian complexity theory @xcite .
conceptually , the area law is also reminiscent of the holographic principle @xcite .
it has thus been of crucial importance to find out the general mechanism and criteria of the area law . in particular ,
one of the prominent open problems has been whether the area law is generally obeyed in gapped local systems in high dimension , since its one - dimensional problem was solved in ref .
@xcite . in this context , the fundamental question in hand is concerning the entanglement entropy of the ground state @xmath5 of a general local @xmath6-body hamiltonian @xmath7 having a finite spectral gap @xmath8 for sufficiently large @xmath6 @xcite .
as it turned out , in an arbitrary spatial dimension , proving ( or disproving ) the area law in such a general case is a daunting task at present . from a practical point of view
, however , if we are in a position to tackle ordinary many - body systems ( e.g. , as in the context of classical simulations of quantum systems @xcite ) , we may bring in a few empirical assumptions without sacrificing much of the generality , thereby significantly relaxing the technical difficulties and furthermore offering a clear - cut insight into the problem .
specifically , we note that a many - body system is generally defined in terms of its microscopic details ( i.e. , its constituent particles , mutual interactions , external potential , etc . ) , while the system size @xmath6 is actually variable . formally speaking ,
when @xmath7 is given , it is implicitly taken for granted that there also exist hamiltonians @xmath9 with @xmath10 and importantly all different @xmath9 share the common defining characteristics of the system .
for example , when we say a certain system is gapped , it generally means that @xmath11 for any @xmath12 with a lower bound @xmath13 . here ,
all the different - sized ground states @xmath14 represent essentially the same matter and in the thermodynamic limit , if exists , any local observable can not discriminate between different @xmath6 as it becomes an intensive quantity .
this makes it reasonable to assume that @xmath14 and @xmath15 have a finite overlap ( i.e. , they are nonorthogonal ) after the boundary effect is properly washed out ( see fig .
[ fig1 ] ) so that they share a common subspace that encapsulates the characteristic features of the system , e.g. , the order parameters , correlation functions , and so on . otherwise , the bulk properties of the system would be utterly unpredictable in practice as they will be drastically altered by a microscopic change of the system size , rendering the system unstable . in this paper
, we consider such a practical situation and prove the entanglement area law in arbitrary - dimensional gapped local spin systems under two general conditions drawn from the above reasoning . before proceeding , let us first clarify our notation .
we consider arbitrary systems of @xmath6 locally - interacting finite - dimensional spins placed on a @xmath16-dimensional lattice with one spin per site . for the ` localness ' of the interaction to make sense , the lattice has two properties .
first , the euclidean distance @xmath17 and the graph distance @xmath18 between sites @xmath19 and @xmath20 satisfy @xmath21 for some constant @xmath22 .
second , the number of sites in a unit volume @xmath23 is bounded by @xmath24 for some constant @xmath25 .
for given site @xmath19 , we define sets of neighboring sites as follows : @xmath26 the interaction being local means that @xmath6-spin hamiltonians can be written as a sum of local terms supported on @xmath27 for @xmath28 with @xmath29 being a constant bounding the range of the interaction . the trace distance between density matrices @xmath30 and @xmath31
is denoted by @xmath32 with @xmath33 being the trace norm .
the operator norm is denoted by @xmath34 .
the two conditions we impose are as follows .
first , there exist a sequence of hamiltonians with different systems sizes @xmath35 having nondegenerate ground states @xmath36 and finite spectral gap @xmath11 for all @xmath12 . here , we index the spins in such a way that the @xmath12-spin system is constructed by adding the @xmath12-th spin on the boundary of the @xmath37-spin system ( see fig .
[ fig1 ] ) .
as the interaction is local , @xmath9 and @xmath38 differ by a local term @xmath39 supported on @xmath40 ( not @xmath41 in general because there can be distinct boundary terms ) .
the interaction strength is finite , which means @xmath42 for some constant @xmath43 . -spin and
the @xmath12-spin systems differ only locally .
the reduced density matrices of the ground states @xmath44 and @xmath45 are obtained by tracing out a local neighborhood of spin @xmath12 . in this illustration , @xmath46 . ]
second , for all @xmath12 , the ground states @xmath47 and @xmath48 have a finite overlap in the sense that there exist constants @xmath49 and @xmath50 such that @xmath51 where @xmath52 ( see fig .
[ fig1 ] ) . for @xmath53
, this condition means that when the ground state @xmath47 is written in terms of the eigenstates of @xmath38 and the states of spin @xmath12 as @xmath54 , the coefficient @xmath55 associated with the ground state @xmath48 is non - zero .
although this seems intuitively natural , one can imagine a counterexample , albeit quite artificial , in which the entire system undergoes a quantum phase transition by a single change of the particle number at the boundary ( see the appendix ) . as mentioned
, such an exceptional case is not of our interest in this work .
the main result of this paper is the following theorem .
consider a lattice spin system satisfying the above two conditions .
take a subregion @xmath1 , which is a @xmath16-dimensional ball of radius @xmath56 .
for the ground state , the entanglement entropy of this region is bounded as @xmath57 where @xmath58 s are constants determined by above - defined system parameters @xmath13 , @xmath43 , @xmath16 , @xmath22 , @xmath25 , @xmath29 , @xmath50 , and @xmath49 .
hence , the upper bound of @xmath3 scales as @xmath59 , satisfying the area law . here
, we took the ball - shaped region to simplify the proof .
generalization to the case of a different shape is straightforward .
note that the area law makes sense only for simple - shaped regions .
for example , the surface area of a fractal shape can be arbitrarily large .
the underlying idea of the proof is as follows .
take a subregion @xmath60 to be a @xmath16-dimensional ball of radius @xmath61 centered at the origin of region @xmath1 with positive constant @xmath62 to be chosen later ( see fig .
[ fig2 ] ) .
suppose there are @xmath63 spins in region @xmath1 , @xmath64 spins in region @xmath60 , and @xmath65 spins in the rest .
our strategy is to take a particular sequence of ground states @xmath66 in the following way .
( 1 ) @xmath67 is the ground state of the @xmath68-spin system corresponding to region @xmath60 .
( 2 ) @xmath69 is the ground state of the whole system . ( 3 ) @xmath47 ( @xmath70 ) is the ground state of an intermediate system .
here we choose the @xmath12-th spin , among @xmath71 remaining ones , to be the one having the shortest euclidean distance to the origin , whereby the shape of the system is retained as far as possible for all @xmath12 ( see fig .
[ fig2 ] ) . for each ground state in the sequence , we can obtain the reduced density matrix for region @xmath1 , @xmath72 . note that @xmath73 is bounded by the logarithm of the hilbert space dimension for region @xmath74 .
our initial bound @xmath75\\ & = n_0v_dr_0dr_0^{d-1}+\mathcal{o}(r_0^{d-2 } ) \end{split } \label{sinitial}\ ] ] thus exhibits the area - law scaling , where @xmath76 is the volume factor for @xmath16-dimensional balls .
we thus find that @xmath77 \label{sfinal}\end{gathered}\ ] ] exhibits the area - law scaling if the summation on the right - hand side also does so for appropriately chosen @xmath78 . in order to show this , we need to understand how @xmath48 is mapped to @xmath47 . , the system is gradually enlarged while retaining its shape as far as possible . ]
the ensuing procedure is based on the following intuitive picture .
suppose we try to identify @xmath47 with @xmath79 , where @xmath80 is a unitary operator supported on @xmath81 and @xmath82 is any state of spin @xmath12 .
in general , the exact identity would be obtained only for sufficiently large @xmath83 , i.e. , @xmath84 . however
, one can see that the spectral gap can play a role here in approximating @xmath85 .
note that both @xmath38 and @xmath9 are gapped . _
if _ @xmath86 also remains gapped for @xmath87 , where @xmath88 , we can then consider an adiabatic passage from @xmath89 to @xmath47 .
intuitively , @xmath90 is likely to be gapped for appropriately chosen @xmath82 since only a small portion of the hamiltonian is varied and @xmath48 and @xmath47 are essentially the same kind of states and hence no quantum phase transition occurs . for the moment
, suppose it is the case .
if so , as the adiabatic passage can be done in a finite time scale ( inversely - proportional to @xmath13 ) and the hamiltonian is varied only locally , the lieb - robinson bound implies that the adiabatic process can affect the system only ( quasi-)locally , which means that @xmath91 is approximately local .
one can thus write @xmath92 for some constant @xmath93 up to a small error .
_ if _ we neglect the error for the moment , we can choose @xmath78 to satisfy @xmath94 so that every term in the summation of eq . vanishes because @xmath95 does not act on region @xmath1 , resulting in the area law .
conceptually , what happens is that the boundary effect spreads no further than @xmath96 away and thus @xmath97 converges to @xmath98 once @xmath12 reaches the point where region @xmath1 does not recognize the existence of a boundary any more . in the above picture ,
we have made two logical jumps to be resolved .
first , it should be ensured that @xmath90 is indeed gapped .
second , @xmath91 is only approximately local and thus we need to work out how the errors add up . as a preliminary step , note that condition implies there is a unitary operator @xmath99 acting on @xmath100 such that @xmath101 , which follows from the uhlmann s theorem @xcite .
let @xmath102 .
this hamiltonian preserves the gap condition and the ground state is @xmath103 . for convenience , let @xmath104 and @xmath105 .
letting @xmath106 be the sum of all terms commonly appearing both in @xmath107 and @xmath108 , we can write @xmath109 , where @xmath110 are supported on @xmath111 .
we are now ready to proceed .
we have local hamiltonians @xmath109 , which have a gap lower - bounded by @xmath13 .
their ground states are @xmath112 and @xmath113 , respectively , and @xmath114 . if we can find an adiabatic path from @xmath112 to @xmath113 , we can also find one from @xmath89 to @xmath47 up to a local unitary operator @xmath99 that does not affect @xmath115 as long as @xmath116 .
the key lemma for our proof is the following .
introduce an ancillary two - level system @xmath117 and consider a local hamiltonian @xmath118 where @xmath119\otimes{|1\rangle}_a{\langle1|}\\ & \quad\quad\,+[h_2^{(n)}+(1-\lambda)\delta]\otimes{|2\rangle}_a{\langle2|},\\ h_a^{(n)}(\lambda)=&f(\lambda)\delta({|1\rangle}_a{\langle2|}+{|2\rangle}_a{\langle1| } ) \end{split}\ ] ] with real non - negative smooth function @xmath120 for @xmath121 .
@xmath120 vanishes at @xmath122 and @xmath123 , and is maximized at @xmath124 with @xmath125 .
there exists @xmath120 with @xmath126 such that @xmath127 is gapped for all @xmath128 and the minimal gap is at least @xmath129 . .
( b ) by adding a coupling term , the ground state degeneracy is lifted . ] the underlying idea for the proof of this lemma is as follows .
let @xmath130 for the moment .
@xmath131 alone is then readily diagonalized by @xmath132 and @xmath133 , where @xmath134 and @xmath135 are the eigenstates of @xmath136 and @xmath137 , respectively . as shown in fig .
[ fig3](a ) , the two lowest energy levels @xmath138 and @xmath139 become degenerate only at @xmath124 .
this degeneracy can be lifted by adding a term that couples the two levels .
@xmath140 plays this role as @xmath141 .
the detailed proof is presented in the appendix . note that the ground state of @xmath142 is @xmath143 and that of @xmath144 is @xmath145 .
the lemma thus implies that there exists an adiabatic path from @xmath146 to @xmath147 up to an irrelevant local unitary transformation , where the hamiltonian varies locally on @xmath111 and @xmath117 .
such a local adiabatic change preserves the area law , as discussed in ref .
@xcite .
we tailor their method in the remainder of our proof .
let @xmath148 be the ground state of @xmath127 . by slightly modifying the derivation in ref .
@xcite , one can construct an artificial hamiltonian governing the change of @xmath148 in @xmath128 , which turns out to be approximately local in the following sense .
there exists an integer constant @xmath149 such that @xmath150{|\tilde{\psi}_0^{(n)}(\lambda)\rangle } , \label{adiabatic}\ ] ] where @xmath151 and @xmath152 are hermitian , @xmath151 is supported on @xmath153 , @xmath152 is supported on @xmath154 , and furthermore @xmath155 for some constant @xmath156 . here , @xmath157 and @xmath158 are determined by @xmath159 , @xmath160 , and @xmath161 .
the last ingredient of our proof is the small incremental entangling theorem presented in ref .
@xcite .
consider a many - body system in a pure state @xmath162 .
the system is divided into four regions @xmath163 , @xmath164 , @xmath165 , @xmath166 , and evolves by a hamiltonian @xmath167 supported on region @xmath168 .
the reduced density matrix for region @xmath169 at time @xmath170 is given by @xmath171 .
the growth rate of the entanglement entropy @xmath172 at any @xmath170 is bounded as @xmath173,\ ] ] for some constant @xmath174 , where @xmath175 is the hilbert space dimension for region @xmath176 .
we are now ready to finish up our proof .
let us choose @xmath177 .
we can bound @xmath178 as follows .
suppose spin @xmath12 has euclidean distance @xmath179 to the origin of region @xmath1 with @xmath180 .
as @xmath181 and @xmath182 yield entanglement entropies @xmath183 and @xmath184 for region @xmath1 , respectively , we can examine @xmath148 to obtain the bound . note that @xmath182 is obtained by evolving @xmath181 through eq . during unit time . by performing the trotter expansion
, one realizes that only @xmath152 s with @xmath185 can affect the entanglement entropy . by using the small incremental entangling theorem
, we find that @xmath186 it thus follows that the summation on the right - hand side of eq . is bounded by @xmath187 which is @xmath188 ( note that @xmath189 ) .
@xmath190 in eq . can be obtained by expanding this integral and eq . , which completes the proof of the theorem .
as a final remark , we note that our approach in its current form is not applicable to topologically ordered systems in general as our inherent physical setting based on an open boundary condition and nondegenerate ground states is not compatible with the nontrivial topology of the space and the topological degeneracy , which are the essential attributes of topological quantum phases @xcite .
for such systems governed by frustration - free hamiltonians , the local topological quantum order , if exists , leads to the area law @xcite .
the general proof without such restrictions is however yet to be given .
|
in models with one universal extra dimension ( ued ) @xcite , all of the standard model particles are promoted to 5 dimensional fields , propagating in a flat extra dimension ( for earlier ideas closely related to ued models see ref .
@xcite ) . in order to allow for the existence of chiral zero - mode fermions , the extra dimension is modded out by the orbifold , @xmath2 . due to the @xmath3 symmetry ,
the interactions between kaluza - klein ( kk ) modes respect what is known as kk - parity , which implies that kaluza - klein particles can only be produced or destroyed in pairs . among other consequences
, this leads to a lower bound on the kk mass scale of only @xmath4 @xcite .
furthermore , kk - parity guarantees the stability of the lightest kaluza - klein particle ( lkp ) , thus providing us with a potentially viable dark matter candidate . for a review of the ued model and its phenomenology , see ref .
@xcite .
despite its minimal field content , the ued model contains a large number of undetermined parameters .
in addition to the higgs mass , @xmath5 , and the compactification radius , @xmath6 , the geometry also allows for operators of the 5d fields at the orbifold fixed points whose magnitudes are independent of the bulk parameters ( conservation of kk - parity forces the boundary localized operators to be included symmetrically on both fixed points , however ) . in the absence of such boundary localized terms ,
one expects the first kk excitation of the hypercharge gauge boson , @xmath0 , to be the lkp @xcite .
the inclusion of such terms , however , can affect the kk mass spectrum @xcite , and allow for other states to potentially serve as the lkp .
as shown in ref .
@xcite , the first kk excitation of the neutral @xmath7 gauge boson , @xmath1 , can be the lkp if small but non - vanishing boundary kinetic terms in the electroweak sector are considered .
other identities for the lkp are more difficult to realize .
the first kk excitation of the higgs boson , @xmath8 , could be the lkp , but would require large boundary localized terms and is strongly constrained for low @xmath8 masses @xcite .
no experimentally viable realizations are known in which the lkp takes the form of a kk - pseudoscalar higgs or kk - neutrino @xcite .
the hypothesis that dark matter takes the form of stable , weakly interacting massive particles ( wimps ) , is gradually being tested by direct and indirect detection experiments . at present ,
the strongest direct detection constraints on the spin - independent scattering cross section of wimps off nuclei have been placed by the cdms @xcite and xenon @xcite collaborations ( the dama collaboration has also reported an annual modulation of their rate which they interpret as a positive detection of dark matter @xcite ) .
in addition , wimps can produce potentially observable particles in their annihilations .
wimps which scatter with nuclei in the sun can be captured and accumulate in the center of sun , where they ultimately annihilate to produce high energy neutrinos that can be observed with large volume neutrino telescopes , such as icecube @xcite . assuming that the time scale for the annihilation and capture rates to equilibrate in the sun is smaller than the age of the solar system , the present annihilation rate of wimps will be determined only by the capture rate of wimps which , in turn , can be determined by the wimp - nucleon scattering cross section @xcite
. therefore both direct detection experiments and neutrino telescopes are sensitive to similar elastic scattering processes . in this article , we study the impact of these indirect and direct detection constraints on the ued parameter space for lkps in the form of a @xmath0 or @xmath1 .
the phenomenology of a @xmath0 lkp has been studied extensively @xcite , including the prospects for direct @xcite and indirect detection @xcite .
the analysis for detecting @xmath0 dark matter via high energy neutrinos from wimp annihilation in the sun was first performed in ref .
@xcite , which we closely follow in this article .
the thermal relic abundance of a @xmath1 lkp , and the related constraints from direct detection experiments , have been calculated in ref .
@xcite which , to the best of our knowledge , is the only study to consider this dark matter candidate ( for discussion of dark matter phenomenology in models with two universal extra dimensions , see ref .
@xcite ) . in this article
, we apply the analysis of ref .
@xcite to the cases of @xmath0 and @xmath1 dark matter . in each case , the lkps scatter off nucleons through the @xmath9-channel exchange of kk - fermions .
the experimental limits from direct detection experiments and icecube lead to strong constraints on the @xmath10 plane , where @xmath11 is the lkp mass and @xmath12 is the fractional splitting between the lkp mass and that of the kk - quarks .
the constraints that derive from cdms and xenon lead to limits on the wimp - nucleon spin - independent cross sections , while those from icecube primarily lead to bounds on the wimp - nucleon spin - dependent cross sections .
following ref .
@xcite , the spin - independent dark matter - nuclei elastic scattering cross section is given by _
si = ^2 , [ dmsisigma ] where @xmath11 is the wimp mass , @xmath13 is the mass of the target nucleus , @xmath14 is the number of protons in the nucleus , and @xmath15 is the number of nucleons in the nucleus .
the wimp - nucleon couplings , @xmath16 , can be written in terms of the couplings to individual quarks : f_p , n = _ q = u , d , s m_p , n f_t_q^p , n , [ fpn ] where @xmath17 are the respective neutron and proton mass , and the nucleon matrix elements are @xmath18 , @xmath19 , @xmath20 , @xmath21 and @xmath22 @xcite . following ref .
@xcite , we have conservatively neglected the couplings to gluons through heavy quark loops . the quark - level couplings for a @xmath0 lkp are given ( in the limit of small mixing between left- and right - handed kk - fermions and small splittings between the kk - fermions mass and the lkp mass ) by @xcite _ q ( + ) or , for a @xmath1 lkp , by _ q ( ) where r_q_l , r , [ rqdef ] @xmath23 is the electric charge , @xmath24 is the weinberg angle , @xmath25 are the appropriate hypercharges , and @xmath26 are the left- and right - handed kk - quark masses . from this
, it is evident that the scattering cross section is greatly enhanced in the case of small mass splittings .
although direct detection experiments are currently most sensitive to spin - independent scattering , spin - dependent scattering can in many cases dominate the capture rate of wimps in the sun . within the context of ued , in particular , spin - dependent scattering
typically dominate this process .
this is , in particular , true for the dark matter candidates considered in this article .
the spin dependent scattering off protons ( _ ie . _
hydrogen nuclei ) arises from the @xmath9-channel exchange of kk - quarks and is given by @xcite ^h_sd , b^(1)&=&(4^p_u+ ^p_d+^p_s)^2 + & & 1.7 10 ^ -6 pb ( ) ^4 ( ) ^2 , + ^h_sd , w^3(1)&=&(^p_u+ ^p_d+^p_s)^2 [ esxw ] + & & 0.36 10 ^ -6 pb ( ) ^4 ( ) ^2 , [ esxb ] where @xmath27 , @xmath28 , and @xmath29 quantify the spin - content of the quarks within the proton @xcite . as only left - handed quarks couple to @xmath1 , @xmath30 depends only on @xmath31 . for @xmath0 , both @xmath32 and @xmath33 contribute , but the cross section is dominated by @xmath33 due to the larger @xmath34 charges of right - handed fermions . therefore in eq .
( [ esxb ] ) we neglect the @xmath32 contributions . in addition
we would also like to emphasize that for @xmath30 the spin - content is given by @xmath35 , where as for @xmath36 it is of @xmath37 .
thus despite the fact that the @xmath38 appearing in @xmath30 is about twice as large as the @xmath39 in @xmath36 , @xmath30 is suppressed relative to @xmath36 .
using the scattering cross sections described in the previous section , we can estimate the rate at which lkps will be captured in the sun . assuming a local dark matter density of 0.3 gev/@xmath40 , the rate of wimps captured in the sun is given by @xcite c_(3.35 10 ^ 18 s^-1 ) ( ) ^3 ( ) ( ) ^2 where @xmath41 is the local rms velocity of the dark matter halo distribution and @xmath42 is the effective elastic scattering cross section , given by @xmath43 in this expression , @xmath44 , @xmath45 and @xmath46 denote the relative degree of form factor suppression , solar abundances , and kinematic suppression for the various species of nuclei .
these factors are evaluated for the case of hydrogen and helium nuclei in the lower expression .
spin - independent scattering with oxygen nuclei can also be important and may even dominate over spin - independent scattering with other nuclei for sufficiently heavy wimps . any spin - independent scattering cross section that is not already excluded by direct detection experiments , however , will generate an unobservably small flux of high energy neutrinos @xcite .
therefore , we focus throughout the remainder of this article on wimp capture in the sun through spin - dependent scattering , which is far less stringently constrained . for the range of masses , elastic scattering cross sections , and annihilation cross sections we consider in this paper
, we have explicitly checked that the processes of dark matter capture and annihilation in the sun reach equilibrium @xcite . in all cases shown , the rate of wimps annihilating in the sun is very close to the rate of wimps captured .
this is in contrast to wimp capture and annihilation in the earth , which does not reach equlibrium in the amount of time available @xcite . to calculate the flux of neutrinos from the sun
, we need to know the fraction of lkp annihilations which proceed to various standard model states . in table
[ tab : br ] , we show the annihilation ratios for @xmath0 and @xmath1 lkps annihilations . for the @xmath0 ,
annihilations occur largely through the exchange of kk - quarks and leptons , and the fraction to each final state is determined by the @xmath34 charges of the exchanged and final state particles . in contrast , the @xmath1 dominantly annihilates to @xmath47 via the @xmath9-channel exchange of a kk-@xmath48 .
.the approximate fraction of @xmath0 and @xmath1 annihilations which proceed to various standard model final states .
the @xmath49 channel is summed over all neutrino species .
the @xmath50 channel is summed over @xmath51 and @xmath52 quarks .
[ cols="^,^,^,^,^,^,^,^",options="header " , ] ( top ) or @xmath1 ( bottom ) dark matter annihilations in the sun . in each frame
, we also show the contributions to the neutrino spectrum from various annihilation channels . in the case of @xmath0 dark matter
, the neutrino flux is dominated by annihilations to taus and neutrinos , whereas for @xmath1 dark matter , annihilations to @xmath53 produce the most neutrinos . in each case
we have used @xmath54 and @xmath55.,title="fig:",scaledwidth=48.0% ] + ( top ) or @xmath1 ( bottom ) dark matter annihilations in the sun . in each frame
, we also show the contributions to the neutrino spectrum from various annihilation channels . in the case of @xmath0 dark matter
, the neutrino flux is dominated by annihilations to taus and neutrinos , whereas for @xmath1 dark matter , annihilations to @xmath53 produce the most neutrinos . in each case
we have used @xmath54 and @xmath55.,title="fig:",scaledwidth=48.0% ] these ratios determine the injection spectrum of high energy neutrinos in the sun . using this injection spectrum
we calculate the @xmath56 energy spectrum at the earth using the tabulated results of ref . @xcite . in fig .
[ fig : channels ] we present the flux of high energy neutrinos , @xmath57 , due to annihilation of @xmath0 or @xmath1 lkps in the sun for @xmath58 gev .
neutrinos coming from @xmath59 s dominate the neutrino spectrum for the @xmath0 case , while for @xmath1 dark matter the dominant contribution is from @xmath47 .
the neutrino spectrum from both of these dark matter candidates is generally hard , enabling us to use the more stringent icecube limits , as opposed to the less stringent limits for annihilations to @xmath60 @xcite . in the case of @xmath0 annihilations ,
the spectrum is actually harder than those considered in the icecube analysis , making the results presented here conservative .
as electroweak precision constraints require the compactification scale to be greater than @xmath61 gev , we do not consider the constraints from super kamiokande which apply mostly to the case of low mass wimps .
to calculate the rate of neutrino induced muon tracks from wimp annihilations in a large volume neutrino telescope such as icecube , we use the following expression : r= de _ dy a_eff n_a , [ nevnts ] where @xmath62 is the effective area of the detector , @xmath63 is the differential neutrino - nucleon charged current cross section , @xmath64 is the fraction of the neutrinos s energy that is transferred to the muon in the interaction , @xmath65 is the depth of the detector ( @xmath66 km for horizontal muons in icecube ) , and @xmath67 is the muon range which is approximately given by @xcite @xmath68 here , @xmath69 is the density of ice , @xmath70 mev @xmath71 g@xmath72 , @xmath73 mev @xmath71 g@xmath74 and we take the threshold energy to be @xmath75 gev , although lower values will likely be attained with the future inclusion of the deepcore array within the larger icecube volume @xcite .
the charged current neutrino and antineutrino - nucleon cross sections are given by @xcite @xmath76 where @xmath77 , @xmath78 , @xmath79 , @xmath80 , and the corresponding expressions for anti - neutrinos can be found by @xmath81 , @xmath82 , @xmath83 , @xmath84 . in fig .
[ fig : detrate ] we show the event rate per square kilometer of effective area as function of lkp mass for both the @xmath0 and @xmath1 lkps , and for different fractional mass splittings , @xmath12 . as expected smaller splittings and lower lkp masses
lead to larger event rates at icecube .
( top ) or @xmath1 ( bottom ) dark matter annihilations in the sun .
the six contours shown in each frame correspond to mass splittings between the lkp and the kk - quarks of @xmath850.01 , 0.02 , 0.05 , 0.1 , 0.2 and 0.3 , from top - to - bottom.,title="fig:",scaledwidth=48.0% ] ( top ) or @xmath1 ( bottom ) dark matter annihilations in the sun .
the six contours shown in each frame correspond to mass splittings between the lkp and the kk - quarks of @xmath850.01 , 0.02 , 0.05 , 0.1 , 0.2 and 0.3 , from top - to - bottom.,title="fig:",scaledwidth=48.0% ] we next turn our attention to placing constraints on the ued parameter space from the existing icecube data ( from 2007 , when icecube consisted of 22 strings of detectors ) .
icecube s current limits on wimp annihilations can be found in ref .
@xcite . rather than being given in terms of event rates ,
the limits are given in terms of the dark matter annihilation rate , or the wimp s elastic scattering cross section with nuclei ( for two choices of the annihilation channel ) . in fig .
[ fig : bounds ] , we translate these bounds into constraints on the ued parameter space .
the black solid lines denote the current icecube constraint on the ued parameter space for the case of a @xmath0 ( top ) or @xmath1 ( bottom ) lkp .
these are compared to the current constraints from the cdms and xenon10 collaborations , as calculated in ref .
@xcite . from the top frame of fig .
[ fig : bounds ] , we conclude that the current constraints from icecube are considerably more stringent than those from direct detection experiments for the case of a @xmath0 lkp . icecube s constraint in the case of a @xmath1 lkp is somewhat less stringent than those obtained from cdms or xenon10 . to estimate the ultimate reach of icecube ( consisting of a full 80 strings of detectors , and occupying approximately a cubic kilometer of instrumented volume ) ,
we have calculated the rate of neutrino - induced muon events from lkp annihilations in the sun , and compared this to the corresponding rate from atmospheric neutrinos . for this calculation , we have adopted the energy dependent effective area for icecube as described in ref .
@xcite , and the spectrum of atmospheric neutrinos given in ref .
@xcite . within a @xmath86 window around the direction of the sun
, we estimate that a ( completed ) icecube will observe 40 events per year arising from atmospheric neutrinos .
this background implies that after three years , icecube should be able to place an upper limit on the event rate from dark matter annihilations in the sun of approximately @xmath87 events per year at the 95% confidence level .
note that these rates are not per square kilometer of effective area , as are those shown in fig .
[ fig : detrate ] , but instead represent the actual rates estimated in icecube . for a 1 tev ( 100 gev )
muon , the effective area of icecube is approximately 0.75 km@xmath88 ( 0.6 km@xmath88 ) .
below @xmath89100 gev , the effective area of icecube drops rapidly . in fig .
[ fig : bounds ] , along side the current constraints from icecube , cdms and xenon10 , we show as dashed lines the projected constraint on the @xmath90 parameter space plane from the icecube experiment . for comparison
, we also show the projected constraints from the direct detection experiments super - cdms and xenon100 . from this figure
, we see that for a @xmath0 lkp , icecube will likely provide the most stringent constraint , where as in the case of a @xmath1 lkp , icecube s constraint will be comparable to or slightly weaker than those from super - cdms and xenon100 .
( top ) or the @xmath1 ( bottom ) , respectively .
the regions below the solid lines are presently excluded by the corresponding experimental limit , while the parameter space below the dashed lines is predicted to be probed by the corresponding experiment in the future.,title="fig:",scaledwidth=48.0% ] ( top ) or the @xmath1 ( bottom ) , respectively .
the regions below the solid lines are presently excluded by the corresponding experimental limit , while the parameter space below the dashed lines is predicted to be probed by the corresponding experiment in the future.,title="fig:",scaledwidth=48.0% ]
in this article , we have studied the constraints from the high energy neutrino telescope icecube on kaluza - klein dark matter within the context of models with one universal extra dimension .
we have considered two kaluza - klein dark matter candidates , the first kaluza - klein excitations of the hypercharge and isospin gauge bosons , @xmath0 and @xmath1 , and calculated the constraints on the ued parameter space from current icecube data .
we have also estimated the future reach of icecube to such dark matter candidates .
we find that in the case of @xmath0 dark matter , icecube currently provides the most stringent constraint on the dark matter mass and its mass splittings with kaluza - klein quarks . in the case of @xmath1 dark matter ,
direct detection constraints are slightly stronger than those from neutrino telescopes .
icecube is more sensitive to @xmath0 annihilations in the sun than to dark matter most other forms because of its large spin - dependent elastic scattering cross section with nuclei , and its propensity to annihilate to tau leptons or neutrinos . _
acknowledgements : tf is supported in part by the federal ministry of education and research ( bmbf ) under contract number 05h09wwe .
tf , am and kf are supported by the doe under grant de - fg02 - 95er40899 and by mctp .
dh is supported by the us department of energy , including grant de - fg02 - 95er40896 , and by nasa grant nag5 - 10842 .
_ 99 t. appelquist , h. c. cheng and b. a. dobrescu , phys .
d * 64 * ( 2001 ) 035002 [ arxiv : hep - ph/0012100 ] .
i. antoniadis , phys .
b * 246 * , 377 ( 1990 ) ; i. antoniadis , n. arkani - hamed , s. dimopoulos and g. r. dvali , phys .
b * 436 * , 257 ( 1998 ) [ arxiv : hep - ph/9804398 ] .
t. appelquist and h. u. yee , phys .
d * 67 * ( 2003 ) 055002 [ arxiv : hep - ph/0211023 ] ; i. gogoladze and c. macesanu , phys .
d * 74 * , 093012 ( 2006 ) [ arxiv : hep - ph/0605207 ] ; k. agashe , n. g. deshpande and g. h. wu , phys . lett .
b * 514 * ( 2001 ) 309 [ arxiv : hep - ph/0105084 ] ; a. j. buras , m. spranger and a. weiler , nucl .
b * 660 * ( 2003 ) 225 [ arxiv : hep - ph/0212143 ] ; a. j. buras , a. poschenrieder , m. spranger and a. weiler , nucl .
b * 678 * ( 2004 ) 455 [ arxiv : hep - ph/0306158 ] ; u. haisch and a. weiler , phys .
d * 76 * , 034014 ( 2007 ) [ arxiv : hep - ph/0703064 ] ; p. colangelo , f. de fazio , r. ferrandes and t. n. pham , phys .
d * 73 * , 115006 ( 2006 ) [ arxiv : hep - ph/0604029 ] ; phys .
d * 74 * , 115006 ( 2006 ) [ arxiv : hep - ph/0610044 ] ; t. m. aliev and m. savci , eur .
j. c * 50 * , 91 ( 2007 ) [ arxiv : hep - ph/0606225 ] .
g. r. dvali , g. gabadadze , m. kolanovic and f. nitti , phys . rev .
d * 64 * , 084004 ( 2001 ) [ arxiv : hep - ph/0102216 ] ; m. s. carena , t. m. p. tait and c. e. m. wagner , acta phys .
b * 33 * ( 2002 ) 2355 [ arxiv : hep - ph/0207056 ] ; f. del aguila , m. perez - victoria and j. santiago , jhep * 0302 * , 051 ( 2003 ) [ arxiv : hep - th/0302023 ] ; f. del aguila , m. perez - victoria and j. santiago , jhep * 0610 * , 056 ( 2006 ) [ arxiv : hep - ph/0601222 ] .
t. flacke , a. menon and d. j. phalen , arxiv:0811.1598 [ hep - ph ] .
z. ahmed _ et al .
_ [ cdms collaboration ] , phys . rev .
lett . * 102 * , 011301 ( 2009 ) [ arxiv:0802.3530 [ astro - ph ] ] . j. angle _ et al . _ [ xenon collaboration ] , phys .
lett . * 100 * , 021303 ( 2008 ) [ arxiv:0706.0039 [ astro - ph ] ] .
r. bernabei _ et al . _
[ dama collaboration ] , eur .
j. c * 56 * , 333 ( 2008 ) [ arxiv:0804.2741 [ astro - ph ] ] .
r. abbasi , arxiv:0902.2460 [ astro-ph.co ] .
m. kakizaki , s. matsumoto , y. sato and m. senami , arxiv : hep - ph/0508283 ; m. kakizaki , s. matsumoto , y. sato and m. senami , phys .
d * 71 * , 123522 ( 2005 ) [ arxiv : hep - ph/0502059 ] .
f. burnell and g. d. kribs , arxiv : hep - ph/0509118 ; k. kong and k. t. matchev , jhep * 0601 * , 038 ( 2006 ) [ arxiv : hep - ph/0509119 ] .
g. bertone , g. servant and g. sigl , phys .
d * 68 * , 044008 ( 2003 ) [ arxiv : hep - ph/0211342 ] ; l. bergstrom , t. bringmann , m. eriksson and m. gustafsson , phys . rev .
lett . * 94 * , 131301 ( 2005 ) [ arxiv : astro - ph/0410359 ] ; l. bergstrom , t. bringmann , m. eriksson and m. gustafsson , jcap * 0504 * , 004 ( 2005 ) [ arxiv : hep - ph/0412001 ] ; t. bringmann , jcap * 0508 * , 006 ( 2005 ) [ arxiv : astro - ph/0506219 ] ; a. barrau , p. salati , g. servant , f. donato , j. grain , d. maurin and r. taillet , phys . rev .
d * 72 * , 063507 ( 2005 ) [ arxiv : astro - ph/0506389 ] ; e. a. baltz and d. hooper , jcap * 0507 * , 001 ( 2005 ) [ arxiv : hep - ph/0411053 ] ; d. hooper and g. d. kribs , phys .
d * 70 * , 115004 ( 2004 ) [ arxiv : hep - ph/0406026 ] ; d. hooper and k. m. zurek , arxiv:0902.0593 [ hep - ph ] .
d. hooper and g. d. kribs , phys .
d * 67 * , 055003 ( 2003 ) [ arxiv : hep - ph/0208261 ] .
b. a. dobrescu and e. ponton , jhep * 0403 * , 071 ( 2004 ) [ arxiv : hep - th/0401032 ] ; g. burdman , b. a. dobrescu and e. ponton , jhep * 0602 * , 033 ( 2006 ) [ arxiv : hep - ph/0506334 ] ; b. a. dobrescu , d. hooper , k. kong and r. mahbubani , jcap * 0710 * , 012 ( 2007 ) [ arxiv:0706.3409 [ hep - ph ] ] ; g. bertone , c. b. jackson , g. shaughnessy , t. m. p.
tait and a. vallinotto , arxiv:0904.1442 [ astro-ph.he ] .
g. k. mallot , in _ proc . of the 19th intl .
symp . on photon and lepton interactions at high energy lp99 _ ed .
jaros and m.e .
peskin,_in the proceedings of 19th international symposium on lepton and photon interactions at high - energies ( lp 99 ) , stanford , california , 9 - 14 aug 1999 , pp 521 - 537 _ [ arxiv : hep - ex/9912040 ] ; w. m. alberico , s. m. bilenky and c. maieron , phys .
rept . * 358 * , 227 ( 2002 ) [ arxiv : hep - ph/0102269 ] ; d. r. tovey , r. j. gaitskell , p. gondolo , y. ramachers and l. roszkowski , phys .
b * 488 * , 17 ( 2000 ) [ arxiv : hep - ph/0005041 ] .
m. cirelli , n. fornengo , t. montaruli , i. sokalski , a. strumia and f. vissani , nucl .
b * 727 * , 99 ( 2005 ) [ erratum - ibid .
b * 790 * , 338 ( 2008 ) ] [ arxiv : hep - ph/0506298 ] ; http://www.marcocirelli.net/dmnu.html k. freese , phys .
b * 167 * , 295 ( 1986 ) ; a. gould , j. a. frieman and k. freese , phys . rev .
d * 39 * , 1029 ( 1989 ) ; j. lundberg and j. edsjo , phys .
d * 69 * , 123505 ( 2004 ) [ arxiv : astro - ph/0401113 ] .
v. barger , w. y. keung , g. shaughnessy and a. tregre , phys .
d * 76 * , 095008 ( 2007 ) [ arxiv:0708.1325 [ hep - ph ] ] ; r. gandhi , c. quigg , m. h. reno and i. sarcevic , phys .
d * 58 * , 093009 ( 1998 ) [ arxiv : hep - ph/9807264 ] .
m. c. gonzalez - garcia , f. halzen and m. maltoni , phys .
d * 71 * , 093010 ( 2005 ) [ arxiv : hep - ph/0502223 ] .
m. honda , t. kajita , k. kasahara , s. midorikawa and t. sanuki , phys .
d * 75 * , 043006 ( 2007 ) [ arxiv : astro - ph/0611418 ] .
|
in models with one universal extra dimension ( ued ) , the first kaluza - klein excitations of the hypercharge gauge boson , @xmath0 , and the neutral component of isospin gauge boson , @xmath1 , are each viable dark matter candidates . in either case , such particles are predicted to accumulate in the core of the sun , where they annihilate to generate a potentially observable flux of high energy neutrinos . in this article , we calculate the flux of neutrinos produced in this model and determine the constraints that can be placed on the ued parameter space from current icecube data . for the case of @xmath0 dark matter , we find that the present limits from icecube are stronger than those from direct dark matter detection experiments such as cdms and xenon10 . for @xmath1 dark matter ,
the present icecube data provides a constraint slightly weaker than direct detection experiments .
in addition , we also present the projected regions of ued parameter space that can be probed by icecube / deepcore in the near future and compare them to the prospects for future direct detection experiments .
|
matrix completion is the task to reconstruct low - rank matrices from a subset of its entries and occurs naturally in many practically relevant problems , such as missing feature imputation , multi - task learning @xcite , transductive learning @xcite , or collaborative filtering and link prediction @xcite .
almost all known methods performing matrix completion are optimization methods such as the max - norm and nuclear norm heuristics @xcite , or optspace @xcite , to name a few amongst many .
these methods have in common that in general ( a ) they reconstruct the whole matrix and ( b ) error bounds are given for all of the matrix , not single entries .
these two properties of existing methods are in particular unsatisfactory in the scenario when one is interested only in predicting ( resp . imputing ) one single missing entry or a set of interesting missing entries instead of all - which is for real data a more natural task than imputing all missing entries , in particular in the presence of large scale data ( resp .
big data ) .
indeed the design of such a method is not only desirable but also feasible , as the results of @xcite suggest by relating algebraic combinatorial properties and the low - rank setting to the reconstructability of the data .
namely , the authors provide algorithms which can decide for one entry if it can be - in principle - reconstructed or not , thus yielding a statement of trustability for the output of any algorithm . in this paper
, we demonstrate the first time how algebraic combinatorial techniques , combined with stochastic error minimization , can be applied to ( a ) reconstruct single missing entries of a matrix and ( b ) provide lower variance bounds for the error of any algorithm resp .
estimator for that particular entry - where the error bound can be obtained without actually reconstructing the entry in question . in detail ,
our contributions include : * the construction of a variance - minimal and unbiased estimator for any fixed missing entry of a rank - one - matrix , under the assumption of known noise variances * an explicit form for the variance of that estimator , being a lower bound for the variance of any unbiased estimation of any fixed missing entry and thus yielding a quantiative measure on the trustability of that entry reconstructed from any algorithm * the description of a strategy to generalize the above to any rank * comparison of the estimator with two state - of - the - art optimization algorithms ( optspace and nuclear norm ) , and error assessment of the three matrix completion methods with the variance bound note that most of the methods and algorithms presented in this paper restrict to rank one .
this is not , however , inherent in the overall scheme , which is general .
we depend on rank one only in the sense that we understand the combinatorial - algebraic structure of rank - one - matrix completion exactly , whereas the behavior in higher rank is not yet as well understood .
nonetheless , it is , in principle accessible , and , once available will can be `` plugged in '' to the results here without changing the complexity much .
in @xcite , an intricate connection between the algebraic combinatorial structure , asymptotics of graphs and analytical reconstruction bounds has been exposed .
we will refine some of the theoretical concepts presented in that paper which will allow us to construct the entry - wise estimator .
an matrix @xmath0 is called mask . if @xmath1 is a partially known matrix , then the mask of @xmath1 is the mask which has @xmath2-s in exactly the positions which are known in @xmath1 ; and @xmath3-s otherwise .
let @xmath4 be an @xmath5 mask .
we will call the unique bipartite graph @xmath6 which has @xmath4 as bipartite adjacency matrix the _ completion graph _ of @xmath4
. we will refer to the @xmath7 vertices of @xmath6 corresponding to the rows of @xmath4 as blue vertices , and to the @xmath8 vertices of @xmath6 corresponding to the columns as red vertices . if @xmath9 is an edge in @xmath10 ( where @xmath10 is the complete bipartite graph with @xmath7 blue and @xmath8 red vertices ) , we will also write @xmath11 instead of @xmath12 and for any @xmath5 matrix @xmath1 . a fundamental result , ( * ? ?
* theorem 2.3.5 ) , says that identifiability and reconstructability are , up to a null set , graph properties .
[ thm : closure ] let @xmath1 be a generic is sampled from a continuous density , then the set of non - generic @xmath1 is a null set . ] and partially known @xmath5 matrix of rank @xmath13 , let @xmath4 be the mask of @xmath1 , let @xmath14 be integers .
whether @xmath12 is reconstructible ( uniquely , or up to finite choice ) depends only on @xmath4 and the true rank @xmath13 ; in particular , it does not depend on the true @xmath1 . for rank one ,
as opposed to higher rank , the set of reconstructible entries is easily obtainable from @xmath6 by combinatorial means : [ thm : closure ] let @xmath15 be the completion graph of a partially known @xmath5 matrix @xmath1 .
then the set of uniquely reconstructible entries of @xmath1 is exactly the set @xmath11 , with @xmath16 in the transitive closure of @xmath17 . in particular , all of @xmath1 is reconstructible if and only if @xmath17 is connected .
we extend theorem [ thm : closure ] s theoretical reconstruction guarantee by describing an explicit , algebraic algorithm for actually doing the reconstruction
. this algorithm will be the basis of an entry - wise , variance - optimal estimator in the noisy case . in any rank
, such a reconstruction rule can be obtained by exposing equations which explicitly give known and unknown entries in terms of only known entries due to the fact that the set of low - rank matrices is an irreducible variety ( the common vanishing locus of finitely many polynomial equations ) .
we are able to derive the reconstruction equations for rank one .
let @xmath18 ( resp .
@xmath19 ) be a path ( resp .
cycle ) , with a fixed start and end ( resp .
traversal order ) .
we will denote by @xmath20 be the set of edges in @xmath21 ( resp . @xmath22 and @xmath23 ) traversed from blue vertex to a red one , and by @xmath24 the set of edges traversed from a red vertex to a blue one . that directs all edges from blue to red , and then taking @xmath20 to be the set of edges traversed forwards and @xmath24 the set of edges traversed backwards .
this convention is convenient notationally , but any initial orientation of @xmath10 will give us the same result . ] from now on , when we speak of `` oriented paths '' or `` oriented cycles '' , we mean with this sign convention and some fixed traversal order .
let @xmath25 be a @xmath5 matrix of rank @xmath2 , and identify the entries @xmath12 with the edges of @xmath10 .
for an oriented cycle @xmath23 , we define the polynomials @xmath26 where for negative entries of @xmath1 , we fix a branch of the complex logarithm .
[ thm : circpoly ] let @xmath25 be a generic @xmath5 matrix of rank @xmath2 .
let @xmath19 be an oriented cycle .
then , @xmath27 .
_ proof : _ the determinantal ideal of rank one is a binomial ideal generated by the @xmath28 minors of @xmath1 ( where entries of @xmath1 are considered as variables ) .
the minor equations are exactly @xmath29 , where @xmath23 is an elementary oriented four - cycle ; if @xmath23 is an elementary @xmath30-cycle , denote its edges by @xmath31 , @xmath32 , @xmath33 , @xmath34 , with @xmath35 .
let @xmath36 be the collection of the elementary @xmath30-cycles , and define @xmath37 and @xmath38 . by sending the term @xmath39 to a formal variable @xmath40
, we see that the free @xmath41-group generated by the @xmath42 is isomorphic to @xmath43 . with this equivalence , it is straightforward that , for any oriented cycle @xmath44 , @xmath45 lies in the @xmath41-span of elements of @xmath46 and , therefore , formally , @xmath47 with the @xmath48 .
thus @xmath49 vanishes when @xmath1 is rank one , since the r.h.s .
exponentiating , we see that @xmath50 if @xmath1 is generic and rank one , the r.h.s .
evaluates to one , implying that @xmath51 vanishes.@xmath52 [ cor : circpoly ] let @xmath25 be a @xmath5 matrix of rank @xmath2 .
let @xmath53 be two vertices in @xmath10 .
let @xmath54 be two oriented paths in @xmath10 starting at @xmath55 and ending at @xmath56 .
then , for all @xmath1 , it holds that @xmath57 .
it is possible to prove that the set of @xmath58 forms the set of polynomials vanishing on the entries of @xmath1 which is minimal with respect to certain properties .
namely , the @xmath58 form a universal grbner basis for the determinantal ideal of rank @xmath2 , which implies the converse of theorem [ thm : circpoly ] . from this , one can deduce that the estimators presented in section [ sec : est.estim ] are variance - minimal amongst all unbiased ones .
in this section , we will construct an estimator for matrix completion which ( a ) is able to complete single missing entries and ( b ) gives universal error estimates for that entry that are independent of the reconstruction algorithm . in all of the following
, we will assume that the observations arise from the following sampling process : [ ass : sampling ] there is an unknown fixed , rank one , matrix @xmath1 which is generic , and an @xmath5 mask @xmath0 which is known .
there is a ( stochastic ) noise matrix @xmath59 whose entries are uncorrelated and which is multiplicatively centered with finite variance , non - zero . ]
variance ; i.e. , @xmath60 and @xmath61 for all @xmath62 and @xmath63 .
the observed data is the matrix @xmath64 , where @xmath65 denotes the hadamard ( i.e. , component - wise ) product .
that is , the observation is a matrix with entries @xmath66 .
the assumption of multiplicative noise is a necessary precaution in order for the presented estimator ( and in fact , any estimator ) for the missing entries to have bounded variance , as shown in example [ ex : variance ] below .
this is not , in practice , a restriction since an infinitesimal additive error @xmath67 on an entry of @xmath1 is equivalent to an infinitesimal multiplicative error @xmath68 , and additive variances can be directly translated into multiplicative variances if the density function for the noise is known ; this adds an independent combinatorial problem for the estimation of the sign which can be done by maximum likelihood . in order to keep the exposition short and easy
, we did not include this into the exposition . ] .
the previous observation implies that the multiplicative noise model is as powerful as any additive one that allows bounded variance estimates .
[ ex : variance ] consider the rank one matrix @xmath69 the unique equation between the entries is @xmath70 solving for any entry will have another entry in the denominator , for example @xmath71 thus we get an estimator for @xmath72 when substituting observed and noisy entries for @xmath73 .
when @xmath74 approaches zero , the estimation error for @xmath72 approaches infinity . in particular ,
if the density function of the error @xmath75 of @xmath74 is too dense around the value @xmath76 , then the estimate for @xmath72 given by the equation will have unbounded variance . in such a case , one can show that no estimator for @xmath72 has bounded variance . in this section ,
we construct the unbiased estimator for the entries of a rank - one - matrix with minimal variance .
first , we define some notation to ease the exposition : we will denote by @xmath77 and @xmath78 the logarithmic entries and noise . thus , for some path @xmath21 in @xmath10 we obtain @xmath79 denote by @xmath80 the logarithmic ( observed ) entries , and @xmath81 the ( incomplete ) matrix which has the ( observed ) @xmath82 as entries .
denote by @xmath83 the components of the estimator will be built from the @xmath84 : [ lem : explpb ] let @xmath85 be the graph of the mask @xmath4 .
let @xmath86 be any edge with @xmath55 red .
let @xmath21 be an oriented path , then @xmath21 can also be the path consisting of the single edge @xmath16 . ] in @xmath6 starting at @xmath55 and ending at @xmath56 .
then , @xmath87 is an unbiased estimator for @xmath88 with variance @xmath89 _ proof : _ by linearity of expectation and centeredness of @xmath90 , it follows that @xmath91 thus @xmath92 is unbiased .
since the @xmath93 are uncorrelated , the @xmath94 also are ; thus , by bienaym s formula , we obtain @xmath95 and the statement follows from the definition of @xmath96 + in the following , we will consider the following parametric estimator as a candidate for estimating @xmath97 : [ not : calpxalpha ] fix an edge @xmath86 .
let @xmath98 be a basis for the set of all oriented paths starting at @xmath55 and ending at @xmath56 , in the free abelian group generated by the @xmath40 , subject to the relations @xmath99 for all cycles @xmath23 in @xmath100 .
independence can be taken as linear independence of the coefficient vectors of the @xmath101 . ] and denote @xmath102 by @xmath103 .
for @xmath104 , set @xmath105 furthermore , we will denote by @xmath106 the @xmath8-vector of ones .
the following lemma follows immediately from lemma [ lem : explpb ] and theorem [ thm : circpoly ] : @xmath107 in particular , @xmath108 is an unbiased estimator for @xmath109 if and only if @xmath110 .
we will now show that minimizing the variance of @xmath108 can be formulated as a quadratic program with coefficients entirely determined by @xmath88 , the measurements @xmath94 and the graph @xmath6 .
in particular , we will expose an explicit formula for the @xmath111 minimizing the variance . before stating the theorem
, we define a suitable kernel : let @xmath112 be an edge . for an edge @xmath16 and a path @xmath21 , set @xmath113 if @xmath114 otherwise @xmath115 let @xmath116 be any fixed oriented paths .
define the ( weighted ) _ path kernel _
@xmath117 by @xmath118 under our assumption that @xmath119 for all @xmath112 , the path kernel is positive definite , since it is a sum of @xmath103 independent positive semi - definite functions ; in particular , its kernel matrix has full rank .
here is the variance - minimizing unbiased estimator : [ prop : varest ] let @xmath120 be a pair of vertices , and @xmath98 a basis for the @xmath121@xmath122 path space in @xmath17 with @xmath103 elements .
let @xmath123 be the @xmath124 kernel matrix of the path kernel with respect to the basis @xmath98 .
for any @xmath104 , @xmath125 moreover , under the condition @xmath126 , the variance @xmath127 is minimized by @xmath128 _ proof : _ by inserting definitions , we obtain @xmath129 writing @xmath130 as vectors , and @xmath131 as matrices , we obtain @xmath132 by using that @xmath133 for any scalar @xmath134 , and independence of the @xmath94 , an elementary calculation yields @xmath135 in order to determine the minimum of the variance in @xmath111 , consider the lagrangian @xmath136 where the slack term models the condition @xmath137 .
an elementary calculation yields @xmath138 where @xmath106 is the vector of ones . due to positive definiteness of @xmath123
the function @xmath127 is convex , thus @xmath139 will be the unique @xmath111 minimizing the variance while satisfying @xmath126.@xmath52 the above setup works in wider generality : ( i ) if @xmath140 is allowed and there is an @xmath121@xmath122 path of all zero variance edges , the path kernel becomes positive semi - definite ; ( ii ) similarly if @xmath141 is replaced with any set of paths at all , the same may occur . in both cases
, we may replace @xmath142 with the moore - penrose pseudo - inverse and the proposition still holds : ( i ) reduces to the exact reconstruction case of theorem [ thm : closure ] ; ( ii ) produces the optimal estimator with respect to @xmath141 , which is optimal provided that @xmath141 is spanning , and adding paths to @xmath141 does not make the estimate worse .
an estimator for rank @xmath143 and higher , together with a variance analysis , can be constructed similarly once all polynomials known which relate the entries under each other .
the main difficulty lies in the fact that these polynomials are not parameterized by cycles anymore , but specific subgraphs of @xmath6 , see ( * ? ? ?
* section 2.5 ) . were these polynomials known , an estimator similar to @xmath108 as in notation [ not : calpxalpha ] could be constructed , and a subsequent variance ( resp .
perturbation ) analysis performed . in this section ,
we describe the two main algorithms which calculate the variance - minimizing estimate @xmath144 for any fixed entry @xmath12 of an @xmath5 matrix @xmath1 , which is observed with noise , and the variance bound for the estimate @xmath144 .
it is important to note that @xmath12 does not necessarily need to be an entry which is missing in the observation , it can also be any entry which has been observed . in the latter case ,
algorithm [ alg : xalpha ] will give an improved estimate of the observed entry , and algorithm [ alg : var ] will give the trustworthiness bound on this estimate .
since the the path matrix @xmath23 , the path kernel matrix @xmath123 , and the optimal @xmath111 is required for both , we first describe algorithm [ alg : alpha ] which determines those . find a linearly independent set of paths @xmath98 in the graph @xmath6 , starting from @xmath62 and ending at @xmath63.[alg : alpha.step1 ]
determine the matrix @xmath145 with @xmath146 set @xmath147 if @xmath114 , otherwise @xmath115 define a diagonal matrix @xmath148 with @xmath149 for @xmath150 compute the kernel matrix @xmath151 calculate @xmath152 output @xmath153 and @xmath111 .
the steps of the algorithm follow the exposition in section [ sec : est.estim ] , correctness follows from the statements presented there .
the only task in algorithm [ alg : alpha ] that is nt straightforward is the computation of a linearly independent set of paths in step [ alg : alpha.step1 ] .
we can do this time linear in the number of observed entries in the mask @xmath4 with the following method .
to keep the notational manageable , we will conflate formal sums of the @xmath40 , cycles in @xmath154 and their representations as vectors in @xmath155 , since there is no chance of confusion .
if @xmath156 is not an edge of @xmath4 , and @xmath62 and @xmath63 are in different connected components , then @xmath98 is empty .
output @xmath157 .
otherwise , if @xmath156 is not an edge , of @xmath4 , add a `` dummy '' copy .
compute a spanning forest @xmath158 of @xmath4 that does not contain @xmath156 , if possible . for each edge
@xmath159 , compute the fundamental cycle @xmath160 of @xmath16 in @xmath158 .
if @xmath156 is an edge in @xmath4 , output @xmath161 . otherwise , let @xmath162 .
output @xmath163 .
we prove the correctness of algorithm [ alg : calp ] .
algorithms [ alg : xalpha ] and [ alg : var ] then can make use of the calculated @xmath164 to determine an estimate for any entry @xmath12 and its minimum variance bound .
the algorithms follow the exposition in section [ sec : est.estim ] , from where correctness follows ; algorithm [ alg : xalpha ] additionally provides treatment for the sign of the entries .
calculate @xmath23 and @xmath111 with algorithm [ alg : alpha ] .
store @xmath81 as a vector @xmath165 and a sign vector @xmath166 with @xmath167 .
calculate @xmath168 the sign is @xmath169 if each column of @xmath170 ( @xmath171 component - wise ) contains an odd number of entries @xmath172 , else @xmath173 .
return @xmath174 calculate @xmath123 and @xmath111 with algorithm [ alg : alpha ] .
return @xmath175 .
note that even if observations are not available , algorithm [ alg : var ] can be used to obtain the variance bound .
the variance bound is relative , due to its multiplicativity , and can be used to approximate absolute bounds when any reconstruction estimate @xmath144 is available - which does not necessarily need to be the one from algorithm [ alg : xalpha ] , but can be the estimation result of any reconstruction .
namely , if @xmath176 is the estimated variance of the log , we obtain an upper confidence bound ( resp .
deviation ) bound @xmath177 for @xmath178 and a lower confidence bound ( resp .
deviation ) bound @xmath179 , corresponding to the log - confidence @xmath180 also note that if @xmath12 is not reconstructible from the mask @xmath4 ( i.e. , if the edge @xmath156 is not in the transitive closure of @xmath6 , see theorem [ thm : closure ] ) , then the deviation bounds will be infinite .
for three different masks , we calculated the predicted minimum variance for each entry of the mask .
the multiplicative noise was assumed to be @xmath181 for each entry .
figure [ fig : heatmaps ] shows the predicted a - priori minimum variances for each of the masks .
notice how the structure of the mask affects the expected error ; known entries generally have least variance , while it is interesting to note that in general it is less than the starting variance of @xmath2 .
i.e. , tracking back through the paths can be successfully used even to denoise known entries .
the particular structure of the mask is mirrored in the pattern of the predicted errors ; a diffuse mask gives a similar error on each missing entry , while the more structured masks have structured error which is determined by combinatorial properties of the completion graph and the paths therein .
+ + we generated @xmath182 random mask of size @xmath183 with @xmath184 entries sampled uniformly and a random @xmath185 matrix of rank one .
the multiplicative noise was chosen entry - wise independent , with variance @xmath186 for each entry .
figure [ fig:3algs.noisemse ] compares the mean squared error ( mse ) for three algorithms : nuclear norm ( using the implementation @xcite ) , optspace @xcite , and algorithm [ alg : xalpha ] .
it can be seen that on this particular mask , algorithm [ alg : xalpha ] is competitive with the other methods and even outperforms them for low noise .
the data are the same as in section [ sec : noise ] , as are the compared algorithm .
figure [ fig:3algs.linearity ] compares the error of each of the methods with the variance predicted by algorithm [ alg : var ] each time the noise level changed .
the figure shows that for any of the algorithms , the mean of the actual error increases with the predicted error , showing that the error estimate is useful for a - priori prediction of the actual error - independently of the particular algorithm .
note that by construction of the data this statement holds in particular for entry - wise predictions .
furthermore , in quantitative comparison algorithm [ alg : var ] also outperforms the other two in each of the bins .
in this paper , we have introduced an algebraic combinatorics based method for reconstructing and denoising single entries of an incomplete and noisy matrix , and for calculating confidence bounds of single entry estimations for arbitrary algorithms .
we have evaluated these methods against state - of - the art matrix completion methods .
the results of section [ sec : exp ] show that our reconstruction method is competitive and that - for the first time - our variance estimate provides a reliable prediction of the error on each single entry which is an a - priori estimate , i.e. , depending only on the noise model and the position of the known entries .
furthermore , our method allows to obtain the reconstruction and the error estimate for a single entry which existing methods are not capable of , possibly using only a small subset of neighboring entries - a property which makes our method unique and particularly attractive for application to large scale data .
we thus argue that the investigation of the algebraic combinatorial properties of matrix completion , in particular in rank @xmath143 and higher where these are not yet completely understood , is crucial for the future understanding and practical treatment of big data .
9 [ 1]#1 [ 1]`#1 ` urlstyle[1]doi : # 1 e. acar , d.m .
dunlavy , and t.g .
link prediction on evolving data using matrix and tensor factorizations . in _ data mining workshops , 2009 .
ieee international conference on _ , pages 262269 .
ieee , 2009 .
a. argyriou , c. a. micchelli , m. pontil , and y. ying .
a spectral regularization framework for multi - task structure learning . in j.c .
platt , d. koller , y. singer , and s. roweis , editors , _ advances in nips 20 _ , pages 2532 .
mit press , cambridge , ma , 2008 .
emmanuel j. cands and benjamin recht .
exact matrix completion via convex optimization . _ found .
comput . math .
_ , 90 ( 6):0 717772 , 2009 .
issn 1615 - 3375 .
doi : 10.1007/s10208 - 009 - 9045 - 5 .
url http://dx.doi.org/10.1007/s10208-009-9045-5 .
a. goldberg , x. zhu , b. recht , j. xu , and r. nowak .
transduction with matrix completion : three birds with one stone . in j.
lafferty , c. k. i. williams , j. shawe - taylor , r.s .
zemel , and a. culotta , editors , _ advances in neural information processing systems 23 _ , pages 757765 . 2010 .
raghunandan h. keshavan , andrea montanari , and sewoong oh . matrix completion from a few entries .
_ ieee trans .
inform . theory _
, 560 ( 6):0 29802998 , 2010 .
issn 0018 - 9448 .
doi : 10.1109/tit.2010.2046205 .
url http://dx.doi.org/10.1109/tit.2010.2046205 .
franz j. kirly , louis theran , ryota tomioka , and takeaki uno . the algebraic combinatorial approach for low - rank matrix completion .
preprint , arxiv:1211.4116v3 , 2012 .
url http://arxiv.org/abs/1211.4116 .
a. menon and c. elkan .
link prediction via matrix factorization .
_ machine learning and knowledge discovery in databases _ , pages 437452 , 2011 .
n. srebro , j. d. m. rennie , and t. s. jaakkola .
maximum - margin matrix factorization . in lawrence
k. saul , yair weiss , and lon bottou , editors , _ advances in nips 17 _ , pages 13291336 . mit press , cambridge , ma , 2005 .
ryota tomioka , kohei hayashi , and hisashi kashima . on the extension of trace norm to tensors . in
_ nips workshop on tensors , kernels , and machine learning _ , 2010 .
we adopt the conventions of section [ sec : theory ] , so that @xmath17 is a bipartite graph with @xmath7 blue vertices , @xmath8 red ones , and @xmath16 edges oriented from blue to red .
recall the isomorphism , observed in the proof of theorem [ thm : circpoly ] of the @xmath187-group of the polynomials @xmath188 and the oriented cycle space @xmath154 .
define @xmath189 ( the first betti number of the graph ) .
some standard facts are that : ( i ) the rank of @xmath154 is @xmath190 ; ( ii ) we can obtain a basis for @xmath154 consisting only of simple cycles by picking any spanning forest @xmath158 of @xmath17 and then using as basis elements the fundamental cycles @xmath160 of the edges @xmath191 .
this justifies step 4 .
let @xmath156 be an edge of @xmath17 .
define an @xmath62@xmath63 to be the set of subgraphs such that , for generic rank one @xmath1 , @xmath192 . by theorem [ thm : circpoly ]
, we can write these as @xmath187-linear combinations of @xmath193 and oriented cycles . from this , we see that the rank of the path space is @xmath194 and the graph theoretic identification of elements in the path space with subgraphs that have even degree at every vertex except @xmath62 and @xmath63 . thus , if @xmath156 is an edge of @xmath17 , step 5 is justified , completing the proof of correctness in this case . if @xmath156 was not an edge , step 1 guarantees that the dummy copy of @xmath156 that we added is not in the spanning tree computed in step 3 .
thus , the element @xmath162 computed in step 6 is a simple path from @xmath62 to @xmath63 .
the collection of elements generated in step 6 is independent by the same fact in @xmath195 and has rank @xmath196 and does not put a positive coefficient on the dummy generator @xmath193 . @xmath52
|
we propose a general framework for reconstructing and denoising single entries of incomplete and noisy entries .
we describe : effective algorithms for deciding if and entry can be reconstructed and , if so , for reconstructing and denoising it ; and a priori bounds on the error of each entry , individually . in the noiseless case
our algorithm is exact . for rank - one matrices ,
the new algorithm is fast , admits a highly - parallel implementation , and produces an error minimizing estimate that is qualitatively close to our theoretical and the state - of - the - are nuclear norm and optspace methods .
= 1
|
the riemann zeta - function @xmath2 is defined by the dirichlet series @xmath3 for @xmath4 , @xmath5 and @xmath6 .
the functional equation of @xmath2 is given by @xmath7 where @xmath8 this allows us to perform a meromorphic continuation to the whole complex plane except at @xmath9 where @xmath2 has a simple pole with residue equal to @xmath10 .
the connection with number theory comes from the euler product @xmath11 for @xmath12 , and where the product is taken over all the primes @xmath13 .
it is well - known from riemann and from von mangoldt that the non - trivial zeros @xmath14 of @xmath2 are located inside the critical strip @xmath15 .
moreover , if @xmath16 denotes the number of such zeros up to height @xmath17 then @xmath18 where @xmath19 as @xmath20 , see e.g. @xcite for properties of @xmath2 . to state the results , we let @xmath21 denote the number of non - trivial zeros up to height @xmath22 such that @xmath23 . similarly , let @xmath24 denote the number such zeros which are also simple .
we then define @xmath25 the history behind the value of @xmath26 can be found in @xcite .
the main breakthroughs were as follows . in 1942
, selberg @xcite established that @xmath27 .
levinson later showed in 1974 that @xmath28 .
this was improved by conrey to @xmath29 in 1989 and later refined by bui , conrey and young @xcite to @xmath30 , and shortly afterward by feng @xcite to @xmath31 .
it should be noted that both results are improvements of @xmath29 and are independent of each other .
+ the second author , roy and zaharescu @xcite as well as bui @xcite brought up a point regarding the strength of feng s result . in @xcite , it was explained that @xmath32 , unconditionally , using feng s mollifier . however , the computation of the mixed terms of the mollifiers of conrey and of feng was not carried through explicitly .
+ it should also be remarked that bui @xcite suggests that the bound obtained in this paper can be attained using the twisted second moment of the riemann zeta - function due to balasubramanien _ et .
_ @xcite and that he also suggests an alternative argument that could lead to the bound @xmath33 .
+ in this paper , we close this gap and we explain feng s brilliant choice in the context of the powerful technology developed in @xcite .
these ideas come from the ratios conjectures of @xmath1-functions due to conrey , farmer and zirnbaeuer @xcite as well as to conrey and snaith @xcite .
it should be noted that feng s methodology to obtain the main terms of his theorem consisted on an ingenious combination of elementary methods , namely induction and mertens formula , applied to conrey s result @xcite . on the other hand , this choice of methods blurred a bit the length the mollifier was allowed to take .
other than choosing the same mollifier , our computations do not overlap and the methods are quite different . + lastly , the closing of this gap will clarify the situation of the percentage of non - trivial zeros on the critical line when one attaches feng s second - piece mollifier to conrey s .
let @xmath34 be a real polynomial satisfying @xmath35 , @xmath36 , and define @xmath37 where for large @xmath38 , @xmath39 if @xmath40 is a mollifier , then it is well - known from the work of levinson @xcite and of conrey @xcite that littlewood s lemma ( * ? ? ?
* @xmath419.9 ) followed by the arithmetic and geometric mean inequalities yields @xmath42 where @xmath43 , and @xmath44 is a bounded positive real number to be chosen later . following feng @xcite , we will choose a mollifier of the form @xmath45 where @xmath46 is the mollifier considered by conrey .
let @xmath47 be a certain polynomial satisfying @xmath48 , @xmath49 , and let @xmath50 where @xmath51 .
we adopt the notation @xmath52 = p_1 \bigg ( \frac{\log(y_1/n)}{\log y_1 } \bigg)\ ] ] for @xmath53 . by convention , we set @xmath54=0 $ ] for @xmath55
. then @xmath56 is given by @xmath57,\end{aligned}\ ] ] where @xmath58 is the mbius function .
for the second mollifier , we take @xmath59}. \end{aligned}\ ] ] here @xmath60 is an integer of our choice and @xmath61 are distinct primes .
also we need @xmath62 for @xmath63 . in this case @xmath64 where @xmath65 .
it will become clear in the calculation of the crossterm integral between @xmath46 and @xmath66 that one needs @xmath67 .
therefore , if @xmath68 increases , then @xmath69 decreases unless some difficult work is done to push @xmath69 back to its original ( or higher ) value .
see the comments between theorem [ theorem1212 ] and theorem [ theorem4737 ] for more details .
the reason behind this choice is that feng wishes to mollify not only @xmath2 but also @xmath70 , which is the second term coming from .
this is accomplished by looking at @xmath71 when @xmath72 is a square - free positive integer , then one has @xmath73 where @xmath74 denotes the dirichlet convolution of arithmetic functions @xmath75 and @xmath76 . here
@xmath77 stands for convolving the von mangoldt function @xmath78 with itself exactly @xmath79 times .
if @xmath72 contains a square divisor , then , as remarked by feng @xcite , the coefficients @xmath80 resulting from contribute a lower order to the mean value integrals @xmath81 , @xmath82 and @xmath83 related to @xmath26 in ( see below for exact definitions of these @xmath84-integrals ) .
we will prove the following .
[ theorem1212 ] we obtain with @xmath85 @xmath86 unconditionally . using the work of iwaniec and deshouillers @xcite , conrey @xcite was able to push the size of the mollifier @xmath46 to @xmath87 . in the light of lemma [ lemmaafe ] and below , we require @xmath88 in our argumentation .
the points brought up in @xcite and @xcite show that some difficult work is needed if one takes @xmath89 .
theorem [ theorem1212 ] utilizes @xmath90 .
however , if we take @xmath87 and @xmath91 , then we get [ theorem4737 ] we obtain with @xmath87 and @xmath91 @xmath92 unconditionally .
it should therefore be stressed that theorem [ theorem1212 ] is an improvement of the last theorem to ever use @xmath93 , namely the first corollary of @xcite , where it was shown that @xmath94 . + the method sketched in @xcite to deal with multiple piece mollifiers carries through and our main result is as follows .
[ unsmoothed ] suppose that @xmath95 with @xmath96 and @xmath97 and @xmath98 small .
then @xmath99 where @xmath100 and the @xmath101 are given by , and .
we use ` mathematica ` to numerically evaluate @xmath102 with the following choices of parameters . with @xmath103 , @xmath104 , @xmath105
we have @xmath106 . to get @xmath107 , we take @xmath103 , @xmath108 , @xmath109 we also use ` mathematica ` to numerically evaluate @xmath110 with the following choices of parameters . with @xmath103 , @xmath111 , @xmath112
we have @xmath113 . to get @xmath114
, we take @xmath103 , @xmath115 , @xmath116 an interesting question to ask is : what would have happened if feng had published his mollifier before conrey s increment of @xmath68 from @xmath117 to @xmath118 .
since this has not been remarked before in the literature , we take the chance to answer it .
if @xmath46 and @xmath66 are kept at @xmath119 , then feng s piece adds an additional @xmath120 to conrey s @xmath121 as shown in the table below . .%
according to sizes of @xmath122 [ cols="^,^,^,^,^ " , ] since @xmath66 is the perturbation of @xmath46 , it behooves us to take @xmath68 as large as possible ( @xmath118 ) at the cost of sacrificing @xmath69 to @xmath123 which only adds @xmath124 .
the idea of smoothing the mean value integrals was introduced in @xcite and it helps substantially in our calculations .
let @xmath125 be a smooth function satisfying the following properties : 1 .
@xmath126 for all @xmath6 , 2 .
@xmath127 has compact support in @xmath128 $ ] , 3 .
@xmath129 , for each @xmath130 and where @xmath131 .
this allows us to re - write theorem [ unsmoothed ] as follows .
[ smoothed ] suppose that @xmath93 and @xmath132 for @xmath133 small . for any @xmath127 satisfying conditions ( a ) , ( b ) and ( c ) and @xmath134 , @xmath135 uniformly for @xmath136 , where @xmath100 and the @xmath101 are given by , and . how to deal with a two - piece mollifier was explained in @xcite . in @xcite
a 4-piece mollifier was studied .
the idea is to open the square in the integrand to get @xmath137 we will compute these integrals in the next sections .
the integral @xmath82 is asymptotically real , thus @xmath138 follows from @xmath82 , i.e. @xmath139 . the main terms coming from integrals @xmath81 , @xmath82 and @xmath83 are now stated as theorems .
[ c11theorem ] suppose @xmath140 .
then @xmath141 uniformly for @xmath136 , where @xmath142 let @xmath143 denote the pochhammer symbol . [ c12theorem ] suppose @xmath144 where @xmath140 and @xmath145 .
then @xmath146 uniformly for @xmath136 , where @xmath147 .
\end{aligned}\ ] ] [ c22theorem ] suppose @xmath145 .
then @xmath148 uniformly for @xmath136 , where @xmath149 .
\end{aligned}\ ] ] note that in @xcite , @xmath150 , @xmath151 and @xmath152 are all mixed into one single theorem and it is not immediately clear how to separate each individual @xmath153-term . the smoothing argument is helpful because we can easily deduce theorem [ unsmoothed ] from theorem [ smoothed ] and so on . by having chosen @xmath125 to satisfy conditions ( a ) , ( b ) and ( c ) and in addition to being an upper bound for the characteristic function of the interval @xmath154 $ ] , and with support @xmath155 $ ] ,
we get @xmath156 note that @xmath157 .
we similarly get a lower bound . summing over dyadic segments
gives the full result . rather than working directly with @xmath160
, we shall instead consider the following three general shifted integrals @xmath161 the computation is now reduced to proving the following three lemmas .
[ lemmac11 ] we have @xmath162 uniformly for @xmath163 , where @xmath164.\end{aligned}\ ] ] [ lemmac12 ] we have @xmath165 uniformly for @xmath163 , where @xmath166.\end{aligned}\ ] ] [ lemmac22 ] we have @xmath167 uniformly for @xmath163 , where @xmath168.\end{aligned}\ ] ] to get theorems [ c11theorem ] , [ c12theorem ] and [ c22theorem ] we use the following technique .
let @xmath169 denote either of the integrals in questions , and note that @xmath170 since @xmath171 and @xmath172 are holomorphic with respect to @xmath173 small , the derivatives appearing in the equation above can be obtained as integrals of radii @xmath174 around the points @xmath175 , using cauchy s integral formula .
since the error terms hold uniformly on these contours , the same error terms that hold for @xmath171 also hold for @xmath169 .
that the above differential operator on @xmath172 does indeed give @xmath176 follows from @xmath177 note that from the above equation we get @xmath178 as well as @xmath179 hence using the differential operators @xmath180 and @xmath181 on the last line of @xmath182 we get @xmath183 .
\nonumber\end{aligned}\ ] ] theorem [ c12theorem ] then follows by setting @xmath184 and using @xmath185 .
similarly , when we use the differential operators @xmath180 and @xmath181 on the last line of @xmath186 it becomes @xmath187.\ ] ] the same substitutions yield theorem [ c22theorem ] .
the following results are needed throughout the paper .
[ lemmaafe ] suppose that @xmath125 satisfies conditions ( a ) , ( b ) and ( c ) and that @xmath188 , @xmath72 are positive integers with @xmath189 with @xmath190 , and @xmath163 . moreover , set @xmath191 as well as @xmath192 then one has @xmath193 where @xmath194 see lemma 5 of @xcite .
they key point is that non - diagonal terms @xmath195 can safely be absorbed in the error terms .
[ integralwithlogderivative ] suppose @xmath196 , @xmath197 and @xmath198 . for some @xmath199
we have @xmath200 where @xmath201 and the contour is a circle of radius one enclosing the origin and @xmath202 .
this follows a similar procedure to lemma 6.1 of @xcite where the zero - free region of @xmath203 is used .
let @xmath204 be a large parameter to be chosen later .
by cauchy s theorem , @xmath205 is equal to the sum of residues at @xmath206 and @xmath207 plus integrals over the line segments @xmath208 , @xmath209 , and @xmath210 , where @xmath153 is some fixed positive positive constant such that @xmath211 has no zeros in the region on the right - hand side of the contour determined by the @xmath212 s . in the proof of lemma [ integralwithlogderivative].,title="fig : " ]
( -61,92 ) ( -47,188 ) ( -47,30 ) ( -149,118 ) another requirement on @xmath153 is that the estimate ( see ( * ? ? ?
* theorem 3.11 ) ) @xmath213 holds in this region and @xmath214 ( see ( * ? ? ? * theorem 6.7 ) ) .
then , one has @xmath215 since @xmath216 . moreover , since @xmath217 @xmath218 and finally @xmath219 appropriately choosing @xmath220 gives an error of size @xmath221 .
the next step is to sum the residues .
this sum can now be expressed as @xmath222 where the contour is now a small circle @xmath223 of radius @xmath224 around the origin such that @xmath225 .
since the radius of the circle is tending to zero , we can use the laurent expansions @xmath226 to finally obtain @xmath227 using the binomial theorem and a direct estimate gives , we get that the above is equal to @xmath228 which is the desired main term of the lemma .
this integral can be computed exactly . to do this , note that for any integer @xmath229 , one has @xmath230 hence , one arrives at and where we temporarily set @xmath231 @xmath232 by cauchy s integral theorem . when computing the crossterm of @xmath46 and @xmath66 the following result will be needed .
this generalizes ( * ? ? ?
* lemma 8) which is the particular case @xmath233 . [ lemmacombinatorics ] for @xmath234 and @xmath235 square - free
, we have @xmath236 here the @xmath13 s , the @xmath237 s and the @xmath238 s are all distinct primes . we may write @xmath239 using the definition of the binomial coefficient completes the proof .
recall that for a positive integer @xmath72 , the generalized von mangoldt function @xmath240 is defined @xcite by the dirichlet convolution @xmath241 so that @xmath242 .
the generating series is @xmath243 for @xmath12 and here @xmath244 stands for the @xmath72-th derivative of @xmath203 with respect to @xmath245 . by looking at @xmath246 for @xmath12 ,
we see that @xmath247 and in particular for @xmath248 @xmath249 [ lemma1pk ] we have for smooth functions @xmath250 and @xmath251 in the interval @xmath252 $ ] , @xmath253 , and @xmath254 @xmath255 start by setting @xmath256 by applying the abel summation formula , one gets @xmath257 since @xmath258 for @xmath259 by the prime number theorem with remainder , see e.g. @xcite . [ lemma36rrz01 ] we have for smooth functions @xmath250 and @xmath251 in the interval @xmath252 $ ] , @xmath253 , and @xmath254 @xmath260 where @xmath261 denotes the number of ways an integer @xmath262 can be written as a product of @xmath263 fixed factors .
note that @xmath264 and @xmath265 , the number of divisors of @xmath262 .
this can be proved by using induction over @xmath79 and euler - maclaurin summation .
one starts with @xmath266 and then uses ( * ? ? ?
* lemma 4.4 ) .
the exact details can be found in ( * ? ? ?
* lemma 3.6 ) .
[ lemma2pk ] we have for smooth functions @xmath250 and @xmath251 in the interval @xmath252 $ ] , @xmath253 , and @xmath254 @xmath267 same as in the beginning of the proof of lemma [ lemma36rrz01 ] but instead we use lemma [ lemma1pk ] . [ lemmaem2lambdas ] we have for smooth functions @xmath250 and @xmath251 in the interval @xmath252 $ ] , @xmath253 , and @xmath254 @xmath268 this follows by induction on @xmath269 and by using lemma [ lemma2pk ] combined with .
although this was already explained in @xcite , the mean value integral @xmath271 builds up from @xmath272 which in turn is a refinement of @xmath273 .
therefore , careful analysis will repay itself by going over the main points of the evaluation of @xmath273 briefly .
for our purposes , we shall illustrate this for @xmath274 ; however , in @xcite it was shown that one could take @xmath275 .
we start by inserting the definition of the mollifier @xmath46 in @xmath81 so that @xmath276 according to lemma [ lemmaafe ] , we write @xmath277 , where @xmath278 is given by @xmath279 notice that @xmath280 is obtained by replacing @xmath158 with @xmath202 , @xmath159 with @xmath281 and multiplying inside the integrand by @xmath282 . in other words , @xmath283 let us then look at @xmath278 more closely . using the mellin representations @xmath284
= \sum\limits_i { \frac{{{a_i}i!}}{{{{\log } ^i}{y_1 } } } } \frac{1}{{2\pi i}}\int_{(1 ) } { { { \left ( { \frac{{{y_1}}}{h } } \right)}^s}\frac{{ds}}{{{s^{i + 1 } } } } } \quad \textnormal{and } \quad { p_1}[k ] = \sum\limits_j { \frac{{{a_j}j!}}{{{{\log } ^j}{y_1 } } } } \frac{1}{{2\pi i}}\int_{(1 ) } { { { \left ( { \frac{{{y_1}}}{k } } \right)}^u}\frac{{du}}{{{u^{j + 1 } } } } } , \ ] ] we then get @xmath285 we now evaluate the arithmetical sum @xmath286 in the integrand .
this is done @xmath13-adically as follows .
we denote by @xmath287 the number of times the prime number @xmath13 appears in @xmath262 , and without risk of confusion we write @xmath288 .
this means that @xmath289 where the arithmetical factor @xmath290 is given by an absolutely convergent euler product in some product of half - planes containing the origin
. it will be important to remark that when @xmath291 and @xmath292 we have @xmath293 for all @xmath294 , by the mbius inversion formula . inserting this into @xmath278 we get @xmath295
now we deform the path of integration to @xmath296 where @xmath297 small and @xmath298 . by doing this
, we pick up a simple pole coming from @xmath299 at @xmath300 only , since @xmath301 vanishes at the pole of @xmath302 .
the new path of integration with respect to @xmath294 contributes an error of the size @xmath303 thus , we end up with @xmath304 where @xmath305 using the dirichlet series representation for @xmath306 , we can separate the complex variables @xmath245 and @xmath307 . the next step is to use the laurent expansion @xmath308 since @xmath309 for all @xmath294 , in particular for @xmath300 . by the use of lemma [ integralwithlogderivative ]
, we can deform the line integrals into contour integrals around circles of radius @xmath10 around the origin .
thus , @xmath310 these integrals can be computed by the use of , so that @xmath311 let us note that @xmath312 now sum over @xmath313 to get @xmath314 = \sum\limits_i { { a_i}{{\left ( { x + \frac{{\log ( { y_1}/n)}}{{\log { y_1 } } } } \right)}^i}}\ ] ] and similarly over @xmath315 so that @xmath316 + o(t / l ) \nonumber \\ & = \frac{{\widehat w(0)}}{{(\alpha + \beta ) { { \log } ^2}{y_1}}}\frac{{{d^2}}}{{dxdy}}\bigg[y_1^{\alpha x + \beta y } \nonumber \\ & \quad \times \sum\limits_{n \leqslant { y_1 } } { \frac{1}{n}p\left ( { x + \frac{{\log ( { y_1}/n)}}{{\log { y_1 } } } } \right)p\left ( { y + \frac{{\log ( { y_1}/n)}}{{\log { y_1 } } } } \right ) } { \bigg|_{x = y = 0}}\bigg ] + o(t / l ) \nonumber \\ & = \frac{{\widehat w(0)}}{{(\alpha + \beta ) { { \log } ^2}{y_1}}}\frac{{{d^2}}}{{dxdy}}\bigg[y_1^{\alpha x + \beta y } \nonumber \\ & \quad \times \int_1^{{y_1 } } { { r^ { - 1}}p\left ( { x + \frac{{\log ( { y_1}/r)}}{{\log { y_1 } } } } \right)p\left ( { y + \frac{{\log ( { y_1}/r)}}{{\log { y_1 } } } } \right)dr } { \bigg|_{x = y = 0}}\bigg ] + o(t / l ) \nonumber \\ & = \frac{{\widehat w(0)}}{{(\alpha + \beta ) \log { y_1}}}\frac{{{d^2}}}{{dxdy}}\bigg[y_1^{\alpha x + \beta y}\int_0 ^ 1 { p(x + u)p(y + u)du } { \bigg|_{x = y = 0}}\bigg ] + o(t / l ) .
\nonumber \end{aligned}\ ] ] in the second equality we made use of @xmath317 , in the third equality we used the euler - maclaurin formula , and the in the fourth equality we employed the change of variables @xmath318 . by adding and subtracting the same quantity
we find that @xmath319 + i{'_{11 } } ( - \beta , - \alpha ) ( { t^ { - \alpha - \beta } } - 1 ) + o(t / l).\end{aligned}\ ] ] for the term in square brackets we have @xmath320 for the other term in we have @xmath321 by the use of @xmath322 the additional restriction that @xmath323 is dealt with the holomorphy of @xmath324 and @xmath325 with @xmath163 which implies that the error term is also holomorphic in this region .
the maximum modulus principle extends the error term to this enlarged domain .
this proves lemma [ lemmac11 ] .
this is the term involving conrey s and feng s mollifiers . to compute this term ,
let us follow the same strategy as in @xmath273 .
we first insert the definitions of @xmath46 and @xmath66 into the mean value integral @xmath82 so that @xmath326 \sum_{k
\le y_2 } \frac{\mu(k)}{k^{1/2+it } } \sum_{\ell = 2}^k \sum_{p_1 \cdots p_\ell
| k } \frac{\log p_1 \cdots
\log p_\ell}{\log ^\ell y_2 } p_{\ell}[k ] dt \nonumber \\ & = \sum_{\ell = 2}^k \sum_{h , k } \frac{\mu(h)\mu(k)}{(hk)^{1/2 } } p_1[h ] \sum_{p_1 \cdots p_\ell | k } \frac{\log p_1 \cdots \log p_\ell}{\log ^\ell y_2 } p_{\ell}[k ] \nonumber \\ & \quad \times \int_{-\infty}^{\infty } w(t ) \zeta(\tfrac{1}{2}+\alpha+it ) \zeta(\tfrac{1}{2}+\beta - it ) \bigg ( \frac{k}{h } \bigg)^{-it } dt .
\nonumber\end{aligned}\ ] ] as for @xmath273 , we use at this point lemma [ lemmaafe ] to write @xmath327 , where @xmath328 and @xmath329 correspond to the two sums of lemma [ lemmaafe ] and @xmath330 is the error term .
specifically , one has @xmath331 \sum_{p_1 \cdots p_\ell | k } \frac{\log p_1 \cdots \log
p_\ell}{\log ^\ell y_2 } p_{\ell}[k ] \nonumber \\ & \quad \times \sum\limits_{hm = kn } \frac{1}{{{m^{1/2 + \alpha } } { n^{1/2 + \beta } } } } \int _ { - \infty } ^\infty { v_{\alpha , \beta } } ( mn , t)w(t)dt , \end{aligned}\ ] ] and for reasons of symmetry , @xmath329 can be obtained from @xmath328 by switching @xmath158 and @xmath202 and multiplying by @xmath332 for @xmath333 .
we thus see that it is enough to compute @xmath328 .
the error term is given @xmath334 \sum_{p_1 \cdots p_\ell | k } \frac{\log p_1 \cdots \log p_\ell}{\log ^\ell y_2 } p_{\ell}[k ] \nonumber \\ & \ll t^{-a } \sum_{\ell = 2}^k \sum_{h \le y_1 } \sum_{k \le y_2 } \frac{1}{(hk)^{1/2 } } \sum_{p_1 \cdots p_\ell | k } 1 \ll t^{-a } \sum_{\ell = 2}^k \sum_{h \le y_1 } \sum_{k \le y_2 } \frac{1}{(hk)^{1/2 } } ( d(k))^{\ell } \nonumber \\ & \ll t^{-a } \sum_{h \le y_1 } \frac{1}{h^{1/2-\varepsilon } } \sum_{k \le y_2 } \frac{1}{k^{1/2-\varepsilon } } \ll t^{-a } y_1^{1/2-\varepsilon } y_{\ell}^{1/2-\varepsilon } \nonumber \\ & = t^{-a } t^{\theta_1 ( 1/2-\varepsilon ) \theta_{2}(1/2-\varepsilon ) } = t^{-a + ( \theta_1+\theta_\ell)/2 - \varepsilon } \nonumber\end{aligned}\ ] ] for any @xmath335 .
we remark that the above computation works for @xmath336 arbitrarily large but the error term @xmath337 coming from lemma [ lemmaafe ] is only valid for @xmath338 .
the next step is to use the mellin integral representations of the polynomials @xmath339 @xmath284 = \sum\limits_i { \frac{{{a_i}}}{{{{\log } ^i}{y_1 } } } } { ( \log ( { y_1}/h))^i } = \sum\limits_i { \frac{{{a_i}i!}}{{{{\log } ^i}{y_1 } } } } \frac{1}{{2\pi i}}\int_{(1 ) } { { { \left ( { \frac{{{y_1}}}{h } } \right)}^s}\frac{{ds}}{{{s^{i + 1}}}}},\ ] ] and @xmath340 @xmath341 = \sum\limits_j { \frac{{{b_{\ell , j}}}}{{{{\log } ^j}{y_2 } } } } { ( \log ( { y_2}/k))^j } = \sum\limits_j { \frac{{{b_{\ell , j}}j!}}{{{{\log } ^j}{y_2 } } } } \frac{1}{{2\pi i}}\int_{(1 ) } { { { \left ( { \frac{{{y_2}}}{k } } \right)}^u}\frac{{du}}{{{u^{j + 1 } } } } } , \ ] ] and the definition of @xmath342 in lemma [ lemmaafe ] to write @xmath343 we now have to compute the arithmetical sum @xmath344 .
further details on this procedure can be found in @xcite .
let us define @xmath345 we start by inverting the order of the sum so that @xmath346 where @xmath347 and where we define the inner sum to be @xmath348 recall that @xmath349 denotes the number of times the prime number @xmath13 appears in @xmath262 .
we can write the above as @xmath350 where we define @xmath351 as well as @xmath352 and finally @xmath353 hence we arrive at the following expression for @xmath354 @xmath355 where the arithmetical factor @xmath290 is given by an absolutely convergent euler product in some product of half - planes containing the origin .
therefore , when we go back to the expression for @xmath356 in , we obtain the following @xmath357 where @xmath358 at this stage , we compare in its exact form ( that is , with big-@xmath359 terms replaced by their exact expressions ) against and in its exact form , and we use the fact that for @xmath291 and @xmath292 , the ratio of zeta functions @xmath360 reduces to @xmath10 . in other words , reverting the @xmath13-adic analysis in @xmath361 we find that @xmath362 following , we know that @xmath363 and thus , we find that @xmath364 for all @xmath294 .
let us denote the last part of by @xmath365 , specifically @xmath366 we now employ the principle of inclusion - exclusion to write @xmath367 where @xmath368 to end the computation , we must identify the logarithms of the prime numbers with the signature of the von mangoldt function @xmath78 and hence match the resulting expressions to logarithmic derivatives of the riemann zeta - function by the use of @xmath369 for @xmath370 . with this in mind
, @xmath371 becomes @xmath372 where @xmath373 are terms of smaller order and where @xmath374 we also have that @xmath375 all of these terms are analytic in a larger region of the complex plane , thus we are only interested in the term @xmath376 .
consequently , the end result of this is that @xmath377 the next step is to deform the path of integration to @xmath378 where @xmath379 is small , fixed and @xmath380 as well as @xmath381 . by doing this
, we pick up the contribution of the residue of the simple pole of @xmath299 at @xmath300 only , since , as before in the @xmath273 case , @xmath301 vanishes at the pole of @xmath302 .
the new path of integration with respect to @xmath294 contributes @xmath382 by keeping @xmath383 ( since @xmath50 and @xmath64 ) .
we now write @xmath384 where @xmath385 corresponds to the residue at @xmath300 .
then @xmath386 where @xmath387 let us now use the binomial theorem to write @xmath388 where we have used the dirichlet convolution of @xmath389 for @xmath370 . here
@xmath390 for all @xmath262 denotes the identity function .
next , we take @xmath391 and bound the integral trivially to get @xmath392 .
this means that we can use a taylor series so that @xmath393 to write @xmath394 , say .
we recall that we have shown earlier that @xmath309 for all @xmath294 , in particular @xmath395 .
this implies that the complex variables @xmath245 and @xmath307 are now separated as @xmath396 where @xmath397 and @xmath398 the first of these two integrals was dealt with in the @xmath273 case and its main term is @xmath399 for the second integral we will need the following lemma [ integralwithlogderivative ] and equation .
hence , one gets @xmath400 this means that when we insert these results into @xmath401 we obtain @xmath402 by making the changes @xmath403 we can write this in the more convenient form @xmath404 telescoping back to we obtain that @xmath405 + o(t / l ) , \nonumber \end{aligned}\ ] ] where the sum over @xmath313 has been identified to the polynomial @xmath339 , and the sum over @xmath315 to the polynomials @xmath340 .
we now perform the summation over @xmath262 by using lemma [ lemma36rrz01 ] .
to do so , we now set @xmath406 .
the lemma yields @xmath407 therefore , the resulting expression for @xmath408 is @xmath409 + o(t / l ) \nonumber \end{aligned}\ ] ] now we must go back to @xmath82 .
we recall that @xmath272 was formed by adding @xmath328 and @xmath329 , where @xmath410 is formed by taking @xmath411 , switching @xmath158 and @xmath202 , and then multiplying by @xmath412 .
note that @xmath413 and thus only the case @xmath414 contributes to the main term . therefore @xmath415
+ o(t / l ) .
\nonumber \end{aligned}\ ] ] we now use @xmath416 we first take a look at the first term in the brackets @xmath417\\ = & \ ( \alpha+\beta)\log y_1 \int_0 ^ 1 { { ( 1 - u)}^{\ell}}{p_1}\left ( { 1 - ( 1 - u)\frac{{{\theta _ 2}}}{{{\theta _ 1 } } } } \right){p'_\ell } ( y)du\\ & + ( \alpha+\beta)\log y_2 \int_0 ^ 1 { { ( 1 - u)}^{\ell}}{p'_1}\left ( { 1 - ( 1 - u)\frac{{{\theta _ 2}}}{{{\theta _ 1 } } } } \right){p_\ell } ( y)du\\ = & \ ( \alpha+\beta)\log y_1 \left(\int_0 ^ 1 { { ( 1 - u)}^{\ell}}{p_1}\left ( { 1 - ( 1 - u)\frac{{{\theta _ 2}}}{{{\theta _ 1}}}}\right){p'_\ell } ( y)du + \int_0 ^ 1 { { ( 1 - u)}^{\ell}}{p'_1}\left ( { 1 - ( 1 - u)\frac{{{\theta _ 2}}}{{{\theta _ 1 } } } } \right){p_\ell } ( y)du\right)\\ & - ( \alpha+\beta)(\theta_1-\theta_2)\log t \int_0 ^ 1 { { ( 1 - u)}^{\ell}}{p'_1}\left ( { 1 - ( 1 - u)\frac{{{\theta _ 2}}}{{{\theta _ 1 } } } } \right){p_\ell } ( y)du \end{aligned}\ ] ] since we had that @xmath418 it follows that @xmath419 we can therefore write @xmath420 combining these observations , we see that @xmath421 for the expression @xmath422 , we use to get @xmath423 \nonumber \\ & + o(t / l ) .
\end{aligned}\ ] ] by using similar arguments for the holomorphy of the error terms as in the section 3.1 , we end the proof of lemma [ lemmac12 ] .
this is the hardest case .
once again , we insert the definitions of the feng mollifiers @xmath66 in the mean value integral @xmath271 so that @xmath424 \nonumber \\ & \quad\times \sum\limits_{{h_2 } \leqslant { y_2 } } { \frac{{\mu ( { h_2})}}{{h_2^{1/2 - it } } } } \sum\limits_{{\ell _ 2 } = 2}^{{k } } { \sum\limits_{{q_1}{q_2 } \cdots { q_{\ell_2}}|{h_2 } } { \frac{{\log { q_1}\log { q_2 } \cdots \log { q_{{\ell _ 2}}}}}{{{{\log } ^{{\ell _ 2}}}{y_2 } } } } } { p_{{\ell _ 2}}}[h_2]dt \nonumber \\ & = \sum\limits_{{h_1},{h_2 } \leqslant { y_2 } } { \frac{{\mu ( { h_1})\mu ( { h_2})}}{{\sqrt { { h_1}{h_2 } } } } } \nonumber \\ & \quad \times \sum\limits_{{\ell _ 1 } = 2}^{{k } } { \sum\limits_{{\ell _ 2 } = 2}^{{k } } { \sum\limits_{{p_1}{p_2 } \cdots { p_{{\ell _ 1}}}|{h_1 } } { \sum\limits_{{q_1}{q_2 } \cdots { q_{\ell_2}}|{h_2 } } { \frac{{\log { p_1}\log { p_2 } \cdots \log { p_{{\ell _ 1}}}\log { q_1}\log { q_2 } \cdots \log { q_{{\ell _ 2}}}}}{{{{\log } ^{{\ell _ 1 } + { \ell _ 2}}}{y_2 } } } } } } } \nonumber \\ & \quad \times p_{\ell_1}[h_1]p_{\ell_2}[h_2]\int _ { - \infty } ^\infty { w(t)\zeta ( \tfrac{1}{2 } + \alpha + it)\zeta ( \tfrac{1}{2 } + \beta - it){{\left ( { \frac{{{h_1}}}{{{h_2 } } } } \right)}^ { - it}}dt } . \nonumber \end{aligned}\ ] ] we already explained in the computation of @xmath272 how to deal with this integral , namely write @xmath425 , where @xmath426 can be obtained from @xmath427 by switching @xmath158 and @xmath202 and multiplying by @xmath428 we now use the mellin integral representations of the polynomials @xmath429 = \sum\limits_i { \frac{{{b_{i,{\ell _ 1}}}}}{{{{\log } ^i}{y_2 } } } } { ( \log ( { y_2}/{h_1}))^i } = \sum\limits_i { \frac{{{b_{i,{\ell _ 1}}}i!}}{{{{\log } ^i}{y_2 } } } } \frac{1}{{2\pi i}}\int_{(1 ) } { { { \left ( { \frac{{{y_2}}}{{{h_1 } } } } \right)}^{{u}}}\frac{{d{{u}}}}{{{u^{i + 1}}}}},\ ] ] and @xmath430 = \sum\limits_j { \frac{{{b_{j,{\ell _ 2}}}}}{{{{\log } ^j}{y_2 } } } } { ( \log ( { y_2}/{h_2}))^j } = \sum\limits_j { \frac{{{b_{j,{\ell _ 2}}}j!}}{{{{\log } ^j}{y_2 } } } } \frac{1}{{2\pi i}}\int_{(1 ) } { { { \left ( { \frac{{{y_2}}}{{{h_2 } } } } \right)}^{{s}}}\frac{{d{{s}}}}{{{{{s}}^{j + 1 } } } } } .\ ] ] this leaves us with @xmath431 we now have to compute the arithmetical sum @xmath432 with @xmath13-adic analysis .
the first step is to consolidate the two sums over primes into a single sum .
this is accomplished by the use of lemma [ lemmacombinatorics ] .
let us define @xmath433 the next step is to swap the order of the sums so that @xmath434 by making the changes @xmath435 implying that @xmath436 so that @xmath437 here the @xmath13 s , the @xmath237 s and the @xmath238 s are all distinct primes .
let us define the inner sum to be @xmath438 and let us recall that @xmath349 is the number of times the prime @xmath13 appears in @xmath262 so that @xmath439 each product is evaluated to @xmath440 then @xmath441 followed by @xmath442 as well as @xmath443 and finally @xmath444 this leaves us with @xmath445 where @xmath446 is an arithmetical factor that is given by an absolutely convergent euler product in some product of half - planes containing the origin . from our previous analysis of the @xmath272 case , we know that @xmath309 for all values of @xmath294
. therefore we end up with @xmath447 where @xmath448 and @xmath449 and finally @xmath450 we define @xmath451 to be the last part of @xmath452 .
this means that @xmath453 for some polynomial @xmath75 .
applying the inclusion - exclusion principle we then have @xmath454 where @xmath455 as in the previous crossterm , we now need to identify the logarithms of the primes with the signature of the von mangoldt functions @xmath78 and @xmath456 . with this in mind , we first write @xmath457 and @xmath458 for @xmath12 .
this means that @xmath459 where @xmath460 are terms of smaller order and where @xmath461 moreover , we also have that @xmath462 all of these terms are analytic in a larger region of the complex plane , thus we are only interested in the term @xmath463 .
consequently , the end result of this computation is that @xmath464 as in the calculation of @xmath411 , we now take the @xmath465 contours of integration to @xmath297 small and fixed with @xmath380 , and then move @xmath294 to @xmath466 , crossing a simple pole at @xmath300 only ( since , yet again , @xmath301 vanishes at the pole of @xmath302 ) .
the new line of integration with respect to @xmath294 contributes @xmath467 since @xmath132 .
write @xmath468 , where @xmath469 corresponds to the residue at @xmath300 , i.e. @xmath470 where @xmath471 the next step is to employ the binomial theorem in the part of the integrand that involves @xmath203 functions . calling this part @xmath472
, we then have @xmath473 where we have used the dirichlet convolution of @xmath474 for @xmath12 .
now we take @xmath391 and bound the integral trivially to get @xmath475 .
this means that we can use a taylor series expansion so that @xmath476 to write @xmath477 , say .
we recall that earlier we proved that @xmath309 for all @xmath294 , and hence @xmath395 .
this has the effect of separating the complex variables @xmath245 and @xmath307 as follows @xmath478 where @xmath479 and @xmath480 these two integrals are identical , up to the symmetries in @xmath481 , @xmath482 , @xmath483 and @xmath484 and they were in fact treated in the @xmath272 case .
the end results for the main terms are @xmath485 and @xmath486 next , we insert these results into @xmath487 and we end up with @xmath488 to make matters easier , we again employ the change of variables @xmath489 and this produces @xmath490 we are now ready to insert this into @xmath491 so that
@xmath492 + o(t / l ) , \nonumber \end{aligned}\ ] ] where we have used @xmath317 .
we now sum over @xmath313 and @xmath315 , e.g. @xmath493 thereby getting @xmath494 + o(t / l ) .
\nonumber \end{aligned}\ ] ] lemma [ lemmaem2lambdas ] gives us @xmath495 so that we we are left with @xmath496 + o(t / l ) .
\nonumber \end{aligned}\ ] ] note that @xmath497 and @xmath498 .
thus only the cases @xmath499 and @xmath500 contribute to the main term .
we therefore have @xmath501 + o(t / l ) .
\nonumber \end{aligned}\ ] ] recall that @xmath502 and that @xmath503 therefore @xmath504 we first take a look at the first term in the brackets @xmath505\\ & = ( \alpha+\beta)\log y_2 \bigg(\int_0 ^ 1 { { { ( 1 - u)}^{\ell_1+\ell_2 } } { p'_{{\ell _ 1}}}(u){p_{{\ell _ 2}}}(u)du } + \int_0 ^ 1 { { { ( 1 - u)}^{\ell_1+\ell_2 } } { p_{{\ell _ 1}}}(u){p'_{{\ell _ 2}}}(u)du}\bigg ) . \end{aligned}\ ] ] since @xmath506 , we have also @xmath507 this implies @xmath508 combining these observations gives @xmath509 for the expression @xmath510 , we again use to get @xmath511 + o(t / l ) .
\nonumber \end{aligned}\ ] ] by using similar arguments for the holomorphy of the error terms as in the section 3.1 , we end the proof of lemma [ lemmac22 ] .
the authors are grateful to arindam roy and alexandru zaharescu for fruitful discussions .
the first author wishes to acknowledge partial support from snf grant pp00p2 138906 .
|
the mollification @xmath0 put forward by feng is computed by analytic methods coming from the techniques of the ratios conjectures of @xmath1-functions .
the current situation regarding the percentage of non - trivial zeros of the riemann zeta - function on the critical line is then clarified .
|
population studies of x - ray binaries in nearby galaxies have been made possible thanks to the superb spatial resolution and excellent sensitivity of the _ chandra x - ray observatory _ @xcite .
the x - ray luminosity functions ( xlfs ) of point sources have been obtained for many galaxies , and they are found to be environment dependent ( see * ? ? ? * for a review ) . in young normal galaxies , high - mass x - ray binaries ( hmxbs )
dominate , and the xlfs follow a simple power law ( pl ) @xmath10 with @xmath11 over a large range of luminosity : @xmath12 erg s@xmath2 @xmath13 erg s@xmath2 @xcite .
in old normal galaxies or in the bulge of young normal galaxies , low - mass x - ray binaries ( lmxbs ) dominate , and the xlfs seem relatively complicated , showing both a high - luminosity break at @xmath14 erg s@xmath2 @xcite and a low - luminosity break at @xmath15 erg s@xmath2 @xcite .
the slope between these two breaks is @xmath161.82.2 . above the high - luminosity break ,
the xlfs decrease sharply . below the low - luminosity break
, the xlfs might flatten to @xmath17 .
the high - luminosity break may be due to the eddington limit of neutron star ( ns ) lmxbs .
the low - luminosity break has been attributed to either the transition from mass transfer driven by magnetized stellar wind at high luminosities to mass transfer driven by gravitational wave emission at low luminosities @xcite , or different types of donor stars with the high - luminosity ones being giants and the low - luminosity ones being main - sequence stars @xcite .
the former predicts the slope below the break to be about 1.0 and the slope above the break to be about 2.0 .
for the latter explanation , the steepening of the xlf at high luminosity is due to the short life time of binary systems with giants .
the xlfs of lmxbs have often been obtained by combining multiple galaxies in order to improve the statistics . however , this method is subject to the limitation that the normalizations of the xlfs in different galaxies show a scatter of more than a factor of two @xcite .
the xlfs of lmxbs well below @xmath18 erg s@xmath2 are still only obtained for very few old galaxies , most notably centaurus a and ngc 3379 @xcite , and the bulge of m31 @xcite .
a low detection limit is critical for constraining the low - luminosity break , which has been shown mostly for the old populations in young galaxies @xcite .
deep observations of old galaxies are also needed for the investigation of the differences between the xlfs of lmxbs in globular clusters ( gcs ) and in the stellar field , which can be used to check whether they have the same origin .
for instance , there is a relative underabundance of faint lmxbs in gcs when compared with field lmxbs @xcite .
[ sec : intro ] ellipse of the galaxy and the approximate fov of the _ hst_/acs mosaic observation ( dashed box ) are also shown .
[ fig : colorimage],width=326 ] was selected as the target of a 1 megasecond _
chandra _ x - ray visionary project ( xvp ) in cycle 13 .
one main goal was to study the gas flow inside the bondi radius of the central supermassive black hole ( bh ) , which has been reported in @xcite .
the other goal was to have a deep look at the x - ray binary population of a normal early - type galaxy .
the detailed analysis of the data , including the source list and detailed properties of special sources , will be presented in ( * ? ? ?
* paper i hereafter ) . in the present paper ,
we concentrate on the xlf of lmxbs , especially on its faint end below @xmath18 erg s@xmath2 .
is a lenticular ( s0 ) galaxy with an age of @xmath19 gyr @xcite and at a distance of 9.7 mpc @xcite . including previous observations , the total exposure time of _ chandra _ on this galaxy is @xmath01.1 ms , reaching a limiting luminosity of @xmath20 erg s@xmath2 .
thus it is one of the best observed normal early - type galaxies by _ chandra_. accompanying the _ chandra _ xvp observation , there is a six pointing _ hubble space telescope _ ( _ hst _ ) mosaic observation in the f475w and f850lp filters ( hereafter @xmath21 and @xmath22 filters , respectively ) using the advanced camera for surveys ( acs ) .
it provides the information of gcs in the galaxy @xcite , which will be used by us to investigate the dependence of the xlf of lmxbs on the stellar environment .
the paper is organized as follows . in section [ sec : reduction ] , we describe the x - ray data reduction , construction of xlfs , incompleteness correction , and cross - correlation of the x - ray and optical sources . in section [ sec : res ] , we show the spatial distributions of different populations , present the total lmxb xlf , compare the xlfs of gc and field lmxbs , and investigate the gc lmxb properties . in section [ sec : dis ] , we discuss various caveats on the xlfs that we obtain and the implication of our results for the nature and formation of lmxbs .
our conclusions are given in section [ sec : con ] .
rrccc 1 & 2040 & 2001 - 06 - 14 & 35.8 & 1.5 + 2 & 11268 & 2010 - 01 - 27 & 40.6 & 0.1 + 3 & 12095 & 2010 - 01 - 29 & 75.6 & 0.1 + 4 & 13817 & 2012 - 01 - 18 & 171.9 & 0.0 + 5 & 13822 & 2012 - 01 - 21 & 156.6 & 0.0 + 6 & 13819 & 2012 - 01 - 26 & 72.9 & 0.0 + 7 & 13820 & 2012 - 01 - 31 & 184.1 & 0.0 + 8 & 13821 & 2012 - 02 - 03 & 157.9 & 0.0 + 9 & 14383 & 2012 - 04 - 04 & 119.4 & 0.3 + 10 & 14419 & 2012 - 04 - 05 & 46.3 & 0.3 + 11 & 14384 & 2012 - 04 - 06 & 69.7 & 0.3 the _ chandra _ observations of ngc 3115 are listed in table [ tbl : obslog ] .
there are 11 observations in total from essentially three epochs : one in 2001 , two in 2010 and nine in 2012 .
all observations used the imaging array of the axaf ccd imaging spectrometer ( acis ; * ? ? ?
the reduction of the data and the creation of the source list were presented in detail in paper i , and here we briefly summarize the procedure adopted .
the data were analyzed with the _ chandra _ interactive analysis of observations ( ciao , version 4.6 ) package .
the data were reprocessed to apply the latest calibration ( caldb 4.5.9 ) and the subpixel algorithm @xcite using the ciao script _
chandra_repro_. some short background flares seen in observations
2040 , 13819 and 13822 were excluded . the final exposure used for each observation
is given in table [ tbl : obslog ] .
the relative astrometry between observations was corrected , and the source detection was performed on individual observations as well as on the merged one using the 0.57 kev energy band with the ciao _ wavdetect _ wavelet - based source detection algorithm @xcite .
we used two different image binning resolutions : one at single sky pixel resolution ( 0492 ) over the full field of view ( fov ) and the other at 1@xmath238 sky pixel resolution covering an area of 3@xmath24@xmath253@xmath24 centered at the center of ngc 3115 .
the subpixel binning images were used to improve the spatial resolution of the crowded field near the center of the galaxy .
sources detected from the merged observation and individual observations were cross - correlated to create the final unique source list .
for each unique source , we extracted the source and background spectra and created the response file for each individual observation .
they were then merged to create the spectra and response files for the merged observation .
the source region was set to be a circle enclosing 90% of the point spread function ( psf ) at 2.3 kev .
the background region was set to be a concentric annulus , with inner and outer radii of two and five times the source radius , respectively .
the background - subtracted 0.57 kev count rates were then converted to unabsorbed fluxes and luminosities , with the conversion factors based on the corresponding response files and assuming an absorbed pl spectral shape with a photon index of 1.7 and the galactic absorption @xmath26 @xmath27 @xcite .
the point - source detection sensitivity varies across the _ chandra _ image , owing to the position dependence of the diffuse x - ray emission in the galaxy , the psf extent , the exposure , and ccd efficiency .
therefore it is necessary to carry out the incompleteness correction for the xlf .
the @xmath28 region of ngc 3115 has a semi - major axis of @xmath29 ( 10.2 kpc ) , a semi - minor axis of @xmath30 ( 3.5 kpc ) and a position angle of @xmath31 @xcite . to limit the incompleteness effects and the cosmic x - ray background ( cxb ) contribution
, we defined our study field for the xlf of field lmxbs as the region inside @xmath28 . further considering that the central region is very crowded and has strong diffuse x - ray emission , we excluded the central @xmath32 elliptical region ( the eccentricity and position angle follow the @xmath28 ellipse ) for all xlfs throughout the paper .
we calculated the incompleteness function @xmath33 , the fraction of pixels weighted by the assumed spatial distribution of sources , in which a source with the luminosity @xmath34 or higher would be detected , using the backward correction method @xcite . in this method , sources are simulated with marx and added one by one to the real observed image , which is then checked to see whether each one could be detected with _
wavdetect_. the source spectral shape was assumed to be a pl with a photon index of 1.7 and the galactic absorption .
the simulations were carried out for a series of luminosities with an increasing factor of 1.1 and 1.21 below and above the 90% completeness luminosity , respectively .
the positions of simulated sources for each luminosity were specified as follows .
the @xmath28 ellipse was divided into elliptical annuli with a series of ellipses that have the eccentricity and the position angle following @xmath28 and semi - major axis @xmath35 for the @xmath36th ( @xmath37 to 56 ) ellipse assuming @xmath38 , where @xmath39 and @xmath40 .
each elliptical annulus was then divided into 80 cells with an equal area , and the simulated source position was specified at the center of each cell , with 4560 in total . in some cases the simulated sources coincided with the real sources , and we assumed that the simulated sources were detected by _
wavdetect _ only if the simulated sources dominate the flux over the real sources , which is to take into account the source confusion effect .
we calculated @xmath33 for the cxb sources and field lmxbs separately , because the cxb sources have a flat distribution and the field lmxbs are expected to follow the @xmath41-band light @xcite , for which we used the 2mass large galaxy atlas data @xcite .
we also calculated @xmath33 for gc lmxbs .
we used all the gcs detected in the optical ( section [ sec : crcorrel ] ) as the parent spatial distribution of gc lmxbs and assumed that they have equal probability of hosting an lmxb . to limit the incompleteness effects and spurious rate of the gc lmxb identification , our study field for the xlf of gc lmxbs is set to be the _
hst_/acs field of view ( fov ) .
fov reached @xmath01.3@xmath28 and @xmath02.5@xmath42 in the major - axis and minor - axis directions of the @xmath28 ellipse , respectively ( figure [ fig : colorimage ] ) .
the differential xlf of lmxbs in a given region can be calculated as follows ( refer to , e.g. , * ? ? ?
* ) : @xmath43 where @xmath44 is the total number of observed sources and @xmath45 is the distance to ngc 3115 .
the quantity @xmath46 is @xmath47 , the @xmath48@xmath49 distribution of the cxb sources .
we used the full band ( 0.510 kev ) @xmath48@xmath49 distribution of cxb sources from @xcite , with their 0.510 kev flux converted to our 0.57 kev band assuming a pl spectrum with a photon index of 1.4 . for the xlf of field
lmxbs only , we filtered out gc lmxbs and had the cxb contribution estimated as described above .
for the xlf of gc lmxbs , the cxb contribution was not corrected because it is negligibly small .
we did not correct xlfs for the hmxb contribution .
following @xcite , we estimated the star formation rate in ngc 3115 to be 0.07 yr@xmath2 ( see their equation 9 , which is based on the uv and ir emission ) .
based on their xlf for hmxbs ( their equation 18 ) , we can estimate the number of hmxbs in ngc 3115 above @xmath50 to be 2.3 , which is one order of magnitude less than the cxb contribution and is thus negligibly small . to compare with previous studies ,
some xlfs presented in this study will be divided by ( thus normalized to ) the stellar mass enclosed in our study region of the field lmxb xlf ( i.e. , within @xmath28 and outside the central @xmath32 ellipse ) .
following @xcite , we used the @xmath51-band luminosity and estimated the stellar mass in our study region to be @xmath52 ( the total stellar mass within @xmath28 is @xmath53 ) .
we cross - correlated our x - ray sources with optical sources from _
mosaic imaging and subaru / suprime - cam imaging to search for the gc lmxbs .
@xcite compiled 360 gc candidates from _
hst_/acs mosaic imaging and an additional 421 from subaru / suprime - cam imaging @xcite .
before cross - correlation , we first carried out absolute astrometry correction on x - ray sources by cross - correlating their positions with the 360 gc candidates from _
mosaic imaging , whose astrometry was registered to the usno - b1.0 catalog @xcite .
we only used x - ray sources detected at @xmath54 significance and with off - axis angles @xmath55 ( the limit of _
fov ) in the cross - correlation .
we found 30 matches with a median separation residual of 0.06@xmath56 .
we searched for the _ hst_/acs gc counterparts to our x - ray sources using the 99.73% ( i.e. , 3@xmath57 ) positional uncertainty that combines both x - ray and optical components . for the _ hst_/acs sources
, we assumed the half light radius as the 1-@xmath57 positional uncertainty .
we also included a systematic uncertainty which was assumed to be 0.05@xmath56 ( 1@xmath57 , in both r.a . and
decl . ) based on the above matches in the absolute astrometry correction .
this systematic uncertainty is probably overestimated , but it is so small that the number of gc matches remained the same even if we did not include this systematic uncertainty .
the offsets of all matches ( 37 in total ) are @xmath58 ( only 3 have offsets @xmath59 ) and have a median of 0.07@xmath56 .
to estimate the spurious rate , we rotated the _
hst_/acs field by @xmath6010@xmath61 , 180@xmath61@xmath6010@xmath61 , and 180@xmath61 around the galaxy center and carried out the cross - correlation in the same way and found the spurious rate to be about 3% .
gc distribution from @xcite decreases sharply within @xmath0@xmath62 ( see section [ sec : spatialdis ] ) .
therefore their gc list is probably fairly incomplete in this region owing to strong stellar light .
we tried to match our x - ray sources with the sources that were detected by @xcite but not classified as gcs , and we found three extra matches ( s12 , s53 and s79 in paper i ) within @xmath63 .
one more source ( s65 ) seems to have an optical match from our visual inspection but it is not detected by @xcite due to its being too close to the galaxy center ( 5.5@xmath56 , within the @xmath32 elliptical exclusion region ) .
we classify these four sources as gc lmxb candidates .
they are all bright ( @xmath6410@xmath65 erg s@xmath2 ) and are not expected to be cxb sources ( the expected cxb source number at this luminosity is 0.5 and is much smaller if only cxb sources with bright optical counterparts are considered ) .
another source ( s92 in paper i ) outside @xmath62 also has an optical match , but it was classified as a star by @xcite due to the measurement of a radial velocity ( 238 km s@xmath2 ) much lower than the threshold of 350 km s@xmath2 that they adopted to define gcs .
considering that the size and color of this optical match are consistent with typical gcs , we treat it as a gc lmxb candidate too .
we did not include the above five lmxbs in either field or gc xlfs ( although s65 is outside the study region and would not be included anyway ) . in our search for the gc optical counterparts to our sources detected only in the subaru / suprime - cam imaging ( i.e. , not in the _
hst_/acs imaging ) , we also used the 99.73% positional uncertainty . the 1-@xmath57 positional uncertainty of the optical sources was assumed to be 0.1@xmath56 in both r.a . and decl .
the 1-@xmath57 systematic uncertainty was assumed to be 0.1@xmath56 in both r.a . and decl .
( paper i ) . to limit the spurious rate
, we have a maximum searching radius of 2@xmath56 .
the spurious rate was estimated to be 5% ( paper i ) .
only eight subaru / suprime - cam gc matches were found . they are used only for the study of the spatial distribution of gc lmxbs in section [ sec : spatialdis ] but not for the study of xlfs .
( i.e. , excluding the central @xmath32 elliptical region ) and ends at @xmath66 . to reduce the incompleteness effects , only sources with @xmath67 erg s@xmath2 are used .
[ fig : spatialdist],width=326 ] . the grey dot - dot - dashed line uses the maximum luminosity of each source .
the distributions are not corrected for incompleteness or the cxb contribution .
[ fig : lumdis_ind ] , width=326 ] , width=326 ] 0.0460.2 ( filled circles ) and those within @xmath68 0.21.0 ( red open circles ) .
their best - fitting broken pl is shown as a dotted line and a red solid line , respectively.[fig : fieldlmxbxlf_sub],width=326 ] within ( 0.81.3)@xmath69 erg s@xmath2 ) or circles ( others ) , whose size is proportional to the logarithm of the lmxb luminosity .
the red dotted line at @xmath70 mag was used by @xcite as the dividing line between blue / metal - poor and the red / metal - rich subpopulations .
[ fig : gcpro],width=326 ] figure [ fig : colorimage ] shows the _ chandra _ x - ray image of ngc 3115 . from the merged observation , we detected 145 x - ray sources above the 50% completeness luminosity @xmath71 erg s@xmath2 ( section [ sec : xlfs ] ) within @xmath28 and outside the central @xmath32 ellipse ( 23 within this exclusion region ) .
the expected cxb source number is 26.5 .
figure [ fig : spatialdist ] shows the cumulative radial distributions of different classes of objects detected from the merged observation . due to the high inclination ( @xmath72 , * ? ? ?
* ) of ngc 3115 , we plot the distributions with respect to @xmath73 , where @xmath74 is the angular separation between the source and the galaxy center and @xmath75 is the elliptical radius of the @xmath76 isophotal ellipse in the direction from the galaxy center to the source .
the cumulation starts at @xmath77 because we have excluded the central @xmath32 elliptical region in the calculation of xlfs . to reduce the incompleteness effects
, we used only sources above @xmath78 erg s@xmath2 , which is the 82% completeness luminosity for cxb sources and the 94% completeness luminosity for field lmxbs within @xmath66 ( but outside the central @xmath32 elliptical region ) .
the non - gc x - ray sources , expected to consist of field lmxbs and cxb sources mostly ( red solid line ) , and gc lmxbs ( red dot - dashed line ) are plotted separately .
the thick blue dotted line models the spatial distribution of non - gc x - ray sources using two components , one for cxb sources and the other for field lmxbs , which was assumed to follow the ir light in the @xmath51 band .
the normalization of the field lmxb component was determined so as to give the same number of sources , after adding the cxb component , as observed within @xmath79 .
the observed distribution roughly follows this model , with the kolmogorov - smirnov ( k - s ) test giving a probability of 30% .
the radial distribution of gc lmxbs seems to approximately follow that of gcs detected in the optical as well , with the k - s test giving a probability of 62% .
figure [ fig : spatialdist ] shows a dramatic difference between the spatial distributions of field and gc lmxbs .
the field lmxbs tend to cluster toward the galaxy center , while gc lmxbs tend to be more spread out . only 7 out of 360
_ hst_/acs gcs are within @xmath80 . however , we note that the gc detection is most probably fairly incomplete near the galaxy center due to strong stellar light .
some gcs could also be destructed near the galaxy center due to mass segregation .
there are 50 non - gc x - ray sources above @xmath78 erg s@xmath2 observed between @xmath76 and @xmath81 , while the expected number of cxb sources is 40.2 after incompleteness correction and the expected number of field lmxbs are 1.5 based on the ir light in the @xmath51 band .
the above 50 sources include two special sources : s109 and s179 ( paper i ) .
the former is a supersoft x - ray source ( sss ) at 1.53@xmath76 .
the latter is a transient with 0.57 kev long - term variability factor @xmath82 and relatively soft x - ray spectra ( classified as a bh x - ray binary candidate in paper i ) , and it is at 1.04@xmath76
. therefore these two sources are most probably in ngc 3115 , instead of being cxb sources
. then we have 48 left , which is 19% more than the expected number of cxb sources . if we concentrate on the region between @xmath83 and @xmath81 ( there is little ir light outside @xmath83 from the galaxy ) , we have 33 non - gc x - ray sources above @xmath78 erg s@xmath2 , excluding s109 , and this number is very close to the expected number of cxb sources ( 30.6 , incompleteness corrected ) .
lccccccc & & & @xmath84&@xmath85&@xmath86&@xmath87 + & & & @xmath88&@xmath89&@xmath90&@xmath91 + & & & @xmath92 & @xmath93 & @xmath94 & @xmath95 + & & & @xmath96 & @xmath97 & @xmath98 & @xmath99 + field@xmath100 & 44 & 1.0 & @xmath101 & @xmath102 & @xmath103 & @xmath104 + field@xmath105 & 70 & 25.5 & @xmath106 & @xmath107 & @xmath108 & @xmath109 + gc@xmath110 & 36&0 & @xmath111&@xmath112&@xmath113&@xmath114 + gc@xmath115 & 27 & 0 & @xmath116 & @xmath112&@xmath117&@xmath118 because a large fraction of our sources are variable , we first check whether there is any difference in xlfs between observations .
we plot the observed cumulative xlfs for the merged observation and the longest observation in each of the three epochs in figure [ fig : lumdis_ind ] .
we compared their xlfs using the k - s test .
we focused on luminosities above the 90% completeness limit , which are @xmath119 erg s@xmath2 , @xmath120 erg s@xmath2 , and @xmath121 erg s@xmath2 for observations 2040 , 12095 and 13820 , respectively .
we found that the xlfs obtained in these individual observations are consistent with the same distribution as that obtained in the merged observation with the k - s probability of 48% , 98% , and 82% , respectively .
@xcite obtained a similar conclusion for centaurus a. we also constructed the xlf using less variable sources ( the 0.57 kev flux long - term variability @xmath122 , paper 1 ) in the merged observation ( gray solid line in figure [ fig : lumdis_ind ] ) and compared it with the xlf using all sources . above the luminosity at the 90% completeness level ( @xmath123 erg s@xmath2 ) , the k - s test also indicates no obvious difference between them , with the probability of 100% . finally , we constructed the xlf using the maximum luminosity of each source ( gray dot - dot - dashed line in figure [ fig : lumdis_ind ] ) .
when we compared it with the xlf from the merged observation and limited to @xmath124 erg s@xmath2 , the k - s test gave a probability of 37% .
from all the above comparison , we see no significant effect of the source long - term variability on the xlfs .
therefore , we will use the mean luminosities of the sources detected from the merged observations for the xlfs hereafter . the incompleteness functions for ( field )
lmxbs and cxb sources are shown in figure [ fig : incompletefun ] . while they appear to be similar to each other , this is a coincidence , as they can be very different if different regions other than the whole @xmath76 ( excluding the central @xmath32 ellipse ) are used .
the luminosity corresponding to the 50% completeness level is about @xmath125 erg s@xmath2 , which is much lower than typical values of @xmath126 erg s@xmath2 achieved for other galaxies by _ chandra _ ( e.g. , * ? ? ?
above this luminosity limit , the sources detected in the merged observation within @xmath28 are all @xmath642.9@xmath57 .
the incompleteness corrected and cxb contribution subtracted xlf above @xmath127 for all lmxbs is plotted in the left panels in figure [ fig : lumdis_sum_subcxb ] .
the xlf is steep down to a break around @xmath128 erg s@xmath2 , below which the xlf flattens clearly .
we fitted the differential form of xlf with a small @xmath129 bin size of @xmath130 using the c statistic ( it is maximum likelihood - based ) in the xspec fitting package @xcite . in this way ,
the observed xlf was used in the fit , but the fitting model was modified by the incompleteness function ( through a response file ) .
the model that we adopted is a broken pl : @xmath131 where @xmath132 .
we obtained @xmath133 , @xmath134 , and @xmath135 ( table [ tbl : xlffit ] and figure [ fig : lumdis_sum_subcxb ] ( green solid line ) ) .
@xcite fitted the average xlf of 20 early - type galaxies with the template introduced by @xcite , which is essentially a double broken pl . because we do not have enough statistics above the second break that they obtained ( @xmath136 erg s@xmath2 ) , we do not need to introduce the second break to fit our xlf .
our fit is fully consistent with that obtained by @xcite , who reported @xmath137 , @xmath138 , and @xmath139 .
the main difference is the normalization @xmath140 , with ours being about 76% of that of @xcite , after being normalized by stellar mass ( our @xmath141 per @xmath142 m@xmath143 versus their @xmath144 per @xmath142 m@xmath143 ) .
this is consistent with the result obtained by @xcite that ngc 3115 has a relatively low number of lmxbs per unit stellar mass compared with other galaxies studied by them ( ngc 3115 was included in their galaxy sample with only the first three observations analyzed ) .
their fit with the normalization decreased by 24% is also shown in figure [ fig : lumdis_sum_subcxb ] ( blue dashed line ) .
we note that we also found a high - break fit ( table [ tbl : xlffit ] , purple dot - dashed line in figure [ fig : lumdis_sum_subcxb ] ) that gives a higher break luminosity ( @xmath145 erg s@xmath2 ) and a much steeper second slope ( @xmath146 , 1-@xmath57 lower limit ) and has a c statistic only larger than the above low - break fit by 1.1 .
the initial slope of the high - break fit ( @xmath147 ) is slightly higher than that of the low - break fit .
the incompleteness corrected and cxb contribution subtracted xlf above @xmath127 for lmxbs in the field is plotted in the middle panels in figure [ fig : lumdis_sum_subcxb ] .
the results of our fit with a broken pl are given in table [ tbl : xlffit ] .
as for the xlf of all lmxbs , we also find a low - break fit and a high - break fit to the xlf of field lmxbs .
the low - break fit has a c statistic higher than the high - break fit by only 0.3 .
the parameters of both fits are very similar to the corresponding fits to the xlf of all lmxbs ( section [ sec : xlfs ] ) , probably due to relatively few gc lmxbs in the total sample .
figure [ fig : lumdis_sum_subcxb ] shows both the low - break fit ( green solid line ) and the high - break fit ( purple dot - dashed line ) to the field lmxb xlf . to check whether there is spatial variation of the xlf of field lmxbs
, we divided our study field into an inner ( ( 0.0460.2)@xmath28 ) and an outer ( ( 0.21.0)@xmath28 ) region ( the boundary was chosen to ensure enough statistics in both regions ) and created two corresponding xlfs .
they are shown in figure [ fig : fieldlmxbxlf_sub ] .
it is nontrivial to use the k - s test to compare these two xlfs due to their different cxb and incompleteness corrections .
therefore , we also fitted them with a broken pl for comparison .
the fitting results are shown in the figure and given in table [ tbl : xlffit ] .
the xlf of field lmxbs in the inner region seems to be flatter at low luminosities , and the break seems to be at a lower luminosity than that in the outer region .
we see no clear degeneracy in the fit any more , but the fit to the xlf from the inner region is similar to the low - break fit to the xlf from the whole region , and the fit to the xlf from the outer region is similar to the high - break fit to the xlf from the whole region , especially the break and second slope .
the best - fitting @xmath148 , @xmath149 , and @xmath150 of the xlfs in the inner and outer regions differ at the 1.5@xmath57 , 1.8@xmath57 , and 1.5@xmath57 confidence levels , respectively . such differences are marginally significant , and we will discuss the possible origin in section [ sec : dis ] .
the xlf for the lmxbs detected in _
hst_/acs gcs are shown in the right panels in figure [ fig : lumdis_sum_subcxb ] .
it is incompleteness corrected using the incompleteness function ( green dashed line ) shown in figure [ fig : incompletefun ] .
the xlf of gc lmxbs seems flatter than that of field lmxbs .
the fit with a broken pl is given in table [ tbl : xlffit ] .
we obtained @xmath151 , @xmath152 , and @xmath153 , which are different from those from the field lmxbs ( the low - break fit ) by 2.5@xmath57 , 1.6@xmath57 and 2.3@xmath57 , respectively .
our results agree with previous finding that the xlf of the field lmxb is steeper than that of gc lmxbs @xcite , though the difference is not very significant in our case , owing to the relatively few gcs in our sample .
previous studies of the xlf of gc lmxbs often just used sources within the @xmath28 ellipse .
we also obtained such an xlf for gc lmxbs , and the fitting results with a broken pl are given in table [ tbl : xlffit ] .
we obtained @xmath148 , @xmath150 , and @xmath149 different from those from the field lmxbs ( the low - break fit ) by @xmath154 , 1.5@xmath57 , and 1.1@xmath57 , respectively . due to fewer sources used ,
these differences are less significant than those reported above using all _ hst_/acs gcs .
figure [ fig : gcpro ] plots the color - magnitude diagram of _
hst_/acs gcs . to indicate the presence of lmxbs and their luminosity we circle the 37 gcs containing lmxbs , with the size of the circle proportional to the logarithm of the lmxb luminosity , but for 13 gcs whose @xmath155 clusters within ( 0.81.4)@xmath69 erg s@xmath2 , they were enclosed with squares instead .
we note that all our gc lmxbs have @xmath156 erg s@xmath2 .
the median ( @xmath157 ) colors of gcs with lmxbs and without lmxbs are 1.23 and 1.06 , respectively . based on the nonparametric wilcoxon rank sum test
, the difference is at a significance level of 2.7@xmath57 .
concentrating on luminous lmxbs with @xmath158 erg s@xmath2 ( 24 in total ) , we found that the median ( @xmath157 ) colors of gcs with luminous lmxbs and without luminous lmxbs are 1.31 and 1.05 , respectively , corresponding to a wilcoxon rank sum difference of 3.5@xmath57 .
if we follow @xcite and use @xmath70 mag as the dividing line ( dotted line in figure [ fig : gcpro ] ) between the red / metal - rich and blue / metal - poor subpopulations , we find that 23 out of 169 ( i.e. , 13.6% ) red gcs contain lmxbs , while there are 14 out of 191 ( i.e. , 7.3% ) blue gcs containing lmxbs .
thus , the fraction of red gcs hosting lmxbs is about twice of that of blue gcs hosting lmxbs .
concentrating on luminous lmxbs with @xmath158 erg s@xmath2 , we find 18 ( i.e. , 10.7% ) red gcs and 6 ( i.e. , 3.1% ) blue gcs hosting luminous lmxbs .
the former fraction is 3.5 times of the latter , which is consistent with previous studies using a sample of galaxies with limiting x - ray luminosity around @xmath18 erg s@xmath2 @xcite .
figure [ fig : gcpro ] also shows that gcs hosting lmxbs tend to be bright / massive , as found previously for many galaxies ( * ? ? ?
* ; * ? ? ?
* including ngc 3115 , but using only the _ chandra _ observation in 2001 ) .
the median @xmath159 is @xmath160 for gcs with lmxbs and is @xmath161 for gcs without lmxbs , corresponding to a wilcoxon rank sum difference of 6.1@xmath57 .
similar results can be obtained if we just focus on luminous lmxbs with @xmath158 erg s@xmath2 , with the median @xmath159 of @xmath162 for gcs with luminous lmxbs and @xmath163 for gcs without luminous lmxbs ( the wilcoxon rank sum difference is 4.4@xmath57 ) . separating the gc subpopulations and concentrating on bright gcs with @xmath164 mag
, we find 16 out of 53 ( i.e. , 30% ) red gcs and 10 out of 36 ( i.e. , 28% ) blue gcs hosting lmxbs . for
even brighter gcs with @xmath165 , we find 10 out of 14 ( i.e. , 71% ) red gcs and 3 out of 8 ( i.e. , 38% ) blue gcs containing lmxbs .
these fractions are significantly higher than those obtained above for all gcs ( i.e. , 13.6% and 7.3% for red and blue gcs , respectively ) .
we have obtained the xlfs of lmxbs in ngc 3115 down to @xmath166 erg s@xmath2 , which has only been achieved for one other old galaxy , i.e. , centaurus a @xcite .
we have carried out careful corrections to the xlfs to account for the incompleteness effects and cxb contribution .
we found no large discrepancy between the cxb density in our field and that estimated by @xcite , who used data from six large _ chandra _ surveys . even assuming a possible 20% enhancement of the cxb density in our field ( section [ sec : spatialdis ] )
, we found no noticeable effect on the xlfs . in some studies ,
the cxb contribution was taken into account by directly excluding cxb sources identified from the optical cross - correlation ( e.g. , * ? ? ?
we did not show the results using this method because our _ hst _ imaging is not deep enough .
however , we also tested this method by excluding the agns that we identified in paper i ( about 50% of the expected number ) and obtained xlfs very similar to the ones that we have shown .
this is mainly because of the high inclination of ngc 3115 so that the cxb contribution is less significant for this galaxy than other typical ones .
our xlf of field lmxbs can be fitted with parameters typically seen in the literature for other old normal galaxies or the bulge of spiral galaxies , with a possible break around @xmath167 erg s@xmath2 .
such a break has been attributed to different mechanisms of removal of orbital angular momentum ( magnetized stellar wind versus gravitational wave emission , * ? ? ? * ) or different types of donor stars ( giants versus main - sequence stars , * ? ? ?
* ) in the high and low luminosities . however , the interpretation of our xlf of field lmxbs is complicated by the presence of a degenerate solution with a high break at about @xmath168 erg s@xmath2 .
this degeneracy appears to be associated with a spatial variation of the xlf , with the xlf in the inner region ( ( 0.0460.2)@xmath28 ) being flatter and having a lower break ( @xmath169 erg s@xmath2 ) than that in the outer region ( ( 0.21.0)@xmath28 ) , which has a break around @xmath8 erg s@xmath2 .
therefore , the xlf of field lmxbs in the inner region is closer to the xlf of gc lmxbs .
one possible cause for this is that our field lmxb sample in the inner region could include some gc lmxbs that we can not identify due to significant incompleteness effects near the galaxy center that limit our ability to detect gcs in the optical . in the outer region we found 25 lmxbs from 219 gcs and 44.5 in the field ( after excluding the cxb contribution ) above @xmath127 , but
only 2 gc lmxbs from 7 gcs and 43.0 field ones in the inner region . to have the same ratio of gc lmxbs to field lmxbs
, we would have missed 22 gc lmxbs in the inner region . however , it is well known that the spatial distribution of gcs is more extended than the stellar light , as can also be seen in figure [ fig : spatialdist ] for the outer region .
the distribution of gcs within ( 0.31.3)@xmath28 , if fitted with a pl , is found to follow @xmath170 , which would indicate that in the inner region there should be 93 gcs .
assuming the same detection rate of lmxbs ( @xmath171 ) as in the outer region , we would have missed 6.6 gc lmxbs , given that we have detected 2 gc lmxbs and 2 candidates .
thus , the number of gc lmxbs that we missed in the inner region is probably small , compared with the total number of sources observed ( 43.0 after subtracting the cxb contribution ) , and their effect on the xlf of field lmxbs in the inner region should be insignificant .
alternatively , the spatial variation of the xlf of field lmxbs might be real and can be explained if the field lmxbs in the inner region have a dynamical origin similar to gc lmxbs .
there are two scenarios : one is the dynamical formation of lmxbs in the dense stellar environment near galaxy nuclei , and the other is the destruction of gcs that drift toward the galaxy center due to mass segregation , leaving behind the remnant lmxbs .
the former was argued to be the dominant mechanism to account for the high specific frequency of x - ray sources , per unit stellar mass ( following the @xmath172 dependence on the stellar density ) , near the center ( @xmath173@xmath174 ) of m31 by @xcite .
however , we do not see increasing high specific frequency of ( non - gc ) x - ray sources at the very center , compared with the outer region ( figure [ fig : spatialdist ] ) .
the stellar density of m31 is around 30 pc@xmath175 at @xmath174 from the center @xcite .
based on the stellar density model by @xcite , we expect that ngc 3115 reaches a similar stellar density at @xmath74@xmath08@xmath56 in the major - axis direction .
therefore the former mechanism is probably still not significant in our inner region , which excludes the central @xmath32 elliptical region .
there is large uncertainty in estimating the level of the second mechanism .
the specific frequency of ( non - gc ) x - ray sources seems to peak around @xmath176 0.10.2 .
if some part of it is due to the second mechanism , the remnant lmxbs should gain some momentum to reach this region during the destruction of gcs , or the destruction should be able to occur there .
considering the possible contamination of the lmxbs from undetected or disrupted gcs in the inner region , the xlf in the outer region is probably a better representation of the characteristics of primordial field lmxbs .
its break at about @xmath8 erg s@xmath2 ( table [ tbl : xlffit ] ) seems somewhat higher ( at the 90% confidence level ) than typical values of around @xmath128 erg s@xmath2 reported in other studies that normally used most of the field ( that is , no differentiation between the inner and outer regions ) @xcite .
@xcite obtained xlfs combining 20 early - type galaxies for the inner and outer regions separately .
their inner and outer regions were defined as ( 0.23)@xmath177 and ( 410)@xmath177 , respectively , where @xmath177 is the @xmath51-band half - light radius .
our inner and outer regions for ngc 3115 are approximately ( 0.31.3)@xmath177 and ( 1.36.5)@xmath177 , respectively .
visually it appears that the break of the xlf in their outer region is higher than the xlf in their inner region ( see their figure 4 ) .
however , they did not carry out the fit , and the significance of this variation is not clear .
moreover , they did not exclude gc lmxbs , making it difficult to compare directly . in the future
, more galaxies should be used to investigate the xlf of field lmxbs from the outer region to check whether the high break that we observed in the xlf of field lmxbs in the outer region of ngc 3115 is universal or due to statistical fluctuation .
if it is real , the best explanation is probably the eddington limit of ns lmxbs . in paper
i , we have shown that most of our bright lmxbs ( above several 10@xmath178 erg s@xmath2 ) are ns lmxbs in the soft state . in summary ,
the flatter xlf of field lmxbs in the inner region compared to that in the outer region is unlikely due to dynamically formed lmxbs in the dense stellar environment near the galaxy nucleus , but could be due to contamination of lmxbs from undetected and/or ( more likely ) disrupted gcs in the inner region .
the field lmxbs in the outer region are more likely to be primordial .
the break of their xlf could be due to the eddington limit of ns lmxbs , agreeing with our finding in paper i that most of our bright sources are ns lmxbs in the soft state .
considering the large detection rate of lmxbs ( @xmath0@xmath179 , section [ sec : gclmxbs ] ) in the most metal - rich and the most massive gcs , some of these gcs probably in fact host multiple lmxbs which can not be resolved by _
chandra_. we follow the method of @xcite to study such source blending effects .
they found the dependence of the expected number @xmath180 ( assuming poisson statistics ) of lmxbs per gc on the gc properties to be : @xmath181 where the gc mass is @xmath182 ( @xmath183 ) and the half - mass radius is @xmath184}$ ] . to match our observation of 37 gcs hosting lmxbs
, the normalization , @xmath185 , should be at least 0.16 , without taking into account the incompleteness effects .
the expected median @xmath159 is then @xmath186 , slightly fainter than the observed value ( i.e. @xmath160 , section [ sec : gclmxbs ] ) .
concentrating on the region of @xmath187 and @xmath188 where the lmxb detection rate is the highest , with 10 out of 14 gcs observed to host lmxbs ( section [ sec : gclmxbs ] ) , the expected number of gcs hosting lmxbs from equation [ eq : gcprob ] with @xmath189 is 8.0 , with the number of gcs expected to host multiple lmxbs @xmath190 .
the total expected number of lmxbs is @xmath191 . to match the observed value of 10 gcs hosting lmxbs
, @xmath185 should be 0.25 , resulting in @xmath192 and @xmath193 in this region . a similar exercise for the rest of the region ( i.e. , @xmath194 or @xmath195 )
, we find that @xmath196 could match the observed 27 gcs hosting lmxbs , with @xmath197 and @xmath198 .
gcs hosting multiple lmxbs are expected to be more luminous and show less long - term variability in x - rays @xcite . in the dense lmxb region of @xmath187 and @xmath188 , there are four with the maximum 0.57 kev luminosity @xmath199 erg s@xmath2 .
they have 0.57 kev long - term luminosity variability of @xmath200 , 1.4 , 2.1 , and 14.9 . among the field lmxbs within (
0.0461.0)@xmath76 , there are 27 with maximum 0.57 kev luminosity @xmath199 erg s@xmath2 .
their median variability is 2.3 , which is not significantly larger than that found above for the gc lmxbs with high likelihood of blending .
thus we can not confirm any effect of blending on the variability . to investigate the source blending effects on the xlf
, we carried out monte carlo simulations using equation [ eq : gcprob ] with @xmath201 if @xmath187 and @xmath188 and @xmath196 elsewhere , as obtained above .
when a gc was simulated to host multiple lmxbs , we assumed them to have equal luminosities .
the results from 1000 simulations are shown in figure [ fig : gc_sim ] , where we plot the xlf using the mean ( open circles ; the standard deviation is shown as the error bar ) of lmxbs in each luminosity bin from these simulations .
the simulated xlfs seem steeper than the observed one overall , as expected . however , the difference is small , which is mainly due to two reasons .
one is that about one third of gc lmxbs are observed to cluster within a very narrow luminosity range ( 0.81.4)@xmath69 erg s@xmath2 , but these gcs are widely spread around in the color - magnitude diagram , thus producing no significant source blending effects .
the second reason is that in the area of parameter space where gcs preferentially host lmxbs , both faint and luminous lmxbs were observed .
that is , the source blending effects are seen at different luminosity levels .
therefore we conclude that the source blending effects on the xlf are negligible , within the uncertainties of our data . instead of assuming equal weights to construct the xlf ( section [ sec : gcxlf ] )
, we explored the option of weighting the incompleteness function to reflect the tendency of lmxbs to be detected in metal - rich and massive gcs .
we recalculated the xlf of gc lmxbs with the incompleteness function weighted by the probability of hosting one or more lmxbs , based on equation [ eq : gcprob ] again with @xmath201 if @xmath187 and @xmath188 and @xmath196 elsewhere ( the source blending effects can not be taken into account simultaneously though ) .
we find that the xlf obtained in this way shows no significant difference from that shown in section [ sec : gcxlf ] .
therefore the observed paucity of faint gc lmxbs below the xlf break @xmath0@xmath18 erg s@xmath2 in ngc 3115 should be real .
it could be explained if there is a transition from persistent sources to transients around this break @xcite .
one main class of gc lmxbs could be ultracompact x - ray binaries ( ucxbs ) .
these are nss accreting from white dwarf ( wd ) companions with very short orbital periods ( @xmath202 hr ) and might be effectively produced in gcs through direct stellar collisions between nss and red giants @xcite .
@xcite first suggested that ucxbs with @xmath20310 min could explain the xlf of gc lmxbs at high luminosities .
the second slope @xmath152 that we obtained in the broken pl fit to the xlf is consistent with their prediction ( @xmath150=1.77 ) .
however , according to @xcite , for the transition from persistent to transient behavior for a he - rich x - ray irradiated accretion disk to occur at around @xmath18 erg s@xmath2 , systems with @xmath204 min are preferred , at least below the break luminosity .
it has been long - debated whether the entire population of lmxbs in galaxies , including those in the field , was formed dynamically in gcs @xcite .
our result that the xlfs of gc and field lmxbs appear to be different agrees with previous findings ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the difference indicates that they are formed through different channels that result in different system configurations ( orbital period , mass ratio , etc . ) and thus with different mass accretion rates .
therefore , our result supports the idea that not all field lmxbs are formed dynamically in gcs . in summary ,
source blending should occur in gc lmxbs , but we do not expect it to significantly affect the xlf .
the observed paucity of faint gc lmxbs below the xlf break @xmath0@xmath18 erg s@xmath2 , compared with field lmxbs , is likely real , and one explanation is that gc lmxbs are dominated by accreting neutron stars with white dwarf donors that show a transition from persistent sources at high luminosity to transients at low luminosity around this break .
the different xlfs of gc and field lmxbs suggest that field lmxbs are not all formed dynamically in gcs .
, width=326 ]
we have studied the xlf of lmxbs in the early - type galaxy ngc 3115 , using the the megasecond _ chandra _ xvp observation of this galaxy . including three previous observations
, we obtained a total exposure of @xmath01.1 ms and reached a detection sensitivity of @xmath205 erg s@xmath2 , much lower than typically achieved for other early - type galaxies ( @xmath206 erg s@xmath2 ) by _ chandra_. our fit to the xlf of all lmxbs supports the presence of the low - luminosity break of the xlf at around @xmath4 erg s@xmath2 , with the differential pl slope of @xmath207 and @xmath208 below and above the break , respectively , as seen in many previous studies ( e.g , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , we can not exclude the possibility that the break is at around @xmath6 erg s@xmath2 and is sharp , with the differential pl slope of @xmath209 and @xmath89 below and above the break , respectively .
we further created the xlfs for field and gc lmxbs separately .
due to relative few gc lmxbs , the xlf of the field lmxbs is very similar to the xlf of all ( gc+field ) lmxbs and still shows degeneracy in the broken pl fit ( one fit with a break at around @xmath167 erg s@xmath2 and the other fit with a sharp break at around @xmath210 erg s@xmath2 ) .
the field lmxb xlf seems to show spatial variation , with the slopes and the break in the inner region ( @xmath211@xmath212 ) being smaller than those in the outer region ( @xmath213@xmath214 ) .
this could be due to the incompleteness effects of the optical gc detection in the inner region and/or contamination of remnant lmxbs left behind from the destruction of gcs that drift toward the galactic center due to mass segregation .
the xlf from the outer region is thus probably more close to the xlf of primodial field lmxbs .
it has a differential pl slope up to a break at around @xmath8 erg s@xmath2 , which is close to the eddington limit of ns lmxbs .
the detection of spatial variation explains the degeneracy in our fit to the xlf from the whole study region .
the xlf of gc lmxbs overall is flatter than that of field lmxbs .
our observation of the difference between the xlfs of gc and field lmxbs casts doubt on the idea that all lmxbs in the galaxy are formed dynamically in gcs .
the break of the gc lmxb xlf is at around @xmath9 erg s@xmath2 and might be due to a transition from persistent sources at high luminosity to transients at low luminosity , which can be explained if gc lmxbs are dominated by accreting nss with wd donors . as in previous studies
, we found that metal - rich / red gcs are more likely to host lmxbs than the metal - poor / blue ones , an effect that is more significant for more luminous lmxbs , and that more massive gcs are more likely to host lmxbs .
although source blending is likely to occur , our simulations indicate that it should not significantly affect the xlf . while at the end of the preparation of this paper , @xcite also reported the study of three old normal galaxies .
their main goal was to test the evolutionary model of lmxbs , but they also obtained the xlfs of lmxbs in ngc 3115 using the same data presented here . in the appendix , we briefly compared our study with theirs .
we found no large discrepancy between our results and theirs , if factors such as the fitting degeneracy and the possible spatial variation are taken into account . the work is supported by _
xvp grant go2 - 13104x .
this material is based upon work supported in part by the national science foundation under grants ast-1211995 and ast-1308124 .
this material is based upon work supported in part by hst - go-12759.02-a and hst - go-12759.12-a .
grs acknowledges support from an nserc discovery grant .
in their study of the evolutionary model of lmxbs , @xcite also presented the xlfs of lmxbs in three normal galaxies , including ngc 3115 .
they used the same data presented here for ngc 3115 , but there are many differences between their analysis method and ours .
lehmer et al .
used the limiting significance level of @xmath215 for _ wavdetect _ and kept sources with false binomial probability less than 0.004 ( see their equation 1 ) , while we adopted the limiting significance level of @xmath216 for _ wavdetect _ ( paper i ) and used sources above @xmath127 , which results in using only sources with the signal to noise ratio @xmath2172.9 .
moreover , lehmer et al .
used sources detected from the merged and individual observations ( though the false binomial probability was calculated exclusively from the merged photometry ) , while we used sources detected from the merged observation only .
therefore we expect that lehmer et al
. could detect more real faint sources but also more spurious faint sources than we did .
the region studied is also different .
lehmer et al .
used an elliptical region of a semi - major axis of 2.7@xmath24 , a semi - minor axis of 1.1@xmath24 and a position angle of @xmath218 ( based on the @xmath140-band galaxy emission ) , excluding a central circular region of radius @xmath219 , but we use the slightly larger @xmath28 ellipse , excluding the central @xmath32 elliptical region .
we excluded a smaller central region because of our use of subpixel binning images for source detection in the central region . within their study region
, we have 90 sources above @xmath127 , among which 17.4 are expected to be cxb sources , while in our study region , we have 145 sources above @xmath127 , 26.5 of which are expected to be cxb sources . the handling of the cxb contribution in the fits to the xlfs is also different .
lehmer et al .
excluded all agns that they could identify from the _ hst_/acs imaging ( they found 9 such sources , which is about 50% of the expected number ) , while we estimated the cxb contribution following @xcite .
we tried to check whether we can reproduce their xlf fitting results ( they also fitted the xlfs with a broken pl ) based on the sources that we detected but using their study region .
we followed their technique to exclude all agns that we could identify ( 10 from paper i ) from the _
hst_/acs imaging instead of estimating the cxb contribution following @xcite .
we found that we generally obtained slightly lower values of the initial slope @xmath148 by ( 12)@xmath57 , most probably due to their inclusion of more very faint sources than were in our sample .
specifically , in the fit to the total xlf , they obtained @xmath220 , and we obtained @xmath221 if we chose a high - break solution with @xmath222 erg s@xmath2 similar to their @xmath223 erg s@xmath2 . we still see the degenerate low - break ( @xmath224 erg s@xmath2 ) fit , which has a c statistic larger than that of the high - break fit by only 0.4 ( or by only 0.1 if the cxb contribution was subtracted from modeling instead ) . for the field lmxb xlf , we preferred a high - break fit , which was also adopted by lehmer et al . and is similar to our xlf of field lmxbs in the outer part of the @xmath28 region ( section [ sec : fieldxlf ] ) . for the gc lmxb xlf , they also obtained a high break luminosity @xmath225 erg s@xmath2 , but we can not constrain it well ( @xmath226 erg s@xmath2 ; we have 22 gc lmxbs ( including three candidates ) above @xmath127 in their region , while they have 25 gc lmxbs detected ) .
overall , we see no large discrepancy between our results and theirs for lmxbs in their study region .
no large discrepancy is seen either between our results for our study region and theirs , if factors such as the fitting degeneracy and the possible spatial variation are taken into account . , m. w. , pivovaroff , m. , baganoff , f. , et al .
1998 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 3444 , society of photo - optical instrumentation engineers ( spie ) conference series , ed .
r. b. hoover & a. b. walker , 210224 , g. , de vaucouleurs , a. , corwin , jr .
, h. g. , et al .
1991 , third reference catalogue of bright galaxies , ed .
de vaucouleurs , g. , de vaucouleurs , a. , corwin , h. g. , jr . ,
buta , r. j. , paturel , g. , & fouque , p.
|
we have studied the x - ray luminosity function ( xlf ) of low - mass x - ray binaries ( lmxbs ) in the nearby lenticular galaxy ngc 3115 , using the megasecond _ chandra _ x - ray visionary project observation . with a total exposure time of @xmath01.1 ms
, we constructed the xlf down to a limiting luminosity of @xmath0@xmath1 erg s@xmath2 , much deeper than typically reached for other early - type galaxies .
we found significant flattening of the overall lmxb xlf from @xmath3 above @xmath4 erg s@xmath2 to @xmath5 below it , though we could not rule out a fit with a higher break at @xmath0@xmath6 erg s@xmath2 .
we also found evidence that the xlf of lmxbs in globular clusters ( gcs ) is overall flatter than that of field lmxbs .
thus our results for this galaxy do not support the idea that all lmxbs are formed in gcs .
the xlf of field lmxbs seems to show spatial variation , with the xlf in the inner region of the galaxy being flatter than that in the outer region , probably due to contamination of lmxbs from undetected and/or disrupted gcs in the inner region .
the xlf in the outer region is probably the xlf of primordial field lmxbs , exhibiting @xmath7 up to a break close to the eddington limit of neutron star lmxbs ( @xmath0@xmath8 erg s@xmath2 ) .
the break of the gc lmxb xlf is lower , at @xmath0@xmath9 erg s@xmath2 .
we also confirm previous findings that the metal - rich / red gcs are more likely to host lmxbs than the metal - poor / blue gcs , which is more significant for more luminous lmxbs , and that more massive gcs are more likely to host lmxbs .
|
recently five experiments announced discovery of a narrow , exotic baryonic state called @xmath4 @xcite .
most probably this state belongs to the positive parity baryon antidecuplet , which naturally emerges in chiral soliton models @xcite .
early prediction of its mass @xcite obtained in the skyrme model @xcite extended to three flavors @xcite is in surprising agreement with the present experimental findings .
moreover , if the discovery of the heaviest members of the antidecuplet , @xmath5 , announced by na49 collaboration @xcite at 1860 mev is confirmed , again the same model will be off only by 70 mev @xcite .
however , the most striking experimental result is perhaps a very small width @xmath6 which is estimated to be of the order of a few tens mev or less @xcite .
such a narrow width was predicted in a seminal paper by diakonov , petrov and polyakov @xcite within the su(3 ) chiral quark soliton model ( @xmath7qsm ) .
the smallness of @xmath6 in this model is due to the cancellation of the coupling constants which enter the collective decay operator .
it is , however , at first sight to some extent unnatural that the cancellation occurs between the constants which are of different order in the number of colors , @xmath8 . if only the leading term were retained , the width of @xmath0 would be of the same order as @xmath9 @xcite . in this note
we show that in fact this cancellation occurs order by order in @xmath8 .
this is due to the fact that the additional factor of @xmath8 appears when one properly takes into account the su(3 ) flavor representation content of the lowest lying baryonic states . indeed , for arbitrary number of colors
, baryons do not fall into ordinary flavor octet , decuplet and antidecuplet , but are members of large su(3 ) representations @xcite , which reduce to the standard ones only for @xmath10 .
taking this extra @xmath8 dependence into account makes the su(3 ) clebsch - gordan coefficients depend on @xmath8 .
we also find that group - theoretical factors suppress @xmath11 with respect to @xmath9 .
this suppression is , however , `` undone '' by the phase space factor , which scales differently with @xmath1 in the chiral limit .
both in the skyrme model @xcite and in the @xmath7qsm @xcite baryons emerge as rotational states of the symmetric top which rotates in the su(3 ) collective space @xcite .
this rotation is described in terms of a rotational su(3 ) matrix @xmath12 .
baryonic wave functions @xcite are given as su(3 ) wigner @xmath13 matrices @xcite : @xmath14 in this notation @xmath15 with @xmath16 being hypercharge , @xmath17 and @xmath18 isospin and its third component . left index @xmath19 and @xmath20 is a state conjugate to @xmath21 .
the fact that the right index runs only over the su(2 ) subspace of the su(3 ) representation @xmath22 follows from the form of the hedgehog ansatz for the static soliton field and the wess - zumino term @xcite .
@xmath23 is the charge of state @xmath21 .
note that under the action of left ( flavor ) generators @xmath24 @xmath25 transforms like a tensor in representation @xmath22 , while under the right generators @xmath26 like a tensor in @xmath27 rather than @xmath22 . .
bold dots and circles denote physical states , squares and dots at the bottom denote spurious states which disappear in the limit @xmath28.,width=516,height=136 ] for @xmath10 @xmath29 and the lowest allowed su(3 ) representations are those of triality zero : @xmath30 , @xmath31 , @xmath32 , @xmath33 etc . in large @xmath8 limit , however , these representations are generalized in the following way @xcite ( other possible choices are discussed in ref.@xcite ) : @xmath34 where@xmath35 note that in this notation anti - ten is not a conjugate of deccuplet , neither is an octet selfadjoint
. therefore we shall denote a complex conjugate by a @xmath36:@xmath37 for large representations like ( [ reps ] ) physical flavor states have hypercharge different from the one in the real world , however isospin , charge and strangeness take the physical values .
this is depicted in fig . 1 .
baryon masses are given in terms of the effective hamiltonian@xmath38 where @xmath39 is the casimir operator for representation @xmath40 and @xmath41 is the classical soliton mass .
it is easy to convince oneself that in the chiral limit nonexotic splittings scale like @xmath42 while exotic - nonexotic like @xmath43 , e.g.:@xmath44 the fact that @xmath45 mass difference is of the order of 1 was used to argue that the rigid - rotor quantization of the chiral soliton is not valid for exotic states @xcite .
the discussion of this point is beyond the scope of the present paper , let us however note , that arguments have been also given in favor of the rigid - rotor quantization @xcite despite eq.([ncscaling ] )
the baryon - meson ( @xmath46 ) coupling operator can be written in terms of the collective coordinates as @xcite:@xmath47 p_{a}. \label{op}%\ ] ] where@xmath48 here @xmath49 are generalized spin right su(3 ) generators related to the known `` isospin '' , @xmath50-spin and @xmath51-spin operators in the following way@xmath52 note that these operators act on the right index of the wave function ( [ wf ] ) , for which `` isospin '' is related to the physical spin .
we have adopted here a convention that greek indices run over all possible values : @xmath53 @xmath54 , capital latin indices over the su(2 ) subgroup : @xmath55 and small latin indices @xmath56 . in order to calculate the width for the decay @xmath57 we have to evaluate the matrix element of @xmath58 between the baryon wave functions , square it , average over initial and sum over final spin and isospin @xcite:@xmath59 coupling constants @xmath60
are related to the axial - vector couplings by a goldberger - treiman relation and scale differently with @xmath8:@xmath61 finally , in order to get the width one has to multiply ( [ mav ] ) by the phase space volume and the final result reads@xmath62 where@xmath63 is the momentum of meson @xmath46 . in ref.@xcite eq.([width ] ) was multiplied by a ratio of the baryon masses which is important for the numerical results , which , however , scales as @xmath43 with @xmath8 and therefore is irrelevant for further discussion .
the action of the @xmath64 functions entering the collective operators can be calculated with the help of the su(3 ) clebsch - gordan coefficients
@xcite : @xmath65{cc}% $ \mathcal{r}_{1}$ & $ \mathcal{r}_{2}$\\ $ b_{1}$ & $ b_{2}$% \end{tabular } \right| \begin{tabular } [ c]{c}%
$ \mathcal{r}_{3}^{\gamma}$\\ $ b_{3}$% \end{tabular } \right ) \left ( \left .
\begin{tabular } [ c]{cc}% $ \mathcal{r}_{1}$ & $ \mathcal{r}_{2}$\\ $ j_{1}$ & $ j_{2}$% \end{tabular } \right| \begin{tabular } [ c]{c}%
$ \mathcal{r}_{3}^{\gamma}$\\ $ j_{3}$% \end{tabular } \right)\end{aligned}\ ] ] where @xmath66 is the degeneracy index . in order to calculate matrix elements of ( [ ops ] ) between wave functions ( [ wf ] ) we shall also use the action of the operators @xmath67 and @xmath68 on the spin states @xcite:@xmath69{ccccc}% u_{+ } & \nwarrow & & \nearrow & v_{+}\\ & & & & \\ v_{- } & \swarrow & & \searrow & u_{-}% \end{array}\ ] ] note that the spin states belong to @xmath27 .
the relevant action is depicted in fig.[fig2 ] . and
@xmath67 operators on spin @xmath70 states belonging to antidecuplet @xmath71 and decuplet @xmath72 . in the latter case
the upper entries refer to the transitions to the spurious states @xmath73 . ]
finally we shall need clebsch - gordan coefficients for large su(3 ) representations ( [ reps ] ) . here
we list the clebsch - gordan series for the highest weights in two cases relevant to the present calculation @xcite:@xmath74 remaining clebsch - gordan coefficients can be found by applying lowering operators to ( [ hw ] ) . the result of the calculations are as follows ( for spin down states and @xmath75 ) : * @xmath76@xmath77{cc}% $ 8 $ & $ " 8"$\\ $ \pi$ & $ n$% \end{tabular } \ \right| \begin{tabular } [ c]{c}% $ " 10"$\\ $ \delta$% \end{tabular } \
\right ) \sqrt{\frac{q+3}{3(q+4)}}\nonumber\\ & \cdot\left [ g_{0}+\frac{1}{2}g_{1}\right ] \cdot p , \label{ddcy}%\end{aligned}\ ] ] * @xmath78@xmath79{cc}% $ 8 $ & $ " 8"$\\ $ k$ & $ n$% \end{tabular } \
\ \right| \begin{tabular } [ c]{c}% $ " 10"$\\ $ \theta^{+}$% \end{tabular } \ \
\right ) \sqrt{\frac{q+1}{2(q+2)(q+4)}}\nonumber\\ & \cdot\left [ g_{0}-\frac{q+1}{2}g_{1}-\frac{1}{2}g_{2}\right ] \cdot p. \label{thdec}%\end{aligned}\ ] ] for @xmath3 decay we get:@xmath80 ^{2}% p^{2}\nonumber\\ & \frac{3}{(m_{n}+m_{\delta})^{2}}\frac{(n_{c}-1)(n_{c}+5)}{2(n_{c}% + 1)(n_{c}+7)}\left [ g_{0}+\frac{1}{2}g_{1}\right ] ^{2}p^{2}. \label{mdav}%\end{aligned}\ ] ] whereas for @xmath4we have:@xmath81 ^{2}p^{2}\nonumber\\ & \frac{3}{(m_{n}+m_{\theta})^{2}}\frac{3(n_{c}+1)}{(n_{c}+3)(n_{c}% + 7)}\left [ g_{0}-\frac{n_{c}+1}{4}g_{1}-\frac{1}{2}g_{2}\right ] ^{2}p^{2}. \label{mthav}%\end{aligned}\ ] ] two important remarks are here in order .
first of all , and this is our main result announced in the introduction , for @xmath0 decay constant @xmath82 is enhanced by a factor of @xmath8 and therefore the second term in eq.([mthav ] ) is of the same order as @xmath83 .
these two terms cancel against each other yielding numerically @xmath0 width much smaller than the width of @xmath3 .
this cancellation is therefore consistent with @xmath8 counting and justifies the use of nonleading terms in the decay operator ( [ op ] ) .
secondly , the overall factor in front of @xmath84 \times p^{2}$ ] is @xmath43 for @xmath76 and @xmath85 for @xmath78 .
this effect , as can be seen from es.([ddcy],[thdec ] ) , is entirely due to the spin part of the matrix elements @xmath86 . indeed , flavor clebsch - gordan coefficients in es.([ddcy],[thdec ] ) scale as @xmath43 with @xmath8 , as can be inferred from es.([hw ] ) .
this is , however , not a complete @xmath8 dependence , since momentum @xmath87 also depends on @xmath8 .
we shall come back to this dependence in a moment . after multiplying by the phase factors we get:@xmath88
^{2}p^{3},\nonumber\\ \gamma_{\theta } & = \frac{3}{2\pi(m_{n}+m_{\theta})^{2}}\frac{3(n_{c}% + 1)}{(n_{c}+3)(n_{c}+7)}\left [ g_{0}-\frac{n_{c}+1}{4}g_{1}-\frac{1}{2}% g_{2}\right ] ^{2}p^{3 } \label{widths}%\end{aligned}\ ] ] where @xmath87 is given by eq.([mom - p ] ) .
in the chiral limit @xmath89@xmath90 since the difference @xmath91 scales differently with @xmath8 for @xmath3 and @xmath0 decays ( [ ncscaling]):@xmath92 and the overall scaling for the widths reads@xmath93 it is interesting to ask at this point how well the @xmath1 scaling arguments work numerically in nature . for the mass differences we get ( assuming @xmath94 , @xmath95 mev , which gives @xmath96 mev for the average antidecuplet mass ) : @xmath97 ( in mev ) which is in a reasonable agreement with an expected factor of @xmath98 , see eq.([ncscaling ] ) .
as far as the phase - space factor is concerned , the physical value of the meson momentum reads ( in mev ) @xmath99 and it is hard to argue that the scaling of eq.([pscale0 ] ) really holds . formally for @xmath100 meson momentum @xmath87 scales in both cases @xcite as @xmath43 . in that case:@xmath101 which would explain parametrically the narrowness of @xmath4 with respect to @xmath3
coupling constants @xmath102 are related to the axial couplings through goldberger - treiman relation . on the other hand we know explicit model formulae for these couplings @xcite:@xmath103 up to the same proportionality factor of the order of @xmath104 .
explicit formulae and numerical values of the inertia parameters @xmath105 and @xmath17 can be found in ref.@xcite .
if in the @xmath7qsm one artificially sets the soliton size @xmath106 , then the model reduces to the free valence quarks which , however , remember the soliton structure @xcite . in this limit ,
many quantities , like the axial - vector couplings , are given as ratios of the group - theoretical factors @xcite:@xmath107 with these values we get that the nucleon axial coupling @xcite@xmath108 which is the well known naive quark model result @xcite .
for the antidecuplet decay strength we get:@xmath109 we see that the cancellation is exact in each order in @xmath8 @xcite .
let us speculate that the cancellation of the leading order terms in ( [ g10bar ] ) is exact .
then @xmath110 while @xmath111 .
in this case we would get@xmath112{ccc}% \mathcal{o}(\frac{1}{n_{c}^{2 } } ) & & m_{\kappa}=0\\ \mathcal{o}(n_{c } ) & & m_{\kappa}\neq0 \end{array } \right .
\nonumber\\ \gamma_{\theta } & \sim\mathcal{o}(\frac{1}{n_{c}^{2}})\times p^{3}% \rightarrow\left\ { \begin{array } [ c]{ccc}% \mathcal{o}(\frac{1}{n_{c}^{2 } } ) & & m_{\kappa}=0\\ \mathcal{o}(\frac{1}{n_{c}^{2 } } ) & & m_{\kappa}\neq0 \end{array } \right.\end{aligned}\ ] ] which would mean that in the chiral limit both decay widths scale as @xmath113 while for @xmath100 @xmath4 decay is damped by a factor @xmath114with respect to @xmath3 .
our primary goal was to show that the cancellation which takes place in the case of the @xmath4 width is consistent with the @xmath8 counting . indeed , by employing correct generalizations of standard su(3 ) representations for arbitrary number of colors
, we have shown that there is additional @xmath8 enhancement of the constant @xmath82 which is formally one power of @xmath8 less than @xmath83 .
this enhancement comes entirely from the spin part of the matrix elements @xmath115 and carries over to the decays of all particles in antidecuplet .
we have also found that there is @xmath85 suppression factor in the @xmath4 width with respect to @xmath3 , coming from the same source , namely from the @xmath8 dependence of the su(3)@xmath116 glebsch - gordan coefficients .
unfortunately , this suppression is
undone by the phase space factor @xmath117 , which in the chiral limit scales differently for @xmath4 and @xmath3 decays ( [ pscale0 ] ) .
if we assume that meson masses are nonzero , then the suppression survives ( [ gscalem ] ) .
this kind of noncommutativity of the chiral limit and large @xmath8 expansion is well known and there are many examples where it creates problems . finally , we have shown that in the nonrelativistic quark model limit , i.e. in the limit where we artificially squeeze the soliton , the cancellation in the decay strength for @xmath4 is exact and occurs independently at each order of @xmath8 . in this
limit @xmath0 width vanishes identically .
the author would like to thank dmitri diakonov for stimulating discussion and thomas cohen for useful remarks .
this work has been partially supported by the polish state committee for scientific research under grant 2 p03b 043 24 and it has been authored under contract no .
de - ac02 - 98ch10886 with the u. s. department of energy .
99 t. nakano _ et al .
_ [ leps collaboration ] , phys . rev .
* 91 * ( 2003 ) 012002 [ hep - ex/0301020 ] ; s. stepanyan _ et al . _
[ clas collaboration ] , hep - ex/0307018 ; v. v. barmin _ et al . _ [ diana collaboration ] , hep - ex/0304040 ; j. barth [ saphir collaboration ] , hep - ex/0307083 ; a.e .
asratyan , a.g .
dolgolenko , m.a .
kubantsev , hep - ex/0309042 .
diakonov and v.yu .
petrov , sov .
jetp lett .
* 43 * ( 1986 ) 57 ; d.i .
diakonov , v.yu .
petrov and p.v .
pobylitsa , nucl .
* b306 * ( 1988 ) 809 ; h. reinhardt and r , wuensch , phys . lett . * b215 * ( 1988 ) 577 ; th .
meissner , f. gruemmer and k. goeke , phys . lett . *
b227 * ( 1989 ) 296 ; m. wakamatsu and h. yoshiki , nucl .
a524 * ( 1991 ) 561 .
a. blotz , d.i .
diakonov , k. goeke , n.w .
park , v. petrov and p.v .
pobylitsa , phys . lett .
* b287 * ( 1992 ) 29 ; nucl . phys .
* a555 * ( 1993 ) 765 ; h. weigel , r. alkofer and h. reinhardt , phys .
b284 * ( 1992 ) 296 .
|
in the chiral quark soliton model the smallness of @xmath0 width is due to the cancellation of the coupling constants which are of different order in @xmath1 .
we show that taking properly into account the flavor structure of relevant su(3 ) representations for arbitrary number of colors enhances the nonleading term by an additional factor of @xmath1 , making the cancellation consistent with the @xmath1 counting .
moreover , we show that , for the same reason , @xmath0 width is suppressed by a group - theoretical factor @xmath2 with respect to @xmath3 and discuss the @xmath1 dependence of the phase space factors for these two decays .
tpju-11/2003 + bnl - nt-03/38
|
the main source of a star s energy is the nuclear fusion reactions occurring either in its core or in thin burning shells around the core .
this energy is transported from the production site to the surface by radiative transfer or convection .
the first transports energy in the form of photons , while the second is a cyclic macroscopic mass motion that carries the energy in bulk . in regions where convection is stabilised
, radiative transfer leads to the equation of stellar structure , @xmath0 where @xmath1 is the luminosity at radius @xmath2 , @xmath3 the density , @xmath4 the stefan - boltzmann constant , @xmath5 the temperature and @xmath6 is the rosseland mean opacity .
opacity @xmath7 is a measure of the degree to which matter absorbs photons .
there are four main sources of opacity in stars * bound - bound transitions are the transitions of an electron in an atom , ion or molecule between energy levels which are accompanied by either the absorption or emission of a photon . only a photon with the correct wavelength
can cause a given transition , the process is wavelength dependent . *
bound - free transitions , or photoionisation , occur when an incoming photon has enough energy to ionise an atom or ion and free an electron .
the reverse process is the capture of an electron by an atom or ion .
it will not occur until photons above a threshold energy are available and falls off as @xmath8 where @xmath9 is the frequency .
* free - free transitions are scattering processes which occur when an electron and photon interact near an atom or ion . the process is also known as bremsstrahlung .
again it is proportional to @xmath8 . *
electron scattering is wavelength independent at low temperatures where it is thomson scattering .
the electron , with its low cross - section , is a small target and so only dominates at high temperatures when most atoms are ionised . at very high temperatures relativistic effects are important and compton scattering dominates . from this list
it is possible to see that calculating the opacity in a stellar model is a difficult process and depends on the composition , temperature and density of the material .
deep in the star local thermodynamic equilibrium ( lte ) is achieved and an average for radiative transfer over all wavelengths requires the rosseland mean opacity @xmath6 @xcite expressed as @xmath10 where @xmath11 is the opacity at frequency @xmath9 , @xmath5 is the temperature and @xmath12 is the flux per unit area into unit solid angle per unit frequency in lte .
the most comprehensive opacities available today are from the opal @xcite or op @xcite groups who have made detailed models of the above processes .
they provide tables of the rosseland mean opacity variation with temperature , density and composition with the metal abundance usually scaled to solar compositions .
@xcite took a step forward for the opal project team by providing tables that include mixtures enhanced in carbon and oxygen ( c and o ) relative to the base solar composition .
we have incorporated their full range of composition tables in the @xcite evolution code . in the last implementation of this code @xcite there only 10 of the 265 tables
available were used .
this provides us with the opportunity to refine our models so as to accurately follow the changes of opacity which occur in the later stages of evolution .
the effects are expected in many types of stars .
while these may be small for the evolution of main - sequence stars and white dwarfs we expect larger differences to be found in agb and wolf - rayet stars .
* asymptotic giant branch ( agb ) stars undergo third dredge - up which mixes helium burning products to the surface and forms carbon stars . using the enhanced mixture tables we shall be able to model the thermal pulses and envelope evolution more accurately .
* wolf - rayet ( wr ) stars are massive and have lost their hydrogen envelopes exposing the helium cores . as time progresses helium burning products
are slowly exposed at the surface and in some cases the stars are eventually mostly composed of carbon and oxygen .
we present our method of opacity table construction and detail its implementation in the eggleton stellar evolution code .
we discuss the effects on main - sequence stars , red giants and white dwarfs .
we then present three tests of our work .
the first is the effect of including extra carbon and oxygen on the structure and evolution of a low - mass population - iii star , the second is a @xmath13 thermally pulsing agb star and the third a wolf - rayet star of @xmath14 with mass loss .
we start with the opal tables @xcite as the framework around which we construct our full tables in an approach similar to that employed by @xcite .
however we choose to use the variable @xmath15 rather than the density .
this has the advantage that our tables can be a third smaller in this dimension .
this partly compensates for increased memory requirement of the full tables .
the opal tables extend in @xmath16 from @xmath17 to @xmath18 in steps of @xmath19 to @xmath20 and in @xmath21 from @xmath22 to @xmath23 in steps of @xmath24 .
we choose the range of the final tables to cover from @xmath25 to @xmath26 in @xmath16 with increments of @xmath19 , and from @xmath27 to @xmath28 in @xmath21 with increments of @xmath24 .
this gives us tables of 141 by 31 elements . to account for the effect of composition ,
the opal tables include typically 265 @xmath5 and @xmath29 grids for each metallicity .
these tables are split into 5 groups with different hydrogen mass fractions x of @xmath30 , @xmath31 , @xmath32 , @xmath33 and @xmath34 .
the tables in each group have different compositions in helium , carbon and oxygen .
the lowest temperature in the opal tables is @xmath35 and we use the tables of @xcite , which extend from @xmath25 to @xmath36 in @xmath16 , to complete them at low temperature . where the tables overlap they match well @xcite .
it should be noted that the tables provided by @xcite do not include enhanced carbon and oxygen mixtures .
therefore at temperatures when @xmath37 our tables do not follow these mixtures .
this is only important in the surface of agb stars .
we plan in future to deal with low temperature enhanced mixtures by including the work of @xcite before application to the structure of agb star envelopes . to complete each table we fill the region from @xmath18 to @xmath26 in @xmath16 , where electron scattering dominates , according to @xcite .
we then combine with a full table of the effective opacity owing to electron conduction for which we use the fits of @xcite to the tables of @xcite and @xcite .
we combine the radiative opacity with the effective conductive opacity by reciprocal addition , @xmath38 before the complete tables are implemented in the stellar evolution code we perform one last process .
we rescale the tables over the c and o plane to a more regular grid . in the opal grid
the spacing used requires interpolation within irregular polygons , so to make the interpolation simpler and faster , we include extra tables in a similar but fully rectangular grid system .
this means we have 61 tables for each x abundance or 305 tables for each metallicity .
we implement the tables in the stellar evolution code first described by @xcite and most recently updated by @xcite .
further details can be found in the references therein .
the main drawback with implementation of the new tables is the increased memory requirement .
whereas before there were only 10 opacity tables there are now 305 .
the new tables are each a third of the size of the old tables so the overall increase is a factor of ten .
this corresponds to a memory requirement increase from 17@xmath39 mb to 170@xmath39 mb but the average specification of today s desktop computers comfortably accommodates this .
one of the reasons why this memory requirement is so large is that we also store the spline coefficients for each of the tables .
when we use the tables within the code we interpolate in the @xmath29 and @xmath5 plane with bicubic splines .
this ensures the smoothness of our tables .
we then use linear interpolation in the 3 composition dimensions to derive the final opacity value at each point . to test the effect of these tables
we use three different interpolation schemes , * method a uses 5 tables with @xmath40 and @xmath34 and @xmath41 . where @xmath42 is the hydrogen mass fraction , @xmath43 the helium mass fraction and @xmath44 the initial , solar mixture , metallicity . we add a further two tables , both with @xmath45 , the first with @xmath46 and @xmath47 . the other with @xmath48 and @xmath49
where @xmath50 and @xmath51 are the mass fractions for the enhanced amount of carbon and oxygen above that included in the metallicity mass fraction so the total carbon mass fraction is @xmath52 and for oxygen @xmath53 .
this gives a total of 7 tables allowing simple interpolation in composition similar to the old method .
* method b is similar to method a but we include two extra tables for each hydrogen mass fraction
. again both have @xmath54 .
one has @xmath55 and @xmath47 and the other @xmath48 and @xmath56 .
this leads to a total of 15 tables in all . *
method c incorporates all 305 tables in the opacity calculations .
we have 61 tables for each hydrogen abundance .
when interpolating in composition we use three variables @xmath42 , @xmath57 and @xmath58 .
the hydrogen abundances are as for method a. we take values of @xmath59 and @xmath60 for the carbon and oxygen grid planes . on this grid
we perform a three - dimensional linear interpolation .
the new tables give almost identical results to the old for main - sequence stars and red giants .
the timescale of core helium burning is altered . from method
a to method c it increases by a factor of @xmath32 percent for a @xmath13 star .
this is reassuring .
current stellar evolution models describe well the evolution of the main sequence and the red giant branch .
however we do find that the opacity tables enhance numerical stability in the later stages of evolution because they resolve changes with composition in greater detail .
this leads to a smoother variation in opacity .
a similar result is found for white dwarfs . between methods a and c for a carbon and oxygen white dwarf the radius increases by 0.2 percent . while conditions in the atmosphere do pass through regions of the greatest difference between the two methods ( the dotted line in figure [ fc ] ) there is not enough mass at these temperatures and densities to make a significant global difference . for our first detailed test
we take a zero metallicity @xmath62 star and pollute it uniformly with carbon and oxygen .
such a star could have been formed just after the death of the first stars if the products of these stars were mainly carbon and oxygen .
we choose such a low - mass star to limit the effect on the results of burning via the cno cycle .
our initial model has the following properties , @xmath63 , @xmath64 , @xmath65 , @xmath66 and @xmath67 .
the tables provided by opal do not include one with @xmath68 , we could extrapolate to obtain the opacity but for this test we prefer to remain within the tables . from the initial model
we perform three sub - tests . in the first we pollute the star at a constant c / o ratio of 1 ,
then evolve each model and compare the time from zero - age main sequence to the helium flash for methods a , b and c. we also compare a sequence of models for stars of the same mass but with a metallicity of the same mass fraction as the carbon and oxygen .
these results are presented in table [ ta ] .
we see that with the pollution fraction increase , the timescale of hydrogen burning increases because the energy produced at the centre takes longer to reach the surface .
however when comparing the difference between methods b & c we see only a comparatively small increase in timescale .
when we compare with the scaled solar models we see a greater difference .
this is due to the much larger opacity of heavier elements such as iron . with the second set of models we look at the radius , surface temperature and luminosity of the star on the main sequence as we pollute it with either only carbon or only oxygen ( tables [ tb ] and [ tc ] ) .
we use method a as the base to which we compare methods b and c. a similar trend is seen in the results in tables [ tb ] and [ tc ] .
we therefore see it is important to include the affect of enhanced c and o mixtures before hydrogen exhaustion .
similar mixtures are encountered in the envelopes of carbon stars of very low metallicity .
finally we consider the effect of varying the c / o ratio of the pollution ( table [ td ] ) at a constant pollutant mass fraction of @xmath69 .
table [ td ] demonstrates the need to use method c to follow accurately the variation in opacity as the c / o ratio changes .
this is important in agb evolution during the thermal pulses when the c / o ratio changes gradually as carbon becomes dominant .
the difference between methods b and c in this case comes about because of non - linear structure in the c and o opacities that can only be resolved with the full set of tables .
our thermally pulsing agb model is of a @xmath13 star . to follow a number of pulses we include convective overshooting @xcite during the pre - agb evolution
this leaves the star with a larger core which reduces the rate at which the thermal pulses grow in strength .
this means we do not find dredge - up in our models but we can evolve through the pulses .
we obtain the average pulse period as the arithmetic mean of the times between the first ten helium shell flashes .
results are recorded in table [ te ] .
first we find that the different methods affect the age of the star when it starts to undergo thermal pulses .
second we find that methods a and b have similar interpulse periods while method c s is @xmath70 percent less .
we attribute this to the change in opacity due to varying he / c / o ratios in the inter - shell region which alters the time evolution of the pulse .
we have used the mass - loss rates of @xcite to manufacture a wolf - rayet star of solar metallicity and an initial mass of @xmath14 .
some details are described by @xcite .
our tables are expected to make a difference because wolf - rayet stars become naked helium stars and helium burning products are mixed throughout and exposed at the surface of the star .
we only use methods a and c and not method b because the carbon and oxygen mass fractions do not rise above those included in the base metallicity until after the hydrogen envelope has been removed and so our tables only begin to make a difference after this . in figure
[ fa ] we present the variation of opacity through the star , just after core carbon ignition , with mesh point , the grid over which we solve the equations of stellar structure .
the solid line is with method a. we then take this structure and use method c to calculate the opacities to give the dashed line .
this model is not in hydrostatic equilibrium and the dot - dashed line shows the relaxed new structure in hydrostatic equilibrium from method c. comparing the physical parameters of these calculations we find the main change to be an increase in radius of 3.0 percent .
the surface temperature has decreased by 1.5 percent .
we then made two complete runs with similar initial conditions by methods a and c. in figure [ fb ] we display the difference between wolf - rayet evolution as the hydrogen envelope is removed .
the greatest changes are in the radius and surface temperature .
the star is in the wn phase between @xmath71 to @xmath72 , after which it enters the wc phase . in the wn phase
there are only small differences but upon entering the wc phase the differences build up so that the temperature is lower by around 2 percent and the radius about 1 percent larger .
the luminosity also tends to be larger by around 0.5 percent .
figure [ fc ] is a plot of where the difference between methods a and c are greatest in @xmath3 and @xmath5 when @xmath73 .
the plot is of the mean difference in the opacity calculated by methods a and c at all possible carbon and oxygen abundances divided by the opacity when @xmath74 . from the plot
we see that the regions of greatest difference are small .
the lines show where the wolf - rayet model calculated by method c lies at certain points of its evolution .
the solid line is for the star at @xmath75 and the dashes at @xmath76 . from figure [ fb ]
we see that the largest changes begin around the @xmath77 point .
this is when the model has moved into the region where the differences between a and c are greatest .
we find the opacity tables alter the timescales for the late burning stages .
method c increases the helium burning timescale by 0.1 percent while the carbon - burning time scale increases by 2.5 percent relative to method a. these changes come about because the opacity varies with composition and effects the temperature structure and burning rates .
while there is no large difference in a single variable the overall effect on structure and evolution is more significant .
our major conclusion is that the changes induced by properly including opacities for varying c and o mixtures are small .
however they are a good thing to include in stellar evolution codes because they help numerical stability by removing sharp changes in the variation of opacity with composition that are encountered when we use fewer tables and add another level of detail to models without much loss of computational speed .
the main computational cost is the extra memory to store the tables .
method c requires ten times the memory ( about 200@xmath39 mb ) compared to methods a and b. however there is little computational speed cost : relative to method a , methods b and c require about 4 percent more time for the evolution of a wolf - rayet star even though we must evaluate four times as many opacities for each @xmath29 and @xmath5 value than with method a. the tables and stellar evolution code are freely available from http://www.ast.cam.ac.uk/@xmath78stars .
interested users are welcome to contact the authors for details on how to download and implement our tables .
the authors would like to thank onno pols for informative suggestions .
jje would like to thank pparc for his funding and his parents for helping to provide the computing power for this work .
cat thanks churchill college for his fellowship .
99 alexander d.r .
, ferguson j.w . , 1994 , apj , 437 , 879 buchley j.r . ,
yueh w.r . , 1976 ,
apj , 210 , 440 canuto v. , 1970 , apj , 159 , 641 cunto w. , mendoza c. , ochsenbein f. , zeippen c.j . , 1993 , a&a , 275 , 5 dray l.m . ,
, 2003 , mnras , 341 , 299 eggleton p.p . , 1971 , mnras , 151 , 351 hubbard w.b . , lampe m. 1969 , apj , 163 , 18 ( supplement series ) iben i. , 1975 , apj , 196 , 525 iglesias c.a . ,
rogers f.j . , 1993 , apj , 412 , 752 iglesias c.a . , rogers f.j . , 1996 , apj , 464 , 943 marigo p. , 2002 ,
a&a , 387 , 507 nugis t. , lamers h.j.g.l.m . , 2000 ,
a&a , 360 , 227 pols o.r . ,
, eggleton p.p .
, han z. , 1995 , mnras , 274 , 964 rosseland s. , 1924 , mnras , 84 , 525 schrder , k.p . , pols , o.r . ,
eggleton , p.p .
, 1997 , mnras , 285 , 696 .variation of evolutionary time scale with increasing carbon and oxygen mass fraction . here
@xmath79 and @xmath80 are the evolution times to the helium flash for methods a , b and c and @xmath81 is the evolution time for models with an equivalent metallicity . [ cols="^,^,^,^,^ " , ]
|
we have included opacity tables in our stellar evolution code that enable us to accurately model the structure of stars composed of mixtures with carbon and oxygen independently enhanced relative to solar .
we present tests to demonstrate the effects of the new tables .
two of these are practical examples , the effect on the evolution of a thermally pulsing asymptotic giant branch star and a wolf - rayet star .
the changes are small but perceptible .
[ firstpage ] radiative transfer stars : carbon stars : evolution stars : general stars : agb stars : wolf - rayet
|
neutron stars are often thought of as exciting cosmic laboratories .
this is natural since their description depends on much complex physics @xcite : with a mass of about one and a half times that of the sun compressed inside a radius of ten kilometers or so , they are compact enough to require a fully general relativistic description . with central densities several times the nuclear saturation density it may be energetically favourable for exotic phases of matter , like kaon condensates , hyperons and/or deconfined quarks , to be present in the core . with temperatures much below the fermi temperature for the various constituents , neutron stars are cold on the nuclear scale which means that the presence of both solid ( the outer layers form a kilometer sized nuclear lattice ) and superfluid regions is expected .
neutron star dynamics is also intriguing .
the obvious example of this is the glitches , which are taken as evidence of at least two weakly coupled interior components , observed in a number of radio pulsars @xcite .
in addition , the evolution of the sample of the neutron star population which is spinning rapidly may be affected by both hydrodynamical and/or radiation - driven instabilities @xcite .
this is particularly interesting from the point of view of gravitational - wave observations . with a generation of large - scale gravitational - wave interferometers now reaching design sensitivity ,
it is worth emphasising that a signal from an ( unstable ? ) neutron star pulsation mode could provide a unique , and perhaps the only , probe of the bulk motion of matter inside such stars .
the information that could be gleaned from such data would be truly unprecedented @xcite . as a useful illustration of a problem which plays a key role in neutron star dynamics ,
let us consider the coupling between the elastic crust of nuclei and the fluid core .
the relevant coupling timescale impacts on , for example , the damping rate of oscillations driven unstable by gravitational - radiation emission ( eg . the r - modes @xcite ) , the relaxation following a pulsar glitch @xcite , and possible neutron star free precession @xcite . in its simplest form
the problem is analogous to the classic `` spin - up problem '' in fluid dynamics , where one studies the rotation rate of a viscous fluid following a change in rotation of the container .
it is well - known that the fluid velocity changes due to the formation of a so - called ekman layer at the interface .
the role of this viscous boundary layer is to ensure that the fluid motion satisfies a no - slip condition at the interface .
its presence results in a coupling to the bulk of the fluid on a timescale t _ ~t _ ( _ r ) _ = ( ) ^1/2 where @xmath0 is the width of the ekman layer , @xmath1 is the radius of the container , @xmath2 is the rotation rate , @xmath3 is the mass density at the interface and @xmath4 is the shear viscosity coefficient . for a neutron star , where @xmath5 it is clear that the resulting timescale is much shorter than that of viscous diffusion , i.e. t_~ r^2 analogously , in the case of an oscillation in the fluid the ekman layer provides an efficient damping mechanism , see @xcite for a recent review .
however , it is a serious oversimplification to model the core - crust interface as a viscous fluid / solid wall transition . in particular ,
the fact that the crust is expected to be permeated by superfluid neutrons should also be considered .
similarly , the multi - fluid nature of the expected superfluid neutron / superconducting proton mixture in the outer core must be accounted for .
this involves many difficult issues associated with the presence of rotational vortices in the neutron superfluid , and magnetic fluxtubes in the proton fluid ( if it forms a type - ii superconductor ) .
the interactions involving the vortices , leading to the non - dissipative entrainment effect @xcite as well as the mutual friction @xcite , and the potential pinning of vortices to the nuclear lattice @xcite are not yet understood in detail .
it is clear that one faces a problem which stretches our understanding of neutron star physics considerably . in a series of papers ,
we aim to make progress by considering each of the relevant issues in turn .
the purpose of the present paper is to discuss the shear viscosity in a superfluid neutron star . in doing this
, we are taking a small step towards a fully consistent description of dissipative neutron star cores .
as a first step in this development we revisit the standard shear viscosity ( ignoring for the moment the multi - fluid dynamics aspects ) , and ask how well understood the relevant viscosity coefficients are .
this leads to a set of useful formulas that can readily be used in future work .
in addition , our discussion identifies a number of challenging research problems that require further attention .
we begin by making the following observation : in their by now classic paper on the effect of viscosity on damping of oscillations in superfluid neutron stars , cutler and lindblom @xcite state : it is interesting to note that contrary to our experience with other superfluids like @xmath6 , neutron star matter becomes more viscous in the superfluid state than it was in the normal state . "
this , at first sight , counter - intuitive result provides ample motivation for our investigation .
estimates of the effect that shear viscosity has on the fluid motion in neutron stars have so far almost exclusively been based on the work of flowers and itoh @xcite , who calculated the relevant transport coefficients at supranuclear densities .
based on the results of these papers , cutler and lindblom @xcite discussed the viscous damping of oscillations in the neutron star fluid . their analysis was based on , first of all , a fit to the result in @xcite for the total shear viscosity in normal fluid neutron stars , which can be written @xmath7 where @xmath8 and @xmath9 k. as stated by flowers and itoh ( and as is also evident from figure 2 in @xcite ) , the dominant contribution to @xmath10 is due to neutron - neutron scattering .
this result makes sense since the neutrons make up the bulk of the fluid at the relevant densities ( and more exotic massive particles like hyperons are not being considered ) .
however , the neutrons in the outer core are expected to become superfluid as soon as the star cools below ( say ) @xmath11 k , i.e. soon after its birth in a supernova core collapse . below the neutron superfluid transition temperature
the neutron - neutron scattering is suppressed and the dominant contribution to the shear viscosity is made by scattering processes involving the relativistic electrons .
cutler and lindblom @xcite argue that the relevant estimate ( for electron - electron scattering ) in a superfluid neutron star is @xmath12 this estimate for @xmath13 has since been used in a number of contexts .
the puzzling fact that @xmath14 throughout a typical neutron star does not seem to have caused much concern .
in addition to possible confusion about this result , the estimate ( [ eta_cl ] ) is not quite satisfactory .
one reason is that it is associated with a particular supranuclear equation of state ( derived by baym , bethe and pethick in 1971 @xcite ) . in practise
, one would like to have an estimate which explicity includes the dependency on the electron ( = proton ) fraction .
this is important since one may then be able to better understand the effect of varying the supranuclear equation of state ( and the proton fraction ) within the current range of uncertainty .
to derive the relevant estimate , we take as our starting point the general formula ( cf .
e = n_e p_e^2 5 m_e^ [ etaformula]where @xmath15 is the electron number density , @xmath16 is the corresponding fermi momentum , @xmath17 is the effective electron mass and @xmath18 is the timescale for momentum transfer due to the various scattering processes in which the electrons are involved . in a normal fluid neutron star core , scattering off of the ( nonrelativistic ) protons provides the main channel for momentum exchange . however , if the protons become superconducting then electron - electron scattering takes over as the dominant contribution to @xmath18 @xcite . in order to model the resultant shear viscosity , we use the timescale for electron - proton scattering derived by easson and pethick @xcite : _ ep = 4 ^2 ^2 ( _ fp k_b t ) ^2 k_ft c k_f^2 where @xmath19 is the fine - structure constant , @xmath20 is the proton fermi energy and @xmath21 is the thomas - fermi screening wave - vector ( to be discussed later ) . note that we are assuming that the matter is charge neutral , i.e. we take @xmath22 .
we can extract an analogous formula for electron - electron scattering from eq .
( 91 ) in @xcite .
this equation can be written = 152 ^4 ^2 k_f^3 ( k_b t p_e c ) ^2 ( 2 k_fk_ft ) [ shear]noting that 810 ^ -4 ( n_e 1 ^-3 ) ^-1/3 4.5 10 ^ -3 ( x_p 10 ^ -2 ) ^-1/3 _ 15 ^ -1/3 where @xmath23 is the proton ( electron ) fraction , we see that we only need to retain the leading order contribution ( in @xmath24 ) . by comparing ( [ etaformula ] ) and ( [ shear ] ) we then find that _ ee = 4 10 ^2 ^2 ( _ fe k_b t ) ^2
k_ft c k_f^2 where @xmath25 for relativistic electrons . in order to combine the two results for @xmath26 and @xmath27 , we need to note that individual scattering processes add like `` parallel resistors '' , cf
. @xcite , which means that we have = ^-1 [ tautot]this shows that the most important contribution to the shear viscosity comes from the most frequent scattering process ( as one would intuitively expect ) . using = 1 10 _ fe _ fp = 1 10 2m_p^c k_f where @xmath28 and @xmath29 is the effective proton mass
, we find that @xmath30 under normal circumstances .
thus the electron - proton scattering provides the dominant contribution to the electron shear viscosity , in accordance with the discussion of flowers and itoh @xcite .
we now want to account for the likely possibility that the protons are superconducting . in that case
the electron - proton scattering will be suppressed , essentially because there will be fewer states available for the protons to scatter into . in order to allow for the transition to proton superconductivity
, we introduce a suppression factor @xmath31 such that _ ep far below the critical transition temperature at which the protons become superconducting we should have @xmath32 , and we see from ( [ tautot ] ) that the electron - electron scattering then dominates the shear viscosity @xcite . combining the above results we obtain the following final formula = 4 10 ^2 ^2 ( _ fe k_b t ) ^2 k_ft
^-1 [ tau_eta]which should be used together with ( [ etaformula ] ) .
it should be noted that we have neglected the contribution from electron - neutron scattering .
this should always be a valid approximation .
let us now turn to the screening factor @xmath21 .
electron scattering is screened by both electrons and protons , and the relevant screening wave vector is given by k_ft^2 = 4 k_f^2 ( 1 + m_p^c k_f ) the first term in the bracket is due to the electron screening , while the second is due to the protons .
it is straightforward to show that the latter tends to dominate under the conditions that prevail in a neutron star core .
having introduced the key ingredients , we can discuss two limiting cases . in the case of normal protons we take @xmath33 in ( [ tau_eta ] ) .
this leads to _ ep 1.8 10 ^ 18 ( x_p 0.01 ) ^13/6 _ 15 ^ 13/6 t_8 ^ -2 [ eta_est ] here , and in the following , we have taken ( for simplicity ) @xmath34 in evaluating @xmath21 .
our estimate ( [ eta_est ] ) agrees with the result derived by easson and pethick @xcite . in the opposite limit
the protons are ( strongly ) superconducting , which means that @xmath35 , and we arrive at the estimate _
ee 4.4 10 ^ 19 ( x_p 0.01 ) ^3/2 _ 15 ^ 3/2 t_8 ^ -2 [ eta_ee]this result should be compared to the formula used by cutler and lindblom @xcite in deriving ( [ eta_cl ] ) .
one can easily show that the two results are compatible .
this means that the differences that can be seen in figure [ ee - fig ] , where we compare the various coefficients , is entirely due to the fact that different equations of state ( eg .
proton fractions ) are used .
this figure shows clearly that the ( potentially puzzling ) result that the electron - electron viscosity in the neutron star core is stronger than that due to normal neutron - neutron scattering remains true .
we also learn that the key reason for the emergence of the dominant electron shear viscosity is the superconductivity of the protons .
k. the thick solid line shows our estimate @xmath13 while the thick dashed curve corresponds to @xmath36 ( where we have taken @xmath34 in evaluating the screening factor ) .
these estimates are compared to i ) the result of cutler and lindblom @xcite for electron - electron viscosity ( cl87 , thin solid line ) , ii ) the total shear viscosity in a normal fluid neutron star , which is dominated by neutron - neutron scattering @xcite ( fi79/cl87 , thin dashed line ) , iii ) an alternative calculation of the neutron shear viscosity due to mornas @xcite ( m94 , dot - dashed thin line ) , iv ) the results obtained by flowers and itoh @xcite for the electron - electron shear viscosity in the crust region ( crust , fi76 , thick dotted curve ) and v ) a recent estimate by yakovlev and chugunov @xcite which accounts for the fact that in the crust the electrons mainly scatter off of nuclei in the crust ( yc04 , double dot - dashed line ) .
the crust region is indicated by the grey region , and we have assumed that the core - crust transition takes place at a density of @xmath37 . all estimates ( apart from the crust results ) were determined using the simple pal equation of state model discussed in the text .
it is interesting to note that the flowers and itoh crust result dips to join our estimated @xmath13 as the base of the crust is approached .
it is also worth noting the several order of magnitude discontinuity between the dominant viscosity contributions at the core - crust interface . comparing the result of yakovlev and chugunov to the other estimates ,
we see that there will be a step of more than two orders of magnitude even when the protons are normal and provide the main scattering agent for electrons in the core .
, height=302 ] the results illustrated in figure [ ee - fig ] ( and others discussed throughout the paper ) were obtained using one of the phenomenological pal equations of state ( with compression modulus @xmath38 mev ) @xcite . in order to obtain a suitably simple model we have followed kaminker et al @xcite who provide the following fit for the two nucleon number densities ; n_x = a _ 14^b/ ( 1 + c _ 14 + d _
14 ^ 2 ) [ palfit]where @xmath39 and @xmath40 the total number density @xmath41 and the proton fraction @xmath23 follow immediately from @xmath42 and @xmath43 , respectively . in one of the few alternative calculations of the neutron shear viscosity ( that we are aware of )
, mornas @xcite derives the following expression [ cf .
( 4.9b ) ] _ n 2.410 ^ 19 _ 15 ^ 3/5 t_8 ^
-2 [ morn ] it is notable that the scaling with the density is very different from that in eq .
( [ eta_n ] ) , possibly because of the density dependence of the effective neutron mass . as can be seen from figure [ ee - fig ] , eq .
( [ morn ] ) leads to a considerably weaker viscosity at high density .
it should also be noted that mornas s formula leads to results that are very close to the electron - electron scattering results for the crust region determined by flowers and itoh @xcite .
we have no rational explanation for why this should be the case ( in fact , it must be a coincidence ) , but it is a potentially useful observation .
the formulas given in the previous section provide a useful step towards modelling the dynamics of realistic neutron stars since they allow us to determine a consistent electron viscosity for various supranuclear equations of state and core temperatures . of course , in order to use these formulas we need not only the total number density and the proton fraction , we also need to i ) determine whether the protons and/or neutrons are superfluid , and ii ) if so , quantify the associated suppression factors .
this forces us to venture into the thorny area of nucleon superfluidity ( see @xcite for a recent review ) . in this section ,
we summarise the current understanding of the relevant superfluid energy gaps and discuss the suppression factors we require to complete our model . from a survey of the vast superfluid gap - literature @xcite it is clear that the determination of the relevant parameters constitute a serious challenge . in many ways
this is not too surprising .
after all , significant uncertainties concerning the supranuclear equation of state remain .
given the fact that the superfluid energy gaps are sensitive to the internal constitution of the star , we have to accept a range of possibilities at the present time .
it is natural to adopt a strategy similar to that used in the ( closely related ) discussion of neutron star cooling , see for example the work by kaminker and colleagues @xcite and the recent discussion of a `` minimal cooling paradigm '' by page et al @xcite .
that is , to consider the plausible range of superfluid parameters suggested by the theoretical studies and ask whether observations provide useful constraints on the models . since our focus is on the shear viscosity in a neutron star core , the relevant observations concern pulsar glitches and potential future data for core fluid oscillations .
in particular , it is known that gravitational waves from internal dynamics could provide an excellent probe of the core physics @xcite . in principle , superfluidity in the singlet state appears if the formation of cooper pairs from particles with opposite momenta and spins leads to a lowering of the ground state energy .
a key parameter in any discussion of nucleon superfluidity is the energy gap @xmath44 , which corresponds to the energy needed to create a quasiparticle of momentum @xmath45 in the superfluid @xcite .
the gap on the fermi surface @xmath46 can be interpreted as half the energy required to break the cooper pairs . in the following
we will focus on estimates of this quantity since it allows us to approximate both the critical temperature at which the matter becomes superfluid and the reduction factors required in our analysis of shear viscosity .
the strong interaction provides several channels in which pairing is possible .
the indications are that cooper pairs with zero angular momentum in a spin - singlet state @xmath47 should form at low momenta .
however , since the effective interaction is expected to become repulsive near the nuclear saturation density , the corresponding energy gap will disappear as the density increases @xcite . as a result , because of the comparatively low proton number density one would expect proton @xmath47 pairing at densities too high for neutron pairing in that phase @xcite .
all existing models suggest that the dominant proton pairing takes place well inside the core - crust transition . at the same time
, the energetic advantage of neutron pairing in the anisotropic @xmath48 state at higher densities may be attributed to the tensor force @xcite .
given the practical difficulties associated with numerical many - body calculations , most studies have been based on the `` standard '' bcs theory , including more and more refined pairing interactions by adding successive corrections to the bare nucleon - nucleon potential @xcite .
the next order beyond the bare interaction in this scheme corresponds to various medium effects .
these are often accounted for by renormalising the single - particle energies , eg . by lowering the effective nucleon mass .
it is by now well - known that this leads to a smaller energy gap . the key factor determining the effective mass is the number of interaction partners of opposite type that a given nucleon has , i.e. the number of neutrons per proton and vice versa .
the greater the number , the greater will be the effect of the medium in reducing the effective mass .
hence , the effective proton mass tends to be smaller than that for neutrons , and one may expect the maximum gap for the @xmath47 proton superfluid to be smaller than that for neutrons . similarly , a major difference between results for pure neutron matter and nuclear matter is due to the induced lowering of the effective neutron mass by the relatively small proton contaminant @xcite . while the results for the bare interaction have converged towards a predicted maximum @xmath47 neutron gap of about 3 mev at @xmath49 @xcite ( model @xmath50 in table [ gaptable ] ) , the higher order calculations still lead to a range of possibilities . particularly important
is the fact that pairing polarises the medium @xcite . after accounting for
the polarisation effects one finds that the maximum gap has been quenched to nearly 1 mev at @xmath51 @xcite ( models @xmath52 in table [ gaptable ] ) . a nice explanation for this effect is provided by lombardo and schulze @xcite who show that low density polarisation effects suppress the bcs gap by roughly a factor @xmath53 .
the @xmath47 proton and @xmath48 neutron gaps provide further challenges . in the former case one must take into account the background neutron influence ( eg . the polarisation terms ) , while in the latter one
must solve the anisotropic gap equations ( in principle ten coupled equations ) in a consistent way .
in the case of the proton gap the result is particularly sensitive to the so - called `` symmetry energy '' .
this is expected since this parameter governs the proton abundance , cf .
the available results suggest a maximum gap for proton @xmath47 pairing of approximately 1 mev at @xmath54 ( models @xmath55 and @xmath56 in table [ gaptable ] ) .
this is roughly a factor of 3 smaller than the maximum neutron gap .
the difference can be understood from the fact that the effective proton mass is smaller than the effective neutron mass .
( in a pure proton medium the proton gap should , because of the charge symmetry of the nuclear interaction , be identical to the neutron gap in a pure neutron fluid @xcite . )
one would expect polarisation effects to further influence the proton gap , leading to a suppression ( at least ? ) similar to that found for the @xmath47 neutron channel ( a factor of a few ) @xcite .
thus the maximum proton gap may be reduced to perhaps 0.20.3 mev ( a level similar to model @xmath57 in table [ gaptable ] ) . in order to study the anisotropic @xmath48 neutron gap in detail
one must extend bcs theory , and replace the single gap equation for the @xmath47 case by ten coupled equations .
in fact , the added attraction from the tensor coupling is essential for the existence of superfluidity in this state @xcite . as in the case of the isotropic gaps
, one finds that the @xmath48 gap is reduced significantly ( by a factor of two or so ) by the lower neutron effective mass in nuclear matter .
the available results for the @xmath48 gap illustrate the extent to which this problem remains to be understood : the calculation of elgary et al @xcite ( models @xmath45 and @xmath58 in table [ gaptable ] ) suggests a much smaller gap than that predicted in the work by baldo and colleagues @xcite ( models @xmath59 in table [ gaptable ] ) .
the difficulties associated with superfluid pairing in the @xmath48 channel are exacerbated by the fact that relativistic effects come into play at the relevant densities . while the @xmath47 results remain largely unaffected by the inclusion of relativistic effects ( model @xmath57 in table [ gaptable ] )
, the associated change in the single - particle energies reduces the @xmath48 gap by about a factor of two @xcite ( model @xmath60 in table [ gaptable ] ) . finally ,
in contrast to the case for the @xmath47 gaps , it has been suggested that polarisation effects ( which have yet to be accounted for ) may increase the @xmath48 neutron gap .
.detailed parameters for our various gap models constructed from ( [ deltf ] ) .
the models are based on calculations in the given references and represent the current range of possibilities .
model @xmath50 is for the bare interaction and is relevant in a pure neutron ( proton ) medium .
models @xmath52 are for the @xmath47 neutron pairing , while models @xmath61 correspond to the @xmath47 proton results and models @xmath62 are for the @xmath48 neutron channel . [ cols="^,^,^,^,^,^,<",options="header " , ] in order to consider various gap - models in our analysis , without complicating things excessively , we take a lead from kaminker and colleagues @xcite and represent the energy gap ( at the fermi surface ) by the phenomenological formula ( k_f ) = _ 0 ( k_f - k_1)^2 ( k_f - k_1)^2 + k_2 ( k_f - k_3)^2 ( k_f - k_3)^2 + k_4 [ deltf ] where @xmath63 is the fermi momentum of the relevant nucleon .
this expression makes sense because we know from the weak - coupling formula of bcs theory that @xmath64 at low densities ( leaving out an exponential factor which further suppresses the gap as @xmath65 ) . the results in the literature also indicate that the bare interaction gap function is roughly symmetric @xcite .
this suggests that @xmath66 , where the gap vanishes for @xmath67 , for higher densities .
the additional parameters @xmath68 and @xmath69 permit us to adapt the shape of @xmath70 according to various results which incorporate , for example , polarisation effects . in table [ gaptable ] we provide data for a set of gap models based on the literature is only well defined in the isotropic case . for triplet state pairing ,
the gap varies over the fermi surface and one must be careful to choose a `` characteristic '' gap . ] .
the models have been selected to span the current range of uncertainty .
the given parameters lead to good representations of the original results , although it should be noted that we have not worried too much about the numerical precision of each individual gap .
the pairing gaps obtained from the data in table [ gaptable ] are shown in figures [ 1s0gaps ] and [ 3p2gaps ] .
pairing , corresponding to the data in table [ gaptable ] , are shown as functions of @xmath63 .
the left panel corresponds to the neutrons , while the right panel shows the proton results .
, title="fig:",height=226 ] pairing , corresponding to the data in table [ gaptable ] , are shown as functions of @xmath63 .
the left panel corresponds to the neutrons , while the right panel shows the proton results .
, title="fig:",height=226 ] neutron pairing , corresponding to the data in table [ gaptable ] , are shown as functions of @xmath63 .
the two panels corresponds to strong and weak neutron superfluidity in the core , respectively .
, title="fig:",height=226 ] neutron pairing , corresponding to the data in table [ gaptable ] , are shown as functions of @xmath63 .
the two panels corresponds to strong and weak neutron superfluidity in the core , respectively .
, title="fig:",height=226 ] having surveyed the current thinking about the nucleon superfluid energy gaps , and modelled the results in a useful way , we want to use the results to derive , first of all , the associated transition temperatures and then turn our attention to the suppression of shear viscosity . as discussed in detail by yakovlev et al @xcite ( see also @xcite for useful approximations ) , the critical temperatures can be approximated by \ { lll 0.5669 ^1s_0 + 0.8416 ^3p_2 ( m_j=0 ) + 0.4926 ^3p_2 ( |m_j|=2 ) .
[ tcr]here the effect of the anisotropy of the @xmath48 channel , for which @xmath71 represents the projection of the angular momentum of the cooper pair on the quantisation axis , is apparent .
( note that , even though the @xmath72 case has not been studied in detail it is expected to be qualitatively similar to the @xmath73 case . ) . for clarity
we only show the @xmath48 neutron results for the @xmath73 pairing .
the results for the @xmath74 case would be suppressed by a factor of roughly @xmath75 , cf .
( [ tcr ] ) . as in figure
[ ee - fig ] the grey region represents the neutron star crust , in which there are no free protons and hence no corresponding @xmath76 .
the inset in the right panel helps identify the two sets of gap - models we combine to describe `` strong '' and `` weak '' superfluidity .
, height=302 ] if we combine these approximations with our various gap models ( and the data for the pal equation of state from ( [ palfit ] ) ) we obtain the results shown in figure [ tcrit ] .
the two panels distinguish between cases that can be refered to as representing `` strong '' and `` weak '' superfluidity .
we also distinguish the crust and core regions . to make this distinction we have assumed that the transition takes place at @xmath75 of the nuclear saturation density @xmath77 , i.e. at @xmath78 .
it should be noted that there are no free protons in the crust , and hence no proton gap in that region ( we are not accounting for the possibility that nuclei may exhibit pairing ) .
it is also worth commenting on the fact that the models for the @xmath47 neutron gap in the crust do not account for interactions with the crust nuclei in detail .
there are many difficulties associated with this analysis which need to be addressed in future research .
anyway , from figure [ tcrit ] we see that superfluid neutrons are expected to be present in the crust for temperatures above a few times @xmath11 k. the same is true for the proton superfluid in the core .
this means that , all but newly born ( less that a month old ? ) neutron stars should contain both superfluid neutrons and superconducting protons in the @xmath47 phase . however , it is not clear whether one should expect the corresponding regions in the star to overlap .
figure [ tcrit ] certainly gives examples of @xmath47 neutron superfluids which are entirely confined to the crust . at this point
it is relevant to mention possible constraints on the neutron star superfluidity provided by glitch data . by assuming that the glitches are associated with the neutron superfluid in the inner crust , one can put constraints on the corresponding moment of inertia required to explain the observations
this leads to a requirement that at least @xmath79 % of the star s moment of inertia is in the superfluid , see for example @xcite
. comparing to the models given in table [ gaptable ] , this constraint could mean that model @xmath80 ( and possibly also model @xmath55 ) is ruled out .
of course , this conclusion has many caveats and one should avoid making definitive statements given the lack of truly quantitative glitch models .
as we have already discussed , the results for the @xmath48 neutron gap are associated with a great deal of uncertainty . from figure [ tcrit ]
we see that the critical temperature may , in fact , be as low as @xmath81 k. this extreme case would mean that the core of an accreting neutron star in a low - mass x - ray binary , for which nuclear burning in the crust is expected to equilibrate the temperature around a few times @xmath81 k , may not contain superfluid neutrons at all .
it should , of course , be emphasised that most of the models have significantly higher critical temperatures for the @xmath48 superfluidity .
it would certainly be surprising , given the present results , if a neutron star with core temperature at the level of @xmath81 k did not contain a region where superconducting protons coexist with superfluid neutrons in the @xmath48 state .
next , introducing the variable @xmath82 , we learn from ref
. @xcite that the temperature dependency of the gap functions can be approximated by y = ( t ) k_b t \ { lll ( 1.456 - 0.157 / + 1.764 / ) ^1s_0 + ( 0.7893 + 1.188 / ) ^3p_2 ( m_j=0 ) + ( 2.030 - 0.4903 ^ 4 + 0.1727
^8 ) / ^3p_2 ( |m_j|=2 ) .
[ tau]combining these results with the critical temperatures from ( [ tcr ] ) we can estimate the pairing gaps at finite temperature .
this is a key ingredient in constructing the superfluid suppression factor @xmath31 , which represents the extent to which the superconducting protons no longer scatter the electrons . to estimate this factor we use eq .
( 45 ) from @xcite , which can be written @xmath83 \right .
\nonumber \\ & + & \left .
0.5051 \left [ \exp(4.273-\sqrt { ( 4.273)^2+y^2 } ) - 1 \right ] \right\ } \exp \left ( 1.187 - \sqrt{(1.187)^2 + y^2}\right ) \end{aligned}\ ] ] the form of this fit to the numerical results was chosen to lead to the limits , @xmath84 as @xmath85 and @xmath86 as @xmath87 . for temperatures above @xmath76 one should simply use @xmath88 .
it is perhaps worth pointing out that the exact form of this suppression factor is only relevant very close to the transition temperature , and that an accurate model may only be important for a small subset of neutron star models .
we need an analogous suppression factor for neutron superfluidity . in this case
we have carried out a fit to the expressions for @xmath89 and @xmath90 given by eqn .
( 45 ) in @xcite .
thus we find _
n ^3 ( 0.1148 - ) obviously with @xmath91 evaluated for the relevant pairing gap according to ( [ tau ] ) .
given this suppression factor , we use _
n for the shear viscosity due to neutron - neutron scattering .
if we now combine all the results we have discussed , we can calculate the electron shear viscosity coefficient as a function of the density for various neutron star temperatures . in figures
[ strongvisc ] and [ weakvisc ] we show the result for strong and weak superfluidity , respectively .
the former model combines gaps @xmath92 , @xmath55 and @xmath93 from table [ gaptable ] , while the latter follows if we use gaps @xmath80 , @xmath57 and @xmath58 .
the two figures show the shear viscosity as a function of the density for four different temperatures @xmath94 k , @xmath95 k , @xmath11 k and @xmath96 k , also indicated by horisontal dotted lines in figure [ tcrit ] .
the illustrated data shows ( very nicely ) how the neutron viscosity becomes strongly suppressed at lower temperatures .
one can also clearly distinguish , eg . from the transition that takes place at a density of about @xmath97 in the last three panels of figure [ strongvisc ] , the region where the protons are superconducting and the ( stronger ) electron - electron viscosity dominates .
taken together , the two figures provide a useful insight into the complexity of the neutron star interior and how the viscosity changes as the star matures .
these figures represent the final results of this paper .
having reached the stage where we can generate this data , we are now able to turn our attention to the implications for neutron star dynamics
. in particular , we should be able to study the impact of nucleon superfluidity on neutron star oscillations in more detail than has been possible so far .
we can also return to well - known problems like the `` spin - up '' problem and re - assess the viscosity due to the core - crust interaction and the associated ekman layer .
hoping to stimulate further work on the many relevant issues , we have discussed the effects of superfluidity on the shear viscosity in a neutron star core .
as should be clear from our analysis , an understanding of this problem requires input from several different areas of research , most notably nuclear physics and relativistic astrophysics . although the results we have collated may not be original , we are not aware of any previous study which combines the data into a consistent description of the shear viscosity in a superfluid neutron / superconducting proton mixture .
this is in sharp contrast to the many discussions of the role of superfluidity in neutron star cooling , see for example @xcite .
we believe that this paper provides a useful update on the core viscosities which should be valuable for future modelling of neutron star oscillations and associated instabilities @xcite .
we have provided a simple model for the electron viscosity which is relevant both when the protons form a normal fluid and when they become superconducting . this result explains ( as in @xcite ) why it is natural that proton superconductivity leads to a significant strengthening of the shear viscosity .
it also clarifies some confusion associated with the description of cutler and lindblom @xcite .
in particular , it is clear that the superfluidity of the neutrons is not the key factor which leads to electron - electron scattering becoming the main shear viscosity agent .
rather , it is the fact that the onset of superconductivity suppresses the electron - proton scattering .
this is an important point . as can be seen from figure [ ee - fig ]
, the electron - electron shear viscosity is not too different from the result for neutrons scattering off of each other .
this means that , in the temperature range where shear viscosity dominates , the damping of neutron star oscillations will be quite similar ( modulo multi - fluid effects ) in the extreme cases when i ) the neutrons and protons are both normal , and ii ) when the neutrons and protons are superfluid / superconducting , respectively .
the contrast with the case when the neutrons are superfluid and the protons normal ( and viscosity is dominated by @xmath36 ) is clear from figure [ ee - fig ] .
this is an interesting observation because it shows that proton superconductivity ( or rather absence thereof ) could have a significant effect on the dynamics of a neutron star core .
our results are in a form which permits the use of data for any given modern equation of state .
in contrast to the often used expression obtained by cutler and lindblom @xcite , our final formulas are explicitly dependent on the proton fraction ( as well as the total mass density ) . this is a crucial difference because of the simple fact that the result in @xcite was obtained for the now seriously outdated equation of state derived by baym , bethe and pethick @xcite in 1971 .
we can obtain an indication of the likely `` error bars '' associated with the electron shear viscosity from ( [ eta_ee ] ) which suggests that @xmath98 .
as the proton fraction varies by up to perhaps a factor of two for different equations of state we can expect the viscosity coefficient to be uncertain at the level of ( at least ) a factor of three .
our discussion also highlights a number of challenges that must be met if we are to make further progress .
first of all , it is clear that an understanding of the various superfluid gaps is crucial .
we believe that the results summarised in table [ gaptable ] , i.e. the `` phenomenological '' models we have constructed from a range of results in the relevant literature ( in line with the philosophy adopted in @xcite ) , provide a useful survey of the theoretical range of possibilities .
these models are also readily included in quantitative studies like that discussed in this paper .
of course , many uncertainties remain concerning the superfluid parameters .
most important are the suppression factors for the various scattering rates which are needed to evaluate the viscosity coefficients ( @xmath31 and @xmath99 in our analysis ) .
any detailed discussion of these coefficients is certainly valuable . especially since it allows us to make the analysis of temperatures near the critical temperature @xmath76 quantitative rather than qualitative ( as in the present discussion ) . in addition to these problems ,
there are many related challenges .
although it is commonly acknowledged that superfluid components play a crucial role in neutron star dynamics ( eg . in pulsar glitches
@xcite ) , our understanding of the multi - fluid aspects and the relevant dissipation mechanisms can be improved considerably . that this area provides exciting possibilities
is nicely illustrated by the demonstration that a two - stream instability may operate in superfluid systems with relative flow @xcite .
of particular importance for the physics in the neutron star core will be the superfluid vortices ( and magnetic fluxtubes in the case of a superconductor ) , leading to i ) dissipation due to mutual friction @xcite and ii ) potential vortex pinning to the crust ( see , for example , the discussion in @xcite ) .
as the parameters required to model dynamical superfluids ( like the entrainment coefficients @xcite ) are not yet well constrained , this is a research area in its infancy .
yet another illustration of this fact is provided by the recent discussion following the likely observation of free precession in psr b1828 - 11 @xcite .
of particular relevance for the present paper is link s suggestion @xcite that the interaction between neutron vortices and proton fluxtubes in a type - ii superconductor would lead to a strong coupling between the two fluids and prevent long - period precession . as an alternative
, link suggests that the protons could form a type - i superconductor , from which the magnetic flux would be expelled .
this idea prompted a recent analysis by buckley et al @xcite , who find that the interaction between neutron and proton cooper pairs may indeed lead to the neutron star core becoming a type - i superconductor . exactly how this would affect the various viscosity estimates , and the dynamics of the neutron star core , is unclear at this point .
models of the neutron star crust almost exclusively assume that it can be described as a regular bcc lattice . at some level
this will be an oversimplification .
in particular , it is likely energetically favourable for the nuclei to have a variety of `` shapes '' , ranging from droplets to rods to plates , as the core - crust phase - transition is approached @xcite .
an effective description of these exotic phases may differ significantly from the standard picture .
in addition to this , one should account for the presence of superfluid neutrons throughout the bulk of the crust .
the relevant parameters for this superfluid phase must be considered as poorly known since the various gap - calculations do not ( yet ) account for the presence of the nuclei .
this constitutes a difficult theory problem , but the result could be crucial for a number of scenarios ranging from the pulsar glitches to gravitational - wave driven mode - instabilities .
an indication of how different the parameters of the crust superfluidity may be compared to those in the core fluid is provided by the work of carter , chamel and haensel @xcite .
their analysis of the entrainment near the base of the crust shows that the results differ greatly from those typically used to model the outer core .
the essential difference is in the effective mass of the neutrons . as in the other cases ,
the dynamical repercussions of this result remain to be investigated .
finally , let us turn to the physics of the neutron star inner core .
it is generally expected that , at densities above a few times nuclear saturation density , it will be energetically favourable for exotic phases of matter to be present .
the two most commonly considered cases concern hyperons and deconfined quarks . in the first of these cases it has been pointed out that hyperons provide an efficient refrigerant , since direct urca reactions will be in operation . these reactions would cool the star much faster than indicated by the observational data . the only way to avoid contradictions with the observations is to appeal to hyperon superfluidity @xcite , which would quench the reactions and slow down the cooling rate .
it is also worth pointing out that the fast nuclear reactions in a hyperon core are thought to lead to rapid damping of oscillations due to the resultant bulk viscosity @xcite .
this effect would be significantly suppressed by superfluidity @xcite . in the context of the present paper
it is relevant to make two comments .
first of all , it is known that an increase of the hyperon population tends to lead to a significant depletion of the electron number density .
if we suppose that all the other components ( neutrons , protons and hyperons ) are superfluid , then the electrons may provide the only channel for shear viscosity .
if their number density is depressed to say @xmath100 then we would estimate that the shear viscosity is something like five orders of magnitude weaker than that illustrated in figure [ ee - fig ] .
the dynamical effects of this could be important .
secondly , it is worth pointing out that neither the many possible vortex - vortex and electron - vortex interactions , nor the multi - fluid aspects of superfluid hyperon mixtures have yet been addressed .
the situation is quite similar for the deconfined quark core .
the mixture of s , d , and u quarks is thought to be able to exhibit a variety of `` colour superconducting '' pairings @xcite .
whether this leads to multi - fluid dynamics is not clear , although it seems likely for at least some of the possible phases .
there have not yet been many attempts to estimate the relevant transport coefficients .
yet , it is clear that such studies are of great importance . especially since they may help unveil the true ground state of matter . in this context
, it is worth mentioning an interesting result .
madsen @xcite has combined estimated viscosity coefficients for strange stars with the radiation reaction results for the r - mode instability to demonstrate that , if the fastest spinning pulsar were a pure colour - flavour locked quark star then it ought to be unstable ( see @xcite for an alternative ) .
one interpretation of this result is that this particular form of quark pairing is not present in the bulk of a compact star .
there are of course many caveats to this statement , but it is nevertheless relevant .
in particular , it demonstrates the need to put current and future models of exotic neutron star physics in an astrophysical context and ask whether observations provide useful constraints on the theory .
it is particularly important to consider this at the present time when gravitational - wave observations of compact object dynamics seem a definite possibility .
taking the situation at face value , neutron star ( astro)physics is a vibrant area of research which provides a number of exciting challenges for the future . in order to meet these challenges
we need a closer dialogue between researchers in different fields , like nuclear physics and relativistic astrophysics .
we hope that this paper will help stimulate such discussion .
we are grateful to dima yakovlev for a number of helpful comments on a preprint version of this paper , and for sharing unpublished results with us .
we also thank mark alford , curt cutler , michael gusakov and pawel haensel for useful discussions . na and
kg are grateful for support from pparc grant ppa / g / s/2002/00038 .
|
we discuss the effects of superfluidity on the shear viscosity in a neutron star core .
our study combines existing theoretical results for the viscosity coefficients with data for the various superfluid energy gaps into a consistent description .
in particular , we provide a simple model for the electron viscosity which is relevant both when the protons form a normal fluid and when they become superconducting .
this model explains in a clear way why proton superconductivity leads to a significant strengthening of the shear viscosity .
we present our results in a form which permits the use of data for any given modern equation of state ( our final formulas are explicitly dependent on the proton fraction ) .
we discuss a simple description of the relevant superfluid pairing gaps , and construct a number of models ( spanning the range of current uncertainty ) which are then used to discuss the superfluid suppression of shear viscosity .
we conclude by a discussion of a number of challenges that must be met if we are to make further progress in this area of research .
|
what drives prices ?
this question has been studied for centuries with quantitative theories dating back at least to bachelier @xcite , who proposed the random walk hypothesis for price trajectories . as prices in general
do not become negative , economist later realized that a more realistic framework was obtained by assuming the random walk hypothesis for the logarithm of the price @xcite .
this has made relative returns the prime focus of financial investigation with only a few exceptions , such as hedge funds focusing on absolute returns , and benchmarking are almost exclusively done by the financial community by comparing relative returns with respect to a fixed time interval . within the current economic paradigm
the efficient market hypothesis ( emh ) @xcite the idea of randomly fluctuating prices in the absence of new hard information has been re - formulated in the framework of hard - working rational traders with complete knowledge of all available information whose continuing effort more or less instantaneously removes any imbalance in prices due to past differences in the expectations of traders . in short ,
the emh states that current prices reflect all available information about the priced commodity , _
i.e. _ , all available information is at any given instant already priced in by the market and any change in prices can only be due to the revelation of new information . in other words
, there is no free lunch .
hence , the emh claims that information drives prices .
unfortunately , this just leave us with another question , namely how to price in the available information ? in the case of a stock , this must be done by considering how the available information affects _ future earnings of the company_. this obviously ( again ) introduces some ambiguity as not only do peoples expectations to a largely unpredictable future differ , but so do their strategies with respect to for example investment horizons , _
i.e. _ how long they intend to hold their investment before taking any profit , and how large a risk they are willing to take over that time period . in order to qualify different pricing models
_ , the financial industry has performed many statistical studies establishing a number of so - called stylized facts @xcite as well as benchmarking for the performance of various financial instruments with respect to investment returns and in its complement , risk taking . due to this focus on returns and risk
, most financial studies essentially amount to measuring two - point correlations in one way or another , most commonly done by studying the distribution of returns calculated over some pre - defined fixed time period @xmath2 @xcite .
empirically it has been established , that for not too long time intervals @xmath2 s , say less than a week , the return distributions are characterized by `` fat tails '' @xcite .
fat tails of a distribution refers to a much larger probability for large price changes than what is to be expected from the random walk or gaussian hypothesis .
as @xmath2 reaches longer time scales , the distribution of returns gradually converges to a gaussian distribution
. however , no real consensus regarding the exact quantitative nature of this distribution exist .
> from the point of view of the present authors , a more reasonable answer to the question of what drives prices , besides the influence of interest rates and other macroscopic economic factors as well as quarterly earnings reports _ etc .
_ of the company in question , is _ human psychology _ with all its facets of bounded rationality , limited information , personal beliefs _ etc . _ in accordance with this view , the economics and financial communities have in the past 510 years witnessed an increased interest in what has been coined `` behaviourial economics '' .
a prime example of such a study is the book by r.j .
schiller entitled `` irrational exuberance '' , published in 2000 , but written _ before _ the crash of april of this year @xcite .
this book gives numerous examples of both present ( at that time ) and past events where the price of stocks , and more relevant , the p / e ( price - earning - ratio ) has more or less exploded based on little more than vague notions of a `` new economic era '' and the only real difference comes with the specific sector driving the bubble prior to the crashes seen in the past 150 years or so .
if we focus on the us stock market , then in the 1860s and 1870s it was rail road era , in the 1920s it was utilities , semi - conductors ( `` tektronix '' ) influenced the development in 1950s and 1960s , in the 1980s it was a general public investment boom driven by liberalization of the financial markets and , most recently , in second half of the 1990s it was first an emerging market boom and then the notorious information technology bubble .
most notable are the `` explosions '' in the p / e in the 1920s and in the period from early 1980s to 2000 . during these periods ,
the p / e of the sp500 went from about @xmath3 to @xmath4 and from about @xmath5 to @xmath6 , respectively for us stocks , which is still the highest in western markets . ] .
this corresponds to a more than six - fold increase in little less that one and two decades for the two respective periods . after the october 1987 crash , r.j .
schiller asked a number of institutional and individual investors the following question : `` which of the following better describes your theory of the decline(s ) : _
( i ) _ a theory about investor psychology , or _
( ii ) _ a theory about fundamentals such as profits and interest rates ? ''
the answers revealed that about 68% of the institutional investors and 64% of the individual investors picked the former alternative . from this r.j .
schiller concluded that : `` it appears that the stock market crash had substantially to do with a _ psychological feedback loop _ among the general investing public from price declines to selling and thus further selling , and so forth '' .
this conclusion was in accordance with his investigation of the news stories published immediately prior to the crash of october 1929 where also no specific event or events could be singled out as responsible for the crash .
> from the point of view of human psychology , the financial communities focus on returns does not make as much sense as one would like . the obvious answer to the question
: `` what return would you like to see a year from now ? '' , is of course `` as large as possible ! ''
or , if the question concerned risk `` as small as possible ! '' .
a more natural investment strategy is thus to settle for some return level and/or maximum risk and then ask how long one must wait in order to achieve this . but
how is one to determine _ a priori _ this investment horizon in an utterly unpredictable world with respect to the future some months away ?
even if one solely focus on macroscopic fundamentals as well as those of the company in question , the predictability of future earnings is a very difficult question . as j.p .
bouchaud said in a lecture in granada of february 2005 : `` i do not know how to price even my own company within a factor of two or perhaps even three though i know everything to be known about it ! ''
another , more philosophical problem , with standard models of financial markets is the following : if one , as these models do , only assumes that the market participants are purely selfish individuals , who optimize their own utility function through fixed time contracts with other nominally identical individuals , then , despite the achievement of mutual benefits , the term `` contract '' would not be defined in a general context .
why is this so ? because , a general and lasting definition of the term `` contract '' requires long - term ( meaning much longer than an ordinary human time span ) institutions , which can only be upheld through non - selfish behavior .
legal institutions , for example , have a life time of centuries and hence can not be put to work within a framework of selfish individuals who s time horizon ca nt possibly be longer that there own life span . hence , most models of the financial markets are lacking a very essential ingredient , which one usually denote by the general word `` culture '' . in conclusion , in our opinion most models of the financial markets lack `` psychology and culture '' .
this does not mean , however , that one can not ask what drives the economy and the stock market on longer time scales , _
i.e. _ over decades , and what behaviour the economy exhibits over these time scales !
an example of such a study is presented on fig .
[ publicdebt ] . here
, the historical public debt of the usa is shown together with a labeling of the most significant historic events , mostly military conflicts , involving the us during the same period .
notice that the korean war and it s formal un - forces did not represent any significant increase in us public debt .
however , the so - called cold war between the usa and the ussr most certainly did .
it is clear that the over - all rise in the public debt is exponential driven by large `` bumps '' signifying rapid increases in public debt .
it is striking , on a qualitative level , that the origin of these large increases in us public debt is simply related to war with two major exceptions ; the purchase of the louisiana territory from napoleon in 1803 on the unauthorized initiative of the us - ambassador in paris , and the keynesian ( in the meaning massive public investments ) `` new deal '' of roosevelt in the 1930s .
note , that the logarithmic scale is the reason why the `` bumps '' belonging to the purchase of the louisiana territory and the spanish war of 1898 do not `` stand out '' in fig .
[ publicdebt ] . if one compares this figure with one of the us stock market , say , the djia ( figs .
[ djiahistory ] ) one clearly sees that on a qualitative level the rises in the public debt , due to wars and new deal , are followed by steep rises in the stock market with one big exception , namely the bubble of the 1920s . in hind - sight
, this may not be so surprising , since increases in public debt normally means large public spending which funnels large amounts into the private sector .
however , one should note that the average growth rate of the us public debt is about @xmath7 , see fig .
[ publicdebt ] .
this should be compared with that of the djia until the 1950s , which is about @xmath8 .
what _ is _ surprising is that most time periods with peace exhibit a significant decline in the public debt , most notably the period after the war of 1812 until the second war with the seminole indians ( 18351842 ) , as well as modest growth in the djia . in the first half of the 19th century ,
the us public debt dropped to a meager us$ 33.733 in 1835 from us$ 75.463.477 in 1791 . in conclusion with respect to long - term growth in the stock market , public spending , especially in the case of war , has played a _ very _ significant role in the long - term growth of the us economy and hence of the djia . in the next sections
, we will turn to the subject of optimal investments horizon distributions for both individual stocks as well as markets .
we will do this in order to qualify an answer to the previous questions re - formulated as `` what a return is reasonable to expect in @xmath9 days from now '' for both individual stocks as well as markets by considering the gain distribution for pre - defined return levels . in order to quantify the risk from historical data
, we also consider the corresponding loss distribution .
the remaining part of this paper is organized as follows . in the next section
, we turn to the short - term behaviour of three us stock markets , namely the djia , the sp500 and the nasdaq .
here , we will re - introduce a conceptually new framework for the analysis of short - term ( in the sense of days and weeks ) price fluctuations .
we call this tool of analysis for _ inverse statistics _ as we fix the return level and let time float , and not _ vice versa _ , as in the usual return statistics .
( it is worth noting that an analysis , where _ both _
return level and time floats conditioned on a stable trend in either direction has been published by the first author @xcite ) .
first , we re - visit previous results for the djia as well as presenting new results for the sp500 and the nasdaq index , showing that the distribution of time horizons needed to obtain a specified return can be excellently parameterized by a generalized gamma - distribution .
such distributions are well - known from various first - passage problem @xcite , but the quality of this parametrization is nevertheless surprising considering the nature of the data .
we then turn to the previously found _ gain / loss asymmetry _ for the djia @xcite and show that a similar asymmetry is present in both the sp500 and nasdaq . in the third section
, we turn to the use of the inverse statistics on the single stocks that are included in the djia in order to further investigate the origin of this gain / loss asymmetry .
surprisingly , we find that the _ gain / loss asymmetry _ obtained for the index vanishes for the individual stocks .
the last section concludes arguing that the gain / loss asymmetry found in the index comes from a cooperative cascade through the various sectors of the economy represented in the djia .
in resent publications @xcite , the present authors have proposed to invert the standard return - distribution problem and instead study the probability distribution of waiting times needed to reach a _
fixed level _ of return @xmath1 for the first time ( see also ref .
@xcite ) . as mentioned previously , this is in the literature known as the `` first passage time''-problem @xcite and the solution is known analytically for a brownian motion as the gamma - distribution @xmath10 ( with @xmath11 ) , where one for large ( waiting ) times recovers the well - known first return probability for a random walk @xmath12 .
historical financial time series such as the djia , sp500 and nasdaq possesses an ( often close to exponential , see however @xcite ) positive drift over long time scales due to the overall growth of the economy modulated with times of recession , wars _ etc .
_ if such a drift is present in the analyzed time series , one can obviously not compare directly the empirical probability distribution for positive and negative levels of return .
as the focus of the present paper will be on such a comparison , we must eliminate or at least significantly reduce the effect of this drift .
one possibility for detrending the data is to use so - called deflated asset prices , but such prices are calculated using certain economic assumptions , which we as physicists are naturally suspicious of . in the present study as well as in those of @xcite ,
we have instead chosen to remove the drift based on the use of wavelets @xcite , which has the advantages of being non - parametric .
this technique has been described in detail elsewhere @xcite and for the present purpose , it suffices to say that this wavelet technique enables a separation of the original time series @xmath13 into a short scale ( detrended ) time series @xmath14 and a ( long time - scale ) drift term @xmath15 so that @xmath16 ( cf .
[ djiahistory](c ) ) .
based on @xmath14 for some historical time period of the djia , the empirical investment horizon distributions , @xmath17 , needed to obtain a pre - defined return level @xmath1 _ for the first time _ can easily be calculated for different @xmath1 s . as @xmath14 is stationary over time scales beyond that of the applied wavelet ( for a time larger than say 1000 days ) it is straightforward to compare positive and negative levels of return .
as the empirical logarithmic stock price process is known not to be brownian , we have suggested to use a generalized ( shifted ) gamma distribution @xmath18 where the reasons behind @xmath19 are purely technical .
the results so far have been very encouraging with respect to excellent parametrization of the empirical probability distributions for three major stock markets , namely djia , sp500 and nasdaq ; cf .
[ invdist_djia](a ) , [ invdist_sp500](a ) and [ fullnasdaq ] for examples using a return level of @xmath20 and the figure captions for values of the fit parameters .
the choice of @xmath21 is not accidental .
it is sufficiently large to be above the `` noise level '' quantified by the historical volatility and sufficiently small to be of quite frequently occurrence .
we have also considered other return levels , showing qualitatively the same features , but there results are not shown . in all three cases , the tail - exponents @xmath22 of the distributions parameterized by eq .
( [ fit - func ] ) are indistinguishable from the `` random walk value '' of @xmath23 , which is not very surprising .
what _ is _
both surprising and very interesting is that these three major us stock markets ( djia , sp500 and nasdaq ) exhibit a very distinct gain / loss asymmetry , _
i.e. _ , the distributions are not invariant to a change of sign in the return @xmath1 ( figs .
[ invdist_djia](a ) , [ invdist_sp500](a ) and [ fullnasdaq ] ) .
furthermore , this gain / loss asymmetry quantified by the optimal investment horizon defined as the peak position of the distributions has for at least the djia a surprisingly simple asymptotically power law like relationship with the return level @xmath1 , see fig .
[ optilevel ] .
see ref .
@xcite for an application of inverse statistics to high - frequency foreign exchange data , specifically the us$ against the deutch mark and the japanese yen .
we are also currently investigating the use of inverse statistics for intra - day stock prices , but the results so far are still preliminary .
as mentioned previously , the purpose of the present paper is to further investigate the origin of the gain / loss asymmetry in djia . we do that by simply comparing the gain and loss distributions of the djia with the corresponding distributions for a single stocks in the djia as well as their average .
an obvious problem with this approach is that the stocks in the djia changes with historical period and hence an exact correspondence between the djia and the single stocks in the djia is difficult to obtain if one at the same time wants good statistics .
this is the trade - off , where we have put the emphasis on good statistics .
however , _ if _ this choice produces interpretational difficulties , then one must ask why analyze the historical djia at all ? the 21 company stocks analyzed and presently in the djia ( by the change of april 2004 ) are listed in table [ complist ] together with the their date of entry into the djia as the time period of the data set analyzed .
as previously mentioned , the empirical distributions for @xmath24 are presented in figs .
[ invdist_djia](a ) , [ invdist_sp500](a ) and [ fullnasdaq ] for the djia , the sp500 and the nasdaq respectively for the entire time span available to the authors . furthermore , in figs . [ invdist_djia](b ) and
[ invdist_sp500](b ) we have truncated the full data sets of the djia and sp500 to a shorter historical time period in order to compare the results for the indices with that of the individual stocks .
what one should primarily note in the comparison between the two sets of figures , _
i.e. _ , the longer data sets with the shorter , is that the only significant difference is to be found in the weaker statistics of the shorter periods ; the positions of the maxima for the positive and the negative gains are in both cases roughly just below @xmath25 and @xmath26 days , respectively , _
i.e. _ , a difference of roughly a factor of 2 for all three indices for a return level of @xmath20 . in figs .
[ fig : djia - companies ] we show the waiting time distributions for 4 companies in the djia , which are representative for the distributions obtained for all the companies listed in table [ complist ] .
we see that , for a return level @xmath27 , the value of the optimal investment horizon , _
i.e. _ the position of the peak in the distribution , ranges from around 2 days to around 10 days depending on the company .
more importantly , it is clear from just looking at the figures that , within the statistical precision of the data , the distributions are the same for positive and negative values of @xmath1 . in order to further quantify this invariance with respect to the sign of @xmath1 , we have averaged the ( company ) gain and loss distributions separately in order to obtain an average behavior for the stocks listed in table [ complist ] .
the result of this averaging process ( fig .
[ avestock ] ) is nothing less that an almost perfect agreement between the gain and loss distributions with a peak position around 5 days for both distributions .
this means that the optimal investment horizon for the average of these selected individual stocks is approximately half that of the loss distribution for the djia and approximately one fourth of that for the gain distribution . in other words , it is twice as slow to move the djia down and four times as slow to move the djia up compared to the average time to move an individual stock in the djia up or down .
how can we rationalize these results ?
what we have done in essence is to interchange the operations of averaging over the stocks in the djia and calculating the inverse statistics for the stocks of this index .
since the djia is constructed such that it covers all sectors of the economy of which it seems quite reasonable to assume that a @xmath20 gain / loss in the shares of for example boeing airways in general has nothing fundamentally to do with a corresponding gain / loss in the shares of coca - cola company _ especially _ since the data are detrended .
in other words , it seems quite reasonable to assume that there is nothing special about a return level of @xmath20 in terms of economic fundamentals _ etc .
_ this assumption is also strongly supported by the results presented in fig .
[ optilevel ] .
this then means that the two operations , _
i.e. _ the averaging and the inverse statistics calculation , do not commute not even approximately .
hence significant inter - stock correlations must exist even for a rather modest return level @xmath28 .
in our view , this is a quite surprising result and especially considering that the large differences in the optimal investment horizons for the distributions of the index and the average of the individual stocks .
we have considered inverse statistics for financial data . it is argued that the natural candidate for such statistics is what we call the investment horizon distribution .
such a distribution , obtained from the historic data of a given market , indicates the time span an investor historically has to wait in order to obtain a predefined level of return .
for the three major us markets , namely the djia , sp500 and nasdaq , the distributions are parametrized excellently by a shifted generalized gamma distributions for which the first moment does not exist . the typical waiting time , for a given level of return @xmath1 ,
can therefore be characterized by _
e.g. _ the time position of the maximum of the distribution which we call the _ optimal _ investment horizon . by studying the behaviour of this quantity for positive ( gain ) and negative ( loss ) levels of return ,
a very interesting and pronounced gain / loss asymmetry emerges for all three markets considered .
the nature of this gain / loss asymmetry was further investigated by calculating the investment horizon distribution for @xmath29 of the @xmath30 individual stocks in the djia .
quite surprisingly the observed gain / loss asymmetry in the djia is _ not present _ in the investment horizon distribution for the individual stocks .
specifically , we have shown that if one first `` average '' the stock prices of , say , the djia in order to create the index and then calculate the inverse statistics of it , then one obtains a pronounced gain / loss asymmetry . reversing the order of these two operations , however , makes this asymmetry _
disappear_. admittedly , this has been done systematically for only a single gain and loss level @xmath1 , but it seems unreasonable to claim anything special about the 5% gain / loss level used .
hence , the investment horizon distribution for the individual stocks is _ invariant _ under a change of sign for @xmath1 .
furthermore , the optimal investment horizon for the average of the distributions for the individual stocks is approximately half that of the loss distribution for the entire market ( djia ) and approximately one fourth of that for the gain distribution for the entire market . in other words , it is twice as slow to move the djia down and four times as slow to move the djia up compared to the average time to move an individual stock in the djia up or down .
there are several possible scenarios which may explain the observed behavior .
however , they all amount to more or less the same thing .
a down / up - turn in the djia may be initiated by a down / up - turn in some particular stock in some particular economical sector .
this is followed by a down / up - turn in economically related stocks in other sectors and so forth .
the result is a cascade , or synchronization , of consecutive down / up - turns in all the sectors covered by the djia .
the initiation of this may be some more general new piece of information , which is considered more crucial for one sector than other sectors , but as argued for in length in @xcite it may also happen for no obvious reason what so ever .
an ( rational ) example would be that intel drops significantly due to bad quarterly earnings in turn , by a cascade process , affecting the stock price of ibm and microsoft and so forth .
as the index , at least from a physicist s point of view , can be compared to an external field , movements in the index due to a single or a few stocks can rapidly spread through most or all sectors , _ if _ psychology in general and specifically feed - back loops are important , thus creating a relatively large movements in the over - all index .
that losses in general are faster than gains must also be attributed to human psychology : people are in general much more risk adverse than risk taking . in conclusion
, the results presented here thus provide further evidence for the presence of cooperative behavior ( or synchronization ) possibly with a psychological origin in the stock market beyond what can be deduced from standard financial models .
i.s . would like to thank the nordic institute for theoretical physics ( nordita ) for kind hospitality where the last part of this work was completed .
i. simonsen , a. johansen , and m.h .
investment horizons : a time - dependent measure of asset performance . in h. takayasu ,
editor , practical fruits of econophysics .
springer verlag , 2005 . to appear .
.[complist ] list of the ( @xmath29 ) djia stocks analyzed in this work ( about @xmath31 of the total number ) .
furthermore , their date of entry into the djia are shown , and the time period covered by the analyzed data set .
all of these companies are also naturally part of sp500 with general electric as the most heavily weighted stock.@xmath32former aluminum corporation of america .
@xmath33former american tel . &
truncated due to anti - trust case in 1984 .
@xmath34former travelers group .
@xmath35former standard oil . [ cols="<,<,<",options="header " , ] .
some historic events are marked by dashed vertical lines in the figure .
they are : ( a ) the 1812 war ( 18121814 ) ; ( b ) the second war with the seminole indians ( 18351842 ) ; ( c ) the mexican - american war ( 18461848 ) ; ( d ) the civil war ( 18611865 ) ; ( e ) the spanish american war ( 1898 ) ; ( f ) the first world war ( 19141918 ) ; ( g ) the second world war ( 19401945 ) ( h ) the cold war ( 19471991).,width=510,height=359 ] same as fig .
[ invdist_djia](a ) , but for the nasdaq .
the historical time period considered is 1971.2 to 2004.6 .
again he solid lines represent fit of the empirical data against eq .
( [ fit - func ] ) with parameters : @xmath36 , @xmath37 , @xmath38 and @xmath39 ( loss distribution ) ; @xmath36 , @xmath40 , @xmath41 and @xmath42 ( gain distribution ) .
note again that the tail exponents @xmath22 are very close to the `` random walk value '' of @xmath23 for both distributions.,width=510,height=340 ] the optimal investment horizon @xmath0 for positive ( open circles ) and negative ( open squares ) levels of return @xmath43 for the djia . in the case
where @xmath44 one has used @xmath45 on the abscissa for reasons of comparison .
if a geometrical brownian price process is assumed , one will have @xmath46 with @xmath47 for all values of @xmath1 .
such a scaling behaviour is indicated by the lower dashed line in the graph .
empirically one finds @xmath48 ( upper dashed line ) , only for large values of the return.,width=510,height=321 ] averaged gain and loss distribution for the companies listed in table [ complist ] .
the fit is eq .
( [ fit - func ] ) with values @xmath49 , @xmath50 , @xmath51 and @xmath52 .
note that the tail exponent @xmath22 is @xmath53 above the `` random walk value '' of @xmath23.,width=510,height=340 ]
|
in recent publications , the authors have considered inverse statistics of the dow jones industrial averaged ( djia ) @xcite .
specifically , we argued that the natural candidate for such statistics is the investment horizons distribution .
this is the distribution of waiting times needed to achieve a predefined level of return obtained from detrended historic asset prices .
such a distribution typically goes through a maximum at a time coined the
_ optimal investment horizon _ , @xmath0 , which defines the most likely waiting time for obtaining a given return @xmath1 .
by considering equal positive and negative levels of return , we reported in @xcite on a quantitative gain / loss asymmetry most pronounced for short horizons . in the present paper , this gain / loss asymmetry is re - visited for 2/3 of the individual stocks presently in the djia .
we show that this gain / loss asymmetry established for the djia surprisingly is _ not _ present in the time series of the individual stocks .
the most reasonable explanation for this fact is that the gain / loss asymmetry observed in the djia as well as in the sp500 and nasdaq are due to movements in the market as a whole , _
i.e. _ , cooperative cascade processes ( or `` synchronization '' ) which disappear in the inverse statistics of the individual stocks .
|
long - period variable stars ( lpvs ) are cool giant stars showing more or less periodic light variations with amplitudes of several magnitudes in the visual and with periods of several hundred days . depending on their visual amplitudes and on the regularity of their variability cycles , they appear in several flavours , namely mira ceti - type variables ( mira stars or miras ) , semi - regular variables ( of the sra or srb subtypes ) or irregular variables ( of the lb subtype ) .
mira and sr variables represent one of the latest stages in the evolution of stars with initial masses in the approximate range 1 to 9 . it is known since long that the brightness variations of lpvs go along with spectral changes : ( i ) hydrogen and some metallic lines turn from absorption to emission after minimum light ( merrill 1921 ) , ( ii ) the velocity of the emission and absorption lines correlates with phase and excitation potential ( merrill 1923a ; adams 1941 ) , ( iii ) several absorption lines appear double around maximum light ( adams 1941 ; merrill & greenstein 1958 ; maehara 1968 ) . in a review on red stars ,
merrill ( 1955 ) was the first to suggest that the bright emission lines sometimes appearing in lpvs may be explained by some kind of `` hot front '' moving outward .
he further suggested that this running hot front may have the structure of a shock wave .
although the shock wave scenario is nowadays quite widely accepted ( see de la reza 1986 and references therein ) , the lack of self - consistent pulsation models for lpv stars ( due to the important role played by convection and the difficulty of modelling it ) prevents from grasping exactly how , why and where emission and double absorption lines form . as a consequence , several ( sometimes conflicting ) theories aiming at explaining
the spectral peculiarities of lpvs have appeared in the literature .
although the most popular models locate the formation of emission lines in the hot wake of the shock ( e.g. , gillet 1988a ) , an alternative model proposed by magnan & de laverny ( 1997 ) associates emission lines with purely radiative non - lte processes independently of any shock wave . in this model , the double - peak nature of the emission lines is not connected in any way with the velocity field , contrary to the shock - wave model .
concerning the absorption lines , their doubling around maximum light is most easily seen in the rather clean near - infrared spectral domain ( e.g. , gillet et al .
1985 ; hinkle et al . 1997 and
references therein ) .
long - term radial velocity measurements of the infrared rotation - vibration lines of co have revealed well - defined variations with phase , regularly repeating from one cycle to the next and following a typical s - shaped curve ( hinkle et al .
these radial - velocity variations are generally believed to reflect the differential bulk motions occurring in the large and tenuous atmosphere of lpvs and associated with its pulsation .
alternatively , the complex absorption profiles have also been explained without resorting to differential velocity fields : gillet et al .
( 1985 ) have argued that the doubling of the metallic absorption lines results in fact from the development of an emission core in the line .
this emission component is caused by the radiative release of the thermal energy dumped into the post - shock layer by the shock front .
this model has the advantage of being able to account for the p cygni line profiles sometimes observed in lpv spectra ( and appearing when the emission core becomes brighter than the local continuum ) .
a similar explanation in terms of the shock - induced temperature inversion has been proposed by karp ( 1975 ) for the line doubling observed in cepheid variables .
detailed radiative - transfer calculations have shown that the formation of double absorption lines does not necessarily require a velocity gradient , since a temperature inversion is in principle sufficient .
recent studies have clearly validated theories accounting for the complex absorption line profiles observed in several kinds of pulsating stars ( rrlyrae : chadid & gillet 1996 , fokin & gillet 1997 ; @xmath0cephei stars : mathias et al .
1998 ; blherculis : gillet et al .
1994 , fokin & gillet 1994 ; rvtauri : gillet et al .
1989a ) in terms of velocity fields .
the validity of the scenario accounting for double absorption lines in terms of a velocity gradient in the photosphere remains however to be demonstrated for lpv stars .
the present paper presents the first clear observational evidence thereof .
a definitive way to conclude that double absorption lines are indeed the signature of a velocity gradient in the atmosphere rather than of a central emission core is to check whether the double absorption - line profiles behave as predicted by the scenario .
although ( 1952 ) proposed this scenario in relation with wvir variables , it should be applicable to any kind of variable stars with a shock wave propagating through its photosphere . according to this scenario ,
the intensity of the blue and red components of a double line should follow ( around maximum light ) the temporal sequence illustrated in fig .
[ fig : schwarzschild ] ( note that this scenario implicitly assumes that the processes radiating away the thermal energy dumped into the post - shock layers by the shock front have a negligible impact on the line profile ) .
the observation of such a temporal sequence for the absorption line profiles in miras would definitely point towards the velocity stratification as the origin of the double absorption lines , and moreover would offer a way to directly probe the velocity field associated with the pulsation .
the extreme difficulty in testing whether the scenario operates in lpv stars comes from ( i ) the need for an intensive mid - term ( i.e. , daily for about 2 months around maximum light ) monitoring of lpv stars , ( ii ) the crowded nature of the spectra of mira stars , particularly in the optical domain .
the cross - correlation technique provides a powerful tool to overcome the second difficulty .
indeed , the information relating to the line doubling ( velocity shift and line shape ) is in fact distributed among a large number of spectral lines , and can be summed up into an average profile , or more precisely into a cross - correlation profile , by a cross - correlation algorithm . if the correlation of the stellar spectrum with a specially designed mask involves many lines , it is possible to extract the relevant information from very crowded and/or low signal - to - noise spectra ( e.g. , queloz 1995 ) .
examples of the use of the cross - correlation technique ( with the coravel spectrovelocimeter ) to derive line profiles may be found in gillet et al .
( 1989a , 1990 ) for rvtauri stars and in udry et al .
( 1998 ) for mira stars .
a monitoring of two mira - type stars , rtcyg and xoph , was performed with the fibre - fed echelle spectrograph elodie ( baranne et al .
1996 ) from 1999 , august 3 to october 4 , with an interruption between september , 3 and september , 16 .
both stars were observed each night ( weather - permitting ) , resulting in 32 spectra for both rtcyg and xoph .
elodie is mounted on the 1.93-m telescope of the observatoire de haute provence ( france ) , and covers the full range from 3906 to 6811 in one exposure at a resolving power of 42000 .
cross - correlation functions ( ccf ) with the default k0iii mask have been computed directly in the ( pixels , orders ) space ( i.e. , without merging or rebinning the spectrum , in order not to lose accuracy ) following the prescription of baranne et al .
( 1996 ) . in fig .
[ fig : suivi ] , the nightly sequence of ccf profiles obtained for rtcyg is presented .
rtcyg is a m2e m8.8e mira star belonging to an old ( extended / thick disk or halo ) population ( alvarez et al . 1997 ) , with a rather short period [ 190.28 d according to the _ general catalogue of variable stars ( gcvs ) _ ; kholopov et al . 1988 ] , typical of old populations , and a visual amplitude of 7 mag . its maximum light was reached around 1999 , september 1 according to the aavso light curve ( mattei 1999 , observations from the aavso international database , private communication ) . due to the uncertainties on the epoch of maximum
light ( 23 days ) and since the period is not constant from one cycle to the other , the transformation from julian dates to phases carries some degree of arbitrariness . in the present paper ,
phases were computed assuming that rtcyg reached maximum light on 1999 september , 1 and adopting the gcvs period .
it is worth noting that the light curve of rtcyg shows a bump on the rising branch .
by analogy with rrlyr variables , this may be a signature of a particularly strong shock wave running through the atmosphere ( hill 1972 ; fokin 1992 ; bessell et al . 1996 ) .
it can be seen in fig .
[ fig : suivi ] that the sequence of ccfs follows the scenario sketched in fig .
[ fig : schwarzschild ] : at phase @xmath10.80 , a single absorption peak is seen at @xmath2113 km s@xmath3 ; then , coming closer to maximum light , the ccf becomes more and more asymmetric ( phases 0.800.90 ) , until the blue component becomes clearly visible around phase 0.90 .
the blue peak continues to strenghten and becomes even stronger than the red peak after phase 0.96 .
the intensity of the red component , that remained almost constant until this phase , then starts to fade away .
a visual inspection of a few clean lines in the spectrum has confirmed that the doubling of the ccf is not an artefact of the method , but reflects the doubling of the spectral lines probed by the considered mask .
the ccfs have been fitted by gaussian or double gaussian functions in order to derive the corresponding radial velocity , with the results shown on fig .
[ fig : vr_rtcyg ] .
the single absorption peak with @xmath2113 km s@xmath3 seen at phase 0.80 splits up into a blue component with @xmath2121 km s@xmath3 and a red component with @xmath2109 km s@xmath3 .
the blue component becomes more and more blueshifted until its velocity reaches @xmath2130 km s@xmath3 around phase 0.0 .
after maximum light , the blue shift of the blue component slowly decreases .
the radial velocity of the red component appears to be much more stable than that of the blue component .
the monitoring stopped at phase 0.158 ( see lebzelter et al .
1999 for radial velocities from infrared co and ti lines covering a more extensive fraction of the light cycle ) .
unfortunately , the center - of - mass ( com ) velocity is not known for that star , as there are no submm co observations available for the com velocity if rtcyg is pulsating in the fundamental mode , or @xmath4 for a first - overtone pulsation .
since these values are strongly model - dependent , they will not be used here . ] .
the chances of detecting co lines in rtcyg are weak , though , since its iras 12 @xmath5 m flux ( 6.4 jy ) is more than 10 times smaller than in stars with detected co lines .
the unknown com velocity prevents us from putting constraints on the strength of the shock ( see , however , the analysis of willson et al . 1982 ) . for the classical scenario to hold true , the com velocity needs to fall in between the red and blue peaks ( as it is the case for all the mira variables monitored in the ir by hinkle et al .
1997 , as well as for the theoretical estimate of the com velocity of rtcyg by willson et al . 1982
; see also lebzelter et al .
1999 ) . in that case
, a new shock wave appears at the bottom of the photospheric layers as they are already moving downwards on the ballistic motion induced by the previous shock ( chadid & gillet 1996 ) .
the blue peak observed in rtcyg would then correspond to ascending matter in the envelope , driven by the coming shock wave .
if on the contrary the com velocity would not fall in between the red and blue peaks but both peaks are ` redshifted ' ( a situation encountered at some phases in radial pulsators like rrlyrae stars ; see chadid & gillet 1996 ) , this situation corresponds to a receding shock ( in an eulerian rest frame ; in lagrangian coordinates , the shock is obviously propagating outwards ) .
such a shock is not strong enough to oppose the infalling atmosphere . in principle , the velocity variations of the blue peak convey information about the propagation of the matter just behind the shock .
the shock may either accelerate as it encounters less and less resistance in an atmosphere of decreasing density , or decelerate as part of its energy is dissipated by radiative losses during its upward motion ( gillet et al .
1985 ) . in rtcyg ( fig .
[ fig : vr_rtcyg ] ) , the two trends are observed in succession : the ascending layer which is first accelerating ( phases 0.87 to 0.98 ) decelerates after phase 0.08 .
the same trend may be observed for the layer where the red peak forms , which reacts with some delay to the motion of the underlying ` blue ' layer . at phase 0.9 ,
when the outward motion of the blue layer is the fastest , the inward motion of the red layer is decelerated ( the slope changes from positive to negative on fig . [
fig : vr_rtcyg ] ) as it is colliding with the rising shock .
the outward acceleration stops at phase 0.0 , and the red layer resumes its inward acceleration ( on a possibly ballistic motion ) as the force exerted by the blue layer has vanished ( since the acceleration of the blue layer is now directed _ inwards _ ) . however , the interpretation presented above is very crude as it neglects possible complications coming from optical - depth or geometrical effects .
the line formed in an expanding atmosphere will be asymmetric even in the absence of a velocity gradient ( e.g. , karp 1975 ) , simply because the projection of the expansion velocity on the line of sight varies across the stellar disk . in the presence of velocity gradients ,
the problem is further complicated by the fact that different lines of sight across the stellar disk probe different geometrical depths ( i.e. , layers with different velocities ) at any given wavelength .
moreover , in the presence of shocks , the matter gets compressed , leading to a steep increase in the optical depth . if the location of this steep optical - depth rise changes with respect to the velocity field , velocity variations unrelated to the shock acceleration or deceleration may be recorded .
dynamic models including radiative transfer are therefore ultimately needed to correctly interpret the results presented on fig .
[ fig : vr_rtcyg ] .
the maximum velocity discontinuity ( observed around phase 0.98 ) amounts to 20 km s@xmath3 , translating into 26 km s@xmath3 after correction for limb darkening ( parsons 1972 ) .
similar discontinuities are reported by hinkle et al .
( 1997 ) for the two components of the double co infrared lines observed in many mira stars , which form however in different atmospheric layers than the optical lines sampled by the elodie mask .
willson et al . (
1982 ) have performed a detailed analysis of the velocity structure of the atmosphere of rtcyg , based on photographic spectra covering the optical domain as does elodie .
these authors claim to have observed _ two _ shocks , the deeper one having an amplitude of about 30 , and the upper one about 13 ( see their fig . 5 ) . denoting by @xmath6 and @xmath7 the post- and pre - shock velocities of the deeper shock , and by @xmath8 and @xmath9 the post- and pre - shock velocities of the upper shock , they obtain @xmath10 , @xmath11 , @xmath12 and @xmath13 , relative to the inferred com velocity
. however , this structure relies on the ability to distinguish lines formed in regions a and c separated by only 4 , close to the resolution of the photographic spectrograms used , and moreover , lines formed at b tend to be weak and are not very numerous .
therefore , willson et al . (
1982 ) acknowledge that a single - shock structure is in principle possible as well .
as weak lines are not sampled by the k0iii mask used by elodie , the presence of two shocks in the atmosphere of rtcyg can not be tested with this mask .
the tomographic method , more appropriate to investigate this question as described in sect .
[ sect : tomo ] , provides however no evidence for a second shock from the mask sampling weak lines ( see ccf on first line of fig . [
fig : tomo ] ) . quite interestingly , xoph does not exhibit the same line - doubling phenomenon as rtcyg .
this star is classified as a m5e - m9e mira star , with a mean period of 328.85 days and a visual amplitude of 3.3 mag ( @xmath14 ) according to the _
gcvs_. its maximum light was reached on 1999 september , 16 ( mattei 1999 , observations from the aavso international database , private communication ) .
xoph is a visual binary with a k1 giant companion ( of magnitude @xmath15 ; merrill 1923b ) in an orbit of semi - major axis 0.34 and of period about 500 y ( baize & petit 1989 ) .
according to the aavso light curve , the faintest magnitude reached by xoph during the survey was 7.7 , so the presence of the companion did not disturb in any way our measurements . the sequence of ccfs is presented in fig .
[ fig : suivi ] .
the ccf remained single during the whole monitoring ( a conclusion also reached by hinkle et al .
1984 from infrared co lines monitored over a large fraction of the light cycle ) , and the velocity stayed almost constant at @xmath273 km s@xmath3 ( it varied between @xmath272.8 and @xmath273.4 km s@xmath3 ) .
nevertheless , fig .
[ fig : suivi ] clearly reveals that the contrast of the ccf decreases with advancing phase . from co millimeter observations ,
groenewegen et al . ( 1999 ) derive for xoph a com velocity of @xmath16 with respect to the local standard of rest , or @xmath273.5 with respect to the solar barycentre .
this com velocity is close to the radial velocity derived from the ccf profiles : surprisingly , the photospheric layers probed by the k0iii mask are almost at rest with respect to the com . at this point , two important conclusions may be drawn : ( i ) _ for the first time , clear evidences have been obtained that the mechanism operates in at least some lpv stars ( such as rtcyg ) _ , ( ii ) _ not all lpvs exhibit line doubling around maximum light though_. the second conclusion is confirmed by a survey of 75 lpvs observed around maximum light , which reveals that about 48% of the observed stars did not exhibit the line doubling phenomenon with the k0iii mask ( alvarez et al . , in preparation ) . nevertheless , this result does not necessarily imply that shock waves are not present in the deep atmosphere of those stars . indeed , balmer emission lines , which are commonly believed to be formed in the hot wake of the shock , are observed , although not very intense , in the spectra of xoph ( see sect .
[ sect : balmer ] ) .
the absence of mechanism may either be due to the combined effects of a weak shock and an insufficient resolving power to resolve the small doppler shift between the red and blue peaks ( that possibility seems however unlikely in the case of xoph as the width of the ccf does not change around maximum light ; see fig .
[ fig : suivi ] ) , to optical - depth effects ( the shock is either optically thin or propagates in optically - thick layers ) , or simply to the use of an inadequate template . the latter possibility will be investigated in a forthcoming paper of this series which will also attempt to identify the key stellar parameter(s ) governing the appearance of the line doubling and/or the strength of the shock .
the cross - correlation technique in the optical spectrum opens the possibility of much more detailed studies of pulsating atmospheres .
we briefly describe here the method that we are currently developing to perform the tomography of the envelope of lpv stars . a more thorough description will be presented elsewhere ( alvarez et al . , in preparation ) .
the method rests on our ability to construct reliable synthetic spectra of late - type giant stars , which has made significant progress in recent years ( plez 1999 ) .
the starting point is the ` depth function ' @xmath17 , derived from models of late - type giant stars , and expressing the geometrical depth corresponding to optical depth 2/3 at the considered wavelength @xmath18 .
the different masks @xmath19 are constructed from the collection of @xmath20 lines @xmath21 such that @xmath22 , where @xmath23 is some constant optimized to keep enough lines in any given mask without losing too much resolution in terms of geometrical depth .
each mask @xmath19 should thus probe lines forming at ( geometrical ) depths in the range @xmath24 in the atmosphere .
these masks are then used as templates to correlate with the observed spectra of the lpv stars .
this procedure should provide the velocity field as a function of depth in the atmosphere of the lpv stars .
one crucial requirement of the method is that the masks constructed from static atmospheric models do not lose their ability to probe a layer of a _ given _ ( geometrical ) depth when applied to a dynamic atmosphere .
this point will be discussed in more details in a forthcoming paper .
[ fig : tomo ] displays a sequence of ccfs probing increasing depths in the atmosphere of rtcyg at four phases . at phase 0.905
, it can clearly be seen that the line doubling occurs only for the 3 deepest masks . at a later phase ( 0.953 ) , the line doubling involves 3 more masks further out , translating the upward motion of the shock . at even later phases ( 0.000 and 0.147 ) ,
the line doubling is seen up to the outermost mask . at phase 0.147 , it is worth noting that the deepest mask displays a _ single blue _
peak , while the outermost mask displays a single red peak ( though somewhat asymmetrical ) . in summary
, the scenario can be observed on both the temporal _ and _ spatial variations of the ccfs displayed in fig .
[ fig : tomo ] , thus clearly revealing the upward motion of the shock wave .
the balmer lines are important diagnostics to which the temporal behaviour of the ccf of absorption lines may be compared .
it is not the purpose of this paper to discuss thoroughly the different theories concerning the origin of the emission lines .
nevertheless , thanks to the two - month - long monitoring , the possibility is offered to present a high - quality temporal sequence of balmer lines , which might later on be used to constrain the different existing models .
figure [ fig : rtcyg_balmer ] displays the evolution of the , , and profiles of rtcyg during the monitoring . the abcissa scale is a velocity scale . since the com velocity is not known for rtcyg , the zero point of the velocity scale has been taken equal to a heliocentric velocity of @xmath2110 km s@xmath3 ( corresponding to the mean velocity of the red component of the ccf profiles ; see fig . [
fig : vr_rtcyg ] ) . when analyzing fig .
[ fig : rtcyg_balmer ] , it should thus be remembered that the zero point of the velocity scale has been adjusted so as to correspond to the velocity of the red component of the absorption lines , which is almost certainly different from the stellar com velocity .
the profile is not shown for phase 0.079 because of the too low signal - to - noise ratio achieved in this order .
the profile is seen in emission during the whole monitoring , although weakly so on the first frame corresponding to phase 0.806 . in the early development of the emission feature
, it is obliterated by an absorption feature falling right on top of it . at later phases ,
a separate red component is very clearly present .
such a structure has been observed in several mira stars ( see gillet 1988a ; woodsworth 1995 ) .
the and profiles are rather similar in the first half of the monitoring , but , at the latest phases , the profile shows no red component but instead a very clear pedestal at the base of the line .
the and emissions are much more complex in the early stages of the monitoring .
mutilations of the and profiles are expected to be less pronounced than in the case of and as molecular absorptions are weaker in their spectral domain ( gillet 1988b ) .
figure [ fig : xoph_balmer ] displays the temporal evolution of the balmer lines of xoph during the monitoring .
the com velocity has been taken equal to @xmath273.5 ( see sect . [
sect : monitoring].3 ) .
the vertical - axis scale is the same as in the case of rtcyg to make the comparison with fig .
[ fig : rtcyg_balmer ] easier .
the , and profiles are not shown for phase 0.857 because of the too low signal - to - noise ratio achieved in the blue orders . although is almost as strong in xoph as in rtcyg ( but not quite so ) , is much weaker ( by at least of factor of 2 ) , and and are only barely visible after phase 0.991 .
it is interesting to note that the relative intensities of the balmer lines are similar in xoph and rtcyg .
therefore , it is not surprising that never gets very strong in xoph , since a look at rtcyg shows that , for to be clearly visible , the intensity of needs to exceed 20 ( relative to the local continuum ) which is never the case in xoph .
this may indicate that the same ( recombination ? ) mechanism is operating in both stars to produce the balmer emission lines and that the variation of the underlying opacity with wavelength is the same as well .
the absence of mechanism in xoph and its occurrence in rtcyg may similarly be related to their respective balmer line intensities .
the ccf profile starts becoming clearly asymmetric at phase 0.884 in rtcyg ( fig .
[ fig : suivi ] ) , when the relative intensity of amounts to 3 . as such intensity levels
are never reached in xoph , it is thus not surprising that the ccf remains single , provided of course that the mechanism and the balmer line intensities are indeed somehow correlated . whether or not this is systematically the case remains however to be confirmed on a larger sample .
at this point , it may just be concluded that rtcyg and xoph differ by the presence / absence of the mechanism _ and _ by the strength of the balmer lines .
the comparison of the and absorption - line velocities illustrates the respective merits of the competing models accounting for the complex line profile .
two different models involving a single spherical shock wave have been proposed to explain the origin of the double emission structure observed in mira stars ( gillet 1988b ) .
the first one is the so - called ` geometrical model ' ( willson 1976 ; gillet et al . 1985 ) stating that , while the emission line is produced within the de - excitation zone behind the shock front , the redshifted component is emitted by the shock propagating far above the photosphere in the hemisphere opposite to the observer whereas the blueshifted component is emitted by the shock propagating in the hemisphere facing the observer . in that model ,
the two emission components should be almost centered on the stellar com velocity .
the second interpretation is the self - reversal model ( bidelman & ratcliffe 1954 ; gillet 1988a ) stating that there is a true absorption obliterating the emission line , caused by the presence of cool hydrogen gas above the shock front . in the latter interpretation
, the central absorption should have the velocity of the layers lying above the shock . in the framework of the model ,
those layers are responsible for the red absorption - line component .
as the velocity of those layers approximately corresponds to the adopted zero - point of the velocity scale for the balmer features , the central absorption should have a velocity close to zero , which indeed appears to be the case , as indicated below in relation with fig .
[ fig : velbalmer ] .
an example of profile obtained during the monitoring ( phase 0.953 ) of rtcyg is shown in fig .
[ fig : rtcyg_ha_mod ] . to evaluate the merits of the above two models ,
the complex line profile has been fitted by gaussian functions in two different ways .
the ` geometrical model ' requires to fit separately the blue and red emission components ( _ case a _ on fig . [
fig : rtcyg_ha_mod ] ) whereas the self - reversal model requires to fit separately the broad emission line and the central absorption ( _ case b _ on fig .
[ fig : rtcyg_ha_mod ] ) . figure [ fig : velbalmer ] displays the velocity structure of the line as a function of phase , for the two cases .
it should also be remembered that there is an atomic or molecular absorption feature unrelated to falling right on top of it , as indicated by the arrow in fig .
[ fig : rtcyg_ha_mod ] .
its velocity is also displayed in fig .
[ fig : velbalmer ] for phases earlier than 0.01 , since it is no more visible afterwards .
because these absorption features mutilate the still weak line during the early stages of the monitoring , the fit of the broad emission feature performed for _ case b _ is rather inaccurate and has therefore not been presented on fig .
[ fig : velbalmer ] until phase 0.91 when the fit becomes more reliable . figure [ fig : velbalmer ] offers the interesting possibility to compare the evolution of the structure of the line with the corresponding evolution of the absorption features ( fig .
[ fig : vr_rtcyg ] ) .
the comparison of these two figures reveals interesting similarities in the framework of _ case b_. first , the velocity of the central absorption feature overlying the broad line has a velocity similar to that of the red component of the absorption lines , as expected if both features form in the layers just above the shock .
second , the velocity difference between the broad emission line and the overlying absorption feature ( @xmath25 ) is of the same order as the difference between the velocities of the red and blue peaks of the absorption lines .
this is again consistent with the hypothesis that the broad emission and the blue absorption lines both form in the ascending , post - shock layers . on the contrary
, there is no clear correspondence between the absorption - line velocities and the velocities of the red and blue emission peaks defined in the framework of _ case a_. at this point , one should mention that the interpretation of the line profile according to _ case a _ or _ b _ also results in very different shock - front velocities ( gillet 1988a ) . the geometrical model ( _ case a _ )
yields rather large shock velocities ( of the order of 60 ) , whereas the self - reversal model ( _ case b _ ) involves much smaller shock velocities ( @xmath2630 ; gillet 1988a ) . on theoretical grounds ,
gillet et al .
( 1989b ) have shown that a shock velocity in excess of 60 is required for the shock to be strong enough to first photodissociate h@xmath27 molecules and then photoionize the resulting hydrogen atoms , to produce the observed balmer emission lines by recombination .
this requirement is consistent with the velocities ( 5070 ) deduced from the observations of fluorescent lines in miras ( willson 1976 ) , and would also favour _ case a _ over _ case b_. in the above discussion , _ case a
_ faced however the difficulty that there is no clear coincidence between the and absorption - line velocities , although one would naively expect some if those lines are formed in the vicinity of the same shock .
the above conclusion of discrepant velocities has been reached assuming that both and absorption lines form in a thin layer , so that they can be assigned the velocity of the layer where they form .
however , this assumption might not hold true for absorption lines which form over a broader region than , so that the velocity of the absorption lines result from an average of the velocity field over the line - forming region .
the discrepant and absorption - line velocities observed in the framework of _ case a _ may possibly be reconciled when such averaging effects are properly taken into account .
detailed dynamic models coupled to radiative transfer are thus ultimately needed to answer the question of the origin of the complex line profile .
the origin of double - peaked emission and absorption lines in the complex lpv atmospheres is a long - standing problem . if the shock wave and the associated velocity gradient are generally believed to be at the origin of the complex line profiles of lpv stars ( de la reza 1986 ; gillet 1988b ; hinkle et al .
1997 ) , alternative models not resorting to differential atmospheric motions are sometimes invoked .
they involve either purely non - lte effects to account for the double - peaked balmer emission ( magnan & de laverny 1997 ) , or a temperature inversion to account for the splitting of the absorption lines ( karp 1975 ; gillet et al . 1985 ) .
thanks to the cross - correlation technique and a daily survey of the mira star rtcyg , the role of velocity fields in the formation of genuinely double absorption lines has been undoubtedly confirmed in the framework of the mechanism .
evidence for the existence of strong shock waves in the outer layers of lpvs becomes compelling , although not all mira stars exhibit the scenario , as demonstrated by the observations of xoph collected in the present paper .
the origin of this distinct behaviour will be investigated in the next paper of this series .
the evolution of the , , and emission - line profiles during 35% ( 20% ) of the cycle of rtcyg ( xoph ) around maximum light has also been discussed .
the comparison of this data set with the temporal evolution of the profile of absorption lines offers interesting perspectives to identify the origin of the complex profile of the emission line .
the present paper had no other ambition than providing a description of the data collected with the elodie spectrograph around maximum light for rtcyg and xoph .
these data may be used as constraints for future detailed models .
firm conclusions on the line - formation mechanisms in mira variables should thus await the advent of such detailed dynamic models coupling hydrodynamics to radiative transfer .
we are grateful to x. delfosse , d. erspamer , a. gomez , m. haywood , e. josselin , m. mayor , d. naef , d. segransan , e. oblak and s. udry for performing part of the observations , as well as to the staff of the haute - provence observatory .
ephemeris and light curves provided by the aavso international database were very helpful for planning the monitoring described in the present paper .
benefits of a tmr `` marie curie '' fellowship at ulb
is research associate from the _ fonds national de la recherche scientifique _ ( belgium ) .
queloz d. , 1995 .
_ echelle spectroscopy with a ccd at low signal - to - noise ratio .
_ in : a.g .
davis philip , k.a .
janes , a.r .
upgren ( eds . ) new developments in array technology and applications ( iau symp .
167 ) , kluwer , dordrecht , p.221
|
this paper is the first one in a series devoted to the study of the dynamics of the atmospheres of long - period variable ( lpv ) stars .
results from a two - month - long monitoring of the mira variables rtcyg and xoph around maximum light with the elodie spectrograph at the haute - provence observatory are presented .
the monitoring covers phases 0.80 to 1.16 for rtcyg and phases 0.83 to 1.04 for xoph .
the cross - correlation profile of the spectrum of rtcyg with a k0iii mask confirms that the absorption lines of rtcyg in the optical domain appear double around maximum light .
no line doubling was found in the optical spectrum of xoph around maximum light , indicating that this feature is not common to all lpvs .
this paper also presents the application to rtcyg of a new tomographic technique deriving the velocity field across the atmosphere by cross - correlating the optical spectrum with numerical masks constructed from synthetic spectra and probing layers of increasing depths .
this technique reveals that both the temporal evolution of the line doubling , and its variation with depth in the atmosphere of rtcyg , are consistent with the ` scenario ' .
this scenario relates the temporal evolution of the red and blue peaks of the double absorption lines to the progression of a shock wave in the atmosphere . the temporal evolution of the balmer , , and emission lines around maximum light is also presented for rtcyg and xoph .
the velocity variations of and of the absorption lines are discussed in the framework of two competing models for the formation of balmer emission lines in lpv stars .
|
we consider ranking and selection of systems based on the average performance of the alternatives .
the particular set - up used here is motivated by problems from optimization under uncertainty .
often in operations research as well as in technical applications the performance of solutions depends on some random influence like market conditions , material quality , or simply measurement errors . we shall call such random influences a random _ scenario_. usually , the aim is then to find a solution with minimal expected costs taken over all scenarios .
we assume here that we can describe the random scenario mathematically by a random variable @xmath0 with a distribution @xmath1 .
let @xmath2 denote the cost of solution @xmath3 if it has been applied to random scenario @xmath4 then the average costs are given by the expectation @xmath5 . except for particularly simple cases , we will not be able to calculate this expression analytically . instead , we have to estimate @xmath5 based on a sample @xmath6 where @xmath7 are observations of the random variable @xmath8 , usually produced by a stochastic simulation model on a computer . with estimated costs , optimization can be performed by heuristic search methods like genetic algorithms or ant algorithms .
typically , these methods take a relatively small set @xmath9 of solutions ( a ` population ' ) and try to improve the quality of @xmath10 iteratively .
the improvement step usually includes a _ selection _ of the best or at least the most promising solutions from @xmath10 with respect to their expected costs @xmath11 .
as @xmath12 can only be estimated , this selection will return sub - optimal solutions with a certain error probability .
often , this probability can be made arbitrarily small by increasing the sample size @xmath13 , i.e. the number of simulations .
the aim of this paper is to find selection mechanisms for good solutions suitable for heuristic optimization that have error probabilities below a given bound and use few simulations only .
most heuristic optimization procedures do not need the exact values of @xmath14 in order to perform a reasonable selection , instead it is sufficient to find the correct _ ranking _ @xmath15 or even only part of it . as it has been noted before ( see e.g. @xcite ) it is much easier ( needs less simulations ) to estimate the correct ranking than to estimate the correct values .
methods of ranking and selection are therefore widely used in heuristic optimization under uncertainty , see e.g. @xcite for an overview . in this paper
we have a fixed set @xmath16 of solutions , the observed cost values are assumed to be normally distributed with unknown mean and unknown covariance matrix .
we use a bayesian approach and estimate the ranks of the solutions in each iteration based on the present posterior means .
roughly , the steps of our procedure to rank and select the solutions can be summarized as follows : 1 .
initialization : observe all solutions for a fixed number @xmath17 of scenarios ( = simulations ) .
2 . determine the ranking and selection based on the posterior means given these observations .
calculate a lower bound lb(pcs ) for the _ probability of correct selection _
pcs 4 . while lb(pcs ) @xmath18 do 1 .
allocate a fixed _ budget _ of additional simulations to the solutions , 2 .
recalculate the posterior means , the ranking and selection and lb(pcs ) based on the extended set of observations .
5 . return the last ranking and selection .
input parameters are the allowed error probability @xmath19 , the simulation budget @xmath20 and the initial sample size @xmath17 .
this is the standard framework of a ( bayesian ) computing budget allocation ( cba ) algorithm , see e.g. in @xcite .
our contributions concern the following points which we label for further reference : depend : : we allow for _ dependent observations _ , i.e. we apply the solutions to common random scenarios ( common random numbers , crn ) observing @xmath21 , a vector of possibly dependent , normally distributed values .
target : : we use a very _ general concept of a target _ for ranking and selection that includes targets as median or span of means and still allows to bound the probability of a correct selection in a simple bonferroni fashion .
alloc : : most importantly , we introduce a new _ heuristic allocation scheme _ for additional simulations based on the posterior distribution of the means and the covariance matrix .
this scheme reuses terms that have been evaluated to calculate the lb(pcs ) in the iteration before and therefore requires little additional effort . in the experiments described below
, our algorithm needed less simulations to guarantee lb(pcs)@xmath22 than competitors from literature .
let us discuss these issues in more detail .
* depend * : classical ranking and selection ( r&s ) procedures evaluate the alternatives based on random scenarios drawn _ independently _ for each alternative , see e.g. @xcite .
only a few authors ( see e.g. @xcite ) use _ dependent _ samples based on common random numbers ( crn ) for simulation .
it is well known , that if the observations are _ positively correlated _ , it is more efficient to use common random numbers ( crn ) , i.e. to compare the solutions on the same scenario ( see e.g. @xcite ) .
positive correlation in our case roughly means that if a solution @xmath23 has , for a scenario @xmath24 costs @xmath2 that are above average , then costs @xmath25 will tend to be over average for all the other solutions @xmath26 also .
in other words , if some scenario @xmath27 is relatively difficult ( costly ) for solution @xmath28 then @xmath29 will tend to be difficult for all solutions .
similarly , a scenario that has small costs ( below average ) for one solution will tend to be an easy scenario for all solutions creating smaller costs for all of them .
as this is the behaviour we would expect from a set of solutions at least after some iterations of improvement , it seems to be justified to expect positive correlation and to use crn specific r&s procedures for heuristic optimization .
the numerical results below show that if the cost are positively correlated huge savings in simulation effort are indeed possible and if there is no positive correlation our approach is not worse than other procedures . *
target * : we fix a set @xmath30 of ranks and try to select ( only ) those solutions that have these ranks with respect to the posterior means ( and sort them if necessary ) .
e.g. if we have @xmath31 then we want to select the @xmath32 best solutions ( with minimal means ) .
we describe the sets of possible correct selections in a way that uses a minimal set of pairwise comparisons of solutions , see @xcite for a similar approach .
this allows to give a bonferroni type of lower bound to the posterior probability of correct selection that involve only one - dimensional @xmath33-distributions .
the important and difficult part of this calculation is the determination of the posterior distributions in the presence of incomplete observations due to the possibly unequal allocation of the budget .
* alloc * : the expressions forming the bonferroni lower bound for the pcs may be used to identify pairs of solutions for which the comparison has a large error probability and which need more data for a clear r&s decision .
this results in a simple allocation rule that is simpler to calculate and turned out to be more efficient in our experiments than greedy ocba allocation rules as they are often used in literature ( see e.g. @xcite ) .
this paper is based on the doctoral thesis of one of the authors ( @xcite ) .
it is organized along the three issues outlined above as follows .
* depend * : in the first section we give the exact mathematical description of the sampling process and describe our bayesian model .
details of the rather involved calculation of the posterior distribution with missing data are summarized in an appendix .
* target * : the second section generalizes the concept of ranking and selection of solutions to the case of partial selections . *
alloc * : in section 3 we introduce several schemes for allocating the simulation budget to solutions based on the posterior distributions .
a precise definition of our complete ranking and selection algorithm is given in section 4 .
we report on extensive empirical tests with this algorithm in section 5 where we compare it to a greedy type of ocba algorithm and to the standard @xmath34 procedure of @xcite .
some conclusions are given in the final section .
let @xmath35 denote the fixed set of solutions .
@xmath36 is the @xmath32-th observation of solution @xmath37 , i.e. the cost when @xmath38 is applied to the @xmath32-th random scenario .
we do not consider cost functions and scenarios explicitely in the sequel .
we collect the @xmath36 into a matrix - like object @xmath39 : a column in our data @xmath40 represents the observations of all solutions with a single scenario , whereas the rows represent all observations from a particular solution .
@xmath40 need not be a complete matrix : due to possibly unequal allocation of simulations we may have a ragged right border , see below .
as we use crn , missing values at the end of a row are not really ` missing ' but are rather ` not yet observed ' .
simulating _ all _ of the solutions from @xmath41 with the @xmath32-th scenario therefore leads to a ( column ) vector of observations @xmath42 as it is often done in ranking and selection , we assume that @xmath43 are independent and identically @xmath44-distributed observations . here , @xmath45 denotes the @xmath46-dimensional normal distribution with mean @xmath47 and the positive definite @xmath48 covariance matrix @xmath49 .
this model includes the case , where the @xmath50 solutions are simulated independently ( i.e. with different scenarios ) , then @xmath51 is a diagonal matrix . @xmath52 and @xmath51 are assumed to be unknown . our overall goal is to extract information about @xmath53 from the observations @xmath39 .
typically , we want to identify solutions @xmath37 that have small mean values @xmath12 .
we take a bayesian point of view , that is we assume that the unknown parameters @xmath53 and @xmath51 are themselves observations of random variables @xmath54 and @xmath55 having a prior distribution with density @xmath56 .
we do not assume any specific prior knowledge about the parameters , therefore we shall use the so - called uninformed prior distribution ( see @xcite , @xcite ) with @xmath57 where @xmath58 means that the right hand side gives the density @xmath59 up to some multiplicative constant that does not depend on @xmath53 or @xmath51 .
@xmath60 is a so - called hyperparameter that allows to control the degree of uncertainty about @xmath51 . instead of simulating all solutions equally often
, we iteratively allocate a given amount @xmath20 of simulations to the solutions , depending on the observations seen so far , see the detailed algorithm in section [ sec : thealgorithm ] below . the result
are samples that may contain different numbers of observations for different solutions .
we use @xmath61 to denote the random ( row ) vector of the @xmath62 observations produced for the @xmath38-th solution , @xmath63 , and similarly @xmath64 for a specific sample .
let @xmath65 be the vector of the @xmath46 present sample sizes , then the overall samples can be collected into a matrix - like scheme with possibly different row lengths @xmath62 : @xmath66 columns @xmath67 may be incomplete , as the @xmath32-th simulation ( scenario ) may not have been allocated to all solutions , see figure [ fig : samplingscheme ] for an illustration .
let @xmath68 denote the set of possible samples of that shape for a particular size vector @xmath69 .
we shall now define the possible samples and allocation rules recursively . at first
( in iteration @xmath70 ) , all solutions are simulated @xmath17 times where @xmath17 is a given fixed number .
with @xmath71 we are therefore observing a ( @xmath72-matrix ) sample @xmath73 from @xmath74 next we apply an _ allocation rule _ @xmath75 @xmath76 that determines the numbers @xmath77 of additional simulations for solution @xmath63 in the first iteration
. then the overall number of simulations seen so far is @xmath78 after these additional simulations have been performed we have a sample @xmath79 let @xmath80 be the set of all possible samples after the first iteration , i.e. @xmath81 where the operation @xmath82 concatenates elements row - wise , i.e. @xmath83 generally , after the @xmath84-th iteration we have a sample @xmath85 where @xmath86 gives the total number of simulations performed with each solution so far .
then , an allocation rule @xmath87 is applied to determine the numbers of additional simulations in the @xmath88-st iteration .
the set of possible samples after that iteration is @xmath89 note , that under crn the simulation of different iterations will in general not be independent as we may have to _ reuse _ scenarios ( random numbers ) for some solutions that have already been used in earlier iterations for other solutions .
a new independent scenario is drawn for the simulation of some solution @xmath38 only if @xmath90 has been applied to all scenarios drawn before .
see figure [ fig : samplingscheme ] for a simple example . and
budget @xmath91 .
the allocations of the second iteration are marked in gray , those of the third in black .
the results in the ( complete ) columns @xmath92 are @xmath93-distributed . ,
width=188 ] the allocation schemes we use below depend on the posterior distribution of the mean @xmath54 given data @xmath94 which is determined in the next subsection .
the likelihood function of incomplete samples as described in the last section are examined in great detail and generality in @xcite .
our case is comparatively simple , as we have consecutive observations for each solution ( row in the sample ) and there are no ` holes ' , data may only be missing at the end of the sample .
this is due to the crn scheme we are using : assume a solution @xmath95 has been simulated with the first @xmath32 scenarios resulting in @xmath96 .
if no simulations are allocated to @xmath38 in the next iterations , but additional simulations are performed in later iterations , these have to use scenarios @xmath97 resulting in observations @xmath98 .
also , the pattern of the missing data is a function only of sample values observed before , this is called a monotone and ignorable pattern ( @xcite ) . there is no closed expression for the posterior density of @xmath54 given data @xmath99 if the covariance matrix @xmath51 of the underlying normal distribution is unknown .
we therefore use a semi - bayesian approach and first assume that the covariance matrix is known .
we determine the conditional distribution of @xmath54 given @xmath100 and @xmath101 exactly and then replace @xmath102 by its ( non - bayesian ) estimate @xmath103 . to simplify notation ,
let the set of solutions @xmath104 be ordered such that for the present data @xmath105 with @xmath106 we have @xmath107 this means that solution @xmath108 has been simulated with @xmath109 different independent scenarios . as we use crn , the other solutions @xmath110 have been observed with subsets of these scenarios , see fig .
[ fig : monotonesample ] for a simple example . and @xmath111 .
, width=226 ] we need projections of vectors and matrices to components corresponding to subsets of the solutions @xmath104 . for @xmath112 and @xmath113 let @xmath114}:= ( y_1 , \ldots , y_{i-1})^t\in { \mathbb{r}}^{i-1},\quad \text { and } \quad { \bm{y}}_{[\le i]}:= ( y_1 , \ldots , y_{i})^t\in { \mathbb{r}}^{i}.\ ] ] similarly , for the @xmath115 matrix @xmath116 we define @xmath117}&:=&(\sigma_{kl})_{k , l = 1 , \ldots , i-1 } , \qquad { \bm{\sigma}}_{[i,<i]}:= ( \sigma_{i1 } , \ldots , \sigma_{i , i-1})\quad \text { and } \label{eq : sigmaproj}\\ \beta_i & : = & \sigma_{[i,<i]}\big(\sigma_{[<i]}\big)^{-1 } \label{eq : defbeta}. \end{aligned}\ ] ] here , @xmath118 is a @xmath119-dimensional row vector .
let @xmath120 be the sample mean of all @xmath62 observations @xmath121 available for solution @xmath122 .
for @xmath123 we have @xmath124 therefore we may define the sample mean of solution @xmath38 restricted to its first @xmath125 observations @xmath126 [ fig : samplemeans ] } $ ] contains all @xmath127 sample means from the boxed values.,title="fig:",width=226 ] then @xmath128}:=({\overline{\bm{x}}}^{(n_i)}_{1 } , \ldots,{\overline{\bm{x}}}^{(n_i)}_{i-1})^t $ ] is the @xmath119-dimensional column vector of sample means restricted to the first @xmath62 observations for solutions @xmath129 , that all have at least @xmath62 observations , see fig .
[ fig : samplemeans ] . the posterior distribution of @xmath54 given an incomplete observation @xmath130 with known prior covariance is given in the following theorem , its rather technical proof is sketched in the appendix . [
the : posterior ] let ( the columns ) @xmath131 be @xmath132 distributed with _ known _
covariance matrix @xmath51 and unknown mean @xmath133 .
assume that the variable @xmath54 for the mean has the uninformative prior @xmath134 .
let the ( possibly incomplete ) data @xmath135 with @xmath136 for some @xmath137 , be given as described in section [ subsec : iterativeallocation ] .
then the posterior distribution of @xmath54 given @xmath100 is an @xmath46-dimensional normal distribution @xmath138 with mean @xmath139 where @xmath140 } - \bar{{\bm{x}}}_{[<i]}^{(n_i)})\quad \text { for } i=2 , \ldots , l,\ ] ] and covariance matrix @xmath141}$ ] where @xmath142}=\frac{\sigma_{11}}{n_1},\qquad { \bm{\lambda}}_{[\le i]}= \left(\begin{array}{c|l } { \bm{\lambda}}_{[<i ] } & { \bm{\lambda}}_{[<i]}\ \beta_i^t \\[0.9ex]\hline \rule[0.5ex]{0pt}{2ex}\beta_i\ { \bm{\lambda}}_{[<i ] } & \frac{1}{n_i}(\sigma_{ii}-\beta_i{\bm{\sigma}}_{[<i]}\beta_i^t ) + \beta_i\ { \bm{\lambda}}_{[<i]}\ \beta_i^t \end{array}\right)\ ] ] for @xmath143 . for the case of
an _ unknown _ covariance matrix @xmath51 we have to replace @xmath144}$ ] and @xmath118 by their estimators @xmath145}({\bm{x}})$ ] and @xmath146 . for the estimation of @xmath147}=\big(\sigma_{kl}\big)_{1\le k , l < i}$ ]
we only consider solutions @xmath148 .
these all have sample sizes @xmath149 we may therefore define for @xmath150 and @xmath151 the estimator @xmath152 and put @xmath153}= \hat{{\bm{\sigma } } } _ { [ < i]}({\bm{x}})&:=&\big ( \hat{\sigma}^{(n_{i-1})}_{kl } \big)_{k , l=1 , \ldots , i-1 } \label{eq : defsigest}\\ \hat{{\bm{\sigma}}}_{[i,<i]}=\hat{{\bm{\sigma}}}_{[i,<i]}({\bm{x}})&:= & ( \hat{\sigma}^{(n_{i-1})}_{i1 } , \ldots , \hat{\sigma}^{(n_{i-1})}_{i , i-1}).\notag\end{aligned}\ ] ] note that we use the maximum - likelihood estimator for the covariances and that @xmath154}$ ] is not necessarily contained in @xmath155}$ ] as both estimators may use different sample sizes @xmath156 and @xmath62
. we also have to estimate the @xmath118 and put @xmath157}(\hat{{\bm{\sigma } } } _ { [ < i]})^{-1},\ ] ] where we have to make sure that @xmath158 } $ ] is non singular . from @xcite it is known , that if the sample size @xmath156 of @xmath158}$ ] fulfills @xmath159 then @xmath158}$ ] is positive definite with probability one .
this could be guaranteed if we require that the initial sample size @xmath160 is larger than the number of solutions @xmath161 as then @xmath162 .
we may now plug these estimators into the definition of the posterior means and obtain @xmath163 } - \bar{{\bm{x}}}_{[<i]}^{(n_i)}),\quad \hat{{\bm{\nu}}}({\bm{x}}):=(\hat{\nu}_1 , \ldots , \hat{\nu}_l).\ ] ] to obtain an adequate estimator @xmath164 we first look at the case of _ complete _ observations , i.e. with @xmath165 .
then it is well - known ( see e.g. @xcite , 10.3 ) that with @xmath51 _ known _ and an uninformative prior distribution for the mean @xmath54 , the posterior distribution of @xmath54 would be normal with covariance matrix @xmath166 .
if @xmath51 is _ unknown _ with the improper prior distribution as in the marginal posterior distribution of the mean @xmath54 in the complete observation case is an @xmath46-dimensional @xmath33-distribution with @xmath167 degrees of freedom , location parameter @xmath168 and scale matrix @xmath169 .
here , the correction term @xmath170 involves the hyperparameter @xmath171 , the sample size and the dimension of the problem . for the present situation of incomplete observations and _ known _ covariance matrix
, the posterior covariance @xmath172 has to be replaced by @xmath173 as in , reflecting the different sample sizes for each solution instead of the fixed size @xmath174 .
then , in the case of _ unknown _ @xmath102 , it seems reasonable to approximate the posterior distribution of the mean @xmath54 by an @xmath46-dimensional @xmath175-distribution with @xmath167 degrees of freedom and location parameter @xmath176 .
the scale matrix should be obtained from @xmath173 with the @xmath177 replaced by their adequate estimators . in particular , the correction terms @xmath170 from the complete observation case should be replaced on the @xmath38-th step of the recursion in by @xmath178 using the number @xmath38 of solutions considered in that step .
we thus obtain @xmath179}=\hat{{\bm{\lambda}}}_{[\le 1]}({\bm{x}})&:=\frac{\hat{\sigma}^{(n_1)}_{11}}{n_1 - 1+\nu_0},\label{eq : lambdahat } \\
\hat{{\bm{\lambda}}}_{[\le i]}= \hat{{\bm{\lambda}}}_{[\le i]}({\bm{x } } ) & : = \left(\begin{array}{c|l } \hat{{\bm{\lambda}}}_{[<i ] } & \hat{{\bm{\lambda}}}_{[<i]}\ \hat{\beta}_i^t \\[0.9ex]\hline \rule[0.5ex]{0pt}{2ex}\hat{\beta}_i\ \hat{{\bm{\lambda}}}_{[<i ] } & \frac{1}{n_i - i+\nu_0}(\hat{\sigma}^{(n_i)}_{ii}-\hat{\beta}_i\hat{{\bm{\sigma}}}_{[<i]}\hat{\beta}_i^t ) + \hat{\beta}_i\ \hat{{\bm{\lambda}}}_{[<i]}\ \hat{\beta}_i^t \end{array}\right ) \notag\end{aligned}\ ] ] for @xmath143 , and @xmath180}({\bm{x}})$ ] . in this paper ,
the posterior distribution of the mean @xmath54 is used mainly to express the probability of a wrong selection , see below .
in particular , we need the posterior probability of the event @xmath181 for @xmath182 and @xmath183 . from the properties of the multivariate @xmath33-distribution we may conclude that @xmath184\ = \
{ \text{\textbf{p}}}[w_i - w_j\le \delta\mid { \bm{x}}={\bm{x}}]\\ & \approx \
g(\delta;\ \min\{n_i , n_j\}-1,\ ; \hat{\nu}_i-\hat{\nu}_j,\ ; \hat{\lambda}_{ii}+\hat{\lambda}_{jj}-2\hat{\lambda}_{ij})\notag\\ & : = p_{ij}^\delta\notag\end{aligned}\ ] ] where @xmath185 is the distribution function of the one - dimensional @xmath186-distribution with @xmath32 degrees of freedom , location parameter @xmath187 and scale parameter @xmath20 .
this approximation will be used below .
we start with a general definition of the ranking and selection schemes we are examining in this paper . it is a generalization of the approach in @xcite .
we want to select solutions from the set @xmath188 according to their ranks under some performance measure , in our case the mean value .
e.g. we want to select the solution with the lowest mean value ( the best solution ) or the @xmath189 best solutions as it is often used e.g. in ant algorithms .
we restrict ourselves to such selections that can be determined using pairwise comparisons of solutions only .
the idea is to formulate _ minimal _
requirements for the correctness of a selection in terms of pairwise comparisons and then use to determine the error probability .
this may allow a tighter error bound compared to a complete ranking of all solutions .
we first introduce an abstract concept which is illustrated by an example below . for a general ranking and selection scheme ( r&s - scheme )
we first define a set of _ target ranks _ @xmath190 .
this means that we want to select those solutions that have ranks from @xmath191 with respect to their mean values . as the mean
values are assumed to be unknown , we use their available estimates @xmath192 instead .
these are ordered in increasing order @xmath193 and the solutions with estimated ranks in @xmath191 are selected , i.e. the solutions @xmath194 the aim is then to determine the error probability of the selection @xmath195 i.e.the posterior probability that a solution @xmath196 does not have a rank from @xmath197 .
as we have continuous posterior distributions for which @xmath198 = 0\ ] ] we may restrict ourselves to the @xmath46-dimensional euclidean space without equal values : @xmath199 let @xmath200 be the set of all @xmath46-dimensional vectors , in which the selected positions from @xmath201 have ranks @xmath191 .
thus , @xmath202 is the set of all potential means for which the selection @xmath201 would be correct for target @xmath191 . recall that @xmath203 is the random variable indicating the unknown mean . with the help of @xmath202 we can therefore express the probability that @xmath201 is a correct selection for target @xmath191 simply as @xmath204 = { \text{\textbf{p } } } [ \{\text{rank}(w_l)\mid l \in b\}=a \mid { \bm{x}}={\bm{x}}].\ ] ] for an easier access to this probability we assume that there is a binary _ relation _ @xmath205\ \cup\ [ ( { \mathcal{l}}-b)\times b]\ ] ] on pairs @xmath206 that relates selected positions @xmath196 to unselected positions @xmath207 such that @xmath202 has the following representation @xmath208 where as usual @xmath209 is true if @xmath210
. thus @xmath211 provides a definition of a correct selection by pairwise comparison of certain pairs @xmath212 .
note that @xmath201 and hence @xmath202 and @xmath211 all depend on the observation @xmath130 .
the following examples explain these definitions .
assume that holds . 1 .
let @xmath213 this means we want to select the best solution . then with , the selection is @xmath214 as @xmath215 is the smallest posterior mean . with @xmath216
we have from @xmath217 @xmath202 is the set of all potential mean value vectors where selection @xmath218 leads to the best ( minimal ) solution .
2 . if @xmath219 for some @xmath220 we want to select the @xmath189 best ( minimal ) solutions .
then from we see that we must choose @xmath221 . as characterizing relation
we obtain @xmath222 and @xmath223 3 .
if , in addition , the @xmath189 best solutions have to be ranked among themselves then we would choose @xmath224 and @xmath225 this example includes complete ranking of the solution with @xmath226 and @xmath227 4 .
if we want to select the median with @xmath228 we take as selection @xmath229 the median solution with respect to posterior means and as auxiliary sets @xmath230 and @xmath231 .
then we use as characterizing relation @xmath232 and obtain @xmath233 as the set of vectors from @xmath234 where about half of the components is less than @xmath235 and the rest is larger , therefore @xmath235 is the median .
5 . as a final example let @xmath236 denote a target that could be used to describe the span of the mean values of the solution .
then the selection should be @xmath237 from and @xmath238 and @xmath239 in the iterative procedure below , the posterior means @xmath240 have to be determined after each iteration based on the new observations .
therefore , @xmath241 and @xmath202 have to be adapted in each iteration also .
we now define the probability of correct selection ( pcs ) for a selection as in simply by @xmath242.\ ] ] to determine pcs directly , one has to solve high dimensional integrals in order to arrive at the marginal distribution of the posterior distribution on the components of the selection @xmath201 .
instead , we use the relation @xmath243 that defines the set @xmath202 to derive a simple lower bound of bonferroni type for the pcs as follows @xmath244\notag\\ & = & { \text{\textbf{p}}}\big [ w_i \le w_j \text { for all } 1\le i , j\le l \text { with } \rho_{ab}(i , j)\mid { \bm{x}}={\bm{x}}\big]\notag\\ & = & 1- { \text{\textbf{p}}}\big [ w_i > w_j \text { for some } 1\le i , j\le l \text { with } \rho_{ab}(i , j)\mid { \bm{x}}={\bm{x}}\big]\notag\\ & \ge & 1- \sum_{(i , j)\in \rho_{ab } } { \text{\textbf{p } } } [ w_i > w_j \mid { \bm{x}}={\bm{x } } ] \label{eq : bonferroni}\\ & = & 1- \sum_{(i , j)\in \rho_{ab } } \big(1-{\text{\textbf{p}}}[w_i - w_j\le 0 \mid { \bm{x}}={\bm{x}}]\big ) . \notag \end{aligned}\ ] ] using we may approximate this lower bound by the following expression @xmath245 it is quite usual in ranking and selection to relax the definition of correctness of a selection . instead of strict comparisons of means
@xmath246 for relevant pairs @xmath210 we are content if @xmath247 for some fixed _ indifference zone _ value @xmath248 .
this allows to take the limited precision of our observations into account .
formally , we then define @xmath249 and have , using @xmath250\notag\\ & \ge & 1- \sum_{(i , j)\in \rho_{ab } } { \text{\textbf{p } } } [ w_i >
w_j+\delta { } \mid { \bm{x}}={\bm{x } } ] \notag \\ & \approx & 1- \negthickspace\negthickspace \sum_{(i , j)\in \rho_{ab } } \negthickspace\big(1-g ( \delta { } ; \min\{n_i , n_j\}-1,\ ; \hat{\nu}_i-\hat{\nu}_j,\ ; \hat{\lambda}_{ii}+\hat{\lambda}_{jj}-2\hat{\lambda}_{ij})\big)\label{eq : deflbdelta}\\ & = & 1- \negthickspace\negthickspace \sum_{(i , j)\in \rho_{ab}}(1-p_{ij}^\delta)\ = : lb^\delta({\bm{x } } ) \notag \end{aligned}\ ] ]
in each of the iterations described in section [ subsec : iterativeallocation ] it has to be decided how many simulations @xmath251 should be performed for each solution @xmath37 .
we first look at a modified version of the so - called optimal computing budget allocation ( ocba ) strategy introduced by chen ( see for example @xcite ) .
ocba allocation strategies try to allocate a fixed simulation budget @xmath252 in such a way that the expected value of the @xmath253 is maximized .
let @xmath254 with @xmath255 denote an allocation of the @xmath20 simulations to solutions @xmath256 . calculating the expected effect of additional simulations and to find @xmath257 that are strictly optimal with respect to that expectation is a difficult non - linear optimization problem , see e.g.@xcite .
therefore , most versions of the ocba in the literature solve a substitute problem and try to maximize an approximation @xmath258 to the lower bound as in for given data @xmath130 and additional simulations @xmath259 . in this sense ,
a myopically optimal allocation is the solution to the following optimization problem : @xmath260 in general , even this simplified problem can be solved exactly only for very small instances of @xmath261 and small simulation budgets @xmath262 . for larger instances ,
heuristic solution approaches must be used , see e.g. @xcite .
we adapt this approach to our environment with dependent observations .
the difficulty lies in estimating the future covariances _
after _ the additional simulations @xmath259 have been performed .
we use the lower bound @xmath263 where the scale parameter @xmath264 is a weighted mean of the posterior covariances : @xmath265\hat{\lambda}_{ij}.\end{aligned}\ ] ] here it is assumed that the posterior variance of @xmath266 after @xmath259 additional simulations can be estimated from the present posterior values @xmath267 and @xmath268 by entering the variances @xmath269 inversely proportional to the relative increases of the sample sizes and by scaling the covariance @xmath270 such that the influence of the sample sizes @xmath271 is proportional to the posterior variances @xmath272 and @xmath273 . a greedy heuristic ( greedyocba ) to find good solutions to the problem
may then be defined similar as in @xcite and @xcite .
greedyocba first measures the benefit of assigning all @xmath20 additional simulations to a single solution @xmath274 : @xmath275+\sum_{i\in{\mathcal{l}}:(i , l)\in\rho_{ab}}\left[p^\delta_{il}(0,b)-p^\delta_{il}(0,0)\right],\end{aligned}\ ] ] where the @xmath276 are defined in and @xmath277 is the @xmath278-th unit vector which has a @xmath108 at the @xmath279-th position and zeros elsewhere . then , greedyocba distributes the simulation budget proportional to the weights @xmath280 , i.e. @xmath281 a remaining budget @xmath282 is allocated according to the largest remainder method .
this allocation strategy requires @xmath283 evaluations of the distribution function @xmath284 in total .
hence its computational effort is about twice as high as the calculation of the bonferroni bound @xmath285 in .
this complexity can be reduced if we do not try to estimate the future pcs but simply look at its present value , more precisely its components @xmath286 $ ] , defined in .
@xmath287 describes the posterior probability of @xmath288-dominance of @xmath289 over @xmath290 .
recall that for @xmath210 we have to make sure the @xmath291-dominance @xmath181 in order for the present selection @xmath201 to be correct with indifference parameter @xmath291 .
hence a small value of @xmath292 for @xmath293 indicates that at least one of the solutions @xmath294 has insufficient data .
we therefore define the weights @xmath295 then the budget is allocated proportionally to these weights as above . here
, those solutions @xmath279 get a larger weight that are part of a pair @xmath296 that has a small dominance probability and where the partner of @xmath279 has a relatively small variance .
this implies that @xmath279 has a relatively large variance and hence needs more simulations to increase the certainty of the dominance relation @xmath297 .
we call this strategy dominance probability weighting ( @xmath298 ) . in empirical tests
@xmath298 proved slightly better than greedyocba with respect to the number of simulations needed , at the same time it needs less computational effort as only the @xmath299 from the bonferroni bound are used .
it turned out in the experiments that the performance could be improved further by stopping simulations for those solutions with high enough dominance probabilities .
therefore we restricted the definition of @xmath300 in to those @xmath287 which are below @xmath301 where @xmath302 is the number of summands in the bonferroni bound .
the allocation strategy @xmath303 therefore uses the weights @xmath304 and then allocates simulations as in .
empirical results on greedyocba and @xmath303 are given in section [ sec : computational - study ] below .
we are now in the position to define our algorithm in detail .
let the following items be given : @xmath104 is the set of solutions , @xmath190 is the target set of ranks to be selected and possibly ranked , @xmath305 is the bound for the error probability , @xmath306 is the simulation budget for each iteration , @xmath307 is an initial sample size and @xmath291 is the indifference zone parameter .
0.5 cm * initialization * : : : observe @xmath308 for @xmath309 i.e. choose @xmath17 random scenarios and apply all solutions to each of them .
let @xmath130 denote the result .
+ determine the posterior means @xmath310 as in , the selection @xmath311 as in , relation @xmath312 as in and and the dominance probabilities @xmath313 , as in .
finally calculate the lower bound @xmath314 for the pcs as in . *
while * : : @xmath315**do * * + * apply an allocation rule @xmath316 to determine the number of additional simulation runs @xmath317 , * perform @xmath251 simulations with solution @xmath37 , taking into account the common random numbers scheme as described in section [ subsec : iterativeallocation ] , let @xmath130 be the extended data including the new simulation results , * update the posterior means @xmath176 , the selection @xmath311 , the relation @xmath243 , the dominance probabilities @xmath313 , and the lower bound @xmath314 .
: : * return * the present selection @xmath311 . in @xcite it is shown that this algorithm terminates if at least one simulation is allocated to each solution in each iteration .
we implemented the algorithm bayesrs and compared its efficiency to other r&s procedures .
the main objectives of the study were : * to compare different allocation strategies within the framework of bayesrs , here we use greedyocba , @xmath303 and equalallocation , the naive equal allocation of the budget to solutions , * to show the increasing efficiency of @xmath303 with increasingly positive correlation of the observations , * to compare bayesrs with allocation rule @xmath303 to another standard from literature , namely @xmath34 , a procedure designed to find the best solution , see @xcite , and * to investigate the behaviour of our approach in cases where there is no positive correlation .
it seems that the ( approximately ) correct determination of the posterior distribution for missing values makes the allocation more precise , even if , in the case of negative correlation , common random numbers may increase the variance of the estimators used .
we measured the performance by the average number of simulations each strategy needs with a common error bound @xmath19 .
also we checked the empirical error probabilities or rather pcs . the tests were implemented using the free statistics software ` r ` . in @xcite further studies show the efficiency of these allocation strategies in the context of heuristic optimization methods like ant algorithms .
as our methods require normally distributed observations , we generated them by a @xmath318-random generator for different values of @xmath53 and @xmath51 .
some values were fixed : the number of solutions @xmath319 ( other values showed similar behaviour ) , the error probability @xmath320 and the indifference zone parameter @xmath321 .
the initial sample size was @xmath322 , the budget to be allocated in each iteration was @xmath323 , and at least one simulation was allocated to each solution in each iteration .
the parameter @xmath171 for the prior probability of the covariance matrix as in showed best results in our setup for @xmath324 and was therefore fixed to this value throughout our tests .
.different values for @xmath325 were used in the @xmath53 -case `` ufc '' . [ cols="<,<",options="header " , ] a _ basic scenario _ consists of the following four variables that are varied in the tests : 1 .
we have examined three different * r&s - cases * : in `` best@xmath326 '' we want to select the best solution , i.e. we use a target set @xmath327 ( see subsection [ subsec : targetandselection ] ) , in `` best@xmath328 '' we want to select the better half of the solutions ( @xmath329 ) and in `` rank@xmath328 '' we also want to rank these ten best solutions .
also three different * @xmath53-cases * have been used for the @xmath318-generator . in the case ``
inc '' , @xmath53 is the increasing sequence @xmath330 . in the unfavourable case `` ufc '' , @xmath53 is adapted to the r&s - case chosen as described in table [ tab : mucase ] . in the case
`` unif '' , @xmath46 values are drawn randomly from the interval @xmath331 $ ] with a minimal distance of at least @xmath332 and then sorted into increasing order .
this is repeated @xmath333 times , so that 15 simulations with different random @xmath53 are performed . as we use a bayesian approach
, the `` unif''-case seems to be the most natural to simulate the uninformative prior but the cases may differ remarkably depending on how well the random values are separated .
3 . as our method claims to work
well when observations are positively correlated , we created covariance matrices @xmath334 for the @xmath318-generator with a given joint correlation @xmath335 @xmath336 . to do so ,
variances @xmath337 were chosen uniformly distributed in the interval @xmath338 $ ] , then we put @xmath339 for @xmath340 . to examine the robustness of our procedure when there are negative correlations , we complemented the above construction for @xmath341 with @xmath342 resulting in a covariance matrix with alternating positive and negative entries . finally we used a random covariance matrix following a wishart distribution with @xmath46 degrees of freedom and a random scale matrix .
these make nine different @xmath51**-cases**. 4 . finally , we distinguish two * crn - case * : we apply the complex calculation for the posterior means and covariances based on the joint distribution as described above ( case `` iscrn '' ) or we simply use the marginal posterior distributions as if no crn had been used . in this case
( `` nocrn '' ) the posterior means and variances are estimated by standard non - bayesian estimators .
this would be correct in the uncorrelated case with @xmath343 , but it is a simplification in the other cases .
this results in 162 different scenarios . for each scenario , @xmath344
random covariance matrices @xmath51 were generated according to the @xmath51-case chosen and , if the @xmath345-case `` unif '' was used , also @xmath346 random vectors @xmath53 were generated .
for each pair of @xmath347 , @xmath348 repetitions of the different allocation strategies were performed .
we kept track of the random seeds so that all strategies used the same random observations .
-case : comparison of the allocation strategies equalallocation , greedyocba and @xmath349 using the full posterior distribution of the unknown means.,width=340 ] we first show the results for @xmath53-case `` ufc '' with crn - case `` iscrn '' . in figure
[ fig : ufc_compfulldistrabs ] the mean number of simulations necessary to obtain a pcs @xmath351 with bayesrs is shown for the three allocation strategies equalallocation , greedyocba and @xmath349 .
the @xmath352-axis shows the nine different @xmath353-cases repeated for each of the three r&s - cases `` best@xmath326 '' , `` best@xmath328 '' and `` rank@xmath328 '' . here
, the @xmath102-case with the random covariance matrix is indicated by rnd. the number of simulations on the @xmath354-axis is the mean over the @xmath355 different covariance matrices , each with the given correlation , and the @xmath348 repetitions for each scenario .
figure [ fig : ufc_compfulldistrabs ] clearly shows that the intelligent allocation rules are much more efficient than the simple equal allocation .
also in all scenarios , our new allocation rule @xmath349 performed better than the classical greedyocba .
this holds for all @xmath356-cases , including the ones involving negative correlations as well as the random covariance matrix .
increasing the correlation from @xmath357 to @xmath358 reduced the necessary effort , for all strategies .
figure [ fig : ufc_compmargdistrabs ] shows @xmath53-case `` ufc '' with crn - case `` nocrn '' .
we use the same multi - normal observations as before , but this time the posterior distribution is based on the marginal distribution of the observation only , neglecting the possible dependence . in particular , the complicated definition of the posterior covariance matrix @xmath359 in is replaced by a diagonal matrix @xmath360 with diagonal @xmath361 as defined in .
note that this is justified only in the @xmath356-case @xmath362 , the only case where the observation for different solutions are independent .
-case : comparison of the allocation strategies equalallocation , greedyocba and @xmath363 . here , the posterior distribution is calculated using marginal distributions only.,width=340 ] nevertheless , figure [ fig : ufc_compmargdistrabs ] shows a similar picture as figure [ fig : ufc_compfulldistrabs ] , @xmath349 is superior in all scenarios , but the absolute values of the number of simulations needed is larger . in figure
[ fig : ufc_compmargfullabs ] we therefore compare the crn - cases `` iscrn '' and `` nocrn '' for the best allocation strategy @xmath349 and @xmath53-case `` ufc '' .
figure [ fig : ufc_compmargfullabs ] shows that it is indeed worth calculating the full posterior distribution in order to save simulations .
-case : comparison of the two crn - cases for strategy @xmath363.,width=340 ] finally , figure [ fig : ufc_emppcs ] shows the relative frequency of correct selections and rankings in the above experiments . as it can be seen , it is always well above @xmath364 , indicating that our setup is quite conservative .
-case : the empirical frequency of correct selections in the above experiments.,width=340 ] next we look at the results for @xmath53-case inc , the results are quite similar .
figure [ fig : inc_compfulldistrabs ] shows the crn - case `` iscrn '' and again , @xmath349 performs best , though for the two first r&s - cases best@xmath326 and best@xmath328 , greedyocba has about the same performance .
-case `` inc '' : comparison of the allocation strategies equalallocation , greedyocba and @xmath349 using the full posterior distribution of the unknown means.,width=340 ] figure [ fig : inc_compfullmarginabs ] again compares the result for @xmath365 for the estimators based on full posterior distributions and the ones using only marginal information .
the advantage of the full distribution is obvious .
the empirical pcs was well above 95% in all cases .
-case `` inc '' : comparison of the crn - cases for the strategy @xmath349.,width=340 ] in the `` unif''-case , results are means over @xmath346 different @xmath53 drawn randomly from @xmath331^l$ ] , over @xmath355 random covariance matrices and @xmath366 repetitions .
results vary considerably depending on the difficulty of the random @xmath53 , so that the mean number of simulations gives only a rough picture of the performance .
figure [ fig : unif_compfulldistrabs ] shows that our new strategy @xmath349 is superior to the other strategies for positively correlated observations and with random covariance matrix , but it uses more simulations than greedyocba for @xmath367 . in figure
[ fig : unif_emppcs ] , it can be seen , that @xmath368 and greedyocba have high empirical pcs except for one of the cases with @xmath367 .
-case `` unif '' : comparison of the allocation strategies equalallocation , greedyocba and @xmath349 using the full posterior distribution of the unknown means.,width=340 ] -case `` unif '' : the empirical pcs.,width=340 ] in @xcite ( see also @xcite ) the sequential procedure @xmath34 is introduced .
it is suitable to identify the best solution by iteratively excluding solutions that seem inferior to at least one of the other solutions . in our tests , @xmath34 needed much more simulation runs than our procedure bayesrs with allocation @xmath368 , in particular in the @xmath53-case ufc , where @xmath34 was not able to find the best solution within our limit of 150000 simulations . to make results better comparable
, we biased the setup in favor of @xmath34 . for @xmath34
, we used a larger indifference zone parameter @xmath369 in the @xmath53-cases ufc and inc , and @xmath370 the actual minimal distance in @xmath53 for unif-case , thus adapting the parameter to the step between best and second best solution .
for @xmath371 , we used @xmath332 for all @xmath53-cases as before .
figure [ fig : compdpwkn ] shows the mean number of simulations needed for the r&s - case best@xmath326 and the three @xmath53-cases . to make the figure better readable , we scaled down the results for @xmath34 in the unif-case by a factor @xmath372 . in the @xmath53-case
unif , we performed only @xmath373 repetitions for @xmath374 covariance matrices and @xmath375 different @xmath53-values to save runtime . with the relaxed indifference zone parameter @xmath369 ,
@xmath34 was considerably faster than @xmath368 in the
ufc-case , but missed the empirical pcs of @xmath376 in most of these cases , as can be seen in figure [ fig : compdpwkn_pcs ] .
hence , one may conclude that in most cases our new approach is superior to @xmath34 with respect to the number of simulations needed and empirical error probabilities observed . and @xmath377 for the r&s - case best@xmath326 .
here , @xmath34 has a larger indifference zone parameter @xmath291 for the first two cases and its
unif-values are shown multiplied by @xmath378 . ,
width=340 ] and @xmath377 for the r&s - case best@xmath326 . here , @xmath34 was below @xmath376 for most covariances in the @xmath53-cases
ufc and unif .
, width=340 ]
in this paper we presented a new sequential bayesian r&s procedure with support for common random numbers .
based on an approximation of the posterior distribution of the unknown mean and covariance , the simulation effort could be allocated to solutions in such a way that , empirically , a given pcs can be obtained with less simulations than are needed by other strategies .
in our future work we will extend this concept to multivariate selection problems .
essential parts of this problem were solved in @xcite .
we need the following well - known facts about multivariate normal distributions , see e.g. @xcite or @xcite , sec .
let @xmath379 be a random vector with distribution @xmath380 .
let @xmath381 and @xmath382 be a partition of @xmath383 for some @xmath384 and define @xmath385 so that @xmath386 note that with the notation from , we may write @xmath387}$ ] and @xmath388}$ ] . then the marginal distributions of @xmath389 and @xmath390 are normal : @xmath391 moreover , the conditional distribution of @xmath390 given @xmath392 is a @xmath393-dimensional normal distribution @xmath394 with @xmath395 the converse does also hold : if @xmath396 are random variables such that @xmath397 has a @xmath398-distribution and the conditional distribution of @xmath399 given @xmath400 is as in , then the joint distribution of @xmath401 is @xmath402 . in particular , if @xmath403 so that @xmath404 then we have @xmath405 @xmath406 and @xmath407 } & ( \sigma_{1d } , \ldots , \sigma_{d-1,d})^t\\ ( \sigma_{d1 } , \ldots , \sigma_{d , d-1 } ) & \sigma_{dd } \end{pmatrix } = \begin{pmatrix } { \bm{\sigma}}_{[<d ] } & { \bm{\sigma}}_{[d,<d]}^t\\ { \bm{\sigma}}_{[d,<d ] } & \sigma_{d , d } \end{pmatrix } , \ ] ] where we have used the symmetry of the covariance matrix @xmath51 and the notation from .
then from , the conditional distribution of @xmath408 given @xmath409 is a one - dimensional normal distribution with mean @xmath410}^{-1}\big((y_1 , \ldots , y_{d-1})^t-(\mu_1 , \ldots , \mu_{d-1})^t\big ) \notag\\ & = \mu_d + \beta_d({\bm{y}}_{[<d]}-{\bm{\mu}}_{[<d]})\label{eq : defmutilde}\end{aligned}\ ] ] and variance ( independent of @xmath411 ) @xmath412}^{-1 } ( \sigma_{d1 } , \ldots , \sigma_{d , d-1})^t \label{eq : defsigmatilde}\\ & = & \sigma_{dd } - \beta_d { \bm{\sigma}}_{[<d ] } \beta_d^t.\notag \end{aligned}\ ] ] then the likelihood function of the unknown mean @xmath53 given data @xmath130 is @xmath417}-\bar{{\bm{x}}}_{[<i]}^{(n_i ) } ) , \frac{\tilde{\sigma}_{(i-1)}}{n_i } ) \ ] ] where @xmath418 is the one - dimensional normal density with mean @xmath187 and variance @xmath20 . as we assume that the covariance matrix @xmath51 is known , the likelihood function for @xmath419 given the sample @xmath420 as in can be written as a product of independent blocks of normal densities of different dimensionality .
let @xmath421 be the number of solutions with at least @xmath62 observations .
@xmath422 is the height of the data blocks with at least @xmath62 observations , see fig .
[ fig : datablocks ] .
then with @xmath424 we have for the likelihood function @xmath425 } , { \bm{\sigma}}_{[\le l_i]}\big),\ ] ] where @xmath426 is the density function of the @xmath279-dimensional normal distribution . here
, the inner product describes the likelihood for the sample values within one of the blocks as marked in fig .
[ fig : datablocks ] . note that may contain empty products ( defined to be @xmath108 ) if @xmath427 .
now each of the @xmath422-dimensional densities in can be factorized into @xmath422 factors of one - dimensional conditional densities .
we write @xmath428 for the density of the conditional distribution of @xmath383 given @xmath429 . for @xmath430
let @xmath431 then using and we obtain @xmath432 } , { \bm{\sigma}}_{[\le l_i]}\big)\notag \\ & = f_{x_{1k}}(x_{1k})\cdot f_{x_{2k}|x_{1k}}(x_{2k}|x_{1k})\cdot \ldots\label{eq : nfactorization } \\ & \hspace*{3cm}\cdot f_{x_{l_ik}|x_{1,k } , \ldots , x_{l_i-1k}}(x_{l_ik}|x_{1k } , \ldots , x_{l_i-1k})\notag\\ & = \phi_1(x_{1k};\ \mu_1,\sigma_{11})\cdot \phi_1(x_{2k};\ \tilde{\mu}_{(1)}(x_{1k}),\tilde{\sigma}_{(1 ) } ) \cdot \ldots \notag\\ & \hspace*{3 cm } \cdot \phi_1\big(x_{l_ik};\ \tilde{\mu}_{(l_i-1)}(x_{1k } , \ldots , x_{l_i-1,k}),\tilde{\sigma}_{(l_i-1)}\big)\notag\\ & = \phi_1(x_{1k};\ \mu_1,\sigma_{11})\cdot \phi_1(x_{2k};\ \tilde{\mu}_{(1)}({\bm{x}}_{[<2,k]}),\tilde{\sigma}_{(1 ) } ) \cdot \ldots \notag\\ & \hspace * { 3 cm } \cdot\phi_1\big(x_{l_ik};\ \tilde{\mu}_{(l_i-1)}({\bm{x}}_{[<l_i , k]}),\tilde{\sigma}_{(l_i-1)}\big)\notag \end{aligned}\ ] ] where @xmath433}:=(x_{1k } , \ldots , x_{i-1,k})^t .
$ ] plugging into and reordering terms we get @xmath434}),\tilde{\sigma}_{(i-1)}).\ ] ] dropping terms that do not depend on @xmath53 ( remember that @xmath51 was assumed to be known ) we obtain from straightforward calculations
@xmath435}),\tilde{\sigma}_{(i-1)})&\propto\notag\\ \phi_1(\bar{{\bm{x}}}_i;&\ \mu_i+\beta_i(\bar{{\bm{x}}}_{[<i]}^{(n_i ) } - { \bm{\mu}}_{[<i ] } ) , \tilde{\sigma}_{(i-1)}/n_i)\label{eq : prop2}\end{aligned}\ ] ] for @xmath436 . from , and we therefore obtain @xmath437}^{(n_i ) } - { \bm{\mu}}_{[<i ] } ) , \tilde{\sigma}_{(i-1)}/n_i ) \notag\\ & = \phi_1(\mu_1;\ \bar{{\bm{x}}}_{1 } , \sigma_{11}/n_1 ) \cdot \prod_{i=2}^l \phi_1(\mu_i;\ \bar{{\bm{x}}}_i + \beta_i
( { \bm{\mu}}_{[<i]}-\bar{{\bm{x}}}_{[<i]}^{(n_i ) } ) , \tilde{\sigma}_{(i-1)}/n_i).\label{eq : posteriorfactorized}\end{aligned}\ ] ] moreover , from we see , that is the factorized representation of a @xmath46-dimensional normal density @xmath440 where the mean @xmath441 and covariance matrix @xmath173 are such that the operations in lead to the representation in . to find @xmath442 and @xmath173 we have to undo the operations from
this is possible since we see from that @xmath443}-\bar{{\bm{x}}}_{[<i]}^{(n_i ) } ) , \tilde{\sigma}_{(i-1)}/n_i).\label{eq : dichtewix } \end{aligned}\ ] ] from this it is also obvious , that @xmath444},w_{[<i ] } } ( { \bm{\mu}}_{[<i]}\mid { \bm{x}}_{[<i,{\,\bm{\cdot}\,}]})\end{aligned}\ ] ] where @xmath445 and @xmath446}$ ] denotes all observations for solutions @xmath447 .
this means that the conditional expressions do not depend on observations for solutions @xmath448 resp .
therefore we can conclude for the means @xmath450 @xmath451\ = \ { \text{\textbf{e}}}[w_1 \mid { \bm{x}}_{1{\,\bm{\cdot}\,}}={\bm{x}}_{1{\,\bm{\cdot}\ , } } ] = \bar{{\bm{x}}}_{1 } \label{eq : nu1}\\ \nu_i&= { \text{\textbf{e}}}[w_i \mid { \bm{x}}={\bm{x}}]\ = \ { \text{\textbf{e}}}[w_i \mid { \bm{x}}_{[\le i,{\,\bm{\cdot}\,}]}={\bm{x}}_{[\le i,{\,\bm{\cdot}\ , } ] } ] \label{eq : nu2}\\ & = { \text{\textbf{e}}}\big[{\text{\textbf{e } } } [ w_i\mid { \bm{x}}_{[\le i,{\,\bm{\cdot}\,}]},w_{[<i ] } ] \
\big|\ { \bm{x}}_{[\le i,{\,\bm{\cdot}\,}]}={\bm{x}}_{[\le i,{\,\bm{\cdot}\,}]}\big]\notag\\ & = { \text{\textbf{e}}}\big [ \bar{{\bm{x}}}_i + \beta_i ( w_{[<i]}-\bar{{\bm{x}}}_{[<i]}^{(n_i ) } ) \ \big|\ { \bm{x}}_{[\le i,{\,\bm{\cdot}\,}]}={\bm{x}}_{[\le i,{\,\bm{\cdot}\,}]}\big]\notag\\ & = \bar{{\bm{x}}}_i + \beta_i\big({\text{\textbf{e}}}\big[w_{[<i]}\ \big|\ { \bm{x}}_{[\le i,{\,\bm{\cdot}\,}]}={\bm{x}}_{[\le i,{\,\bm{\cdot}\,}]}\big ] -\bar{{\bm{x}}}_{[<i]}^{(n_i)}\big)\notag\\ & = \bar{{\bm{x}}}_i + \beta_i ( { \bm{\nu}}_{[<i ] } - \bar{{\bm{x}}}_{[<i]}^{(n_i)})\qquad \text { for } i=2 , \ldots , l. \notag \end{aligned}\ ] ] for the @xmath452 posterior covariance matrix @xmath453=\big({\text{cov}}[w_k , w_l\mid { \bm{x}}={\bm{x}}]\big)_{k , l=1 , \ldots , l}\ ] ] we again put @xmath454 } : = \big({\text{cov}}[w_k , w_l\mid { \bm{x}}={\bm{x}}]\big)_{k , l=1 , \ldots , i}.\ ] ] then we have for @xmath455 that @xmath456 } $ ] equals @xmath457 \\ { \bm{\lambda}}_{[<i ] } & \qquad\vdots\\ & { \text{cov}}[w_{i-1},w_i| { \bm{x}}={\bm{x}}]\\[0.9ex]\hline \rule[0.6ex]{0pt}{2ex}{\text{cov}}[w_i , w_1| { \bm{x}}={\bm{x}}]\ \ldots\ { \text{cov}}[w_i , w_{i-1}| { \bm{x}}={\bm{x } } ] & { \text{cov}}[w_i , w_i| { \bm{x}}={\bm{x } } ] \end{array}\right)\ ] ] to determine the entries of this matrix , we use the following facts about the conditional covariance which are straightforward to prove .
let @xmath458 and @xmath459 be random variables such that all integrals exist and let @xmath460 be a suitable function : @xmath461&= { \text{\textbf{e}}}[u\cdot v\mid z]-{\text{\textbf{e}}}[u\mid z]\cdot { \text{\textbf{e}}}[v\mid z],\notag\\{\text{cov}}\big[u , \big(v+g(z)\big)\ \big|\ z\big]&={\text{cov}}[u , v\mid z],\notag \\ { \text{cov}}[u , v\mid z]&= { \text{cov}}\big[u,\ { \text{\textbf{e}}}[v\mid y , z]\;\big |\ , z\big ] \ \text { if $ u = g(y ) , $ } \label{eq : covprop1}\\ { \text{cov}}[u , u\mid z]&= { \bm{v}}[u\mid z]\label{eq : covprop2}\\ & = { \text{\textbf{e}}}\big[{\bm{v}}[u\mid z , y]\;\big |\ , z\big ] + { \bm{v}}\big[{\text{\textbf{e}}}[u\mid z , y]\,\big |\ ; z\big].\notag \ ] ] we now have from @xmath462}&={\text{cov}}[w_1,w_1\mid { \bm{x}}={\bm{x}}]\notag\\ & = { \bm{v}}[w_1\mid { \bm{x}}={\bm{x}}]= \frac{\sigma_{11}}{n_1}={\text{cov}}[w_1,w_1\mid { \bm{x}}_{1{\,\bm{\cdot}\,}}={\bm{x}}_{1{\,\bm{\cdot}\ , } } ] \label{eq : lambda1}\end{aligned}\ ] ] and similarly from and @xmath463 = { \bm{v}}[w_i\mid { \bm{x}}={\bm{x}}]\notag\\ & = { \text{\textbf{e}}}\big [ { \bm{v } } [ w_i|{\bm{x } } , w_{[<i ] } ] \;\big|\ , { \bm{x}}={\bm{x}}\big]+{\bm{v}}\big[{\text{\textbf{e}}}[w_i\mid { \bm{x } } , w_{[<i]}]\;\big|\ , { \bm{x}}={\bm{x}}\big]\label{eq : covecke}\\ & = { \text{\textbf{e}}}\big [ \frac{\tilde{\sigma}_{(i-1)}}{n_i } \;\big|\ , { \bm{x}}={\bm{x}}\big]\notag\\ & \hspace * { 2cm}+{\bm{v}}\big[\bar{x}_i+\beta_i(w_{[<i]}-\bar{x}_i^{(n_i ) } ) \;\big|\ , { \bm{x}}_{[<i,{\,\bm{\cdot}\,}]}={\bm{x}}_{[<i,{\,\bm{\cdot}\,}]}\big]\notag\\ & = \frac{\tilde{\sigma}_{(i-1)}}{n_i } + { \bm{v}}\big[\beta_i w_{[<i ] } \;\big|\ , { \bm{x}}_{[<i,{\,\bm{\cdot}\,}]}={\bm{x}}_{[<i,{\,\bm{\cdot}\,}]}\big]\notag\\ & = \frac{1}{n_i}(\sigma_{ii}-\beta_i{\bm{\sigma}}_{[<i]}\beta_i^t ) + \beta_i\;{\text{cov}}\big[w_{[<i]}\mid { \bm{x}}_{[<i,{\,\bm{\cdot}\,}]}={\bm{x}}_{[<i,{\,\bm{\cdot}\,}]}\big]\ ; \beta_i^t\notag\\ & = \frac{1}{n_i}(\sigma_{ii}-\beta_i{\bm{\sigma}}_{[<i]}\beta_i^t ) + \beta_i \ { \bm{\lambda}}_{[<i]}\ \beta_i^t\notag.\end{aligned}\ ] ] finally , for @xmath464 we have using and @xmath465 = { \text{cov}}\big[w_k , w_i\ \big|\ { \bm{x}}_{[<i,{\,\bm{\cdot}\,}]}={\bm{x}}_{[<i,{\,\bm{\cdot}\,}]}\big]}\label{eq : covspalte}\\ & = & { \text{cov}}\big[w_k,\ { \text{\textbf{e}}}[w_i\mid w_{[<i]},{\bm{x}}_{[\le i ] } ] \
\big|\ { \bm{x}}_{[<i,{\,\bm{\cdot}\,}]}={\bm{x}}_{[<i,{\,\bm{\cdot}\,}]}\big]\notag\\ & = & { \text{cov}}\big[w_k,\ \bar{x}_i+\beta_i(w_{[<i]}-\bar{{\bm{x}}}_{[<i]}^{(n_i)})\ \big|\ { \bm{x}}_{[<i,{\,\bm{\cdot}\,}]}={\bm{x}}_{[<i,{\,\bm{\cdot}\,}]}\big]\notag\\ & = & { \text{cov}}\big[w_k,\ \beta_i w_{[<i]}\ \big|\ { \bm{x}}_{[<i,{\,\bm{\cdot}\,}]}={\bm{x}}_{[<i,{\,\bm{\cdot}\,}]}\big]\notag\\ & = & \beta_i \big({\text{cov}}[w_k , w_l\ \big|\ { \bm{x}}_{[<i,{\,\bm{\cdot}\,}]}={\bm{x}}_{[<i,{\,\bm{\cdot}\,}]}]\big)^t_{l=1 , \ldots , i-1}\notag\\ & = & \beta_i\ { \bm{\lambda}}_{[<i]}.\notag\end{aligned}\ ] ] from - and we now obtain the recursion @xmath466}=\frac{\sigma_{11}}{n_1},\qquad { \bm{\lambda}}_{[\le i]}= \left(\begin{array}{c|l } { \bm{\lambda}}_{[<i ] } & { \bm{\lambda}}_{[<i]}\ \beta_i^t \\[0.9ex]\hline \rule[0.5ex]{0pt}{2ex}\beta_i\ { \bm{\lambda}}_{[<i ] } & \frac{1}{n_i}(\sigma_{ii}-\beta_i{\bm{\sigma}}_{[<i]}\beta_i^t ) + \beta_i\ { \bm{\lambda}}_{[<i]}\ \beta_i^t \end{array}\right)\ ] ] for @xmath467 which concludes the proof of theorem [ the : posterior ] .
jrgen branke , stephen e. chick , and christian schmidt . . _ management science _ , 530 ( 12):0 19161932 , dec 2007 .
doi : 10.1287/mnsc.1070.0721 .
url http://mansci.journal.informs.org/cgi/doi/10.1287/mnsc.1070.0721 .
chun - hung chen , hsiao - chang chen , and liyi dai . .
in john m. charnes , douglas j. morrice , daniel t. brunner , and james j. swain , editors , _ proceedings of the 1996 winter simulation conference _ , pages 398405 .
ieee , 1996 .
isbn 0 - 7803 - 3383 - 7 .
doi : 10.1109/wsc.1996.873307 .
chun - hung chen , jianwu lin , enver ycesan , and stephen e. chick . .
_ discrete event dynamic systems _ , 100 ( 3):0 251270 , 2000 .
doi : 10.1023/a:1008349927281 .
url http://www.springerlink.com/index/p132u04nm122g459.pdf .
chun - hung chen , donghai he , michael fu , and loo hay lee . . _ informs journal on computing _ , 200 ( 4):0 579595 , sep 2008 .
doi : 10.1287/ijoc.1080.0268 .
url http://joc.journal.informs.org/cgi/doi/10.1287/ijoc.1080.0268 .
hsiao - chang chen , liyi dai , chun - hung chen , and enver ycesan . .
in sigrun andradottir , kevin j. healy , david h. withers , and barry l. nelson , editors , _ proceedings of the 1997 winter simulation conference _ , pages 334341 .
ieee , 1997 .
isbn 0 - 7803 - 4278-x .
doi : 10.1109/wsc.1997.640417 .
stephen e. chick and koichiro inoue . .
_ management science _ , 470 ( 8):0 11331149 , aug 2001 .
doi : 10.1287/mnsc.47.8.1133.10229 .
url http://mansci.journal.informs.org/cgi/doi/10.1287/mnsc.47.8.1133.10229 .
christian schmidt , jrgen branke , and stephen e. chick . .
in franz rothlauf and others , editors , _ applications of evolutionary computing _ , pages 752763 . 2006 .
doi : 10.1007/11732242_73 .
url http://www.springerlink.com/index/jk184206618h1248.pdf .
|
we want to select the best systems out of a given set of systems ( or rank them ) with respect to their expected performance . the systems allow random observations only and we assume that the joint observation of the systems has a multivariate normal distribution with unknown mean and covariance .
we allow dependent marginal observations as they occur when common random numbers are used for the simulation of the systems .
in particular , we focus on positively dependent observations as they might be expected in heuristic optimization where ` systems ' are different solutions to an optimization problem with common random inputs . in each iteration , we allocate a fixed budget of simulation runs to the solutions .
we use a bayesian setup and allocate the simulation effort according to the posterior covariances of the solutions until the ranking and selection decision is correct with a given high probability . here , the complex posterior distributions are approximated only but we give extensive empirical evidence that the observed error probabilities are well below the given bounds in most cases .
we also use a generalized scheme for the target of the ranking and selection that allows to bound the error probabilities with a bonferroni approach .
our test results show that our procedure uses less simulations than comparable procedures from literature even in most of the cases where the observations are not positively correlated .
* keywords : * sequential ranking and selection , common random numbers , bayesian statistics , multiple testing , missing data
|
the problem of the quantum dynamics of a small system interacting with its environment has been extensively studied since the origin of quantum mechanics . however , in spite of the noticeable progresses in the theory of open quantum systems , many conceptual difficulties still remain . indeed
the understanding of the effects of the environment on the physical system of interest , in general , is not an easy task .
the conventional way to afford the problem of the description of the time evolution of an open quantum system consists in eliminating the degrees of freedom of the environment in order to derive an equation of motion for the reduced density matrix of the small system .
this procedure stems from the fact that we are usually only interested in the dynamics of this subsystem and do not really care about the state of the environment and its evolution .
the procedure of tracing over the environmental variables thus amounts at declaring that everything other than the small system is indifferent to us . in this way one reduces the number of variables one needs to take into account to study the problem .
only very few physical systems prove to be amenable to an analytic description of their open dynamics . in this paper
we deal with the simplest open quantum system worth being studied : the damped harmonic oscillator .
such a system is of great conceptual importance because it provides a simple , successful starting point in the theoretical description of many experimental situations in the quantum optics , solid state physics and quantum field theory contexts . for this reason it is also one of the most extensively studied physical systems @xcite .
understanding the dissipative behaviour of open quantum systems is a crucial problem for both fundamental and applicative reasons . on one hand , indeed , during the last few years , the interpretation of zurek , associating the quantum to classical transition with environment induced decoherence effects , has become increasingly popular . from this point of view , studying the open system dynamics of exemplary quantum systems , i.e. harmonic oscillator , would help in identifying the elusive border between quantum and classical descriptions of the world . on the other hand ,
the huge advances in experimental techniques for controlling the evolution of quantum systems have paved the way to the realization of the first quantum logic gates , the key elements of quantum computers .
the biggest obstacle to building quantum computers is the decoherence process due to the unavoidable coupling with the external environment .
thus , studying a decohering quantum computer is an essential step for the identification of realistic quantum error correcting codes @xcite .
a standard method to describe the effects of the environment on a small quantum system is based on the born - markov master equation .
such an approach is valid whenever the environmental correlation time is much shorter than the typical time scales of the system dynamics . under these conditions
, it is possible to derive and , in some cases , to solve analytically the master equation ruling the dynamics .
this is for example the case of a quantum harmonic oscillator coupled through bilinear interactions to a reservoir of harmonic oscillators .
this model is a particular case of the quantum brownian motion ( qbm ) , where we specify the system to be just a quantum harmonic oscillator @xcite .
there exist situations , however , wherein the reservoir correlation time is longer than the system time scales of interest and thus the born
markov approximation does not hold anymore .
this is for instance the case of atoms decaying in photonic band gap materials or atom lasers @xcite .
moreover , very recently , the potential interest of non markovian reservoirs for quantum information processing has been demonstrated @xcite and a non
markovian description of quantum computing , showing the limits of the markovian approach , has been presented @xcite .
several procedures have been developed to treat non markovian processes .
most of them are reviewed in @xcite .
very often , in order to describe the dynamics of an open quantum system in terms of a master equation for its density matrix , a number of approximations involving the relevant system and reservoir parameters are required .
one of the most common is the weak coupling approximation based on the assumption of a weak system - environment coupling strength .
as underlined by paz and zurek in @xcite , a perturbative approach in the coupling strength can always be shown to lead to master equations local in time .
this fact could appear surprising at a first sight since often non
markovian quantum systems are described by generalized master equations involving a non local memory kernel taking into account the past history of the reduced system .
the advantage of dealing with a generalized master equation which is local in time is that the memory effects of the environment are incorporated in its time dependent coefficients @xcite .
this circumstance makes it possible , in some cases , to find an analytic solution of the generalized master equation . in this paper
we present an original method for the derivation of the operatorial density matrix solution of the non markovian master equation for qbm in the weak coupling limit .
we apply our procedure to the case in which the system is a harmonic oscillator interacting with the quantized reservoir through a bilinear coupling .
our method does not rely on any assumption other than the weak coupling limit .
in particular it is independent of the type of environment considered and on the characteristic parameters of the system , provided a perturbative approach to the second order in the coupling constant is valid .
we demonstrate the usefulness and simplicity of our analytic solution discussing , as an example , the dissipative dynamics of the mean energy of the harmonic oscillator . the paper is organized as follows . in section
ii we introduce the mathematical formalism of superoperators and we present some key properties for solving the generalized master equation for our system . in section iii
we describe the non markovian master equation for qbm in the weak coupling limit and in section iv we derive its operatorial solution . in section v we discuss some approximated forms of the density matrix and we present some applications . finally in section
vi we present conclusions .
let us begin introducing the mathematical formalism and properties which we use in the rest of the paper to derive the time dependence of density matrix of the system under scrutiny .
our method is based on the extension of the notion of eigensolution of an operator in the _
_ superoperator__formalism .
+ given two generic operators @xmath0 and @xmath1 , let us define the _ superoperators of the s - type @xmath2 and of the @xmath3-type @xmath4 _ in the following way : [ super ] @xmath5 , \qquad \text{$s$ type}\ ] ] @xmath6 where the squared and curl brackets indicate the commutator and anti - commutator respectively . eqs . and define a particular class of superoperators hereafter also called commutator ( @xmath7 ) and anticommutator ( @xmath3 ) superoperators .
it is important to note that a @xmath7-superoperator @xmath8 may be linearly combined with or multiplied by a @xmath3-superoperator @xmath9 giving rise to superoperators belonging neither to the @xmath7 class nor to the @xmath3 class . from eqs .
and linearity follows immediately @xmath10 and the proprieties [ comrel ] @xmath11=[\hat{a},\hat{b}]^s,\ ] ] @xmath12=[\hat{a},\hat{b}]^{\sigma},\ ] ] @xmath13=[\mathbf{a}^{\sigma},\mathbf{b}^s]=[\hat{a},\hat{b}]^{\sigma}.\ ] ] the main consequence of such relations is that if the commutator of two operators is a c - number then the corresponding @xmath7 ( @xmath3 ) superoperators do commute.for example , for the super - operators @xmath14 and @xmath15 , corresponding respectively to position @xmath16 and conjugate momentum @xmath17 operators , the following relation holds @xmath18=0.\ ] ] moreover , from eq .
one gets @xmath19=[\mathbf{x}^{\sigma},\mathbf{p}^s]=2\imath . \label{scalar2}\ ] ] another superoperatorial identity extensively used in the rest of the paper is : @xmath20 and , in particular , @xmath21 in a certain sense superoperators can be considered as a generalization of the concept of operator used in quantum mechanics .
superoperators act on a space whose elements are operators to give other operators .
let us consider a generalization of the concept of eigenstates .
looking at the equations [ auto ] @xmath22 we can identify @xmath23 as the eigenoperator of the superoperators @xmath24 and @xmath25 with eigenvalues @xmath26 and @xmath27 respectively .
+ it is not difficult to show that the operators @xmath28 define a complete basis @xcite in the operatorial space associated to the single particle hilbert space . such a basis is orthonormal with respect to the scalar product between two generic operators @xmath29 and @xmath30 , defined as follow @xmath31 since @xmath32 it is then possible to express the density - matrix of a single particle in the following form @xcite @xmath33 where the scalar function @xmath34 , known in the literature as _ quantum characteristic function _ ( qfc )
@xcite , satisfies the following properties : @xmath35 another superoperatorial eigenvalue equation of interest in this paper is @xmath36 where @xmath37 .
in particular we have @xmath38 in the following we will use , for the sake of simplicity , a matrix representation of the previous relations . let us define @xmath39 with this notation eqs . , and can be recast in a more compact form @xmath40 @xmath41,\ ] ] @xmath42 in these equations @xmath43 is the following @xmath44 matrix @xmath45 and thus satisfies the properties @xmath46 and @xmath47 .
note that , using this notation , the superoperator @xmath48 can be cast in the form @xmath49 .
in this section we specify the physical system we wish to study , namely a single harmonic oscillator interacting with a quantized environment , and we introduce and discuss the generalized master equation governing its dynamics .
let us consider a harmonic oscillator of frequency @xmath50 surrounded by a generic environment .
we express the total hamiltonian @xmath51 as follows @xmath52 where @xmath53 , @xmath54 and @xmath55 are the system , environment and interaction hamiltonians respectively and @xmath56 is the coupling constant .
the interaction hamiltonian here considered has a simple bilinear form with @xmath16 position operator of the system and @xmath57 generic environmental operator .
+ for the sake of simplicity we have written the previous expressions in terms of adimensional position and momentum operators .
let us denote with @xmath58 the density matrix operator for the oscillator - environment system .
let us now assume that 1 . at @xmath59 system and environment
are uncorrelated , that is @xmath60 , with @xmath1 and @xmath61 density matrices of the system and the environment respectively ; 2 .
@xmath62=0 $ ] ( stationarity of the environment ) ; 3 .
@xmath63 ( as for example in the case of a thermal reservoir ) .
a second order perturbative approach in the coupling constant is possible under these conditions one can show @xcite that the non - markovian generalized master equation describing the harmonic oscillator dynamics , in the schrdinger picture , is the following @xmath64\hat{\rho}(t ) , \label{eqs1}\ ] ] where the superoperator @xmath65 ( not @xmath7 or @xmath3 type ) is defined as @xmath66 also @xmath67 is neither an @xmath7-type nor a @xmath3-type superoperator being defined as follows @xmath68 in eqs . and we have introduced the correlation @xmath69 and the susceptibility @xmath70 functions @xcite .
such quantities , characterizing the environmental temporal behavior , are defined as follows @xmath71 @xmath72 \rangle , \label{eq : seci16b}\ ] ] with @xmath73 .
the form of eq . has a clear physical meaning .
the superoperator @xmath65 , indeed , is strictly related to diffusion ( decoherence ) processes only @xcite .
the superoperator @xmath67 , on the other hand , describes dissipation and frequency renormalization processes @xcite .
such a superoperator arises from a quantum mechanical treatment of the environment and , indeed , vanishes when the environment is treated as a classical quantity ( see also eq . ) .
first of all , we concentrate our attention on @xmath67 . by using the properties of the superoperators introduced in the previous section
, one can show that such a superoperator can be recast in the following form @xmath74^s -\imath\gamma(t)\left(\mathbf{n}+2\right ) , \label{dissi}\ ] ] where the superoperator @xmath48 is defined by eq . and
@xmath75 as for @xmath65 , it is straightforward to see from eq .
that it can be recast in the form @xmath76 where @xmath77 in view of eq . , @xmath24 and @xmath25 do commute . as a consequence
, @xmath65 can be regarded as a quadratic form in the commutative operator variables @xmath24 and @xmath25 .
this means that , using the matrix notation introduced in eq . , we can write @xmath78 it is not difficult to check that @xmath79=0 $ ] , whatever @xmath80 and @xmath81 are . inserting eq . into eq
. one can write the generalized non - markovian master equation describing the dissipative dynamics of our system in the following final form @xcite @xmath82\hat{\rho}(t ) , \label{eqs}\ ] ] with @xmath83.\ ] ] let us note , first of all , that such a master equation is local in time , even if non - markovian .
this feature is typical of all the generalized master equations derived by using the time - convolutionless projection operator technique @xcite or equivalent approaches such as the superoperatorial one sketched in this section .
let us have a closer look at the form of our master equation .
. shows the appearance of two terms , produced by the interaction with the environment , modifying the free hamiltonian of the system . the first one , proportional to @xmath84 , is a time dependent renormalization of the frequency of the oscillator while the second one , proportional to @xmath85 , describes a coupling between the the @xmath16 and @xmath17 operators .
the perturbation of the free dynamics due to these terms can be easily visualized in phase space , as shown in fig .
( [ ellipse ] ) . indeed
, the effect of the terms proportional to @xmath84 and @xmath85 is equivalent to a compression and rotation in phase space of the circle describing the free oscillator dynamics . ) , width=5,height=5 ] from a dynamical point of view one can show that the term proportional to @xmath85 gives rise to both a further frequency renormalization and a dynamical dephasing between position and momentum of the oscillator ( see fig .
[ ellipse ] ) .
note that these features are analogue to the ones present in the dynamics of a classical dissipative oscillator @xcite .
in quantum mechanics there exists a well established procedure to determine the dynamical evolution of a given closed physical system .
it prescribes to write down the schrdinger equation and calculate eigenstates and eigenvalues of the hamiltonian of the system .
once determined these quantities it is possible to describe the dynamics evaluating the action of the corresponding temporal evolution operator on the initial state .
the dynamics of open systems , however , is much more complicated since the description of the state of the system in terms of a state vector is not sufficient anymore . in order to describe completely the physical system
one needs indeed to introduce the density matrix operator whose time evolution is governed by a master equation definitely more difficult to handle than the schrdinger equation . in this section
we present a new method to solve the generalized master equation , given by eq .
, for the system of interest in the paper .
in some sense , the method we describe can be seen as a generalization to open systems of the procedure for solving the schrdinger equation of a closed system , since it is based on the solution of an appropriate eigenvalue equation .
indeed we will deal , in the formalism of superoperator , with generalized eigenvalue equations involving superoperators and operators , instead of operators and vectors respectively . as we will see in this section ,
the existence of some useful algebraic properties of the superoperators will help us in treating the problem of the open system dynamics allowing , in particular , to find the operatorial solution for the density matrix of the system .
a remarkable virtue of the procedure we are going to describe is its independence on the expression of the time - dependent coefficients appearing in the master equation .
this implies that it is applicable to all the master equations presenting the same structure .
our method for solving the master equation consists of two steps which can be summarized as follows : \1 ) singling out the temporal evolution superoperator corresponding to eq . ; \2 ) understanding how this temporal evolution superoperator acts on @xmath86 . in the previous section
we have introduced the master equation related to the dissipative non - markovian dynamics of our system : @xmath82\hat{\rho}(t ) .
\label{eqs2}\ ] ] a formal solution of eq . can be written as @xmath87 where the temporal evolution superoperator @xmath88 is given by @xmath89\:dt_1 \right).\ ] ] in the pedix cstands for the dyson chronological order . in the following
we shall prove that @xmath88 can be factorized as follows : @xmath90 where [ fact2 ] @xmath91 , \\
\mathbf{t}_{\gamma}(t)&= & exp\left[\frac{\gamma(t)}{2}\left(\mathbf{n}+2\right)\right],\\ \mathbf{t}_d(t)&=&exp\left[-\int_0^t e^{\gamma(t_1)}\mathbf{\bar{d}}(t_1)\:dt_1\right],\end{aligned}\ ] ] with @xmath92 to this aim we exploit the feynman s rule @xcite stating that , whatever the operator or superoperator @xmath93 and @xmath94 are , @xmath95=exp_c\left [ \int a(t)\:dt\right]\:exp_c\left [ \int \bar{b}(t)\:dt\right],\ ] ] with @xmath96\right)^{-1}\:b(t)\:\left(exp_c\left [ \int a(t)\:dt\right]\right).\ ] ] applying the feynman s rule to the time evolution superoperator defined in eq . and
taking in consideration that , as shown in appendix a , @xmath97=0 $ ] , one gets @xmath98 where @xmath99\:dt_1\right ] .
\label{f2}\ ] ] applying again feynman s rule in eq . and using the property ( see appendix a ) @xmath100=-2\mathbf{\bar{d}}(t)$ ] , leading to @xmath101 we have @xmath102 finally , inserting eq . into eq .
one obtains the final factorized form of the time evolution superoperator given by eq . .
the key advantage of such a factorized form is that it makes easier estimating the action of the time evolution superoperator on the initial density matrix of the system .
as we shall see in the following , this allows not only to solve the master equation , but also to clarify the physical origin of each of its terms .
it is worth noting that the dyson chronological order is present in only one of the three superoperators defined eqs . ,
of course , this circumstance leads to a further simplification of the calculations .
in fact , while for @xmath103 the reason why one can drop the subscript cis already clear from eq .
, in the case of @xmath104 some more comment is needed . as we will demonstrate in the following ,
the dyson chronological order is not necessary in the expression of @xmath105 $ ] because the superoperator @xmath106 , defined in eq . , is a quadratic form in @xmath107 , that is @xmath108 with @xmath109 @xmath110 matrix of time dependent scalar quantities .
indeed it is straightforward to prove that , when eq
. holds , @xmath111= 0 $ ] , for all @xmath112 . in order to prove eq .
, let us note that , as demonstrated in appendix b , the following chain of equalities hold @xmath113 according to these equations , the transformation operated by @xmath114 on @xmath107 is a linear transformation correspondent to the one operated by @xmath115 on @xmath116
. one can show ( see appendix b ) that @xmath117 the functions @xmath118and @xmath119 are the solutions of the cauchy problems @xmath120y=0 \qquad\text{with initial conditions}\qquad \begin{cases } c(0)=1\qquad \dot{c}(0)=0\\ s(0)=0\qquad \dot{s}(0)=\omega_0 \end{cases}\ ] ] the functions @xmath121and @xmath122 are defined as @xmath123 inserting eq . into eq . and using eqs .
one gets @xmath124 these considerations allow to recast @xmath104 in the form @xmath125,\ ] ] where @xmath126 summarizing , in this subsection we have manipulated the expression of the time evolution superoperator in order to put it in a form more convenient for the calculations .
in fact we have proved that such a superoperator can be factorized in the form given by eq . .
moreover we have shown that two of the superoperators appearing in eq . do not need the dyson chronological order .
as we will see in the next section , this circumstance allows to find an analytic expression for the density matrix solution . in this subsection
we calculate the action of the factorized superoperator given in eq . on the initial density matrix
. having in mind eqs . and , and writing the initial density matrix as follows ( see eq . ): @xmath127 with @xmath128 ,
we have @xmath129 let us firstly introduce two superoperator eigenvalue equations we will use in the following .
indicating with @xmath130 a generic superoperatorial well defined function of the superoperator @xmath131 , from eqs . , and we obtain : @xmath132 where now @xmath133 and @xmath134 are simple scalar functions . we begin noting that , using eqs . and , one can write the term @xmath135 appearing in eq . as follows @xmath136 moreover , using eq .
it is not difficult to prove that @xmath137^n}{n!}= e^{\gamma(t)}exp[e^{\frac{\gamma(t)}{2}}(-\imath\:\vec{z}^t\mathsf{j}\hat{\vec{z } } ) ] .
\label{passo2}\ ] ] finally , remembering eqs .
and we get @xmath138 inserting eqs . , and into eq .
we can express the density matrix solution as follows @xmath139\:d^2\vec{z'}\ ] ] after some algebraic manipulation , reported in appendix d , the previous expression can be recast in the following final form @xmath140\ : e^{-\imath\:\vec{z}^t\mathsf{j}\hat{\vec{z}}}\:d^2\vec{z}=\frac{1}{2\pi}\int \chi_t(\vec{z})\ : e^{-\imath\:\vec{z}^t\mathsf{j}\hat{\vec{z}}}\:d^2\vec{z},\ ] ] where @xmath141 eq .
constitutes the main result of the paper .
it gives the operatorial density matrix solution of the problem of the dissipative dynamics of a harmonic oscillator interacting with a generic environment satisfying properties 2 . and 3 .
of section i , as for example a thermal reservoir at @xmath142 temperature .
our approach to the dynamics of the system relies on the weak coupling limit but does not invoke the born markov and rotating wave ( rw ) approximations . in more detail , we have solved a non - markovian generalized master equation for the harmonic oscillator coincident with that deducible using either the time convolutionless projection operator technique @xcite or the superoperatorial approach of @xcite . at the best of our knowledge
this is the first analytical method allowing to solve such a generalized master equation . in the next section
we derive some approximated forms of the solution given by eq .
and we show its usefulness in calculating the analytic expression of many observables of interest and thus in gaining new insight in the dynamics of one of the most extensively studied physical systems .
in this section we derive an approximated form of the operatorial density matrix solution , given by eq . , valid when @xmath143 with @xmath144 given by eq . .
looking at this equation one sees immediately that eq .
amounts at neglecting the time dependent frequency renormalization and dephasing terms @xmath145 it is worth noting that such an approximation is always justified in the weak coupling regime @xmath146 , provided that the reservoir frequency cut off remains finite , as one can appreciate with the help of eqs .
, and . for the same reason it turns out that @xmath147 . from eq . , it is not difficult to prove that this last inequality allows to write the matrix @xmath148 , defined in eq .
, as follows @xmath149 inserting this expression in the definition of @xmath150 , given by eq . , and
exploiting eqs . , and we get @xmath151dt_1,\ ] ] with @xmath152 the form of the matrices defined in the previous equations suggests a further approximation very common in quantum optics : the rotating wave approximation ( rwa ) .
it basically consists in neglecting rapidly oscillating terms as , in our case , those oscillating at frequency @xmath153 .
this amounts at averaging to zero all the elements of the matrices @xmath154 and @xmath155 so that eq .
reduces to @xmath156 substituting eq . into eq .
we obtain the following expression for the density matrix solution in the rwa @xmath157\ : e^{-\imath\:\vec{z}^t\mathsf{j}\hat{\vec{z}}}\:d^2\vec{z}.\ ] ] it is possible to demonstrate that @xmath158 , as given by eq . , satisfies the following master equation @xmath159\hat{\rho}^{\rm rwa } ( t ) , \label{arwa}\ ] ] which in turn has been derived in @xcite . once obtained the density matrix we are able , at least in principle , to evaluate the mean value @xmath160 of each and every operator @xmath29 of interest for the system .
one of the advantages of having a solution of the density matrix in terms of the characteristic function is the possibility of exploiting the following relations @xcite : @xmath161 by using this relations it is not difficult to calculate the time evolution of the mean energy of the oscillatory system : @xmath162 we emphasize that these solutions depend on the initial state of the oscillator only through the term @xmath163 , the second term of the sum being independent of the initial state .
moreover , looking at eq .
one sees that the exponential factor accounting for energy dissipation does not depend on the initial state of the system but only on the characteristic parameters of the reservoir . + these features , characterizing eq .
, are directly related to the factorized form of the the qcf @xmath164 .
in fact , from both eq . and
eq . it appears evident that such a function is the product of a gaussian factor and the function @xmath165 $ ] .
all the information on the initial state of the system is incorporated in this last function from which the first term in the righthand side of eq . comes from . on the contrary
, the second term derives exclusively from the gaussian factor of the qcf , which depends only on the environment functions and not on the initial state of the system .
note from eqs . and that , whatever the initial state of the system is , @xmath166 this behavior is easily understandable , when the environment is a thermal reservoir , in the light of the thermalization process .
let us now have a closer look at eq . .
if we evaluate @xmath167 with the help of eq . , we find that @xmath168 this means that the time evolution of the oscillator energy is not affected by the contribution of the rapidly oscillating terms neglected in the rwa .
this feature comes directly from the particular structure of the free hamiltonian operator . to better understand this point we consider the gaussian factor appearing in the qcf .
the superoperator corresponding to such a factor has the form @xmath169.\ ] ] with some algebraic manipulation one can recast the superoperator appearing in the exponent as follows @xmath170 + \frac{\lambda(t)}{2}\left[(\mathbf{x}^s)^2-(\mathbf{p}^s)^2\right]+\theta(t)\mathbf{x}^s\mathbf{p}^s,\ ] ] where the time dependent coefficients appearing in the previous equation are @xmath171 with @xmath172 and @xmath173 pauli spin matrices . exploiting the properties of the trace and putting @xmath174
, one can easily show that the following chain of equalities holds @xmath175 where we have defined @xmath176 having these equations in mind it is not difficult to convince oneself that calculating @xmath177 using @xmath178 is equivalent at calculating @xmath179 using @xmath165 $ ] . in the case
@xmath180 we have that @xmath181\hat{h}_0=-2\hbar \omega_0 , \qquad\left[(\mathbf{x}^s)^2-(\mathbf{p}^s)^2\right]\hat{h}_0=0 , \qquad \mathbf{x}^s\mathbf{p}^s\hat{h}_0=0,\ ] ] and thus @xmath182 note that this equation is not affected by the rwa as well as @xmath165 $ ] .
this explains why we obtain for the mean value of the oscillator energy the same result with or without the rwa approximation .
the previous procedure suggests a sufficient condition to single out operators which do not suffer the rwa approximation , indeed we have that @xmath183\hat{a}=0 , \qquad \mathbf{x}^s\mathbf{p}^s\hat{a}=0\qquad\rightarrow { \langle\hat{a}\rangle}_t={\langle\hat{a}\rangle}_t^{\rm rwa}.\ ] ] with the expression operators which do not suffer the rwa approximationwe indicate operators having the property that the time evolution of their mean value is not influenced by the rwa . in other words the counter rotating terms
do not contribute to the dynamics of this class of observables .
examples of operators belonging to such a class are @xmath16 , @xmath17 , @xmath184 and all linear combinations of such operators , as one can easily verify . on the contrary @xmath185 and @xmath186
do not satisfy condition . for such operators , indeed , exploiting eq .
, one gets @xmath187 similarly , for the mean value of the correlationoperator @xmath188 one has @xmath189
in this paper we have developed a method to solve the weak coupling generalized master equation for qbm .
in particular we have considered the case in which the system interacting with the environment is a quantum harmonic oscillator .
the master equation we have solved , given by eq . , although non
markovian is local in time .
such a master equation has been derived in @xcite by using a superoperatorial technique and in @xcite by means of the time convolutionless method and is the weak coupling approximated form of the exact master equation for qbm derived by paz and zurek in @xcite .
the main result of the paper is the derivation of the analytic solution of eq . for the density matrix of the reduced system . to this aim
we have used an approach based on the algebra of superoperators .
our method is independent of the specific form of the environment and does not rely on any approximation apart from the weak coupling one .
we have also studied simpler forms of the density matrix solution obtained neglecting frequency renormalization terms and/or performing the rwa .
we have demonstrated the existence of a class of superoperators whose mean value is not affected by the presence of the counter rotating terms at any time @xmath80 and we have given a sufficient condition to verify if a given operator belongs to such a class .
this circumstance simplifies substantially the calculations since , for operators belonging to such a class , one can use the approximated density matrix solution , given by eq .
, in order to calculate their mean value
. the analytic solution we derive and discuss in the paper is given in terms of the qcf ( see eqs . ) by means of which one can calculate the expectation value of many observable of physical interest in a very direct way , as suggested by eqs .. for example , thanks to the simplicity of the analytic solution we have derived , we succeed in calculating the dissipative time evolution of the mean energy of the system .
finally it is worth noting that from the qcf it is easy to derive the wigner function characterizing the state of the dissipative system .
concluding , we believe that , the non markovian analytic approach we have derived in this paper for qbm can be generalized to other fundamental dissipative systems , such as for example the jaynes cummings model with losses . as for the qbm , we think that the analytic solution of the density matrix may be used for studying new aspects of such a basic model , both for fundamental and for applicative research , under conditions in which , up to now , only numerical approaches were possible .
one of the authors ( s.m . ) acknowledges financial support from finanziamento progetto giovani ricercatori anno 1999 , comitato 02 .
in this appendix we demonstrate some useful properties of the superoperator @xmath48 .
such properties , allowing to factorize the time - evolution superoperator given in eq , are the following @xmath190=0 , \qquad \left[\mathbf{n},\mathbf{\bar{d}}(t)\right]=-2\mathbf{\bar{d}}(t).\ ] ] the first one is the most surprising one : it can be considered as a dynamical invarianceof the dissipation process . as we shall see in the following , the dynamical invarianceoriginates from the particular structure of the hamiltonian term @xmath191 , defined by eq . .
independently on the form of the time dependent functions present in such a term , it can be written , in general , as @xmath192 where @xmath193 is a @xmath44 real symmetric matrix . remembering that @xmath194 one gets @xmath195&= & -\frac{1}{2}\imath\ : \left[(\vec{\mathbf{z}}^{\sigma})^t\mathsf{j}\vec{\mathbf{z}}^{s } , \left(\hat{\vec{z}}^t \mathsf{s}(t)\hat{\vec{z}}\right)^s \right]\nonumber\\ & = & \frac{1}{2}\imath\:\left ( ( \vec{\mathbf{z}}^{\sigma})^t\mathsf{j}\left[\hat{\vec{z}}^t \mathsf{s}(t)\hat{\vec{z}},\hat{\vec{z}}\right]^s + \left(\left [ \hat{\vec{z}}^t \mathsf{s}(t)\hat{\vec{z}},\hat{\vec{z } } \right]^{\sigma}\right)^t\mathsf{j}\vec{\mathbf{z}}^{s } \right)\end{aligned}\ ] ] it is possible to show that @xcite @xmath196=\imath \mathsf{j}\mathsf{s}(t)\hat{\vec{z}}.\ ] ] substituting this equation into eq . and using the properties of the matrices @xmath193 and @xmath43 ( cfr eq . ) one obtains @xmath197&= & -\frac{1}{2}\left ( ( \vec{\mathbf{z}}^{\sigma})^t\mathsf{j}\mathsf{j}\mathsf{s}(t)\vec{\mathbf{z}}^s + ( \vec{\mathbf{z}}^{\sigma})^t\mathsf{s}^t(t)\mathsf{j}^t\mathsf{j}\vec{\mathbf{z}}^{s } \right)\nonumber\\ & = & -\frac{1}{2}\left ( -(\vec{\mathbf{z}}^{\sigma})^t\mathsf{s}(t)\vec{\mathbf{z}}^s + ( \vec{\mathbf{z}}^{\sigma})^t\mathsf{s}(t)\vec{\mathbf{z}}^{s } \right)=0.\end{aligned}\ ] ] the generality of the matrix @xmath193 ensures the validity of this relation for a generic quadratic hamiltonian and thus , in particular , for @xmath191 . as far as the second commutation relation is conserned , remembering that @xmath198 one has @xmath199&= & \left[\mathbf{n } , ( \vec{\mathbf{z}}^{s})^t\:\mathsf{\bar{m}}(t)\:\vec{\mathbf{z}}^{s}\right]\nonumber\\ & = & \left[\mathbf{n } , \vec{\mathbf{z}}^{s}\right]^t\:\mathsf{\bar{m}}(t)\:\vec{\mathbf{z}}^{s}+ ( \vec{\mathbf{z}}^{s})^t\:\mathsf{\bar{m}}(t)\:\left[\mathbf{n } , \vec{\mathbf{z}}^{s}\right]\end{aligned}\ ] ] exploiting the superoperatorial commutation rules given in eq .
yields @xmath200= -\frac{\imath}{2}\left[\left(\mathbf{p}^{\sigma}\mathbf{x}^{s}-\mathbf{x}^{\sigma}\mathbf{p}^{s}\right ) , \vec{\mathbf{z}}^{s}\right]=-\vec{\mathbf{z}}^{s}.\ ] ] finally , substituting into eq .
one gets @xmath201=-(\vec{\mathbf{z}}^{s})^t\:\mathsf{\bar{m}}(t)\:\vec{\mathbf{z}}^{s}- ( \vec{\mathbf{z}}^{s})^t\:\mathsf{\bar{m}}(t)\:\vec{\mathbf{z}}^{s}=-2\mathbf{\bar{d}}(t)\ ] ]
in this appendix we shall demonstrate that the general time evolution of a s- or @xmath3-type superoperator is equivalent to the s- or @xmath3-type superoperator of the time evolution of the corresponding operator . in formulas
this amounts at demonstrating eq . .
first of all let us define @xmath202 as solution of the equation @xmath203 where in general @xmath204\not=0 $ ] .
@xmath205 , as well as @xmath202 , can be either an operator or a superoperator
. the previous equation can be solved in iterative way and its solution is @xmath206\ ] ] it can be shown that @xmath207 satisfies the equation @xcite @xmath208 which can be again solved in iterative way giving the following form of the solution @xmath209\ ] ] let us define @xmath210 from the previous equation we have @xmath211=l^s(t)a_c(t)=l^s(t)\mathcal{t}_c(t)\:a \qquad \rightarrow \frac{d}{dt}\mathcal{t}_c(t)=l^s(t)\mathcal{t}_c(t)\ ] ] this means that @xmath212 satisfies an equation similar to the one given by eq . .
the form of @xmath212 can be obtained from eq replacing @xmath213 and @xmath214 .
now , let us suppose that @xmath29 is s- or @xmath3-type superoperator and @xmath215 is an s - type superoperator . from the properties given in we have that @xmath216\cdots]\:dt_1\cdots dt_n\nonumber\\ & = & \sum_{n=0}\int_0^t\cdots\int_0^{t_{n-1}}[\hat{l}(t_1)\cdots [ \hat{l}(t_n),\hat{a}]\cdots]^{s(\sigma)}\:dt_1\cdots dt_n\nonumber\\ & = & \left(\sum_{n=0}\int_0^t\cdots\int_0^{t_{n-1}}\mathbf{l}^s(t_1)\cdots \mathbf{l}^s(t_n)\hat{a}\:dt_1\cdots dt_n \right)^{s(\sigma)}\nonumber\\ & = & \left(exp_c\left[\int_0^t\mathbf{l}^s(t_1)\:dt_1\right]\hat{a}\right)^{s(\sigma)}= \left(\mathbf{t}_c(t)\:\hat{a}\right)^{s(\sigma)}\nonumber\\ & = & \left(\hat{t}_c(t)\:\hat{a}\:\hat{t}^{-1}_c(t)\right)^{s(\sigma)}=\left(\hat{a}_c(t)\right)^{s(\sigma)}\end{aligned}\ ] ] where we have put @xmath217 $ ] .
now , let us consider @xmath218 the last definition comes directly from the definition of @xmath212 and can be verified applying @xmath212 on @xmath219 and , viceversa , applying the previous definition of @xmath220 on @xmath221 .
+ exploiting eq . and following the same lines of the derivation of eq .
, one can show that @xmath222 let us now derive eq . .
, , and remembering the definition of @xmath223 : @xmath224 one hase @xmath225\:\hat{t}_{os}(t ) = -\mathsf{j}\mathsf{\tilde{s}}(t)\hat{\vec{z}}(t)\qquad\mathsf{\tilde{s}}(t)=\frac{1}{\hbar}\mathsf{s}(t).\ ] ] deriving the previous equation once more we get @xmath226\hat{\vec{z}}(t)= \left[-\mathsf{j}\mathsf{\dot{\tilde{s}}}(t)-det[\mathsf{\tilde{s}}(t)]\right]\hat{\vec{z}}(t).\ ] ] in our case @xmath227 so that the differential equation for the operator @xmath228 has the following form @xmath229 a solution of the previous equation can be written as @xmath230 with @xmath118 and @xmath119 solutions of eq . with the same initial conditions . note that , from , it follows that @xmath231 with th help of eq .
we have for @xmath232 @xmath233 where we have defined , as in eq .
, @xmath234 grouping eqs . , and using the matrix representation we obtain @xmath235
in this appendix we sketch the main steps of the derivation of the final form of the density matrix solution , given by eq .
, from eq . .
let us remind the generic expression for the density matrix in terms of the qcf ( see eq . )
@xmath236 where @xmath237 . from eq .
we have @xmath238\:d^2\vec{z'}\ ] ] comparing these last two equations one gets from eq . and using some properties of the @xmath240 function we obtain @xmath241\big{\ } } } = \delta[\vec{z}-e^{\frac{\gamma(t)}{2}}\vec{z'}(t)]= e^{-\gamma(t)}\delta[\vec{z'}-e^{-\frac{\gamma(t)}{2}}\mathsf{r}^{-1}(t)\:\vec{z}].\ ] ] @xmath242 \:d^2\vec{z'}\nonumber\\ & = & exp[-e^{-\gamma(t)}(\vec{z})^t\:(\mathsf{r}^{-1}(t))^t\mathsf{w}(t)\mathsf{r}^{-1}(t)\:\vec{z } ] \chi_0[e^{-\frac{\gamma(t)}{2}}\mathsf{r}^{-1}(t)\:\vec{z}]\end{aligned}\ ] ] g. w. ford , m. kac and p. mazur , journal of math .
* 6 * , nos . 4 , 504 ( 1965 ) g. w. ford and m. kac , journal of stat .
phys . , vol .
* 6 * , nos .
5/6 , 803 ( 1987 ) a.o .
caldeira and a.j .
legget , ann . of phys . * 149 * , 374 ( 1983 ) .
r.p . feynman and f.l .
vernon , ann . phys .
( n.y . ) * 24 * , 118 ( 1963 ) .
s. bay , p. lambropoulos , and k. mlmer , phys .
, * 76 * , 161 ( 1996 ) ; j.j .
hope , phys .
a , * 55 * , r2531 ( 1997 ) ; g. m. moy , j. j. hope , and c. m. savage , phys .
a , * 59 * , 667 ( 1999 ) ; s. john and t. quang , phys . rev .
* 74 * , 3419 ( 1994 )
|
an original method to exactly solve the non - markovian master equation describing the interaction of a single harmonic oscillator with a quantum environment in the weak coupling limit is reported . by using a superoperatorial approach we succeed in deriving the operatorial solution for the density matrix of the system .
our method is independent of the physical properties of the environment .
we show the usefulness of our solution deriving explicit expressions for the dissipative time evolution of some observables of physical interest for the system , such as , for example , its mean energy .
|
the ibis / isgri survey ( bird et al . 2004 ) contains 123 high energy emitting objects detected with the unprecedented sensitivity of @xmath01 mcrab in the energy range 20 - 100 kev discovered by mosaicing all core program observations performed in the first year of the mission ; this first catalogue contains 23 high mass and 53 low mass binary systems , 5 agn , a few snr / x - ray pulsar systems , a few isolated pulsars and a handful of other objects .
around 30 remain at the moment unidentified and are the main targets of follow up observations .
observations of these and other sources in the ibis survey at x - ray wavelengths are useful in order to assess their nature and overall characteristics ; in this sense the bepposax / pds archive is a powerful tool as it can provide information on any spectral and/or flux variation provided that the pds observation was not contaminated by nearby sources . to this end , we have started a program to analyse all pds observations which contain in the field of view a source detected by ibis : the systematic search of the archive has provided a set of 68 objects which were targets of bepposax observations and so have both mecs / pds ( 2 - 100 kev band ) data . in this case
it is possible to reconsider the pds data in the light of the ibis images in order to exclude or evaluate any contamination present .
the pds field of view is 1.3@xmath1 ( fwhm ) , hexagonal in shape and with no imaging capability ( frontera et al .
the response matrix of this instrument is triangular in both directions , with a flat top of 3 and a reduction in sensitivity of a factor of 2 at 38 from the centre up to zero response at 78 distance .
the mecs has instead a field of view of 30 radius and so covers only about 10% of the pds view .
therefore it is also possible that a source is not seen by the mecs but serendipitously observed by the pds .
in fact , from our search , 21 ibis survey sources were measured by the pds in this mode . for the remaining sources
there is no bepposax pointing .
herein we present a few interesting cases where ibis / pds data can be used together in order to define the high energy characteristics of some of these objects which were serendipitously detected by beppo / sax .
the dataset used in this work consists of all core program observations ( gps and gcde ) performed from revolution 20 to revolution 145 .
isgri images were generated for each selected pointing in the sky in the 20 - 100 kev band by combining in mosaics all available science windows ( scw ) with the isdc offline scientific analysis software ( osa ) version 3 ( goldwurm et al . 2003 ) .
source ghosts in the images have been cleaned using an appropriate catalogue available at the isdc .
more details on the data analysis can be found in bird et al .
+ figure 1 shows the first field studied in this work : it represents a mosaic in the 20 - 100 kev band of the scutum region consisting of 400 science windows ( fixed pointings that last about 37 minutes each ) . in this region
we find three pds pointings : the two snrs , kes75 and snr 021.5 - 00.9 ( case a and b ) and the serendipitous pds detection of a new unidentified x-@xmath2 ray source ax j1838 - 0655 ( case c ) .
the two circles in figure 1 represent the pds field of view pointed at the two snrs , while the case of the new source is discussed further in a separate section .
finally , in the last section , we discuss the case of another new isgri source igr j16207 - 5129 and present its pds spectrum .
it is important to note that none of the bepposax data presented in this paper have been reported in the literature yet .
+ in the following , isgri fluxes are based on the count rates quoted in the survey ( bird et al .
2004 ) and scaled with the crab results
. 880 ksec ) .
the circles represent the bepposax / pds field of view .
the x - point is one of the two offset pointings of g021.5 - 00.9 observation.[fig : single ] ] bepposax pointed at the snr 021.5 - 00.9 three times and high energy emission ( 20 - 100 kev ) has been detected by the pds instrument .
this looks as the simplest case of figure 1 where the only high energy emitter in the pds field of view is snr 021.5 - 00.9 .
since the pds detection looks uncontaminated we have analysed the data : a fit with a simple power law ( @xmath3=2.4 ) gives a 20 - 100 kev average flux of @xmath0 1.2 @xmath4 10@xmath5 erg @xmath6 s@xmath7 .
this flux is lower than that seen by isgri ( bird et al .
2004 ) indicating a possible variability of the source or a contamination in the pds offset pointings . since variability in
unlikely in g021.5 - 00.9 , we have checked if the offset field measurements provide indication for the presence of a high energy emitting source .
indeed a 9@xmath8 source was discovered in one of the two offset pointings : cross - checking this pds offset pointing with the isgri image we found that the contaminating source was igr j18325 - 0756 ( see x point in figure 1 ) .
the corrected spectrum of snr 021.5 - 00.9 is shown in figure 2 .
a fit to this spectrum provides a @xmath3 of about 2.6 and a 20 - 100 kev flux of 4 - 5 @xmath4 10@xmath5 erg @xmath6 s@xmath7 more in line with the isgri detection ( @xmath0 6 @xmath4 10@xmath5 erg @xmath6 s@xmath7 ) .
furthermore , the pds spectrum of igr j18325 - 075 ( see figure 3 ) , is well fitted with a power law with @[email protected] and a 20 - 100 kev flux of 7 @xmath4 10@xmath5 erg @xmath6 s@xmath7 again consistent with the isgri detection ( @xmath0 8 @xmath4 10@xmath5 erg @xmath6 s@xmath7 ) . as clearly shown in figure 1 , in the pds field of view of the bepposax pointing of kes 75 ( also axj1846.4 - 0258 , a snr with an x - ray pulsar )
there is another high energy emitter which was detected for the first time by isgri ( bird et al .
2004 ) and named igr j18483 - 0311 .
this new isgri source is 31 away from kes 75 , and so it is not in the mecs field of view but is likely to have been serendipitously detected by the pds , unless it is a transient object not active at the time of the bepposax measurement .
analysis of the bepposax of kes 75 indicates contamination by this nearby source and in fact the mecs / pds cross - calibration constant is 3.5@xmath9 i.e. outside the nominal range of 0.75 - 0.95 generally observed . in figure 4
the pds spectrum from the pointing containing both kes 75 and igr j18482 - 0311 is reported .
the solid line is the extrapolation of the mecs spectrum of kes 75 and provides 20 - 100 kev flux of 3 @xmath4 10@xmath5 erg @xmath6 s@xmath7 .
the difference between this and the data points is reported in figure 5 which should be taken with caution but can be use as a reference spectrum for this isgri source .
the estimated flux is @xmath06 @xmath4 10@xmath5 erg @xmath6 s@xmath7 consistent with the isgri detection ( @xmath0 6 @xmath4 10@xmath5 erg @xmath6 s@xmath7 ) . ] ] in figure 6 a close - up of the area around the ibis survey source ax j1838.0 - 0655 ( see also figure 1 ) is shown .
ax j1838.0 - 0655 was first detected by the ipc instrument on the einstein satellite ( 1e1835.3 - 0658 , hertz & grindlay 1988 ) with an associated 90% error radius of 49 .
the source was also detected by asca in the 0.7 - 10 kev range during the galactic plane survey ( sugizaki et al .
2001 and bamba et al .
2003 ) with an associated uncertainty of 1 radius .
finally , ax j1838.0 - 0655 was serendipitously detected twice by the pds instrument during the bepposax pointings devoted to the study of the gev source 1837 - 0610 . in figure 6
the cross represents the einsten position , the small circle the asca error box while the large circle is the pds field of view .
it is clear from the ibis image that ax j1838 - 0655 is the only hard x - ray emitter in the pds field of view .
it is detected by isgri at 13@xmath8 in the 20 - 100 kev range in the cumulative exposure with an average flux of about 2 - 3 mcrab .
the source , still unidentified , is a bright ( 1.1 @xmath4 10@xmath5 erg @xmath6 s@xmath7 , flat ( @xmath3=0.5 ) and variable asca source ( sugizaki et al .
the pds data show a steep spectrum ( @xmath3=2.5 , see figure 7 ) and a 20 - 100 kev flux of around 2 mcrab in agreement with the isgri flux .
a detailed discussion on the possible nature of ax j1838.0 - 0655 is reported in malizia et al .
in figure 8 the isgri image ( 20 - 100 kev ) of the field containing another new ibis survey source igr j16207 - 5129 ( atel # 229 , walter et al .
2004 ) is shown .
the image is a zoom of a mosaic of the norma region obtained by summing 470 science windows .
again the circle represents the pds field of view . in this last case ,
bepposax was targeted on the snr 1e161348 - 5055 which is located at the centre of the circle and which obviously has not been detected by isgri .
bepposax pointed at 1e161348 - 5055 three times in 1998/99 and a source was detected in the first and third pointings only , demonstrating variability over a period of 6 months .
the high energy emission definitely does not come from the snr since the mecs / pds cross - calibration factor is higher than 100 . in both the first and the last observations ,
the 20 - 100 kev data are well fitted with a simple power law of photon index @[email protected] with a flux of @xmath0 6 @xmath4 10@xmath5 erg @xmath6 s@xmath7 , much in agreement with the isgri data ( @xmath11 7 @xmath4 10@xmath5 erg @xmath6 s@xmath7 ) . in figure 9 the pds 20 - 100 kev spectrum of the first pointing
the isgri data demonstrates that the source of the high energy emission is not the bepposax target , but is the new isgri source igr j16207 - 5129 . ]
in this paper we have demonstrated that the combination of bepposax / pds and isgri data can produce a clearer understanding of the nature of high energy emitters .
work is continuing on this correlation analysis and will improve as the ibis survey dataset increases .
we acknowledge financial support by asi ( italian space agency ) via contract i / r/041/02 .
|
the first ibis galactic plane survey has provided a list of high energy emitting objects above 20 kev ; these sources have been detected mostly in the crowded region of the galactic centre and partly along the galactic plane . in order to validate the detection procedure , to help in the identification process and to study the nature of these ibis sources ,
this list has been cross correlated with the data archive of the pds instrument on bepposax , which operated in a similar energy band and with a similar sensitivity .
we discover a number of associations whose detailed analysis will be particularly useful for the survey work .
also , thanks to the imaging capability of ibis / isgri , objects which could not be studied by the pds due to contamination from nearby sources can now be associated with a definite source or sources .
[ 2001/04/25 1.1 ( pwd ) ]
|
the interiors of compact stars have been discussed as systems where high - density phases of strongly interacting matter do occur in nature , see @xcite and @xcite .
the consequences of different phase transition scenarios for the cooling behaviour of compact stars have been reviewed recently @xcite in comparison with existing x - ray data .
a particular discussion has been devoted to the idea that a strong attraction in three flavor @xmath1-quark matter may allow for the existence of super - dense anomalous nuclei and strange quark stars @xcite .
thereby , in dependence on the value of the bag constant @xmath2 different possible types of stars were discussed : ordinary neutron stars ( ns ) without any quark core , neutron stars with quark matter present only in their deep interiors ( for somewhat intermediate values of @xmath2 ) , neutron stars with a large quark core ( qcns ) and a crust typical for neutron stars , and quark stars ( qs ) with a tiny crust of normal matter and with no crust ( both for low @xmath2 values ) . by qcns
we understand compact stars in which the hadronic shell is rather narrow in the sense that it does not essentially affect the cooling which is mainly due to the quark core .
however , this hadronic shell plays the role of an insulating layer between quark matter and the normal crust . by qs we mean compact stars for which the hadronic shell is absent what allows only for tiny crusts with maximum densities below the neutron drip density ( @xmath3 g/@xmath4 ) and masses @xmath5 @xcite .
therefore , we suppose that the difference between the models of qs and qcns regarding their cooling behaviour is only in the thickness of the crust .
this results in completely different relations between internal ( @xmath6 ) and surface ( @xmath7 ) temperatures for qs and qcns .
recent works @xcite demonstrate the possibility of diquark condensates characterized by large pairing gaps ( @xmath8 mev ) in quark cores of some neutron stars and in qs and discuss different possible phases of quark matter .
large gaps arise for quark - quark interactions motivated by instantons @xcite and by nonperturbative gluon propagators @xcite . to be specific in our predictions we will consider models of the canonical qs and qcns of 1.4 solar masses ( @xmath9 ) at a constant density .
the constant density profile is actually a very good approximation for qs of the mass @xmath10 , see @xcite .
we consider the model of qcns with a crust of the mass @xmath11 , the model of qs with a tiny crust mass ( @xmath5 ) and the model of qs with no crust .
for qcns we shall use the same @xmath12 ratio as for ordinary ns , see @xcite , whereas for qs we use somewhat larger values for this ratio , namely @xmath13 for a tiny crust @xcite and @xmath14 for a negligible crust @xcite . in the latter case , however , we assume the existence of black body photon radiation from the surface as for the cases of more extended crusts .
we will estimate the cooling of qs and qcns first in absence of color superconductivity and then in presence of color superconductivity for small quark gaps ( @xmath15 0.1 ... 1 mev ) , as suggested by @xcite and for large gaps ( @xmath16 mev ) as obtained in refs .
@xcite . in the latter case
we will consider two phases : the color - flavor - locked @xmath1-phase @xcite and the @xmath0 color superconducting phase @xcite , in which the @xmath17-quark is absent and the @xmath18-diquark condensate selects a direction in color space whereby the color charge has to be compensated by the remaining unpaired quarks . finally , we want to discuss the question whether the hypothesis of a color superconducting quark matter phase in compact star interiors is compatible with existing x - ray data .
a detailed discussion of the neutrino emissivity of quark matter has first been given by @xcite where the possibility of color superconductivity has not been discussed . in this work the quark direct urca reactions ( qdu ) @xmath19 and @xmath20 were suggested as the most efficient processes .
their emissivities were estimated as @xmath21 where at baryon densities @xmath22 the strong coupling constant is @xmath23 and decreases logarithmically at still higher densities @xcite .
the nuclear saturation density is @xmath24 , @xmath25 is the electron fraction , and @xmath26 is the temperature in units of @xmath27 k. the larger the density of the @xmath1-system , the smaller is its electron fraction . for a density @xmath28
one can expect a rather low electron fraction of strange quark matter @xmath2910@xmath30 @xcite and eq .
( [ neutr - du ] ) yields @xmath31 , see @xcite .
we did not yet discuss the strange quark contribution given by the direct urca processes @xmath32 and @xmath33 .
although these processes can occur , their contribution is suppressed compared to the corresponding @xmath18-reactions @xcite by an extra factor @xmath34 , where @xmath35 is the cabibbo angle .
if for somewhat larger density the electron fraction was too small ( @xmath36 , for @xmath37 and @xmath38 , @xmath39 is the electron mass ) , then all the qdu processes would be completely switched off @xcite and the neutrino emission would be governed by two - quark reactions like the quark modified urca ( qmu ) @xmath40 and the quark bremsstrahlung ( qb ) processes @xmath41 .
the emissivities of the qmu and qb processes were estimated as @xcite @xmath42 the latter estimate of qb emissivity is done in the suggestion that the exchanged gluon is screened .
if one suggests that quarks are coupled by transverse non - screened gluons then one gets @xcite @xmath43 . with a nonperturbative gluon exchange @xcite
we expect that the estimate @xmath44 is more appropriate than the one given by @xmath45
. therefore , in evaluating the emissivity of qb processes we will use ( [ neutr - b ] ) . other neutrino processes ( like the plasmon decay @xmath46 which goes via coupling to intermediate electron - electron hole states , and the corresponding color plasmon decay @xmath47 which goes via coupling of the gluon to quark - quark hole states @xcite , see fig .
4 ) have much smaller emissivities in the normal quark matter phase under consideration and can be neglected . among the processes in the crust ,
the electron bremsstrahlung on nuclei gives the largest contribution to the emissivity as estimated in @xcite @xmath48 this contribution can be neglected for qs due to a tiny mass of the qs crust @xmath49 . besides
, one should add the photon contribution @xmath50 where @xmath51 is the surface temperature in units of @xmath52 k , see @xcite .
this process becomes the dominant one for qs at essentially shorter times than for the qcns due to the higher @xmath53 ratios for the former .
internal and surface temperatures are related by a coefficient determined by the scattering processes occurring in the outer region , where the electrons become non - degenerate . for ns with rather thick crust , an appropriate fit to numerical calculations @xcite
is given by a simple formula @xcite @xmath54 where @xmath55 and @xmath6 both are measured in units of k. we shall use this expression when dealing with qcns .
a rough estimate of ( [ temp - rel ] ) yields @xmath56 with @xmath57 in dependence on the value of the internal temperature varying in the interval @xmath58 k of our interest . in the
qs the crust is much more thin than in the ns and the ratio @xmath59 should be significantly larger . therefore , depending on the thickness of the crust we shall use two estimates for qs scenarios : @xmath60 for a tiny crust @xcite , and @xmath61 for negligible crust @xcite . in order to compute the cooling history of the star we still need the specific heat of the electron , photon , gluon and quark sub - systems . in accordance with the estimates of @xcite ,
@xcite we have @xmath62 @xmath63 @xmath64 @xmath65 where @xmath66 is the number of different color states of massless gluons . the very small contribution to the specific heat of the crust
can be neglected @xcite .
the cooling equation reads @xmath67 where the summation is over all contributions to the specific heats and emissivities as discussed above .
the evolution at large times on which we are focusing our interest here is insensitive to the assumed value of the initial temperature @xmath68 .
we checked it using different values of initial temperature . to be specific
we choose the value @xmath69 as a typical initial temperature for proto - neutron stars . in figs .
[ fig1]-[fig3 ] we show the cooling history of qcns with standard thickness of the ns crust ( when internal and surface temperatures are related by eq .
( [ temp - rel ] ) ) , qs with a tiny crust ( @xmath13 ) and qs with negligible crust ( @xmath14 ) , respectively .
solid curves are for the matter suggested to be in the normal state , i. e. in the absence of color superconductivity .
different groups of data points are taken from table 3 of ref .
@xcite where the notations are explained . in the lower panels of figs .
[ fig1]-[fig3 ] , we show the results for @xmath70 taking @xmath71 , @xmath72 , @xmath73 . this is a representative set of parameters for which the qdu processes contribute to the cooling .
we see that two low - temperature pulsars can be explained as qcns being in normal state ( fig . 1 ,
thick solid curve , lower panel ) and many observations can be interpreted as qs in normal state with a tiny crust ( fig .
2 , lower panel ) .
the upper panels of figs . 1 - 3 demonstrate the cooling history of qcns and qs for @xmath74 , namely for @xmath75 , @xmath76 , and @xmath77 .
qcns being in normal state ( fig .
1 , thick solid curve , upper panel ) cool down rather slowly but are still in agreement with the data for a few pulsars .
for qs with a tiny crust ( fig .
2 , @xmath78 ) we get a nice fit of many data points . in both cases @xmath74 and @xmath70 , see fig .
3 , qs with negligible crust cool down too fast in disagreement with the x - ray data .
in the standard scenario of ns cooling the inclusion of nucleon pairing suppresses the emissivity resulting in a more moderate cooling .
now , considering qs and qcns we will show that we deal with the opposite case .
due to the pairing , the emissivity of qdu processes is suppressed by a factor @xmath79 and the emissivities of qmu and qb processes are suppressed by a factor @xmath80 for @xmath81 .
thereby in our calculations we will use expression ( [ neutr - du ] ) for qdu suppressing the rate by @xmath79 and expressions ( [ neutr - b ] ) for qmu and qb suppressing the rates by @xmath80 for @xmath81 .
we also observe that plasmon and color plasmon decay processes are switched off in the superconducting phase when the photons and the gluons acquire masses due to the higgs effect , as it happens for photons in usual superconductors .
@xcite however demonstrated that in superconducting matter there appears a new neutrino neutral current process analogous to the plasmon decay but now due to a massive photon decay .
its emissivity is suppressed by the factor @xmath82 rather than by @xmath83 , as for direct urca processes , or by @xmath84 , as for modified urca and corresponding bremsstrahlung processes .
this results in a large contribution for small but finite values of @xmath85 .
naively , one could expect that in a color superconductor the squared photon mass is proportional to the fine structure constant @xmath86 as it is in ordinary superconductors
. then it would be much smaller than the squared gluon mass since the latter quantity in the color superconducting phase has to be proportional to the corresponding strong coupling constant @xmath87 .
in reality , due to the common gauge transformation for electromagnetic and color fields one deals with mixed photon - gluon excitations .
we demonstrate this using the expression for the free energy density of the color superconducting phase @xcite , @xmath88 where @xmath89 is the normal part of the free energy density , @xmath90 , @xmath91 , @xmath92 , @xmath93 , @xmath94 is chemical potential of ultra - relativistic quark , @xmath95 is the electric charge of a @xmath18-pair , @xmath86 .
we have introduced an interaction with two gauge fields @xmath96 and @xmath97 .
@xmath96 is the electromagnetic field and @xmath97 relates to the gluons .
for simplicity we consider only fluctuations of space - like components of the fields and assume the @xmath97 field to be an abelian field .
variation of ( [ gauge ] ) with respect to the fields gives the corresponding equations of motion .
taking @xmath98 , where @xmath99 is the order parameter and @xmath100 is the fluctuating field , we linearize the equations of motion for the fluctuating fields @xmath101 .
solving these equations in the fourier representation , we get three branches of the spectrum .
the branch @xmath102 corresponds to fluctuations of the order parameter characterized by a large mass @xmath103 mev .
the branch @xmath104 describes a massive photon - gluon excitation with a mass @xmath105 .
the extra branch @xmath106 describes a massless mixed photon - gluon goldstone excitation .
thus in difference with a usual proton superconducting phase of ns where the photon has rather low mass @xmath107 mev for @xmath108 mev , in the color superconductor we , probably , deal with a much more massive mixed photon - gluon excitation ( with @xmath109 being replaced by @xmath110 ) and with the corresponding goldstone boson .
the goldstone boson does not contribute to the mentioned photon - gluon decay process , whereas the massive excitation does .
now , armed with an expression for the photon - gluon mass we may estimate the emissivity of the corresponding processes @xmath111 , where @xmath112 and @xmath113 are the electron hole and the quark hole , respectively , see fig .
4 . using the result for @xmath114 @xcite we easily get @xmath115 for @xmath116 , and using the condition @xmath117 .
as we see , the emissivity of this process is strongly suppressed for the values @xmath118 mev following from eq .
( [ ph - g ] ) . also the specific heat of this mixed photon - gluon excitation
is suppressed by the same exponential factor @xmath119 . for the goldstone excitation
the contribution to the specific heat is given by eq .
( [ gam - g ] ) . for the quark specific heat at @xmath81 we use an expression similar to the one which applies for the case of nucleon pairing @xcite , i.e. @xmath120,\end{aligned}\ ] ] where @xmath121 is related to @xmath122 as @xmath123 for the case of small gaps . for cfl and 2sc phases we will use @xmath124 . actually , in the latter case the relation between @xmath121 and @xmath122 is unsettled .
however , one believes that the coefficient in standard bcs formula @xmath125 is appreciably reduced as impact of instanton - anti - instanton molecules @xcite .
the mentioned uncertainty in the value of @xmath121 for cfl and 2sc phases does not significantly affect the cooling curves since the dominant effect comes from the exponential factor where @xmath122 enters rather than @xmath121 .
now , armed with all necessary expressions we may estimate the cooling of qs and qcns being in @xmath1- or 2sc phases ( except for the crust ) .
we select the following values of the pairing gaps : small gaps @xmath128 mev as suggested to occur in the color superconducting region by @xcite and a large gap @xmath129 mev , as suggested for cfl phase in recent works @xcite .
@xcite have discussed the cooling of superconducting qs and qcns for very small gaps with critical temperatures @xmath130 . however , for qdu processes the suppression factor @xmath80 has been used rather than @xmath79 .
the value of the critical temperature @xmath131 seems to be quite small , therefore we use @xmath132 mev for the case of small gaps .
we calculate the cooling history of qs and qcns using eq .
( [ sum ] ) , where now summation over @xmath133 implies summation of emissivities of qdu , eq .
( [ neutr - du ] ) , suppressed by @xmath79 , qmu and qb , eq.([neutr - b ] ) , suppressed by @xmath80 , emissivity of the crust , eq .
( [ neutr - cr ] ) , and emissivity of photon - gluon decay , eq .
( [ em - ph ] ) .
summation over @xmath134 implies summation of the quark contribution evaluated according to eq .
( [ heat ] ) , the electron contribution , eq .
( [ el ] ) , the massless photon - gluon goldstone contribution which coincides with that given by eq .
( [ gam - g ] ) and the contributions of massive gluons given by eq .
( [ gluon ] ) suppressed by @xmath135 and thus being very tiny .
the contribution of the crust to the specific heat is negligible .
also in the cfl phase exist 9 hadronic quasi - goldstone modes .
although their masses are not known we may roughly estimate them as @xmath136 , where @xmath137 is the bare quark mass with a minimum value of @xmath138 mev . with these masses
the contribution of hadronic quasi - goldstone modes is also very small at temperatures of our interest and can be neglected . the dotted curves in figs . 1 - 3
demonstrate the cooling history of qcns and qs for the gap @xmath139 mev , whereas the dashed lines correspond to the cooling of the cfl phase for @xmath140 mev .
all thin dotted and dashed lines correspond to the case when the process of massive mixed photon - gluon decay is artificially excluded whereas the corresponding thick lines represent the cooling history when this process is taken into account according to eq .
( [ em - ph ] ) .
this new process essentially influences on the early stage of the cooling although the mass of the mixed photon - gluon excitation was supposed to be very high ( @xmath141 mev for @xmath142 , lower panel , and @xmath143 mev for @xmath75 , upper panel ) .
this is due to to a big numerical factor in eq .
( [ em - ph ] ) . in all the cases we obtain very rapid cooling in disagreement with the data .
particularly rapid cooling occurs for the cfl phase . in the latter case contributions of the qdu , qmu and qb
processes to the emissivity are suppressed as well as the quark contribution to the specific heat .
the rate is governed by the photon emissivity from the surface . for @xmath144 ,
the specific heat is determined by the electrons .
as the consequence of this reduction of the specific heat we get a very rapid cooling of the cfl ( @xmath144 ) phase , see lower panel of figs . 1 - 3 . for @xmath75 ( upper panel of figs
. 1 - 3 ) there are no electrons and the specific heat is determined by a very small contribution of the goldstone mode given by eq .
( [ gam - g ] ) , so that both qcns and qs cool down even faster than for the case @xmath142 . in both @xmath144 and @xmath145 cases the cooling time of the cfl phase is extremely small .
this means that in reality the cooling is governed by the heat transport in the thin crust @xcite , which we did not take into account .
thus we see that qcns and qs , if present among the objects measured in x - rays , ca nt be in the cfl phase .
the cooling of this phase is so rapid that one might expect problems not only for the models of qs and qcns but also for the models of ns with quark cores consisting of the cfl phase only in deep interiors .
this phase is probably more reliable for qcns rather than for qs since the cfl phase is energetically favorable in the latter case .
the 2sc phase is characterized by large gaps , @xmath16 mev @xcite . to be specific
we suppose that blue - green and green - blue @xmath18-quarks are paired , whereas red @xmath146- and @xmath147-quarks ( @xmath148 ) remain unpaired .
this has the consequence that the qdu processes on the red ( unpaired ) quarks , as @xmath149 , as well as qmu , @xmath150 , and qb , @xmath151 , are not blocked by the gaps whereas other processes involving paired quarks are blocked out by large diquark gaps .
the qdu process on red quarks occurs in the @xmath70 case only .
its emissivity is given by @xmath152 the extra suppression factor of the rate ( [ neutr - du ] ) comes from the fact that the number of available unpaired color states is reduced .
qmu and qb processes on red quarks are also rather efficient .
although there is no one - gluon exchange between @xmath153 , the qmu and qb processes may go via a residual quark - quark interaction , e.g , via two - gluon exchange .
we roughly estimate the corresponding emissivities as @xmath154 in the 2sc ( @xmath70 ) phase the qdu process on red quarks is the dominant process and qmu and qb processes on red quarks are subdominant processes whereas in the 2sc ( @xmath127 ) phase qdu processes do not occur and qmu and qb processes on red quarks become the dominant processes .
other processes like qdu , qmu and qb with participation of other color and flavor quarks are continued to be appreciably suppressed by large gaps .
the specific heat is also changed in the 2sc phase since the @xmath155 and @xmath156 contributions are not suppressed by a factor @xmath79 whereas color - paired @xmath18-contributions remain to be suppressed . with these findings
we calculate the cooling history of qcns and qs .
the results are presented in fig . 5 for @xmath71 , @xmath157 ( thick lines ) , and @xmath75 , @xmath158 ( thin lines ) .
we see that in both cases the cooling history of qcns and also of qs with a tiny crust ( @xmath13 ) nicely agrees with the x - ray data .
the cooling of qs with negligible crust does not agree with the data .
we have estimated the contributions of various quark processes to the emissivity . among them , the new decay process of the massive mixed photon - gluon excitation is operating at the early stage of the cooling and qdu , qmu and qb processes on red quarks determine the cooling of the 2sc phase .
we discussed the cooling history of qs and qcns taking into account different possibilities : @xmath70 and @xmath127 , the normal quark phase , and various color superconducting phases as the `` @xmath1-phase - small gaps '' suggested by @xcite , the cfl phase , and the 2sc phase , as suggested in recent works @xcite .
in all the cases we see that _ qs and qcns being in the cfl phase cool down extremely fast in disagreement with known x - ray data . _ also the cooling curves for the case of small gaps ( @xmath128 mev ) disagree with the data .
even if the cfl phase would be realised only in the deep interior region of a ns it would be problematic to satisfy x - ray data . in this case
the star would radiate mostly not from the surface but from the cfl region due to its extremely small specific heat related to the goldstone excitation .
thus the cooling time would be determined by the heat transport from exterior regions to the center rather than by the cooling of the hadronic shell . in this respect
the following remark is in order .
it is now believed that quark matter below @xmath159 mev is in the color superconducting state characterized by a diquark condensate with large energy gaps ( @xmath16 mev ) rather than being in the normal state or the superfluid state characterized by small gaps ( @xmath160 mev ) .
if so , one could think that our above discussion of normal quark matter and of the case of a small gap has just pedagogic reasoning .
however , this is not really so . indeed , besides the idea of abnormal strange nuclei and strange stars @xcite there is the very similar idea of abnormal pion condensate nuclei and stars with pion condensate nuclei , see @xcite and the review @xcite , chapters 15 , 16 .
the same relates to the kaon condensate objects .
pion condensate systems cool down at about the same rate as given by qdu processes ( for @xmath161 ) . besides
, they can be in the normal state or in the superfluid state characterized by very small gaps @xmath162 mev .
the cooling history of systems being in normal state is described by thick solid curves on the lower panels of figs . 1 - 3 .
thus we may also conclude that _ the hypothesis of pion condensate nuclei - stars _ ( being in normal @xmath163 state with a crust ) _ does not contradict to the x - ray observations .
_ stars with pion and kaon condensate nuclei being in the superfluid state with gaps @xmath164 mev are ruled out as objects being observed in x - rays .
\(i ) it is conceivable that there are more complex collective effects which essentially affect the specific heat and the luminosity .
e.g. , we calculated the mixed photon - gluon spectrum in a simplified model of two abelian gauge fields and concluded that the mass of the excitation is large , whereas one ca nt exclude that in the realistic non - abelian case there exists a photon - gluon excitation of a small mass that could lead to very efficient cooling via the mixed photon - gluon decay process given by eq .
( [ em - ph ] ) .
the masses of hadronic quasi - goldstone modes in cfl phase should be carefully studied .
\(ii ) as we mentioned above in the discussion of the du process , we have neglected the contribution of strange quarks ( qdu - s ) relative to that of the light quarks . the former contribution is @xmath165 times smaller than the latter in the case of normal matter and in the case when all diquark gaps are identical , the discussion given for the cfl phase applies . however ,
if the pairing gaps for strange diquarks are smaller than those for nonstrange diquarks the qdu - s contribution to the emissivity can be essentially enhanced . up to now
there exist only rough estimates of the values of the gaps and this gives rise to large uncertainties in final estimates of the emissivity .
above , in order to be specific , we made calculations considering qdu on u- and d- quarks only . the inclusion of qdu - s in the case when the strange diquark gaps can be smaller than those for non - strange diquarks is simulated by varying the values of the gaps in a wide interval .
this does not change the qualitative picture of the qs and qcns cooling we have discussed in the present work .
\(iii ) we also would like to point out that , if the compact object formed in the explosion of sn 1987a was a qs or a qcns being in the cfl phase ( @xmath127 ) , it is now so cold that it is already impossible to observe it in soft x - rays .
this becomes particularly interesting if continued observation of sn 1987a would find a pulsar and would not observe it in x - rays .
we acknowledge the important remarks and fruitful discussions by m. alford and k. rajagopal after reading the draft of the paper .
we also thank b. friman , j. e. horvath , e. kolomeitsev , r. pisarski , d. rischke , a. sedrakian , and f. weber for their discussions .
one of us ( dnv ) is grateful for the hospitality extended to him during the visit at rostock university and acknowledges financial support from the max - planck - gesellschaft .
alcock , c. , farhi , e. , and olinto , a. 1986 , , 310 , 261 alford , m. , rajagopal , k. , and wilczek , f. 1998 , phys . lett .
b , 422 , 247 alford , m. , rajagopal , k. , and wilczek , f. 1999 , nucl .
b , 357 , 443 alford , m. , berges , j. , and rajagopal , k. 1999 , preprint ( hep - ph/9903502 ) bailin , d. , and love , a. 1984 , , 107 , 325 blaschke , d. , rpke , g. , sedrakian , a. d. , schulz , h. , and voskresensky , d. n. 1995 , , 273 , 596 blaschke , d. , and roberts , c. d. 1998 , , 642 , 197 blaschke , d. , sedrakian , d. m. , and shahabasyan , k. m. 1999 , , in press ( astro - ph/9904381 ) bloch , j. c. r. , roberts , c. d. , and schmidt , s. m. 1999 , , in press ( nucl - th/9907086 ) bodmer , a. 1971 , , 4 , 1601 carter , g. w. , and diakonov , d. 1999 , , 60 , 16004 diakonov , d. , forkel , h. , and lutz , m. 1996 , phys . lett .
b , 373 , 147 duncan , r. c. , shapiro , s. l. , and wasserman , i. 1983 , , 267 , 358 glendenning , n. k. 1996 , compact stars , new york : springer , chapters 8 - 10 horvath , j. e. , benvenuto , o. g. , and vucetich , h. 1991 , , 44 , 3797 iwamoto , n. 1982 , ann .
, 141 , 1 kisslinger , m. b. , and morley , p. d. 1976 , , 13 , 2771 lattimer j. m. , van riper , k. a. , prakash , m. , and prakash , m. 1994 , , 425 , 802 maxwell , o. v. 1979 , , 231 , 201 migdal , a. b. 1971 , zhetf 61 , 220 ; 1978 , rev .
phys . , 50 , 107 migdal , a. b. , saperstein , e. e. , troitsky , m. a. , and voskresensky , d. n. 1990 , phys .
192 , 179 mhlschlegel , b. 1959 , z. phys .
155 , 313 page , d. 1992 , high energy phenomenology , m. a. perez and r. huerta , singapore : world scientific , 347 pisarski , r. , and rischke , d. 1999 , preprint ( nucl - th/9907041 ) pizzochero , p. m. 1991 , , 66 , 2425 price , c. 1980 , , 22 , 1910 rapp , r. , schfer , t. , shuryak , e. v. , and velkovsky , m. 1998 , , 81 , 53 rapp , r. , schfer , t. , shuryak , e. v. , and velkovsky , m. 1999 , preprint ( hep - ph/9904353 ) de rujula , a. , and glashow , s. l. 1984 , , 312 , 734 schaab , ch . , weber , f. , weigel , m. k. , and glendenning , n. k. 1996 , , 605 , 531 schaab , ch . ,
hermann , b. , weber , f. , and weigel , m. k. 1997 , j. phys .
g , 23 , 2029 ; , 480 , l111 schaab , ch .
, sedrakian , a. , weber , f. , and weigel , m. k. 1999 , preprint ( astro - ph/9904127 ) schfer , t. 1998 , nucl .
a , 642 , 45 schfer , t. , and wilczek , f. 1999 , , 82 , 3956 schfer , t. , and wilczek , f. 1999 , preprint ( hep - ph/9903503 ) shapiro , s. , and teukolsky , s. a. 1983 , black holes , white dwarfs and neutron stars : the physics of compact objects , new york : wiley , ch .
10 tsuruta , s. 1979 , , 56 , 237 voskresensky , d. n. , sorokin , g. a. , and chernoutsan , a. i. 1977 , jetp lett .
, 26 , 465 voskresensky , d. n. , and senatorov , a. v. 1984 , jetp lett .
40 , 1212 ; 1987 , sov .
phys . , 45 , 411 voskresensky , d. n. , and senatorov , a. v. 1986 , jetp 63 , 885 voskresensky , d. n. , kolomeitsev , e. e. , and kmpfer , b. 1998 , jetp , 87 , 211 weber , f. 1999 , pulsars as astrophysical laboratories for nuclear and particle physics , bristol : iop publishing witten , e. 1984 , , 30 , 272
|
the effect of color superconductivity on the cooling of quark stars and neutron stars with large quark cores is investigated .
various known and new quark - neutrino processes are studied . as a result , stars being in the color flavor locked ( cfl ) color superconducting phase cool down extremely fast .
quark stars with no crust cool down too rapidly in disagreement with x - ray data .
the cooling of stars being in the @xmath0 color superconducting ( 2sc ) phase with a crust is compatible with existing x - ray data .
also the cooling history of stars with hypothetic pion condensate nuclei and a crust does not contradict the data .
|
using the wmap - determination of the baryonic density @xcite , one obtains predictions of the abundances of the light element isotopes produced in big bang nucleosynthesis @xcite . while the overall values from theoretical predictions and from the observational determinations of the abundances of d and @xmath4he are in good agreement , the theory tends to predict a higher @xmath0li abundance ( by a factor 2 to 3 ) than is observed in the atmospheres of halo dwarf stars @xcite .
the nacre compilation @xcite provided a new set of reaction rates that were used to update the predictions of contemporary big bang nucleosynthesis ( bbn ) @xcite . at that time , the baryonic densities obtained from cmb observations on the one hand and comparison between bbn calculations and spectroscopic data on the other hand were only marginally compatible @xcite . in order to improve the nuclear network ,
@xcite recently performed a re - analysis of low energy data from the 10 key nuclear reactions involved in bbn , by using r - matrix theory @xcite and evaluating the remaining uncertainties in a statistically robust formalism . using this improved network ,
@xcite have recently calculated bbn light element productions assuming for the baryonic density the very precise value provided by wmap @xcite and obtained @xmath2li / h@xmath5 compared to the observed value li / h @xmath6 , confirming the @xmath2li discrepancy .
however , it has been shown @xcite that the @xmath2be(d , p)2@xmath3 reaction ( which destroys the @xmath2be that is the source of @xmath2li at high baryonic density ) , would solve the @xmath2li problem _ if _ its cross section were much higher than assumed .
importantly , prior to the present work , _ no direct experimental data at bbn energies _ were available ( for @xmath7 gk , the gamow window is @xmath8 mev ) .
in fact , the @xmath2be(d , p)2@xmath3 reaction rate relied on an extrapolation made by @xcite based on experimental data at center - of - mass ( c.m . ) energies of 0.6 to 1.3 mev from @xcite . in this experiment , protons corresponding to the @xmath9be @xmath10 ground state ( g.s . ) and first excited state ( 3.03 mev , @xmath11 ) were detected at 90@xmath12 using a nai(tl ) detector .
assuming an isotropic angular distribution , @xcite multiplied the measured differential cross section by @xmath13 and by a further factor of 3 to take into account the estimated contribution of the higher energy @xmath9be states , not observed by @xcite .
consequently , a constant @xmath14-factor of 100 mev - barn was adopted . in order to obtain @xmath0be(d ,
p)2@xmath3 reaction cross section at bbn energies , we have performed an experiment at the cyclone radioactive beam facility at louvain - la - neuve , belgium , using an isobarically pure @xmath0be radioactive beam . the experimental method and results
are presented in section 2 .
the astrophysical consequences are discussed in section 3 .
the conclusions are given in section 4 .
the measurements were performed using a post - accelerated @xmath2be@xmath15 radioactive beam at a nominal energy of 5.8 mev provided by the cyclone110 cyclotron .
a detailed description of the production of the @xmath2be beam can be found in @xcite . to suppress the contamination from the @xmath0li isobaric beam , the @xmath0be beam
was completely stripped to @xmath0be@xmath16 by transmission through a thin @xmath17c foil , prior to analysis by a dipole magnet .
prior to the @xmath2be(d , p)2@xmath3 measurement , the beam energy was determined using a calibrated si detector situated at 0@xmath12 . a laboratory energy of 5.55 mev ( fwhm @xmath18 ) was determined , including a correction for pulse height defect .
this energy was degraded to 1.71 mev ( fwhm @xmath19 ) using a 6 @xmath20 m mylar foil located at 50 cm upstream of the target .
no @xmath0li contamination was observed , consistent with lithium isotopes being unable to support a 4 + charge state .
the target consisted of a 200 @xmath20g/@xmath21 ( cd@xmath22)@xmath23 self - supporting foil . with this set - up
, we were able to investigate the center - of - mass energy range between 1.00 and 1.23 mev ( for a beam energy of 5.55 mev , without degrader ) and between 0.13 and 0.38 mev ( for 1.71 mev , with degrader ) .
the cross section measurement was averaged over these energy ranges .
in addition to the feeding of the ground and first excited states of @xmath9be @xcite , we were able to observe the @xmath0be+d reaction via other kinematically allowed higher energy levels , mainly through a very broad @xmath24 state ( @xmath25 mev ) situated at an excitation energy of 11.4 mev in @xmath9be @xcite . at the beam energy of 5.55 mev , several states in @xmath9be above the @xmath0be+d - p threshold
are present but due to the coulomb barrier in the final state , their contribution are expected to be negligible .
the @xmath26 value of the @xmath0be(d , p)@xmath9be reaction is 16.49 mev , thus the laboratory energies of protons and @xmath3 particles are high .
for example , a 5.55 mev @xmath0be beam traversing a 200 @xmath20g/@xmath21 ( cd@xmath22)@xmath23 target will lead to the production of protons with energies anywhere between about 7.5 and 22 mev , for the range of angles covered .
thus , to clearly identify the protons from the @xmath0be+d reaction from those arising from reactions on the c content of the target , a stack of two ` leda ' silicon strip detector arrays @xcite were employed covering a laboratory angular range of @xmath27 .
a schematic view of the experimental set - up is shown in figure 1 .
the @xmath28 detector consisted of eight sectors of 0.3 mm thickness , while the @xmath29 detector consisted of four sectors of 0.3 mm thickness and four of 0.5 mm thickness .
they were calibrated using a 3-line @xmath3-source ( @xmath30pu , @xmath31am , @xmath32 cm ) and a precision pulser .
this @xmath33 detector system allowed a clear identification of the protons produced in the @xmath0be(d , p)@xmath9be reaction .
all the particles that are not stopped in the front @xmath28 detector and that are either stopped or that left energy on the back @xmath29 detector are protons ( having an energy of more than 6.5 mev ) that populated levels up to the 11.4 mev state in @xmath9be .
we were able to measure the @xmath2be(d , p)2@xmath3 cross section up to an excitation energy in @xmath9be of @xmath34 mev for a beam energy of 5.55 mev and of @xmath35 mev for 1.71 mev . only about 50% of the contribution of the @xmath24 broad state
was observed at 1.71 mev . other light particles ( p , d , @xmath36he , @xmath4he ) from @xmath0be+@xmath17c reactions , as well as recoils and scattered particles , were completely stopped in @xmath28 .
figures 2 and 3 show spectra obtained at beam energies of 5.55 and 1.71 mev , respectively .
the spectrum obtained at 1.71 mev ( figure 3 ) was accumulated over about 26 hours of running time with an averaged @xmath0be beam intensity of @xmath37 pps .
as can be seen , the proton signals are well separated from the background signals ( @xmath38 mev ) , which are produced by random coincidences of @xmath3 particles , scattered @xmath0be and recoil ions in the @xmath28 detector with @xmath39 particles in @xmath29 .
the locus with negative slope contains protons that have passed through the front @xmath28 detector and stopped in the back @xmath29 detector .
the two loci with positive slope are events in which the proton has sufficient energy to pass through both detectors completely .
there are two bands because of the two different thickness @xmath29 detectors .
the most strongly populated regions at the lower left of these bands correspond to protons losing the least energy in passing through the silicon , and thus to the highest energy events . by considering the kinematics and energy losses in silicon @xcite , together with the straggling of the beam and experimental energy resolution
, one may then identify events on the positive slope locus up to 2.5 ( 3.9 ) mev for the 0.3 ( 0.5 ) mm si thickness wafer , as corresponding to events in which the recoiling @xmath9be nucleus is in either the @xmath10 g.s . or the @xmath11 excited state ( the energy resolution is insufficient to resolve the two ) .
the total statistical error was 10% for the beam energy of 1.71 mev and less than 2% for 5.55 mev ( for protons populating the @xmath10 and @xmath11 states it was 13% and 2.5% , respectively ) .
the absolute normalization was obtained using events arising from the elastic scattering of the @xmath0be on the c content of the target ( as recorded by the @xmath28 detector in which the @xmath0be are stopped ) , and assuming that the @xmath0be+@xmath17c elastic scattering follows the rutherford law .
this assumption is realistic at energies below the coulomb barrier , as is the case here . to calculate the average cross section over the energy ranges and angular coverages @xmath40 ,
the number of counts was corrected for the detector solid angle ( uncertainty @xmath41 ) , the number of deuterons in the target ( @xmath42 ) and the total number of incoming beam particles ( @xmath43 and @xmath44 at the higher and lower beam energies , respectively ) , and transformed into the c.m .
system .
the proton angular distribution over the angular range covered here was found to be isotropic at both energies .
thus , we assumed full isotropy and calculated the average cross sections , @xmath45 mb at the effective energy of 0.37 mev and @xmath46 mb at 1.15 mev .
the summed contribution of the @xmath10 and @xmath11 states was about 64% of @xmath47 at 1.15 mev . at 0.37 mev
, @xmath47 includes the contribution of the ground and @xmath11 states and about 50% of that of the @xmath24 broad state . due to the low penetration probability ( @xmath48 ) , the contribution of the @xmath24 state should be less than 36% .
thus , the @xmath47 value at 0.37 mev corresponded to more than 80% of the total cross section ( for a @xmath24 state with @xmath25 mev ) . this was taken into account in the systematic uncertainty . in nuclear astrophysics
it is usual to present the cross section in the form of the astrophysical @xmath14-factor @xmath49 given by @xcite , @xmath50 where @xmath51 is the sommerfeld parameter ( @xmath52 , with @xmath53 and @xmath54 the charge numbers of the target and beam and @xmath55 is the velocity ) and @xmath56 is the effective c.m . energy . in the absence of sharp resonances , the @xmath14-factor varies smoothly with energy . figure 4 shows the @xmath0be(d , p)@xmath9be astrophysical @xmath14-factor @xmath57 in mev - barn as a function of the c.m . energy . for a comparison with the data of @xcite ( open circles ) , the present data including only contributions from the ground and first excited states of @xmath9be ( filled circles ) are shown .
the agreement with the @xcite data at overlapping energies is satisfactory , given the systematic uncertainties .
the total s - factor is also shown ( filled triangles ) .
the present data show that the higher energy states not observed by @xcite contribute about 35% of the total @xmath14-factor instead of the 300% estimated by @xcite .
hence , the @xmath0be(d , p)@xmath9be reaction rate is smaller by a factor of about 2 at energies in the range 1.0 to 1.23 mev and by about 10 at energies relevant to bbn , than previously estimated .
this excludes a nuclear solution to the primordial lithium abundance problem via the @xmath0be(d , p)@xmath9be reaction as its effect is completely negligible compared to the 7% ( @xmath58 ) nuclear uncertainty on the @xmath0li yield .
nevertheless , these results allow a more accurate determination of the @xmath0li abundance using bbn models .
since the pioneering work of @xcite , who found a value of li / h@xmath59 independent of fe / h ( for [ fe / h]@xmath60 ) there have been many independent observations of li confirming the existence of a plateau and suggesting that this abundance reflects the primordial li value . however , the li abundance extracted from observations depends drastically on the assumed surface temperature of the star ( @xcite ) .
recent observations @xcite have lead to li / h @xmath61 which is very close to the first evaluation @xcite .
the more recent work studied and quantified the various sources of uncertainty : extrapolation , stellar depletion and stellar atmosphere parameters . compared to the wmap+bbn value , the discrepancy is a factor of @xmath62 .
if it is shown that there is a mechanism by which the outer layers of population ii stars are transported deep into the stellar interior , then there are several ways in which li abundances might be depleted over the lifetime of the star . in this context , the current estimates for possible depletion factors may be in the range @xmath63 dex @xcite
. however , the data typically show negligible intrinsic spread in the li abundance leading to the conclusion that depletion in these stars is of the order of 0.1 dex .
recently , @xcite have obtained a higher value for the li plateau abundance ( @xmath64 ) due to a new effective temperature scale which is higher at low metallicity .
this new evaluation diminishes the discrepancy , without canceling it .
the observation of @xmath65li is also of interest , since , because it is more fragile than @xmath0li , it can provide yet more severe constraints upon possible depletion mechanisms @xcite . finally , in spite of the various uncertainties related to li observations and to the stellar models , it is very difficult to reconcile the bbn @xmath0li and the spite plateau which presents a narrow dispersion all along the metallicity scale .
the existence of the spite plateau for li seems to indicate that low metallicity halo stars are indeed representative of the primordial bbn abundance . in particular
, the isotope @xmath0li plays a key role as a bridge between big bang nucleosynthesis , stellar evolution and galactic cosmic - ray nucleosynthesis . at present
there is a significant discrepancy between the bbn - predicted @xmath0li abundance ( assuming a baryon density consistent with the concordance model derived from observations of anisotropies in the microwave background ) and the abundance determined from the observations of li in the atmospheres of halo stars .
the experiment reported here demonstrates that the @xmath2be(d , p)2@xmath3 @xmath14-factor at bbn energies was not underestimated by @xcite but , on the contrary , _
overestimated_. the discrepancy can not therefore be resolved by nuclear physics inputs to bbn calculations .
the remaining conventional options ( those not invoking physics beyond the standard model ) are an adjustment of the stellar input parameters needed to extract the li abundances from observations , or stellar depletion of @xmath0li .
however , models must be constructed to avoid dispersion in the @xmath0li abundances over a wide range of stellar parameters , which is a real challenge .
the origin of the discrepancy in the li abundance remains a challenging issue .
this work was partially supported by the ec under contract no .
hpri - ct-1999 - 00110 , the belgian inter - university attraction poles p5/07 , the in2p3 ( france ) , pics 1076 usa / cnrs , and the uk epsrc .
ev is grateful to k. olive for fruitful discussions .
|
the wmap satellite , devoted to the observations of the anisotropies of the cosmic microwave background ( cmb ) radiation , has recently provided a determination of the baryonic density of the universe with unprecedented precision . using this ,
big bang nucleosynthesis ( bbn ) calculations predict a primordial @xmath0li abundance which is a factor @xmath1 higher than that observed in galactic halo dwarf stars . it has been argued that this discrepancy could be resolved if the @xmath2be(d , p)2@xmath3 reaction rate is around a factor of 100 larger than has previously been considered .
we have now studied this reaction , for the first time at energies appropriate to the big bang environment , at the cyclone radioactive beam facility at louvain - la - neuve .
the cross section was found to be a factor of 10 _ smaller _ than derived from earlier measurements .
it is concluded therefore that nuclear uncertainties can not explain the discrepancy between observed and predicted primordial @xmath0li abundances , and an alternative astrophysical solution must be investigated .
|
loop quantum gravity ( lqg ) proposes a background independent framework for a theory of quantum general relativity @xcite .
it realizes a canonical quantization of general relativity in 3 + 1 space - time dimensions and defines the hilbert space of quantum states of 3d space geometry and their dynamics ( through a hamiltonian constraint ) .
background independence means that there is no assumed background metric at all and that the quantum state of geometry describes the whole metric of space(-time ) and not simply the perturbations of the metric field around a fixed background metric .
the whole geometry of the space(-time ) needs to be reconstructed from the quantum state : all geometric notions , such as the distance , need to be constructed from scratch since you can not rely on a background geometry to define , for instance , a reference notion of distance that you could use to describe the metric perturbations .
the states of 3d space geometry are ( superpositions ) of spin networks .
these are equivalence classes of labelled graphs under spatial diffeomorphism .
the labels come from the representation theory of the gauge group @xmath0 , representation labels on the edges and invariant tensors ( or singlet states ) at the vertices .
these spin networks thus contained both algebraic ( or combinatorial ) data and topological data such as winding numbers of the edges of the graph . in the present paper
, we will neglect this topological information and work with _
abstract spin networks _ which are purely algebraic objects . this erases all traces of the topology of the original 3d space manifold which we quantize .
this is an open issue in lqg : whether the quantum states of geometry describe both the metric and topology of space or whether they should only describe the metric state in some fixed background ( space ) topology .
the standard lqg framework fixes the background topology , uses embedded spin networks and thus forbids topology changes , whereas the more recent spin foam framework ( see for example @xcite ) , which proposes a path integral formalism for lqg , uses abstract spin networks and allows transitions between different spatial topologies . more generally , it seems possible to extract some topological information from the algebraic data encoded in the spin network ( see for example @xcite ) .
however , on the other hand , the topological degrees of freedom allowed in spin networks leads to interesting proposals , such as a possible inclusion of the standard model for particles in lqg @xcite .
it is essential to better understand the quantum geometry defined by spin networks and how a classical metric can emerge .
a first step would be to develop the concept of _ distance _ on a spin network .
since we are in a completely background independent context and that the spin network ought to define the ( quantum ) state of geometry , the algebraic and combinatorial structures of the spin networks must induce a ( natural ) notion of distance on the quantum manifold " defined by the spin network .
moreover , since there is no background metric defining a reference frame , there is no concept of the position " of a part of the spin network and we have to deal with _ relations _ between parts of the spin network .
it is therefore natural to try to reconstruct a notion of distance between two parts of the spin network as a function of the correlations between these parts , as suggested in @xcite .
developing a notion of distance , close and far , is a necessary tool to defining a proper coarse - graining procedure of the spin network , which would not rely on an assumed embedding of the spin network in a fixed background manifold but solely on the spin network state itself .
the goal is not to assume but to derive the embedding of the spin network into a classical manifold , which would define its semi - classical limit .
such coarse - graining would thus allow to study the emergence of a classical metric in lqg and the semi - classical dynamics of the theory . in the next section ,
we introduce the mathematics of the spin networks and we describe the quantum states of geometry defined by lqg .
we insist that the spin networks are wave functions of the geometry / metric and we remind their interpretation as probability amplitudes . in section [ arearenorm ] , we remind the definition of surfaces in lqg and describe the ( quantum ) surface states .
surfaces can be thought as made of elementary quantum surfaces patched up together .
we define a coarse - graining of these quantum surfaces in the spirit of the area renormalisation introduced in @xcite .
we explain how renormalisation flow reflects the structure of surface i.e how the elementary surfaces are patched together to form the big surface . in section [ coarse ]
, we introduce a notion of coarse - grained state of a bounded region of the spin network and we discuss the properties of the coarse - grained state . more details on the relation between the coarse - graining and the usual operation of partial tracing is presented in appendix . in section
[ entanglement ] , we use that tool in order to define the entanglement and correlation between two parts of the spin network .
finally in section [ theidea ] , we argue that these correlations define a natural notion of distance between parts of the spin network . this distance would be determined solely from the algebraic structure of the spin network and would not rely on any background structure .
such a relationship between geometrical quantities ( such as the distance or the metric ) and informational concepts ( such correlations ) supports the growing link between ( loop ) quantum gravity and quantum information @xcite .
loop quantum gravity is a canonical formulation of general relativity , describing the ( quantum ) evolution of the 3d metric @xmath1 on a spatial slice @xmath2 of space - time .
the ashtekar - barbero variables describing the 3d geometry are a @xmath0-connection @xmath3 and a triad @xmath4 ( @xmath5-valued 1-form ) which is its canonical momentum .
the connection @xmath6 describes the parallel transport in the 3d space , while @xmath7 defines the 3d metric via @xmath8 . in these variables , general relativity becomes a @xmath0 gauge theory , with gauss constraints insuring the @xmath0 gauge invariance and other constraints implementing the invariance under ( space - time ) diffeomorphisms .
loop quantum gravity chooses the @xmath6-polarization .
we define wave functions @xmath9 and @xmath7 acts as a derivation operator .
we select _ cylindrical functions _ , which depend on the field @xmath6 through a finite number of parameters .
a cylindrical function is defined with respect to an oriented ( closed ) graph @xmath10 ( embedded in @xmath2 ) .
let us construct the holonomies @xmath11\in{\mathrm{su}}(2)$ ] of the connection @xmath6 along the edges @xmath12 of @xmath10 .
a cylindrical function on @xmath10 is defined as a function of these holonomies : @xmath13 ) , \quad e\in\gamma.\ ] ] moreover , we require that these functions are @xmath0 gauge invariant i.e invariant under gauge transformations of the connection @xmath14 for @xmath15 .
this translates into the requirement of @xmath0 invariance at every vertex @xmath16 of the graph @xmath10 : ( u_e[a])=(g_s(e)^-1u_e[a]g_t(e ) ) , g_v(2 ) , where @xmath17 and @xmath18 denote the source and target vertices of the edge @xmath12 .
therefore , for a fixed graph @xmath10 , we are looking at the space of functions over @xmath19 where @xmath7 and @xmath20 are the number of edges and vertices of @xmath10 respectively .
this space is provided with the haar measure on @xmath21 , which defines the kinematical scalar product of lqg .
this measure defines a hilbert space @xmath22 of wave functions associated to the graph @xmath10 .
the invariance of the theory under spatial diffeomorphisms is taken into account by considering equivalence classes of graphs on @xmath2 under diffeomorphisms .
the ( kinematical ) hilbert space of lqg , @xmath23 , is then constructed as the projective limit of the @xmath24 spaces associated to each ( equivalence class of ) graph , which implements a sum over all possible graphs . for more details
, the interested reader can refer to @xcite .
a basis of @xmath22 is provided by the so - called _ spin network _ functionals .
these are constructed as follows .
to each edge @xmath25 , we associate a @xmath0 representation labelled by a half - integer @xmath26 called _ spin_. the representation ( hilbert ) space is denoted @xmath27 and has a dimension @xmath28 . to each vertex @xmath16 , we attach an intertwiner @xmath29 , which is @xmath0-invariant map between the representation spaces @xmath27 associated to all the edges @xmath12 meeting at the vertex @xmath16 : @xmath30 since the conjugate representation @xmath31 is isomorphic to the original @xmath32 , we can alternatively consider @xmath29 as a map from @xmath33 to @xmath34 .
then one can also call the intertwiner @xmath29 an _ invariant tensor _ or a _
singlet state_. once the @xmath35 s are fixed , the intertwiners at the vertex @xmath16 actually form a hilbert space , which we will call @xmath36 . moreover ,
considering the decomposition into irreducible representations of the tensor product @xmath37 , @xmath38 where the @xmath39 are degeneracy coefficients depending on the @xmath35 s , we call @xmath40 the subspace corresponding to the spin @xmath41 component of the tensor decomposition
. then @xmath42 is actually the space of singlets @xmath43
corresponding to the spin @xmath44 component . a spin network state @xmath45 is defined as the assignment of representation labels @xmath35 to each edge and the choice of a vector @xmath46 for the vertices . to shorten the expressions
, we may omit the indices and use vectorial notations , @xmath47 .
the spin network state defines a wave function on the space of discrete connections @xmath19 , [ spinnetdef ] _ , [ g_e ] = g_e|_v _ e d^j_e(g_e ) _ v _ v , where we contract the intertwiners @xmath29 with the ( wigner ) representation matrices of the group elements @xmath48 in the chosen representations @xmath35 . using the orthogonality of the representation matrices for the haar measure , @xmath49 we can directly compute the scalar product between two spin network states : [ matnorm ] , , |,,= _ e _ v _ v|_v .
therefore , upon choosing a basis of intertwiners for every assignment of representations @xmath50 , the spin networks provide a basis of the space @xmath22 .
summing over all graphs @xmath10 , we then define the spin network basis of lqg .
considering that the triad @xmath7 allows to reconstruct the metric @xmath51 and implementing @xmath7 as a derivation operator on the wave functions , one can compute the action of geometrical operators such as the area and the volume on the spin network basis .
for instance , considering a graph @xmath10 and a spin network state @xmath47 , the area of an elementary surface @xmath52 transversal to a given edge @xmath12 ( and not intersecting any other edges of the graph ) will have a quantized spectrum : _
_e|,,= l_p^2 s(j_e)| , , , where @xmath53 or @xmath54 upon a choice of ordering in the quantization .
note that @xmath53 is the value of the casimir operator @xmath55 in the representation @xmath56 , while @xmath57 .
one can also investigate the action of the volume operator and it would be seen to act only at the vertices of the graph and depend on the intertwiners @xmath29 . to sum up the situation ,
the representation labels can be considered as the quantum numbers for the area while the intertwiners would be the quantum numbers for the ( 3d ) volume .
this provides the spin network states with an interpretation as _
discrete geometries_. we conclude this review section with a few points on the representation theory of @xmath0 . for a given representation @xmath56 ,
the trace of the representation matrices @xmath58 defines the character @xmath59 , @xmath60 where @xmath61 , @xmath62 $ ] is half the angle of the rotation , @xmath63 is a unit vector on the 2-sphere indicating the axis of the rotation and @xmath64 are the standard pauli matrices .
we will write in the following @xmath65 .
note that the @xmath66 s are invariant under conjugation @xmath67 . in these variables ,
the normalized haar measure on @xmath0 reads is actually the usual lebesgue measure on the 3-sphere @xmath68 . indeed with the parametrization @xmath69 , we have : @xmath70 ] : @xmath71 note that @xmath72 and that the angle @xmath73 is defined modulo @xmath74 .
finally , we remark that the following operator is actually the identity on the intertwiner space @xmath75 : _ ( 2 ) dg _
d^j_i(g ) = .
[ identity ] this defines the maximally mixed state on the singlet space @xmath75 up to a normalization factor .
even though spin network states diagonalize geometrical operators such as areas and volumes , one must not forget their definition as wave functions : they define probability distributions on the space on ( discrete ) connections .
more precisely , a normalized cylindrical function @xmath76 defines a probability measure on @xmath19 : p(\{g_e , e})_e dg_e also contain information about the parallel transport on the 3d manifold .
we give two simple examples to illustrate the use of the spin network as a probability amplitude . consider the simplest spin network graph based on a single closed loop ( with a single vertex ) .
the spin network wave function is labelled by a single representation label @xmath41 ( and no intertwiner is needed ) and is defined as : w_j(g)=d^j_mn(g)_mn=_j(g ) .
@xmath77 is invariant under conjugation and the only gauge invariant data is the angle @xmath73 of the rotation defined by @xmath78 .
actually @xmath79 is the usual _
wilson loop_. the character is already normalized @xmath80 so that @xmath79 define the following probability distribution : p()d= ^2 |_j()|^2 = ^2 ( 2j+1 ) , where the @xmath81 factor comes from the haar measure .
this probability distribution is @xmath82-periodic .
its maxima -the most probable parallel transport along the loop- are given by @xmath83 , @xmath84 .
on the other hand , @xmath85 vanishes for @xmath86 and in particular for the flat connection @xmath87 .
the states @xmath79 diagonalize the area of an elementary surface transversal to the loop .
the area spectrum is @xmath88 as states above .
these states are in a sense completely delocalized in @xmath73 .
on the other hand , one can write states @xmath89 localizing the rotation angle of the holonomy @xmath78 defined as the following distributions : @xmath90 we can decompose these states in the @xmath79 basis : _
( g)= _ j/2 _ j()_j(g ) .
they have an infinite expansion in the spin network basis and are not normalisable .
one could write normalized states defined by gaussian packets centered around fixed values of @xmath73 to smooth out the state or by adding some regularization factors like @xmath91 , and study their @xmath41-average and their properties , but this is not the purpose of the present paper .
the @xmath92 diagram has a more interesting structure of maxima
. its has two vertices linked by three edges .
its spin network wave functional is determined by three representation labels @xmath93 and is defined as : _
j_1,j_2,j_3(g_1,g_2,g_3 ) c^j_1j_2j_3_m_1m_2m_3c^j_1j_2j_3_n_1n_2n_3 _
i=1 ^ 3 d^j_i_m_i n_i(g_i ) , where the @xmath94 s are the ( normalized ) clebsh - gordan coefficients of the recoupling theory of @xmath0 representations .
we can re - write this expression in terms of the characters : _
j_1,j_2,j_3(g_1,g_2,g_3 ) = _ ( 2 ) dh _
i=1 ^ 3 _ j_i(g_ih ) .
it is straightforward to check that the norm of @xmath92 is @xmath95 , which is either 1 is the @xmath96 s satisfy the triangular inequality ( so that @xmath97 ) or 0 otherwise .
@xmath92 is invariant under diagonal ( left and right ) @xmath0 multiplication : @xmath98 using this @xmath0 gauge invariance , we can define the loop variables @xmath99 and we have : @xmath100 writing @xmath101 and @xmath102 , it is easyto notice that the previous expression only depends on @xmath103 and the angle @xmath104 between @xmath105 and @xmath106 .
the angles @xmath107 represent all the gauge - invariant data for the discrete connections @xmath108 defined on the @xmath92-graph . taking into account the haar measure and the measure on @xmath109 , the probability distribution on the gauge - invariant space is : p(_1,_2 , ) d_1d_2d= |(_1,_2,)|^2 ^2_1 ^ 2_2 d_1d_2d .
the resulting probability distributions in the case of @xmath110 are shown on figs .
[ figtheta]-[figtheta3 ] . , title="fig:",width=302 ] , title="fig:",width=302 ] , title="fig:",width=302 ] , title="fig:",width=302 ] , title="fig:",width=302 ] , title="fig:",width=302 ] in general , more complicated spin networks will have more complex probability graphs .
a generic feature though is that the flat connection @xmath111 will have a vanishing probability ( density ) . nevertheless , when the spins @xmath35 grows to @xmath112 , it is likely that we obtain maxima arbitrarily close to the flat configuration ( like in the case of a single wilson loop ) .
let us emphasize that these are only kinematical " probabilities , based on the kinematical scalar product .
physical " probabilities would be based on the physical scalar product taken into account the projection onto the kernel of the hamiltonian constraint .
the kinematical probabilities describe the unconstrained 3d space geometry , whereas the physical probabilities should implement the canonical constraints imposed by general relativity on the 3d geometry and should reflect the dynamical transition amplitudes induced by the hamiltonian constraint . in the present work , we aim to understand the geometry and structure of the quantum 3d geometry states , or spin networks , at the kinematical level . indeed , before implementing the hamiltonian constraints or the einstein equations at the classical level
, we first have to grasp the notion of metric . here
, we would like to understand the notion of quantum metric , before studying the solutions to the quantum constraint which should implement quantum gravity .
let us consider a surface @xmath113 in the 3d space @xmath2 and a spin network state @xmath114 based on the graph @xmath10 considered as embedded in @xmath2 .
the surface intersects the graph does not intersect the graph @xmath10 at any vertex .
considering that case would not complicate the mathematics but only the notations .
] at a certain number of points @xmath115 belonging to the graph edges @xmath116 .
we then cut the surface @xmath113 in elementary patches @xmath117 intersecting @xmath10 at the points @xmath118 : @xmath119 with @xmath120 .
the area ( operator ) of @xmath113 is defined as the sum of the areas ( operators ) of the elementary patches : _
|,v , l_p^2_i s(j_e_i)|,v , .
this simple definition of the area operator of a surface @xmath113 in lqg raises two natural issues , due to the background independence of the theory .
the first question is the definition of the surface itself in the diffeomorphism - invariant framework of lqg . indeed , forgetting about the embedding of @xmath113 and @xmath10 in the space manifold @xmath2 , the surface @xmath113 is defined on the spin network state @xmath114 only through its intersections @xmath118 with the graph @xmath10 . starting with a quantum geometry defined solely from the state @xmath114 , independently from any metric or embedding or reference to @xmath2 , one can define a ( quantum ) surface as a set of edges @xmath121 of the graph @xmath10 : the surface is then definedas the union of elementary patches @xmath117 transversal to each edge @xmath122 .
however , how can we know that these patches are close to each over and that they do form a smooth continuous surface ? the second question is related and concerns the semi - classical limit of such a quantum surface .
more precisely , we would like to know how the surface is folded , i.e how the elementary patches are organized with respect to each other .
since these elementary patches might be crumpled giving a fractal - like structure to the surface , we would like to be able to coarse - grain the surface , define its macroscopic structure and its macroscopic area . following the preliminary ideas in @xcite
, we will propose a framework of _ area renormalisation _ to address these issues and we will define coarse - grain area operators to probe the quantum structure of a surface .
let us fix a graph @xmath10 and an arbitrary quantum state defined as the cylindrical function @xmath123 .
let us consider the quantum surface defined as the set of edges @xmath124 .
the area operator @xmath125 of the elementary surface @xmath117 transversal to the edge @xmath122 is defined as the casimir operator of the following @xmath0 action on @xmath123 : ( g_e_i , g_ee_i ) ( gg_e_i , g_ee_i ) . in this simple case
, it does nt matter whether we act by left or right multiplication .
it is natural to generalize this area operator to collections of elementary surfaces .
indeed , let us group together the edges @xmath126 and @xmath127 .
we define the coarse - grained area operator @xmath128 as the casimir operator of the diagonal @xmath0 action : ( g_e_1,g_e_2,g_ee_1,e_2 ) ( gg_e_1,gg_e_2,g_e ) .
this is easily generalized to any collection of edges .
for example , the complete coarse - grained area operator @xmath129 of the surface @xmath113 is defined as the casimir operator of the diagonal @xmath0 action : ( g_e_1,g_e_2, .. ,g_e_n , g_ee_1, ..
,e_n ) ( gg_e_1,gg_e_2, .. ,gg_e_n , g_e ) .
it is now necessary to distinguish the left and right multiplication .
we actually define @xmath130 different operators .
not only we can define the two operators , @xmath131 and @xmath132 , corresponding to the actions @xmath133 and @xmath134 , but we must also consider the two other actions @xmath135 and @xmath136 .
the key point is that @xmath137 and this results in a non - trivial flow of the area of a surface under coarse - graining .
let us consider the simple example of a spin network state with the edges @xmath138 labelled with the representations @xmath139 and @xmath140 .
the microscopic area of the surface @xmath141 is @xmath142 in planck unit . now assuming that these two edges meet at a 3-valent vertex as in fig . [ 1000to12 ] and that the third edge is labelled with the representation @xmath143 , then the macroscopic ( coarse - grained ) area of the same surface @xmath113 will be @xmath144 which is substantially different .
, width=188 ] we propose to use the coarse - graining flow , or renormalisation flow , of the surface area to probe the structure of the surface . a possible coarse - graining of the surface @xmath145
is summarized by a tree @xmath146 which describes how we group the edges together : each level of the tree defines a partition of @xmath124 which is coarser than the previous level , from the initial point with individual edges to the final point when all the edges are grouped together .
the coarse - grained area ( operator ) of the surface @xmath113 at each level depends on the partition and is defined as the sum of the area operators of each collection of edges of the given partition . then each tree @xmath146
describes a possible coarse - graining of the surface @xmath113 from the microscopic realm to the macroscopic level . given a state of geometry @xmath76
, we can analyze the renormalisation flow of the expectation value of the area for each coarse - graining tree @xmath146 .
for some trees the flow may be smooth , meaning that transitions from a level to a neighboring lead to small changes in area in planck units , while it might be jumpy for the other trees .
we expect that if there does not exist any tree with smooth flow , then the set of chosen edges can not describe a true surface in the macroscopic limit and that these edges are likely to be far from each other in the geometry defined by the state @xmath76 . on the other hand
, a tree with a smooth flow should give us the right way to put the elementary surfaces together grouping them by closest neighbors .
moreover , we expect that such a ( smooth ) area flow tells us more about the properties of the surface such as its curvature .
for example , a ( nearly ) flat surface should have a almost trivial constant area flow .
this proposes an explicit relationship between the geometry of a spin network state and its renormalisation properties .
this will be investigated in more details through specific examples in future work @xcite .
a similar analysis of the volume renormalisation flow should carry even more information about the geometry of the quantum state @xmath76 .
this is to be compared with theories on the lattice .
renormalisation la wilson is defined through coarse - graining the lattice and deriving the flow of the hamiltonian and of the various observables under that coarse - graining operation .
usually , the lattice is embedded in flat space and the coarse - graining defined through the resulting notion of nearest neighbor . in our quantum gravity context , spin networks can be thought as a lattice .
however , there are provided without any embedding into a space manifold .
our proposal is to consider all the possible renormalisation flow of geometrical observables ( such as the area ) and to reconstruct a correct "
embedding by identifying the good " renormalisation flow(s ) .
the whole issue is then to define what we mean by good " .
the simplest criteria is the smoothness of the flow .
however , this should be studied in more details . in the case of a ( pure ) spin network state @xmath147
, we can give a simple expression of the expectation value of a coarse - grained area operator .
let us consider the collection of edges @xmath124 and the casimir operator of the diagonal @xmath0 action @xmath148 where we act at the source vertex of all edges .
considering the action where @xmath78 sometimes acts at the target vertex would not change much to the following . using the kinematical scalar product on the spin networks defined by the integration with the haar measure , we show that the area expectation value is given by : _
( e_1, .. ,e_n ) _ = , where @xmath149 is a density matrix describing the state of the surface @xmath121 in the hilbert space @xmath150 .
@xmath149 is defined in terms of the intertwiners @xmath29 of the spin network at the vertices where we act with @xmath78 , i.e @xmath151 : _
a_1 .. a_nb_1 ..
i , where @xmath152 label a basis of the representation space @xmath153 .
the trace is over the representation space associated to the edges @xmath12 not belonging to the surface @xmath124 .
an example is given in fig .
[ 3edgesurf ] .
@xmath149 is hermitian and generically represents a mixed state .
, width=188 ] from the above expression for the surface state , it is clear that the coarse - graining procedure probes the structure of the spin network around the considered surface , for instance the intertwiners attached to the vertex adjacent to the surface .
these intertwiners describe how the surface is folded , i.e the `` position '' of the elementary patches with respect to each other .
if we were to use a more complicated scalar product between spin network states , taking into account some dynamical effect and propagation of the geometry degrees of freedom , it is likely that the coarse - grained area expectation value would involve intertwiners attached to further vertices and would depend in a more complex way on the spin network geometry .
we conclude this section with a simple example of area renormalisation .
we consider three edges @xmath154 all labelled by the fundamental representation @xmath155 .
the representation space @xmath156 is two - dimensional and we write the two basis vector of maximal and minimal weights as @xmath157 .
let us consider the surface state defined as the pure state @xmath158 .
then the microscopic area in planck unit is obviously : @xmath159 where the numerical value is computed using the spectrum @xmath53 .
the coarse - grained areas , the one grouping the edges 1,2 together and the other grouping the edges 2,3 together are easily computed : @xmath160 @xmath161 finally the totally coarse - grained area grouping all three edges 1,2,3 together is given by : @xmath162 we can not really introduce a notion of smooth flow for such a small surface with so few edges . we would need a much larger surface , for which the state @xmath149 would inevitably be much more complicated and analytical computations more involved @xcite .
nevertheless , there is one state which can easily studied .
it is the pure state for @xmath163 edges , @xmath164 .
its renormalisation flow is practically constant .
if we partition / group the set of @xmath163 edges into packets of size @xmath165 , @xmath166 , then the corresponding coarse - grained area is : @xmath167 the microscopic area is @xmath168 while the macroscopic area is @xmath169 .
if we use the simpler spectrum @xmath170 , the renormalisation is actually exactly constant .
considering a bounded region @xmath6 of a spin network state , we are interested by the boundary state(s ) on @xmath171 induced by the quantum state of geometry of the bulk of @xmath6 ( or by the region outside the region @xmath6 ) .
we will see the boundary state can be considered as a coarse - graining of the geometry state of @xmath6 and be constructed from partial tracing of the bulk degrees of freedom .
let us consider quantum states based on a fixed given graph @xmath10 , which we assume closed and connected , and a bounded connected region @xmath6 of this spin network which we define as a set of vertices in @xmath10 .
an edge @xmath12 is considered in @xmath6 if both its source and target vertices are in @xmath6 , @xmath172 .
we define the boundary @xmath173 of the region @xmath6 as the set of edges with one vertex in @xmath6 and one vertex outside @xmath6 . for every couple of vertices
@xmath174 in @xmath6 there exists a path of edges in @xmath6 linking them . moreover , if one cuts in two all the edges in the boundary @xmath171 , then one cuts the graph @xmath10 in two connected open graphs .
we will call @xmath175 the connected open graph with all the vertices in @xmath6 and ending at the boundary @xmath171 . finally , we call @xmath176 the exterior of @xmath6 i.e the set of vertices and edges who do not belong to @xmath6 and its boundary @xmath171 . for the sake of the simplicity of the notations , we assume in the following that all the edges @xmath177 on the boundary are identically oriented outward , with @xmath178 and @xmath179
. the boundary state of @xmath6 is the state of the edges @xmath177 irrespectively of the details of the structure of the graph @xmath175 inside the region @xmath6
. we can therefore think about it as the coarse - grained state of @xmath6 once we coarse - grain the whole interior graph of @xmath6 to a single vertex .
let us thus define the graph @xmath180 $ ] where we have removes the whole graph @xmath175 from @xmath10 and replaced it by a single vertex , as in fig.[gammaa ] .
, width=302 ] we then define the following coarse - graining map .
[ def1 ] we define a coarse - graining map @xmath181}$ ] taking cylindrical functions defined on the full graph @xmath10 to cylindrical functions defined on the reduced graph @xmath180 $ ] : ( ) ( \{g_e , e } ) _
( 2)dg ( \{g_e , gg_ea , g^(0)_ea } ) .
the integration over @xmath78 is to ensure the proper gauge invariance of the resulting functional .
the map @xmath92 actually depends on the group elements @xmath182 , which need to be fixed .
we call them the coarse - graining data .
we call the trivial or canonical coarse - graining map the one defined by @xmath183 . for this canonical choice ,
the integration over @xmath0 is irrelevant and we can drop it . due to the gauge invariance of the initial functional @xmath184
, the coarse - graining data @xmath185 is actually defined up to @xmath0 actions at each vertex of @xmath6 . as a result ,
the coarse - graining data lives in the quotient space @xmath186 , where @xmath187 is the number of vertices in @xmath6 who do not have any edge on the boundary .
it is useful to look at this coarse - graining procedure in the case of a ( pure ) spin network state .
the spin network state labels each edge @xmath12 of the graph with a ( irreducible ) representation @xmath35 .
the state of the bulk of the region @xmath6 is given by the intertwiners attached to the vertices of @xmath6 , and lives in the hilbert space @xmath188 .
the state of the boundary @xmath171 lives in the tensor product @xmath189 .
we require it to be @xmath0 invariant so that the boundary hilbert space is @xmath190 .
the coarse - graining is a map from @xmath191 to @xmath192 : we glue the intertwiners inside @xmath6 together in order to obtain an intertwiner solely between the edges on the boundary of @xmath6 .
this procedure is equivalent to the construction described above and effectively reduces the whole interior graph of @xmath6 to a single vertex .
more precisely , given some fixed holonomies along the edges of @xmath6 , we can define the coarse - grained intertwiner as follow .
[ def2 ] given the coarse - graining data @xmath193 , and intertwiners @xmath194 , we define the following coarse - grained intertwiner in @xmath192 : _ a[g^(0)_ea ] _
( 2)dg _ eathe integration over @xmath0 is to impose @xmath0 invariance and obtain an intertwiner at the end of the day .
the canonical choice @xmath195 amounts to straightforwardly gluing the intertwiners . in that case , we automatically respect gauge invariance and we obtain an intertwiner without having to integrate over @xmath0 . more details on the motivation for this definition of coarse - graining from the point of view of partial tracing and the uniqueness of the procedure are given in appendix . due to the gauge invariance
, the coarse - grained intertwiner actually depends only on the orbit of @xmath193 under @xmath0 transformations : it is defined in terms of @xmath196 .
let us explain the simple example of a 4-punctured boundary @xmath197 with the internal graph being made of two 3-valent vertices and a single internal edge labelled with a fixed spin @xmath41 .
the two intertwiners attached to the two 3-valent vertices are unique and given by the usual clebsh - gordan coefficients .
coarse - graining means assigning a group element to the internal edge and integrating over an orbit under the adjoint action @xmath198 .
an orbit in @xmath199 is defined by an angle @xmath200-\pi,\pi]$ ] and is the equivalence class of the group element @xmath201 .
the corresponding coarse - grained intertwiner @xmath202 becomes ^_m_1m_2m_3m_4=_(2)dg c^j_3j_4j_m_3m_4nd^j_mn(gh_g^-1)c^j_1j_2j_m_1m_2 m .
we can evaluate exactly the integral over @xmath78 : _ ( 2)dgd^j_mn(gh_g^-1)= _ mn .
we obtain that @xmath203 always leads to the same intertwiner defined at @xmath87 up to a normalization factor : the coarse - graining procedure is trivial in this simple setting . if the clebsh - gordan coefficients are normalized , then the norm of @xmath204 is 1 . as we will see later , the reason why the coarse - graining procedure is trivial is that the topology of the interior graph ( here a single edge ) is trivial .
coarse - graining will give a non - trivial result only when the interior graph has a non - trivial topology and contains at least one loop .
we now would like to extend these intertwiner definitions from a ( pure ) spin network state to an arbitrary cylindrical function .
we need to decompose the cylindrical function onto the spin network basis , coarse - grain each spin network component separately , and then sum the coarse - grained components back to obtain the coarse - grained cylindrical function .
this is equivalent to the definition def.[def1 ] of coarse - graining , since the procedure is a linear functional only involving integrating over part of the degrees of freedom of the wave - function ( the holonomies on the edges of @xmath6 ) .
nevertheless , we ought to consider the same coarse - graining data for all the spin network components of the initial functional .
however , it might happen that it seems more suited to consider different data for each component , depending on the structure of each spin network state .
the natural issue following the previous definition is how to choose the coarse - graining data . indeed
a priori , the coarse - grained state is a mixed state reflecting all the possible coarse - grainings given by the density matrix : _
a_x_a d(x ) |_a[x]_a[x]| .
a priori the measure @xmath205 is simply the haar measure on @xmath206 .
this density matrix comes explicitly from tracing out the holonomies along the edges within the region @xmath6 .
a first remark is that this density matrix is in general not the totally mixed state @xmath207 .
therefore , our coarse - graining procedure does not erase all the information about the internal state of the region @xmath6 .
obtaining the totally mixed state by averaging over all possible ways of coarse - graining would thus require a special quantum state .
we expect this to be related to quantum black holes @xcite .
the second remark is that the coarse - grained procedure does not provide us with normalized intertwiners @xmath208||\ne1 $ ] . following the interpretation of the spin network functional as a wave function defining a probability amplitude on the holonomies ,
this norm @xmath209||^2 $ ] actually provides us with a measure of the probability on the space of possible coarse - grainings @xmath210 .
indeed , the coarse grained intertwiner @xmath211 $ ] , or equivalently the coarse grained state defined in def.[def1 ] , are defined simply by evaluating the initial spin network on the holonomies @xmath212 .
this means that , although it seems more natural to coarse - grain using the canonical choice @xmath213 , that might not be the most probable holonomies inside the region @xmath6 .
to summarize the situation , let us assume that we work with normalized spin network functionals i.e that all the intertwiners @xmath29 are normalized , @xmath214 .
now the density matrix describing the coarse - grained geometry of the region @xmath6 is : @xmath215\ra\la \ii_a[x]| = \int_{x_a } ||\ii_a[x]||^2\,d\mu(x)\ , |\wt{\ii}_a[x]\ra\la \wt{\ii}_a[x]|,\ ] ] where @xmath216 $ ] is now the _
normalized _ intertwiner describing the geometry of the region @xmath6 reduced to a single vertex .
clearly the probability distribution on this intertwiner is given by the measure @xmath209||^2\,d\mu(x)$ ] .
we can finally generalize these considerations to an arbitrary cylindrical function .
the coarse - grained state will similarly be described by the density matrix , @xmath217 and the most probable coarse - grainings will be defined as the maxima of the norm of the coarse - grained functional @xmath218 . to understand the structure and relevance ( or irrelevance ) of the coarse - graining data , it is useful to explicitly gauge - fix the initial functional @xmath219 .
this is he simplest way to see how the whole graph inside @xmath6 gets reduced to a single vertex . indeed ,
following @xcite , we choose a ( reference ) vertex @xmath220 in @xmath6 and a maximal tree @xmath146 .
@xmath146 is a set of edges in @xmath6 going through all vertices of @xmath6 without ever making any loop .
it has @xmath221 elements . if @xmath222 , this means that the interior graph @xmath175 has no loop and has a trivial topology .
@xmath223 counts the number of non - trivial loops of @xmath175 .
we know that we can use the gauge invariance to fix to the identity the group elements on the edges belonging to the tree , @xmath224 , @xmath225 to define @xmath226 and @xmath227 , we define for every vertex @xmath228 the unique path @xmath229 along @xmath146 from @xmath220 to @xmath16 .
then we define the group element : @xmath230 finally , @xmath231 and @xmath232 .
the @xmath233 can be considered as irrelevant for the coarse - graining since that they can be re - absorbed in the redefinition of the holonomies @xmath234 on the boundary of @xmath6 . on the other hand ,
the group elements @xmath235 can not be forgotten and actually represent the holonomies around each ( non - contractible ) loop of the interior graph @xmath175 .
they therefore reflect the non - trivial topologyof the quantum state of geometry of @xmath6 .
having set the maximal number of holonomies to the identity inside @xmath6 , we have effectively reduced the whole graph @xmath175 to a single vertex ( the vertex @xmath220 ) with @xmath163 open edges ( with @xmath236 is the number of edges in the boundary ) and @xmath237 loops .
the number of loops is simply @xmath238 .
see fig.[looptopetal ] for an example with two loops .
, width=226 ] contracting all the vertices of @xmath6 to a single one , and gluing all the corresponding intertwiners together putting the identity on the edges of the tree @xmath146 , lead a single intertwiner @xmath239 .
here we write @xmath20 to refer to which ever representation @xmath240 lives on the considered edge ( or loop ) .
the region @xmath6 is then described by the contraction of this single intertwiner with the holonomies @xmath235 along the loops . labelling the loops by @xmath241 and noting the corresponding holonomies @xmath242 , the functional reads : _ _ 1 .. _na_1b_1 .. a_l_ab_l_a_i^l_ad_a_ib_i(g_i ) .
coarse - graining this structure amounts to reduce the whole graph to a single vertex without any loop . from this point of view , it is clear that a non - trivial coarse - graining of the region @xmath6 corresponds to a non - trivial topology of the graph @xmath175 inside the region @xmath6 . the relevant coarse - graining data is an element of the group @xmath243 i.e one holonomy for each loopin @xmath6 .
when @xmath244 , there is no loop in @xmath6 and we are left with the trivial coarse - graining setting all holonomies on the edges in @xmath6 to the identity .
coarse - graining thus amounts to erase any non - trivial topology of the graph of the geometry state of @xmath6 .
we coarse - grain by averaging over the loop holonomies with an arbitrary function @xmath245 : _
i dg_i ( g_1, .. ,g_l_a ) _ _ 1 .. _na_1b_1 .. a_l_ab_l_a_i^l_ad_a_ib_i(g_i ) .
we require that the result after integration is still an intertwiner @xmath246 , i.e invariant under ( the diagonal action of ) @xmath0 ( on @xmath247 ) .
this imposes that @xmath248 be invariant under the adjoint action of @xmath0 , @xmath249 this means that @xmath250 is a spin network functional on the flower graph with @xmath237 petals -see fig.[looptopetal ] for an example with a two - petal flower .
now , one could look for coarse - graining data which maximizes the ( norm of the ) coarse - grained intertwiner , either directly as ( an orbit ) @xmath251 , or equivalently as a @xmath198-invariant functional @xmath250 .
if we do not consider @xmath198-invariant coarse - graining data , we will not obtain an intertwiner as result of the coarse - graining , but more generally a vector in the tensor product @xmath252 .
an alternative to imposing by hand to obtain an intertwiner after coarse - graining would to look for maxima among all coarse - graining data @xmath253 , possibly non - invariant .
if the most probable holonomies along the loops are actually the identity ( or at least very close to it ) , then we get simply set them to @xmath254 in the coarse - graining procedure and erase the non - trivial topology of the graph .
if the probability is actually peaked around non - trivial values of the holonomies , this could mean that the non - trivial topology of the graph truly reflect a non - trivial topology of the region @xmath6 and that we should not neglect the contribution / couplings of the loops to the intertwiner .
in that case , our coarse - graining procedure , which erases the non - trivial topology , might not reflect the true geometry of the region @xmath6 . at the end of the day
, we have seen how the coarse - graining procedure can be understood as erasing " the non - trivial topology of the region @xmath6 ( more precisely , the topology of the graph underlying the quantum state of geometry of the region @xmath6 ) through integrating over the degrees of freedom attached to the internal edges of the region @xmath6 . considering a pure spin network state @xmath114 and given a region @xmath6 of that spin network and an assignment @xmath255 of groups elements to each internal edge ,
we have seen to define a boundary state @xmath256 .
since we are studying a spin network state based on a closed graph , the exterior of @xmath6 defines a ( complementary ) region of the spin network with the same boundary . defining the quotient space @xmath257 of all possible coarse - grainings of the exterior of @xmath6 , each element @xmath258 induces a boundary state @xmath259 , which is a priori arbitrarily different from the states @xmath260 obtained by coarse - graining the interior of @xmath6 .
let us point out that @xmath261 is actually in @xmath262 ( because of the reverse orientation of the edges ) , which is nevertheless isomorphic to @xmath192 since we work with @xmath0 . at the intuitive level , one can see the boundary state @xmath260 as the state of @xmath171 induced by the ( quantum ) metric inside @xmath6 while @xmath261 is the boundary state induced by the outside ( quantum ) metric
. then we can look for transition functions ( or transition amplitudes ) allowing to go from @xmath260 to @xmath261 .
more precisely , we would like to identify a function @xmath263 such that : _
eadg_e _ x , y(\{g_e , ea } ) _
ead^j_e(g_e)_a^(x ) .
this makes sense since @xmath264 lives in the tensor product @xmath265 . actually to be rigorous
, we should write @xmath266 . such a transition amplitude @xmath267 would reflect the curvature at the level of the surface @xmath171 , between the inside and the outside of @xmath6 .
indeed it describes the parallel transport when crossing the surface @xmath173 . a solution is actually given by the spin network functional itself @xmath268 $ ] .
indeed a straightforward calculation gives that a suitable function for @xmath269 and @xmath270 is defined as : _ x , y(\{g_e , ea } ) ( g_ea , g_ea , g_e ) .
since the spin network functional is actually the wave - function defining the probability amplitude for the connection and holonomies , this result fits with the intuition that the transition function " @xmath271 defines the probability amplitude for the curvature at the level of the surface @xmath171 .
we have seen previously that the coarse - graining of a region of a spin network is non - trivial only when the topology of the graph is non - trivial .
we propose to look at this from the perspective of the dimension of the hilbert space and see how the number of degrees of freedom get coarse - grained .
more precisely , we investigate the dimensionality of the hilbert space of bulk states once the structure of the boundary is fixed . considering the bounded region @xmath6 of the spin network ,
we look at the topology of the open graph @xmath175 inside @xmath6 .
if the graph @xmath175 is a single vertex , then the hilbert space is simply the space of intertwiners between the representations @xmath27 attached to the edges on the boundary @xmath177 : @xmath272 if @xmath175 is more complicated and consists of many vertices and internal edges , a basis of the hilbert space is given by a choice of representation labels for the internal edges @xmath273 and of ( orthogonal ) intertwiner states for the internal vertices .
this existence of intertwiners at the vertices imposes some restrictions on the possible representation labels ( such as the triangular inequalities for trivalent vertices ) .
when @xmath175 has a trivial topology , i.e has no loops , then the hilbert space is actually still isomorphic to the same intertwiner space : @xmath274 using a tree decomposition of a vertex is actually a standard way of constructing a basis of the intertwiner space .
as soon as @xmath175 has a non - trivial topology , i.e contains loops , the hilbert space is a priori infinite - dimensional .
indeed , one can attach to each loop @xmath275 with an un - constrained representation label @xmath276 .
this is similar to the freedom of associating an arbitrary holonomy along each loop in the previous coarse - graining procedure .
since @xmath276 is an arbitrary integer , the resulting hilbert space has an infinite dimension .
thus one needs to gauge fix this freedom and fix the representation @xmath276 -the flux around the loop- for each loop .
then the hilbert space becomes finite - dimensional but still has a larger dimension than the intertwiner space .
let us look at this in details in a simple example .
we consider a region @xmath6 whose boundary comprises of @xmath277 edges labelled with the fundamental representation @xmath156 .
it is rather straightforward to show that ( see @xcite for example ) : ( v^1/2)^2n= _ k=0^n ( c^n+k_2n - c^n+k+1_2n ) v^k , where the @xmath94 s are the binomial coefficients .
the dimension of the intertwiner space is the degeneracy of the spin-0 space : ^(2n)_trivial= c^n_2n - c^n+1_2n = c^n_2n . ,
width=207 ] let us now look at the one - loop case .
we consider a loop to which the @xmath277 edges are attached , as shown in fig.[oneloop ] , so that all vertices are 3-valent .
since all vertices are 3-valent , there is no freedom in the choice of intertwiners .
all the freedom resides in the choice of representation label for the ( internal ) edges on the loop .
let us consider one of these edge .
we are free to label it with an arbitrary representation @xmath278 .
indeed , one can then find consistent labels on the remaining internal edges , for example alternating @xmath279 on the @xmath277 edges on the loop . since @xmath41 has no constraint ,
we conclude that the one - loop hilbert space has a priori an infinite dimension . however , we can gauge fix this infinity . indeed , starting with a given consistent labelling @xmath280 of the edges of the loop ,
the freedom is actually to add a spin @xmath281to all the labels .
then the labelling @xmath282 provides a new consistent labelling and thus a new orthogonal spin network . to gauge fix this freedom , we need to fix the spin @xmath35 on a _ single _ edge on the loop ( we do _ not _ fix the labels on all the edges on the loop
, that would be too much ) .
this action by arbitrary shift of the spins @xmath283 along the loop is actually equivalent to the action on the spin network state of the holonomy operator along that loop . indeed acting with the operator multiplication by the character @xmath284 of
the @xmath281-representation amounts to tensor all the representations @xmath27 along the loop by the representation @xmath285 .
therefore , we are gauge fixing the action of the ( gauge invariant ) holonomy operators on loops inside the region @xmath6 . since the action of the hamiltonian constraints is intimately intertwined to the holonomy operators ( see for example @xcite ) , at the end of the day , our gauge fixing could be related to gauge fixing the hamiltonian and therefore counting the number of physical degrees of freedom .
actually , we already know that the gauge fixings of the holonomy operators and of the hamiltonian constraints are closely related for a topological bf - type theory .
however , we expect the situation to be more subtle in 3 + 1 gravity . back to our dimension counting , we fix the spin to @xmath41 on a given edge on the loop and we know count the exact number of consistent labelling of the other internal edges .
we notice that from one edge to the next , we can only move from a representation @xmath281 to @xmath286 . then along the loop , we start at @xmath41 and we move by @xmath287 steps to finally come back to @xmath41 .
thus the dimension of the one - loop hilbert space is given by the number of returns ( to 0 ) of a random walk after @xmath277 steps : ^(2n)_1-loop= c^n_2n .
this actually works only if @xmath288 , else the random walk could possibly reach the trivial representation @xmath289 .
since the representation labels are positive , the dimension of the hilbert space would be smaller .
another possible one - loop configuration is a single vertex with the @xmath277 edges sticking out and one loop ( or petal ) going from the vertex to itself , as shown on fig.[oneloop ] .
it is clear that the loop can carry whatever representation @xmath278 we want . once again , to avoid an infinite number of degrees of freedom , we fix the spin @xmath41 on the loop .
we then need to count the number of intertwiners @xmath290 .
taking into account that @xmath291 , the dimension of the ( extended ) intertwiner space is straightforward to compute : ^(2n)_1-loop= _ k=0^(2j , n ) ( c^n+k_2n - c^n+k+1_2n ) = c^n_2n - c^n+(2j , n)+1_2n , which is always larger than @xmath292 .
as soon as @xmath293 , we recover the previous result , @xmath294 .
we can then compare the number of degrees of freedom for a trivial graph topology to the one - loop case : @xmath295 we further expect the number of degrees of freedom to increase with the number of loops on the graph inside the considered region @xmath6 .
these extra degrees of freedom are erased as we coarse - grain the region @xmath6 and fix the holonomies around the non - trivial loops of the graph to finally reduce the whole region @xmath6 to a single vertex .
let us start by recalling definition of correlations and entanglement between two quantum systems @xcite .
[ defentan ] let us consider a possibly mixed state on @xmath296 defined by the density matrix @xmath149 , @xmath297 , @xmath298 , @xmath299 .
we define the reduced density matrices obtained by partial tracing : @xmath300 the entropy of the state @xmath149 is @xmath301=-\operatorname{tr}\rho\log\rho$ ] .
then we define the _ quantum mutual information _ as : @xmath302+s[\rho_b]-s[\rho].\ ] ] note that when we have a pure state @xmath303 , then the entropies of reduced density matrices are equal , @xmath304=s[\rho_b]$ ] . for a pure state @xmath303
, we define the _ entanglement _ between @xmath6 and @xmath305 as the entropy of the reduced density matrices : ( a|b,|| ) s[_a]=s[_b ] .
note that since @xmath301=0 $ ] , we have @xmath306 . for a generic mixed state @xmath149 , we define the entanglement ( of formation ) as the minimal entanglement over all possible pure state decompositions of @xmath149 .
more precisely , we diagonalize @xmath307 .
this decomposition is not necessarily unique and we define : ( a|b , ) _
\{|_i } _ i _ i ( a|b,|_i_i| ) .
the minimization is necessary in order to obtain a meaningful result .
other measures of entanglement are used in the literature and reflect different aspects of entanglement manipulations . the quantum mutual information @xmath308 quantifies the total amount of ( classical and quantum ) correlations between @xmath6 and @xmath305 .
the entanglement @xmath309 defines a measure of purely quantum correlations between the two systems @xmath6 and @xmath305 .
one can quantify the classical amount of correlations as the difference @xmath310 .
let us consider a quantum state @xmath76 defined on the ( closed and connected ) graph @xmath10 , and two small and @xmath305 is small compared to the total number of vertices of the graph .
] regions @xmath6 and @xmath305 of that graph .
let us first assume that @xmath6 and @xmath305 are totally disjoint , i.e. , that there is no edge directly linking a vertex of @xmath6 to a vertex of @xmath305 .
we are interested in the correlations between these two subsystems .
we could consider the correlations between the intertwiners and edges in @xmath6 and the ones in @xmath305 .
however , there is no reason that they be correlated .
for instance , they are totally uncorrelated on a pure spin network state .
it is actually more interesting to look at the correlations between @xmath6 and @xmath305 induced by the spin network state outside @xmath6 and @xmath305 .
intuitively , we would be looking at the correlations between @xmath6 and @xmath305 due to the space metric ( outside @xmath6 and @xmath305 ) .
such a notion of correlation should be related to the concept of proximity / distance between @xmath6 and @xmath305 in the geometry defined by the spin network state .
we call @xmath94 the region exterior to @xmath6 and @xmath305 .
its boundary is the union of the boundaries of @xmath6 and @xmath305 , @xmath311 ( with opposite orientation ) .
we define the mixed state of the regions @xmath312 induced by the spin network state in the @xmath94 region through a coarse - graining of @xmath94 : @xmath313}\})\ , \overline{\theta\psi}(\{g'_{e\in \gamma[c]}\}),\ ] ] where we recall that @xmath314=a\cup b \cup \pp a \cup \pp b$ ] . the coarse - graining map @xmath92 was defined in the definition [ def1 ] . the integration over @xmath315 averages over the coarse - graining data .
we integrate using the haar measure , but we could integrate using a different measure on the space of holonomies ( on the edges of @xmath94 ) up to gauge transformations . if we were to select a single orbit , i.e. a single choice of coarse - graining data @xmath316 , we would work with the pure coarse - grained state @xmath317\,\ra$ ] .
explicitly , the state reads : @xmath318 where the integration over @xmath78 ensures gauge invariance on the closed boundary @xmath319 .
this state is generally a mixed state which correlates / entangles the regions @xmath6 and @xmath305 .
indeed , a generic state @xmath76 would very unlikely lead to a simple tensor product state @xmath320 .
the important point is that from the point of view of @xmath94 and @xmath321 , the two regions @xmath6 and @xmath305 are behind the same boundary @xmath321 and nothing says that they are indeed separated .
if we were given a particular state @xmath76 , we could now compute the correlation / entanglement between @xmath6 and @xmath305 .
then we could define a notion of distance such that the two regions were close or far away from each other depending whether the correlation in @xmath320 was strong or weak .
let us see how this works on a pure spin network state @xmath322 .
the state of the region @xmath94 is defined by the intertwiners outside @xmath6 and @xmath305 , @xmath323 .
we coarse - grain the whole region @xmath94 to obtain the boundary state on @xmath324 . for a given set of holonomies @xmath316 on the edges of @xmath94 , the resulting intertwiner on @xmath319 is : _
( 2)dg _ ec .
the coarse - grained state is then the mixed state given by the density matrix : _
ec dg^(0)_ec || , where we integrate using the haar measure
. we could integrate using a different measure on the space of holonomies ( on the edges of @xmath94 ) up to gauge transformations . if we were to select a single orbit , i.e a single choice of coarse - graining data @xmath316 , we would work with the pure state @xmath325\,\ra$ ] .
this boundary state lives in the space of intertwiners on @xmath319 , @xmath326 .
if we were looking at the boundary states of the two separated regions @xmath6 and @xmath305 , we would naturally write a state in @xmath327 .
the key point here is that the space of intertwiners on @xmath328 is actually _ not _ the tensor product of the space of intertwiners on @xmath173 with the space of intertwiners on @xmath329 , but is strictly larger : @xmath330 intuitively , one could say that the outside metric described by the state of the region @xmath94 leads to correlations between the two regions @xmath6 and @xmath305 . for instance , it is the metric in the outside region @xmath94 which defines the relative position of @xmath6 and @xmath305 and not the metric inside the regions @xmath6 and @xmath305 themselves .
mathematically , @xmath331 can be decomposed as the direct sum : [ intstructure ] ( _ ec v_j_e ) & = & _ j ( v_j_ea v_j_e ) ( v_j_eb v_j_e ) + & & _ j ^j_a^j_b _ j ^j_ab .
this decomposition can be easily described pictorially , as shown in fig.[jlink ] , with a fictitious new edge linking @xmath6 and @xmath305 and carrying the representation @xmath41 .
this corresponds to unfolding the intertwiner in @xmath332 on a graph with two vertices corresponding to @xmath171 and @xmath329 but with an edge linking these two vertices .
this edge / link is required by the @xmath0 invariance of the original intertwiner .
, width=302 ] the hilbert space @xmath333 only corresponds to the subspace @xmath334 with the internal link labelled by the trivial representation @xmath44 . in general
, the boundary state @xmath335 $ ] will not belong to that particular subspace , but will expand on all possible @xmath41 s .
in @xcite , we studied the case where the state on @xmath319 was the totally mixed state , the density matrix @xmath336 being proportional to the identity @xmath254 on @xmath337 . from these results
, we expect the correlations between @xmath6 and @xmath305 to heavily depend on the dominant value of the internal link @xmath41 .
more precisely , we expect the correlations to increase with @xmath41 .
we will study the precise behavior in the next section .
we would like to stress once more that we are looking at the correlations and entanglement between @xmath6 and @xmath305 as induced by the outside metric i.e the complementary region @xmath94 to @xmath338 in the graph @xmath10 .
indeed , depending on the values of its coefficients , the spin network state @xmath339 may or may not entangled the regions @xmath6 and @xmath305 i.e be entangled across the hilbert space @xmath340 .
this is not the correlation / entanglement which we are investigating : we are looking at the correlations between @xmath6 and @xmath305 induced by the intertwiners @xmath341 .
we expect these latter correlations to be related to the distance between the regions @xmath6 and @xmath305 .
this will be discussed in more details in the final section .
this provides a concrete basis for the ideas underlined in @xcite that two regions of the spin network close to each other in some given embedding of the spin network graph into @xmath342 are not necessarily close in the real physical space .
reversely , two regions of the spin network far to each other in some given embedding could be strongly correlated and entangled , which would indicate that that given embedding does not reflect the structure of the true physical space .
we can similarly consider two regions @xmath6 and @xmath305 which are linked by a single edge @xmath343 labelled by a representation @xmath344 .
that edge will be inside the region @xmath338 .
therefore , the value of the spin @xmath344 will not enter the computation of the global state @xmath345\,\ra$ ] : the spin @xmath41 of the internal / fictitious link between @xmath6 and @xmath305 is completely independent of the spin @xmath344 .
intuitively , one could say that the link @xmath344 reflects the correlations between @xmath6 and @xmath305 as seen from the inside geometry of @xmath338 , while the link @xmath41 describes the correlations between @xmath6 and @xmath305 as induced by the outside geometry ( as seen by an observer outside @xmath6 and @xmath305 ) .
these considerations can be generalized to the situation where @xmath6 and @xmath305 are directly linked through several edges .
as we have shown above , all correlation and entanglement calculations within a spin network can be reduced to the following simple set - up .
we consider the two systems @xmath6 and @xmath305 .
the hilbert space attached to @xmath6 is the tensor product of the representation spaces @xmath346 , technically corresponding to all edges in the boundary of the region @xmath6 .
we similarly define the hilbert space attached to the system @xmath305 : @xmath347 our state is an intertwiner in the full tensor product @xmath348 i.e a singlet or spin-0 state in @xmath349 . more precisely
, we can decompose the tensor products in direct sums of irreducible representations : @xmath350 for both systems @xmath351 .
the space @xmath352 is the degeneracy space of states with spin @xmath281 .
we label the basis vectors of @xmath353 as @xmath354 , with @xmath355 running from @xmath356 to @xmath357 .
similarly we introduce a basis @xmath358 of the degeneracy space @xmath359 .
this decomposition provides us directly with a basis of the intertwiner space @xmath360 .
indeed , we have @xmath361 spin-0 states are necessarily states with the spins @xmath281 and @xmath362 of the systems @xmath6 and @xmath305 are equal : @xmath363 .
therefore , we get : ^0_ab ( _ ia , bv^j_i)= _ j ( d^a_jd^b_j ) . a basis of the intertwiner space is thus given by the vectors @xmath364 .
the representation label @xmath41 is the same one as in the previous section and is attached to the fictitous / internal link between the regions @xmath6 and @xmath305 .
it carries the correlation and entanglement between @xmath6 and @xmath305 .
explicitly , these basis states read : @xmath365 it was shown in @xcite that for a state of the following type , @xmath366 the entanglement between @xmath6 and @xmath305 is : _ ( a|b)=_j ( 2j+1)__j , _ j^(j)__j _
j. in words , we compute the average of @xmath367 over the probability distribution of the spin @xmath41 in the state @xmath149 . in particular ,
@xcite considered the totally mixed state , @xmath368 with @xmath369 .
for this special state , the entanglement takes the simple form : @xmath370 that is the average of @xmath367 over the intertwiner space @xmath371 .
let us now analyze more carefully the case where the spin @xmath41 takes a single fixed value .
then states of the previous type , @xmath372 are states that do not carry any entanglement across the degeneracy spaces . all the entanglement is contained in the spin @xmath41 . indeed applying the previous formula
, we find exactly : @xmath373 if we look allow for arbitrary pure states possibly entangling the degeneracy spaces , @xmath374 then the entanglement is obviously bounded by : @xmath375 where the extra term @xmath376 accounts for the contribution from the degeneracy spaces .
finally , we allow for arbitrary pure states with possible superpositions of different values of the spin @xmath41 , @xmath377 the entanglement is then bounded by _
( a|b)d , d=_j ( 2j+1)d_j .
it is easy to see that @xmath378 , where @xmath379 .
the definition of the entanglement ensures that in all cases the maximal possible entanglement is contained in pure states , so that the bounds derived above hold for arbitrary statistical mixtures .
all these bounds can actually be saturated by pure states .
moreover , for higher dimension ( of the degeneracy spaces ) , random pure states are in average nearly maximally entangled @xcite .
our goal is to understand the ( quantum ) metric defined by a spin network state , without referring to any assumed embedding of the spin network in a ( background ) manifold .
we support the basic proposal that a natural notion of distance between two vertices ( or more generally two regions ) of that spin network is provided by the correlations between the two vertices induced by the algebraic structure of the spin network state .
two parts of the spin network would be close if they are strongly correlated and would get far from each other as the correlations weaken .
our set - up is as follows .
we consider two ( small ) regions , @xmath6 and @xmath305 , of the spin network .
the distance between them should be given by the ( quantum ) metric outside these two regions .
thus we define the correlations ( and entanglement ) between @xmath6 and @xmath305 induced by the rest of the spin network .
this should be naturally related to the ( geodesic ) distance between @xmath6 and @xmath305 .
a first inspiration is quantum field theory on a fixed background .
considering a ( scalar ) field @xmath380 for example , the correlation @xmath381 between two points @xmath382 and @xmath383 in the vacuum state depends ( only ) on the distance @xmath384 and actually decreases as @xmath385 in the flat four - dimensional minkowski space - time . reversing the logic
, one could measure the correlation @xmath381 between the value of a certain field @xmath380 at two different space - time points and define the distance in term of that correlation .
indeed just as the correlations in qft contain all the information about the theory and describes the dynamics of the matter degrees of freedom , we expect in a quantum gravity theory that the correlations contained in a quantum state to fully describe the geometry of the quantum space - time defined by that state .
another inspiration is the study of spin systems , in condensed matter physics and quantum information @xcite .
such spin systems are very close mathematically and physically to the spin networks of lqg .
the key difference is that spin systems are physical systems embedded in a fixed given background metric ( usually the flat one ) while spin networks are supposed to define that background metric themselves .
looking at spin systems , we notice that the ( total ) correlation between two spins usually obey a power law with respect to the distance between these two spins .
we would like to inverse this relation in the context of lqg : using a similar power law , we could reconstruct the distance between two points " in term of a well - defined correlation between the corresponding parts of the spin network .
the entanglement is trickier to deal with .
the entanglement between two spins usually decreases with the distance , but it would ( almost ) vanish beyond the few nearest neighbors . this is due to _ monogamy of entanglement _ @xcite : if two systems are strongly entangled , then none of of them can be strongly correlated with any other system basis and ignored the normalization factors .
this is one of the inequivalent maximally entangled states of three qubits @xcite , but it is easy to check that no two subsystems are entangled ( even if they are maximally correlated in the @xmath386 basis ) . on the other hand , but the two other pairs are entangled . ] .
interest in the role of entanglement in quantum phase transitions @xcite , which are characterized by non - analyticity in various quantities at certain values of externally - varied parameters at zero temperature , was one of the driving forces behind the investigations of entanglement in many - body systems .
the most popular models are hamiltonian lattice spin systems in one , two and three spatial dimensions .
recently it was shown in various models that quantum phase transitions of both first and second kinds are signalled by a non - analyticity of the entanglement measures @xcite .
while the states considered are most typically the ground or thermal states of the lattice hamiltonian operator , some of the results may be usefully applied to the attempts to reconstruct geometry from the spin networks .
those states typically are multipartite entangled states .
one of the results that are relevant to us is that the two - qubit entanglement measures vanish very rapidly with the distance : they may become zero at the distance of four or five sites , and sometimes even the third - nearest neighbors entanglement may be zero . at the same time , the correlations may be strong , and at the critical point the correlation length may actually diverge . on the other hand ,
the overall state is typically entangled , so it is the multipartite entanglement @xcite that becomes dominant .
thus , it is likely that the entanglement between two parts of the spin networks could only be a good notion of distance when the two systems are very close .
anyway , we only expect the entanglement on a spin network to be of the same order as the classical correlation for a pure spin network state with few vertices .
we expect that as soon as we need to coarse - grain over large number of vertices and edges or if we work on superpositions of spin network states ( _ weave states _ ) , then the classical correlation would be much larger than the bipartite entanglement .
an obvious criticism to the proposal of reconstructing the distance from correlations and entanglement is the existence of epr pairs : two systems far from each other can still be strongly entangled .
let us first remind that typical states in quantum field theories , including the vacuum state , are entangled @xcite .
any two points would be naturally entangled in the vacuum state with the correlation depending on the distance .
an epr pair would correspond to the creation an anomalously strong entanglement between two given ( space ) points . nevertheless , is that we are building a statistical notion of distance , which would only hold in average and in a semi - classical setting .
a small quantum effect like an epr pair would be averaged out .
a second comment is that epr pairs concern correlations between matter degrees of freedom evolving in an already defined geometry while we are interested of correlations between the degrees of freedom of the ( quantum ) geometry .
these correlations are supposed to fully describe the geometry at the quantum level .
the equivalent of an epr pair in our context would be that the geometry at two points apparently far from each other could be strongly entangled / correlated .
this would actually look as if there was a wormhole connecting these two points .
this is very close to the idea of the creation of non - local links " in a spin network state as proposed by markopoulou and smolin @xcite , which could explain some features of dark matter in lqg . indeed , taking a spin network based on a regular 3d lattice , one can create a
non - local " link , labelled with some representation @xmath41 , between two points far from each other on the lattice . there is no argument based on locality forbidding the creation of such a link since the spin network is defined without any reference of a background metric : the regular lattice is simply a particular embedding of the spin network into flat space and there exists other embeddings for which the two chosen points are arbitrary close to each other .
nevertheless , this non - local " link will create some strong entanglement / correlation ( especially as @xmath41 grows ) between the two points despite their distance on the lattice : there will effectively be close as if related by a wormhole .
finally , one can argue that the creation of entanglement is always local : two systems are entangled if and only if they were close at some point in time .
therefore , the correlation / entanglement is actually still related to a notion of distance , although it seems more likely to reflect the possibility of being close ( or having been close ) .
finally , our close / correlated " point of view have some simple implications for lqg , which could be ( easily ) checked through analytical or numerical computations .
a first lesson is that a low spin on an edge between two vertices does not necessarily mean that these two vertices are close , since low spins tend to lead to small correlations .
reversely , having no direct link between two vertices does not necessarily mean that these vertices are far , since the coarse - graining of the rest of the spin network will create a link between these two vertices which may carry strong correlations .
the basic set - up for the simplest explicit calculations consists of two vertices @xmath6 and @xmath305 linked by a single edge ( our internal / fictitious link ) .
we fix the spins on all the edges coming out of @xmath6 and @xmath305 and we also fix the internal edge label to a given spin @xmath41 . the state of the vertex @xmath6 is defined as an intertwiner state in @xmath387 while the vertex @xmath305 is described by an intertwiner in @xmath388 . following section [ calculation ] , we can compute the entanglement and total correlation between @xmath6 and @xmath305 .
they tend to grow as the spin @xmath41 increases , but the exact value heavily depends on the choice of intertwiners . then to reconstruct the distance between @xmath6 and @xmath305 , it might be more relevant to compute the ratio of the actual correlation with the maximal correlation between @xmath6 and @xmath305 ( to normalize and get rid of possible @xmath389 factors ) . on the other hand , we could compute the average volume corresponding to the intertwiner states in @xmath6 and @xmath305 . then , since the spin @xmath41 is the area of the cross - section between the two boundary surfaces dual to the vertices @xmath6 and @xmath305 , we should be able to reconstruct a measure of distance from a certain ratio of @xmath390 by the area @xmath391 . at the end of the day
, we could compare this two notions of distance , hoping a close match .
we could start by checking a first expectation : assuming that the @xmath35 s are more or less all equal , the vertices @xmath6 and @xmath305 would be close if @xmath41 is large while they get further away from each other as @xmath41 decreases .
moreover , if the spins @xmath35 are large compared to @xmath41 , we expect the vertices to be far .
this should hold for both notions of distance , the information distance " constructed from the correlation ( correlations grow with @xmath41 ) and the geometric distance " constructed from the volumes and areas ( the cross - section @xmath41 should grow as the vertices get closer ) .
a close relationship between these information distance and geometric distance would strongly support the idea that quantum information tools should be useful to understand and probe the quantum structure of the space(-time ) geometry in ( loop ) quantum gravity .
in this appendix we show how the coarse - graining relates to the partial tracing and fill the missing details that were alluded to in the different places above .
first let us recall the standard definition of a reduced density operator @xcite . to simplify the notations , we consider two subsystems @xmath6 and @xmath305 , with hilbert spaces @xmath191 and @xmath392 , respectively .
let @xmath393 be a wave function on @xmath191 , @xmath394 label its basis , and @xmath395 label the basis of @xmath392 .
the combined system has a hilbert space @xmath296 , with the direct product basis @xmath396 . a generic state @xmath397 is given by a linear combination @xmath398 .
if the matrix elements of the operator @xmath399 are ab|o|ab=_a , a,b , bo_bb_aa,[mateld ] then its expectation value is given by |o|=_a , b , bo_bbc_ab|c_ab=(o_^b),[expred ] where the reduced density operator @xmath400 is obtained by tracing out the subsystem @xmath6 , @xmath401 .
in the coordinate basis the operator @xmath399 is given by o(x , y;x,y)=o(y , y)(x - x),[matelc ] so thanks to the orthonormality of the functions @xmath402 the reduced density operator is ^b_(y , y)=dx dx(x , y)(x,y)(x - x)=c_abc_ab_b(y)|_b(y),[rhoc ] and |o|= dy dyo(y , y ) ^b_(y , y)[contexp ] . in the language of the intertwiners a partial tracing does not present any new features . taking an intertwiner @xmath403 and tracing out @xmath6 produces a basis state of @xmath404 , @xmath405 , with a corresponding expressions for the reduced density matrices of general states .
expectation values of the operators that depend only on the region @xmath305 are calculated according to eq .
( [ expred ] ) .
for example , the volume operator @xcite is a sum of vertex operators , with each vertex operator being a function of the self - adjoint right - invariant vector fields .
the isomorphism between the spin network states and zero - angular momentum states of the corresponding abstract spin system leads to the form of its matrix elements that is analogous to eq .
( [ mateld ] ) and an expectation value of the form of eq .
( [ expred ] ) . in particular ,
if @xmath305 contains a single vertex @xmath406 and the operator s matrix elements are @xmath407 , then for a generic pure state @xmath408 coarse - grained intertwiners have an obvious relation with partially traced ones .
the total coarse - graining of @xmath6 is given by _
a=_va_e(a ) d^j_e()_va_v_a , where @xmath409 denotes the summation over pairs of indices pertaining to the vertices in the region @xmath6 .
it turns a normalized basis state @xmath410 into a non - normalized state ( see sec .
[ probcoarse ] for a discussion ) @xmath411}=\hh_{\pp a}\otimes\hh^0_b$ ] .
it can be decomposed as |_a=_c^_a_|_a^ , where the intertwiners @xmath412 , @xmath413 , form the orthonormal basis of @xmath192 . a general coarse - graining procedure of eq .
( [ def2 ] ) leads to the same equation , but with the coefficients depending on the holonomies @xmath414 .
a normalized pure state @xmath408 becomes a normalized pure state |_[a]=__a,_b , s__a_bc^_a_|_a^|_b/ = _ s_|_a|_b_a_b^0 .
cylindrical functions do not allow a natural separation of variables in @xmath415 that is analogous to eq .
( [ matelc ] ) , since there is no relation between the number of the intertwiners @xmath29 , @xmath416 , that define the tensor product structure of @xmath417 , and the number of edges , that define the structure of @xmath418 . to obtain
a wave functional that corresponds to @xmath419 one needs to contract with the representations @xmath420 that corresponds to all the edges of @xmath180 $ ] .
hence the coarse - graining procedures that were described above allow to introduce the reduced subsystems in the language of cylindrical functions . from the definition 1 a total coarse - graining of a spin network state @xmath421 over a region
@xmath6 results in @xmath422}(g)\equiv\6 g|\tilde{\ii}_a\vio_b\9 $ ] , where @xmath423 .
it is given explicitly by _
( g)= _ e(a)d^j_e(g_e),=_va _ e(a ) d^j_e()_v_v/=_a_b .
consider for simplicity a pure state @xmath424 .
mixed states are treated analogously .
the following lemma establishes that the coarse - grained states @xmath422}(g)$ ] play a role of the basis wave functions @xmath425 in the standard partial tracing .
a coordinate " expression of the reduced density matrix @xmath426 is given by _
^b(g , g)= _ , s__a_b__a_b(dg)^e_a(dg)^e_a_e(a)(g_e , g_e)_(g)_(g)/_a=_,s__a_b__a_b_(g)_(g ) , [ defred ] where @xmath427 , and @xmath428 .
we give a proof for @xmath404 containing a single vertex @xmath406 .
what is needed to be shown is that the expectation value of the vertex operator @xmath429 is indeed given by the formula analogous to eq .
( [ rhoc ] ) and that that eq . ( [ defred ] ) holds .
the total coarse - graining over @xmath430 leads to @xmath431}(g)$]with = _
e(a ) d^j_e ( ) _ va _ v _ a. the functional matrix elements of @xmath407 are o_1(g , g)g| |o_1| |g= _ , _ 1g|_1_2
_v o(_1,_1)_1_2 _v|g , and the totally coarse - grained over @xmath6 matrix elements can be defined as o_1^b(g , g)= _ , _
1_(g ) o(_1,_1)_(g ) , where @xmath432 . hence to get the expectation value @xmath433 of eq .
( [ correct ] ) the reduced density matrix should be defined as _
^b(g , g)=_,s__a_b__a_b_(g)_(g),[goal ] and this expression equals the first formula on the lhs of eq .
( [ defred ] ) due to eq .
( [ matnorm ] ) .
the states @xmath434 are the eigenstates of the area operators @xcite with the eigenvalues that depend only on the representation labels @xmath35 of the edges that are cut by the surface .
hence if the region @xmath6 is defined as an interior of some closed surface @xmath435 , the above definition of the reduced operators is consistent with the expectation values of this operator that are taken on a full state . c. rovelli , _ loop quantum gravity _ , ( cambridge university press , cambridge , 2004 ) ; + t. thiemann , _ lectures on loop quantum gravity _ , lect .
notes phys .
* 631 * 41 - 135 ( 2003 ) , gr - qc/0210094 ; + a. ashtekar and j. lewandowski , _ background independent quantum gravity : a status report _ ,
* 21 * r53 ( 2004 ) , gr - qc/0404018 .
a. perez , _ spin foam models for quantum gravity _ , class .
. grav . * 20 * r43 ( 2003 ) , gr - qc/0301113 ; + l. freidel , _ group field theory : an overview _ , int .
* 44 * ( 2005 ) 1769 , hep - th/0505016 .
livine and d.r .
terno , _ quantum black holes : entropy and entanglement on the horizon _ , gr - qc/0508085 ; + e.r .
livine and d.r .
terno , _ entanglement of zero angular momentum mixtures and black hole entropy _ ,
a * 72 * , 022307 ( 2005 ) , quant - ph/0502043 .
d. bruss , _ entanglement splitting of pure bipartite quantum states _
a * 60 * , 4344 ( 1999 ) , quant - ph/9902023 ; + m. koashi and a. winter , _ monogamy of entanglement and other correlations _
a * 69 * , 022309 ( 2004 ) , quant - ph/0310037 .
a. osterloh , l. amico , g. falci , and r. fazio , _ scaling of entanglement close to a quantum phase transition _ , nature * 416 * , 608 ( 2002 ) ; + t. j. osborne and m. a. nielsen , _ entanglement in a simple quantum phase transition _ , phys .
rev a * 66 * , 032110 ( 2002 ) , quant - ph/0202162 .
f. verstraete , m. popp , and j. i. cirac , _ entanglement versus correlations in spin systems _ , phys .
92 * , 027901 ( 2004 ) , quant - ph/0307009 ; + m. popp , f. verstraete , m. a. martin - delgado , and j. i. cirac , _ localizable entanglement _ , phys .
a * 71 * , 042306 f. markopoulou and l. smolin , _ the problem of non - locality in quantum gravity _ , in preparation ; + f. markopoulou , i. premont - schwarz and l. smolin , _ macroscopic deviations from newtonian gravity from quantum gravity _ , in preparation
. s. sachdev,_quantum phase transitions _
( cambridge university press , cambridge 1999 ) .
w. dr , g. vidal , and j. i. cirac , _ three qubits can be entangled in two inequivalent ways _ , phys .
a * 62 * , 062314 ( 2000 ) , quant - ph/0005115 .
t. roscilde , p. verrucchi , a. fubini , s. haas , and v. tognetti , _ studying quantum spin systems through entanglement estimators _ , phys .
rev . lett .
* 93 * , 167203 ( 2004 ) , cond - mat/0404403 .
|
abstract after a brief review of spin networks and their interpretation as wave functions for the ( space ) geometry , we discuss the renormalisation of the area operator in loop quantum gravity .
in such a background independent framework , we propose to probe the structure of a surface through the analysis of the coarse - graining and renormalisation flow(s ) of its area .
we further introduce a procedure to coarse - grain spin network states and we quantitatively study the decrease in the number of degrees of freedom during this process .
finally , we use these coarse - graining tools to define the correlation and entanglement between parts of a spin network and discuss their potential interpretation as a natural measure of distance in such a state of quantum geometry .
|
more than half of all main sequence stars , and an even larger fraction of pre - main sequence stars , are in binary / multiple star systems ( duquennoy and mayor 1991 ; mathieu et al .
virtually all previous models of planet formation , however , have assumed an isolated single star . of the first 131 extrasolar planet systems that have been confirmed , at least 30 are on so - called s - type orbits that encircle one component of a binary star system , including at least 3 that orbit one member of a triple - star system ( raghavan et al .
the effect of the stellar companion on the formation of these planets , however , remains unclear .
one planet has been detected in a p - type orbit which encircles both members of a binary star system .
this planet , which has a minimum mass of @xmath9 2.5 times the mass of jupiter ( m@xmath10 ) , orbits @xmath9 23 au from the center of mass of psr 1620 - 26 , a radio pulsar binary comprised of a neutron star and a white dwarf in a @xmath9 191 day stellar orbit ( lyne et al .
1988 , sigurdsson 1993 , sigurdsson et al .
the most plausible model for its formation is accretion within a metal - rich disk produced by post - main sequence roche lobe overflow ( lissauer 2004 ) .
planets have not been detected in a p - type orbit around two main sequence stars , but short - period binaries are not included in precise doppler radial velocity search programs because of their complex and varying spectra .
planets in p - type orbits around the eclipsing binary star system cm draconis have been searched for using the eclipse timing variation method ( deeg et al .
2000 ) , but results were not definitive .
two substellar companions have been detected around the g6v star hd 202206 , with minimum masses of 17.4 m@xmath10 ( at 0.83 au ) and 2.44 m@xmath10 ( at 2.55 au ) ( udry et al .
the inner companion is so massive that it is considered to be a brown dwarf , and it is likely that the outer companion formed from within a circumbinary ( star - brown dwarf ) disk ( correia et al . 2005 ) . a more general discussion of the detectability of circumbinary planets is presented by muterspaugh ( 2005 ) . note also that the observation of two small moons orbiting in nearly circular / planar orbits about pluto - charon ( weaver et al .
2006 ) , a system which is like a binary with an 8/1 mass ratio , suggests that accretion can occur in p - type orbits about close binaries .
the main objective of this article is to numerically examine the late stages of terrestrial planet formation around both members of a binary star system .
the existence of earth - like planets in orbit about one or both components of main sequence binary stars has yet to be determined , though ground- and space - based efforts to search for extrasolar terrestrial planets are currently in development .
an additional benefit of understanding the differences between planet formation around single stars and that around close binaries is that for eclipsing binaries , the contrast ratio between brightness of the stars and that of the planet(s ) is reduced during the eclipse . for a total eclipse of identical stars , this reduction is a factor of two ; as lower mass main sequence stars can be just slightly smaller but significantly less luminous , the detectability of the planet can be enhanced by more than a factor of two when the fainter star transits the brighter one . in an evolved close
binary having undergone mass transfer , the fainter star can actually completely eclipse its much brighter companion , leading to an even larger improvement in planetary detectability . in the conventional model of planet formation
, terrestrial planets are believed to have formed by an accretion process from within a disk of gas and dust that has remained around a newly formed star ( safronov 1969 , lissauer 1993 ) .
the coexistence of disks of material with stars that possess a stellar companion support the idea that planet formation within binary star systems may be common .
circumbinary disk material has been detected through millimeter and mid - infrared excess emission around several spectroscopic pre - main sequence binary star systems with stellar semimajor axes @xmath6 @xmath8 1 au .
these systems include gw ori ( mathieu et al .
1995 ) , uz tau e ( jensen et al . 1996 ) , and dq tau ( mathieu et al .
the masses of these disks are each comparable to or exceed the minimum mass of the solar nebula , @xmath9 0.01 solar mass ( m@xmath2 ) ( weidenshilling 1977 ) , and are also comparable to the masses of disks found around single stars
. numerical models of circumbinary disks find that , for binary star systems with binary eccentricities ( @xmath7 ) increasing from 0 0.25 , the inner edge of a gaseous disk is truncated to within @xmath9 1.8 2.6 times the semimajor axis of the binary stars mutual orbit ( @xmath6 ) ( artymowicz and lubow 1994 , lubow and artymowicz 2000 ) .
a star with a giant planet orbiting interior to the terrestrial planet region is dynamically a binary system of extreme mass ratio .
raymond et al .
( 2005 ) showed that planetary embryos can accrete into terrestrial planets around a star that has a close - in ( between 0.15 au 0.5 au ) jupiter - mass planet .
herein , we simulate terrestrial planetary accretion within a circumbinary disk of protoplanets around ` close ' ( @xmath6 = 0.05 0.4 au ) binary star systems that each have a combined stellar mass of 1 m@xmath2 . our numerical method and
the initial states of the systems that we have simulated are given in section 2 .
section 3 examines the regions of stability for test particles orbiting about these binary star systems .
the results of the close binary accretion simulations , including a quantitative analysis of the final planetary systems formed , are presented in section 4 .
our conclusions are discussed in section 5 .
appendix a presents new simulations of the late stages of terrestrial planet formation in the sun - jupiter - saturn system that we have performed to facilitate comparisons between planet growth around single and close binary stars , and simulations using an initial disk of bodies whose eccentricities are forced by the binary stars are presented in appendix b. in appendix c , we discuss the scaling of our results to systems with different planetesimal densities , disk sizes , and stellar masses .
the combined mass of the binary stars is equal to 1 m@xmath11 in all of the simulations , with the stellar mass ratio @xmath12 ( the ratio of the secondary star s mass to the total stellar mass ) equal to either 0.2 or 0.5 .
binary star separations in the range @xmath6 = 0.05 au 0.4 au are examined , while @xmath7 begins at 0 , 1/3 , 0.5 , or 0.8 such that the stellar apastron @xmath13 is 0.05 au @xmath14 0.4 au .
not all combinations of these stellar parameters , however , are used . for most of the simulations ,
the midplane of the circumbinary disk begins coplanar to the stellar orbit , but for one set of binary star parameters a relative inclination of @xmath15 = 30@xmath16 is investigated .
although a stellar companion present during the earlier stages of planet formation would likely force the planetesimal disk into the plane of the binary orbit , many binary stars may originate as unstable triple star systems which could produce a binary star system with an accretion disk at a high relative inclination .
it is also possible that a companion may have been captured around a single star that posesses an accretion disk .
giant planets are included in the simulations , as they are in most simulations of the late stages of terrestrial planet accumulation in our solar system ( chambers 2001 , appendix a ) . in all of the planetary accretion simulations presented herein ,
a jupiter - mass planet is placed in the system at @xmath17 = 5.2 au , with an eccentricity of @xmath18 = 0.048 and an inclination of @xmath19 = 0.36@xmath16 relative to the midplane of the disk .
apart from a single set of runs in which the stars are separated by @xmath6 = 0.05 au and travel on an initially circular orbit coplanar with the midplane of the disk , a second giant planet of saturn s mass , with @xmath20 = 9.54 au , @xmath21 = 0.053 , and @xmath22 = 0.89@xmath16 relative to the midplane of the disk , is also included .
the effect of the stellar pertubations on these giant planets is discussed in section 4.2 .
the initial conditions for the bodies in the circumbinary disk are based upon earlier numerical simulations of the late stages of terrestrial planet formation in the sun - jupiter - saturn ( sjs ) system ( chambers 2001 ) which resulted in the formation of planetary systems with masses and orbits similar to the terrestrial planets in the solar system ( see appendix a ) . in this model , 14 planetary embryos ( each with a mass of 0.0933 times the mass of the earth , m@xmath23 ) are embedded in a disk of 140 smaller planetesimals ( each with a mass of 0.00933 m@xmath23 ) , all lying between 0.36 au and 2.05 au of the center of mass of the binary stars .
the radii of these bodies are calculated assuming a material density of 3 g @xmath24 .
all other initial orbital elements were chosen at random , with the eccentricities ranging from 0 0.01 , and inclinations relative to the mean plane of the disk @xmath25 0.5@xmath16 .
analyses of trajectories around close binaries show that closed , near circular , non - crossing orbits occur in situations such as this ( pichardo et al . 2005 ,
see especially their figures 6 and 9 ) . in appendix b
, we present one set of simulations that was run using a disk of planetesimals and embryos with initial eccentricities determined by perturbations from the binary . in the majority of our simulations
, the initial planetesimal / embryo disk extends closer to the stars than the region in which planetesimals are expected to be able to form within a _ gas - free _ disk ( moriwaki and nakagawa 2004 ) .
however , calculations show that planetesimal growth occurs over a much greater region within _ gas - rich _ circumstellar ( s - type orbits ) disks than within analogous regions of gas - free disks ( kortenkamp and wetherill 2000 , thebault et al .
2006 ) , and we would expect an analogous situation for p - type orbital regions . moreover , it is also possible that at least the initial phases of planetesimal growth can occur farther from the stars , and the planetesimals can then migrate inwards as a consequence of gas drag . to examine both the orbital stability and accretion of bodies in a disk
, we use one of the two symplectic @xmath26-body algorithms that we developed to examine planetary accretion in binary star systems ( chambers et al .
the ` close binary ' algorithm ( used in the simulations presented in this article ) was designed to examine accretion in p - type orbits around binary stars .
the ` wide binary ' algorithm , which follows bodies in s - type orbits within binary star systems , was recently used to simulate the late stages of terrestrial planet formation around each star of the nearest binary star system to the sun , @xmath3 centauri ab ( quintana et al . 2002 , quintana 2003 ) . in these simulations , which began with an initial disk mass distribution virtually identical to the disk discussed above , from 3 5 terrestrial planets formed on stable orbits around each individual component in @xmath3 cen ab
provided the initial inclination of the disk relative to the stellar orbit began at or below @xmath9 30@xmath16 .
numerous simulations of terrestrial planet formation in s - type orbits in main sequence binary star systems , with the aim of examining a larger binary star parameter space , will be presented in quintana et al .
( 2007 ) .
the close binary algorithm calculates the temporal evolution of the position and velocity of each body in the disk with respect to the center of mass of the binary stars , subject to gravitational perturbations from both stars and to gravitational interactions and completely inelastic collisions among the bodies .
bodies are removed if their orbit extends more than 100 au from the more massive star , or if they orbit too close to the center of mass of the binary stars . for
selected simulations with larger stellar separations , material is removed if its distance from the center of mass exceeds the smaller star s apastron distance by less than 0.1 au . a time - step of 7 days is used for the bodies in the disk , while the binary stars are given a time - step that is shorter by approximately the ratio of the binary period to the orbital period of the innermost planetesimal ( see chambers et al .
2002 for details ) .
the period of the innermost body is 55.09 days , so the choice of a 7 day time - step may not accurately follow the evolution of the innermost bodies , which could lead to errors , especially concerning the amount of mass lost inside .
we are comparing our results , however , to the simulations in the sun - jupiter - saturn system ( chambers 2001 ) and to simulations around each star in @xmath3 cen ab , each of which used a 7-day time - step , so the differences that we find should be real .
, @xmath7 , @xmath12 ) equal to ( 0.05 , 0 , 0.5 ) and ( 0.1 , 0.8 , 0.5 ) , from 25 50 test particles were placed around each system at 0.01 au intervals , beginning at the distance from the center of mass of the stars at which bodies are removed . for the binary star systems with ( @xmath6 , @xmath7 , @xmath12 ) equal to ( 0.2 , 0.5 , 0.5 ) , ( 0.3 , 1/3 , 0.5 ) , and ( 0.4 , 0 , 0.5 ) , the smallest semimajor axis for which bodies can be stable was determined , @xmath27 ( described further in section 3 ) , and test particles were placed between ( @xmath27 0.25 au ) and ( @xmath27 + 0.25 au ) , with 0.01 au intervals .
all other orbital elements were chosen at random , with eccentricities @xmath28 0.01 and inclinations @xmath29 0.01 radian , but were kept the same for each set of simulations .
the giant planets were not included and the test particle orbits were followed for 10 myr . for the binary system with @xmath6 = 0.05 au that began on an initially circular orbit , the orbital elements of the surviving test particles were nearly identical .
particles in other systems , especially close to the stability limit , showed larger variations .
the differences in results between the runs with 3.5 day time - steps and 3.6 day time - steps , however , were almost as large as the differences from the 7 day time - step runs , indicating that the dominant source of variation was chaos rather than systematic inaccuracy of the integrations
. moreover , in many of our simulations the innermost planetesimal survived the entire integration intact , with little variation in its principal orbital elements ( as was the case for some of our simulations of planetary growth ; see , for example , fig .
we therefore conclude that the use of a 7 day time - step does not significantly degrade the statistical validity of our integrations . ] because these @xmath26-body systems are chaotic , each binary star system under study is simulated five or six times with slightly different initial conditions for the circumbinary disk .
of the 154 rocky embryos and smaller planetesimals , one planetesimal near 1 au is initially displaced by 1 , 2 , 3 , 4 , or 5 meters along its orbit with all other parameters within a given set remaining the same .
the simulations are labelled as follows : cb_@xmath6_@xmath7_@xmath12_@xmath30 , where @xmath6 is the binary semimajor axis , @xmath7 is the binary eccentricity , @xmath12 is the stellar mass ratio , and @xmath30 (= a , b , c , d , e , or f ) signifies each realization of a given system . if the midplane of the initial planetesimal disk is inclined relative to that of the binary star orbital plane , the inclination angle is listed between @xmath12 and @xmath30 , but when they are coplanar ( as for most of our simulations ) , no value is given .
the first run in each set ( the `` a '' integrations ) begins with the ` standard ' bi - modal circumbinary disk mass distribution . in the remaining runs within a set ,
a planetesimal near 1 au is initially displaced by 1 meter along its orbit ( for the `` b '' integrations ) , by 2 meters along its orbit ( for the `` c '' integrations ) , etc . the evolution of the material in the disk is initially followed for 200 myr .
if it appeared that further collisions among the planets were fairly likely , individual integrations were continued for total simulation times of 500 myr ( or 550 myr in one case ) or 1 gyr . in simulations that resulted in the formation of just one planet
, the integrations were stopped after the last collision or ejection .
the results of each accretion simulation are presented in table 2 and discussed in section 4 .
the dynamical stability of test particles in p - type orbits has been previously examined for binary star systems with @xmath12 ranging from 0.1 0.5 and binary eccentricities @xmath7 between 0.0 0.7 ( holman and wiegert 1999 ) . in that study , test particles were placed in the binary orbital plane between 1.0 @xmath6 and 5.0 @xmath6 ( with 0.1 @xmath6 increments ) at eight equally spaced longitudes per semimajor axis , and the system was evolved forward in time for 10@xmath31 binary periods . the closest distance to the binary at which all eight particles survived ( the ` critical semimajor axis ' , @xmath27 ) was calculated for each system .
depending on @xmath7 and @xmath12 , @xmath27 was found to lie within 2.0 @xmath6
4.1 @xmath6 .
many of these simulations also revealed an unstable region beyond @xmath27 that corresponded to one of the system s @xmath32:1 mean motion resonances , followed by an additional outer region of stability .
we performed a similar analysis for the close binary configurations examined herein , with the integration time extending to 10@xmath33 binary orbital periods .
test particles were placed between 1.8 @xmath6 and 5.0 @xmath6 ( with 0.1 @xmath6 increments ) at eight equally spaced longitudes .
integrations were performed using the close binary algorithm .
particles were removed from the simulation if they fell within 0.01 @xmath6 from the more massive star or if their distance exceeded 100 @xmath6 . for each system ,
table 1 gives the smallest semimajor axis at which bodies at all 8 longitudes survive the integration , @xmath34 . in the close binary systems with @xmath7 = 1/3
, regions of instability were found beyond @xmath34 , and the minimum semimajor axis at and beyond which all test particles survive is listed in brackets in table 1 .
analogous integrations were performed with finer increments of 0.01 @xmath6 to find the innermost semimajor axis for which at least one body survived , given by @xmath35 in table 1 .
these results are consistent with those of holman and wiegert ( 1999 ) , who found a ( roughly ) linear dependence of the critial semimajor axis on @xmath7 .
note that for a given @xmath36 1/3 , the region cleared of test particles is greater for @xmath12 = 0.2 than it is for @xmath12 = 0.5 , presumably because the apastron distance of the smaller star from the center of mass of the system is larger for @xmath12 = 0.2 .
for each close binary accretion simulation , the ratio of the semimajor axis of the innermost final planet ( @xmath37 ) to the innermost stable orbit of the system ( @xmath35 ) is presented in table 2 .
figure 1 displays the accretion evolution of system cb_.05_0_.5@xmath38_a ( the dagger signifies that only one gas giant , a jupiter - mass planet at 5.2 au , is included ) .
the eccentricity of each body is shown as a function of semimajor axis , and the radius of each symbol is proportional to the radius of the body that it represents . throughout the simulation ,
the larger embryos remain on orbits with @xmath39 @xmath8 0.1 , whereas the planetesimals become more dynamically excited with time during the first @xmath9 10 20 myr . between @xmath9 5 50 myr
, a trend occurs in which planetesimal eccentricities increase with increasing semimajor axis as a consequence of perturbations by jupiter .
all but one of the planetesimals are either swept up by the larger embryos or are ejected from the system .
the first planetesimal ejection occurs at @xmath9 12 myr , while the only lost embryo was ejected at @xmath9 37 myr .
no bodies in this system traveled closer to the binary orbit than the initial semimajor axis of the innermost planetesimal ; indeed , the innermost planetesimal was the one that survived without impact .
after 500 myr elapsed , 6 planets with masses between 0.11 0.61 m@xmath23 orbited within 2.3 au ; these planets incorporated 84% of the initial disk mass . despite the apparent crowding , the system appears to be quite stable ; no bodies are lost or accreted between 92.7 myr and the end of the simulation .
figure 2 ( cb_.05_0_.5@xmath38_d ) shows the growth of planets formed around a binary star system identical to that shown in fig . 1 , but
in this case the initial disk mass is slightly different ( one planetesimal near 1 au is shifted by 3 meters along its orbit ) .
note that the evolution of the disk in this figure ( and subsequent figures ) is shown beginning at @xmath40 = 0.2 myr , as the plotted properties of the disk at @xmath40 = 0 are identical to those shown in the first panel in fig .
the stellar and giant planet perturbations have a similar effect on the disk as in the simulation shown in fig .
1 . in this case
, three terrestrial - mass planets have formed within 1.5 au , while one planetesimal remains at 2.2 au , all together incorporating 81% of the initial mass in the disk .
the differences in the final planets formed from two simulations with almost identical initial conditions ( fig . 1 vs. fig .
2 ) demonstrate the chaotic nature of these @xmath26-body systems .
figure 3 shows run cb_.05_0_.5_c , which included a saturn - like giant planet in addition to the jupiter - like one .
nonetheless , the outcome looks intermediate between the systems shown in fig . 1 and fig
a comparison between all of the cb_.05_0_.5@xmath38 runs with the cb_.05_0_.5 runs ( table 2 and fig .
7 ) suggests that the extra perturbations of `` saturn '' may reduce the expected number of terrestrial planets formed , but the effects are small enough that there is considerable overlap among the chaos - broadened distributions of outcomes .
note that these distributions also overlap the results of simulations of terrestrial planet growth around a single star with one or two giant planets ( chambers 2001 , chambers et al .
2002 , appendix a ) and around the individual stars in the @xmath3 centauri ab binary , provided the disk begins close to the @xmath3 cen binary orbital plane ( quintana et al .
2002 , quintana 2003 ) . figure 4 displays the evolution of cb_.075_.33_.5_b , in which the stars orbit one another on more distant and eccentric paths .
the system evolves very differently from any of the systems with @xmath6 = 0.05 and @xmath7 = 0 .
the innermost large planet is quite eccentric by @xmath40 = 20 myr , and it accretes or scatters all of the smaller bodies inwards of 1 au prior to being ejected itself at 106 myr .
run cb_.075_.33_.5_d produces a similar planetary system , whereas runs cb_.075_.33_.5_a , cb_.075_.33_.5_c , and cb_.075_.33_.5_e result in planetary systems resembling those formed in the @xmath6 = 0.05 and @xmath7 = 0 simulations ( table 2 and fig . 8) . in a sense , this change in stellar parameters yields systematic differences comparable to the scatter resulting from chaos .
figures 5 and 6 show the evolution of systems cb_.1_.8_.2_d and cb_.4_0_.5_c , respectively . in each case , the binary apastron distance is much larger than in the runs discussed above , and binary perturbations clear the system of all but one planet .
additional planets remain in the runs with nearly identical initial parameters , but the systems ( table 2 , figs . 7 and 8)
still look much sparcer , especially in the inner regions , than those formed around a single star , very close binaries , or individual stars in the @xmath3 cen ab system . in these cases ,
systematic effects resulting from the different binary parameters exceed typical chaotic variations .
figures showing the temporal evolution of most of the simulations discussed in this paper are presented in quintana ( 2004 ) ; plots of simulations cb_.1_0_.5_c and cb_.2_.5_.5_a are presented in lissauer et al .
( 2004 ) .
figures 7 and 8 show the final planetary systems formed in all of our simulations .
the top left row in each figure shows the solar system s terrestrial planets ( labelled ` mvem ' ) , followed by the 5 6 realizations of each binary system under study , presented in order of increasing @xmath4 .
figure 7 presents the final planetary systems formed around binary stars that began on circular orbits with the disk initially coplanar to the stellar orbit ( labelled in the following format : @xmath6_@xmath7_@xmath12 ) , whereas fig .
8 shows the results for simulations with @xmath15 = 30@xmath16 and the sets of runs with @xmath41 . in these figures ,
the radius of each body is proportional to the radius of the planet that it represents , the horizontal lines through each body indicate the periastron and apastron distances to the center of mass of the binaries ( or the sun in the mvem case ) , the vertical lines represent the inclination relative to the binary orbital plane ( the laplacian plane in the mvem case ) , and the arrows show the orientation of the final spin axes of each planet ( arrows are omitted for planetesimals and embryos that survived the integration without a collision ) .
although the final planetary systems formed vary widely among a given set of binary star parameters due to the chaotic nature of these simulations , general trends are apparent in the planets formed around stars with larger separations and higher eccentricities . in order to quantitatively analyze these effects , we developed a set of formulae that characterize the orbits and distribution of mass for all of the final planetary systems .
these are described in the next subsection and the statistical variations are discussed in subsection 4.2 .
the results of all of our close binary simulations are given in table 2 , which lists the stellar parameters / initial conditions and gives the values of statistical parameters that were developed to help characterize the final planetary systems .
most of these statistics were previously used to compare the outcomes from accretion simulations in the sun - jupiter - saturn ( sjs ) system ( chambers 2001 , appendix a ) and around each star in the @xmath3 cen ab binary star system ( quintana et al .
2002 , quintana 2003 ) , all of which used essentially the same initial planetesimal disk .
the first column lists the name of each close binary simulation ( cb_@xmath6_@xmath7_@xmath12_@xmath30 , as described in section 2 ) .
simulations that include just one giant planet ( jupiter ) are labelled with @xmath38 ; all others include a jupiter - like planet and a saturn - like planet .
when the initial midplane of the circumbinary disk is inclined relative to the stellar orbit , the runs are denoted cb_@xmath6_@xmath7_@xmath12_@xmath15_@xmath30 . the duration of each simulation is listed in column 2 .
columns 3 15 present the following statistics ( see chambers 2001 and quintana et al .
2002 for mathematical descriptions of the statistics given in columns 7 15 ; statistics presented in columns 5 and 6 are new in this work ) .
( 3 ) : : the number of planets , @xmath42 , that are at least as massive as the planet mercury ( @xmath9 0.06 m@xmath23 ) . note that the 14 planetary embryos in the initial disk each satisfy this mass requirement , as do bodies consisting of at least 7 planetesimals .
( 4 ) : : the number of minor planets , @xmath43 , that are less massive than the planet mercury . ( 5 ) : : the ratio of the semimajor axis of the innermost final planet to the closest stable orbit of the system , @xmath37/@xmath44 .
note that in principle , the value of this quantity may be ( slightly ) less than unity , as @xmath44 was estimated using a coarse grid and only considered bodies initially on circular orbits within the binary plane .
( 6 ) : : the ratio of the periastron of the innermost final planet , @xmath45 = @xmath37(1 @xmath46 ) , to the binary apastron @xmath4 = @xmath6(1 + @xmath7 ) .
( 7 ) : : the fraction of ( the final ) mass in the largest planet , @xmath47 .
( 8) : : an orbital spacing statistic , @xmath48 , which gives a measure of the distances between the orbits of the final planets ( that are larger than the planet mercury ) .
larger values of @xmath48 imply more widely spaced final planets .
( 9 ) : : the normalized angular momentum deficit , @xmath49 , which measures the fractional difference between the planets actual orbital angular momenta and the angular momenta that they would have on circular , uninclined orbits with the same semimajor axes .
( 10 ) : : a mass concentration statistic , @xmath50 , which measures the degree to which mass is concentrated in one part of the planetary system . ( 11 ) : : a radial mixing statistic , @xmath51 , which sums the radial migrations of the bodies that form a planet .
( 12 ) : : the percentage of the initial mass that was lost from the planetary system ( came too close to the stars or was ejected to interstellar space ) , @xmath52 .
( 13 ) : : the total mechanical ( kinetic + potential ) energy per unit mass for the planets remaining at the end of a simulation , @xmath53 , normalized by @xmath54 , the energy per unit mass of the system prior to the integration .
( 14 ) : : the angular momentum per unit mass of the final planets , @xmath55 , normalized by @xmath56 , the angular momentum per unit mass of the initial system .
( 15 ) : : the @xmath57 component of angular momentum per unit mass relative to the stellar orbit , @xmath58 , normalized by @xmath59 , the initial @xmath57 component of angular momentum of the system .
following the close binary results in table 2 are analogous statistics for the following systems : the four terrestrial planets in the solar system ( labelled ` mvem ' ) ; the averaged values for 31 accretion simulations in the sun - jupiter - saturn system ( ` sjs@xmath60 ' , which are presented in appendix a ) ; the averaged values of a set of accretion simulations around the sun with neither giant planets nor a stellar companion perturbing the system ( ` sun@xmath60 ' , quintana et al . 2002 , appendix a ) ; the averaged values for the planets formed within 2 au of the sun when neither giant planets nor a stellar companion is included ( ` sun@xmath60 ( @xmath61 2 au ) ' , quintana et al .
2002 , appendix a ) ; and the averaged values for the planetary systems formed around @xmath3 cen a in simulations for which the accretion disk began with an inclination @xmath29 30@xmath16 to the @xmath3 cen ab binary orbital plane ( labelled ` @xmath3 cen ( @xmath29 30@xmath16 ) ' , quintana et al .
note that only the six statistics listed for the terrestrial planets in our solar system ( mvem ) are actual observables .
we use a two - sided kolmogorov - smirnov ( k - s ) test to compare each planetary statistic for each binary star configuration to the analogous planetary statistic from the distribution of 31 sjs simulations ( appendix a ) .
table 2 gives the k - s statistic , d , and the associated probability , p , for each set of simulations .
generally , values of p @xmath8 0.05 indicate that the two sets of data are drawn from different distribution functions , @xmath62 , the effect of the binary stars on the disk is statistically significant .
the orbital spacing statistic , @xmath48 ( column 8 in table 2 ) , is undefined in simulations which resulted in the formation of a single planet ( @xmath42 = 1 , including those systems with @xmath43 @xmath63 0 since these smaller bodies are neglected for this calculation ) .
simulations with @xmath42 = 1 and @xmath43 = 0 have an infinite value for the mass concentration statistic , @xmath50 ( column 7 in table 2 ) . in each case , when calculating d and p , the values for these statistics are replaced by the highest finite value of the statistic within that set of runs in order to minimize biasing of the results .
a discussion of the results from table 2 are presented in the next subsection .
nearly all of the simulations that began with binary stars with @xmath64 0.2 au resulted in distributions of planetary systems that are statistically consistent in most properties with those formed in the sjs simulations .
the mass loss , final specific energy , and final specific angular momentum statistics for the close binary simulations ( columns 12 15 in table 2 ) , however , differ from the corresponding sjs distributions , and will be discussed later in this section .
one set of simulations ( the cb_.1_0_.5 runs ) in which 4 5 terrestrial - mass planets formed ( compared to an average of 3 planets formed in the sjs runs ) have planetary statistics that are inconsistent with the sjs distributions , even though the final planets have masses and orbits that appear upon inspection to be similar to the terrestrial planets in our solar system .
this divergence is possible because the statistical tests are of marginal use for comparing ensembles with only five members .
we include the statistics because they provide a different , albeit not necessarily better , perspective from a visual comparison of the final systems shown in figures 7 and 8 . neglecting to include
a saturn - like planet in addition to the jupiter - like planet in the simulations did not affect the final outcomes of the planets in a statistically significant manner .
the first two sets of runs listed in table 2 ( @xmath6 = 0.05 ) show that the effects of chaos are larger than the effects from the number of giant planets that are included .
nonetheless , it is worth noting that the most crowded final system , run cb_.05_0_.5@xmath38_a , in which 6 planets and 1 planetesimal survived , was subject to the perturbations of only a single giant planet .
one surprising result is that one run ended with 6 terrestrial planets and five runs concluded with 5 planets . in all of these runs , the binary stars mutual orbit was circular with @xmath6 = 0.05 au , 0.1 au or 0.2 au .
in contrast , at most 4 planets remained in the single star simulations listed in table 3 . within each @xmath65 of close binary simulations , at least one planet more massive than the planet mercury remained in an orbit farther from the center of mass of the system than the present orbit of mars ( @xmath9 1.5 au ) . in many of the close binary simulations , mars - sized planets formed ( and/or planetesimals remained ) in orbits beyond @xmath9 2 au from the center of mass of the binary stars . in our solar system and in the sjs simulations ,
the location of the @xmath66 secular resonance restricts terrestrial planet formation to within @xmath9 2 au of the sun .
in contrast , orbital precession induced by the binary displaces secular resonances of the giant planets away from the terrestrial planet zone .
we integrated test particles orbiting 1.5 2.5 au from a single or close binary star that had jupiter and saturn - like planetary companions .
the eccentricities excited in the test particles orbiting the single star reached substantially higher values than did those of test particles around the binary , consistent with expectations .
the variations in the orbital elements of jupiter and saturn are larger in most of the close binary simulations compared to simulations with a single star . in the sjs set , the average peak to trough variations of jupiter s semimajor axis , eccentricity , and inclination are @xmath67 0.006 au , @xmath68 0.038 , and @xmath69 0.32@xmath16 , respectively . for saturn , these variations are @xmath70 0.077 au , @xmath71 0.086 , and @xmath72 0.785@xmath16 . in the close binary simulations , the average peak to trough variations in semimajor axes and eccentricities for both jupiter and saturn are slightly larger ( although the changes are more chaotic systems with increasing @xmath4 ) , with @xmath73 0.01 au , @xmath68 0.06 , @xmath74 0.14 au , and @xmath71 0.11 . the variations in inclination increase with increasing @xmath4 for both jupiter ( up to 2.3@xmath16 ) and saturn ( up to 3.5@xmath16 ) . in the majority of simulations that began with binary stars on circular orbits and @xmath75 0.2 au , the terrestrial planet systems that formed span essentially the entire range of the initial disk .
the stellar perturbations on the inner edge of the disk become apparent in simulations with @xmath76 0.3 au and in many of the simulations with smaller @xmath4 but @xmath77 0 ( fig . 7 and fig .
8) . the statistics in columns 5 and 6 of table 2 give a measure of how close the innermost planet forms to the innermost stable orbit of the system , @xmath37/@xmath44 , and the ratio of the closest approach of the planet to the binary apastron , @xmath45/@xmath78 .
these statistics do not have analogs in the sjs system , but are useful in comparing simulations with differing binary star parameters .
note that for most final systems , @xmath45/@xmath78 @xmath9 2 @xmath37/@xmath44 . in the first two sets of simulations ( @xmath6 = 0.05 au )
, the stars have a minimal effect on the inner edge of the disk , which begins at more than 3.5 times the distance ( from the center of mass of the binary stars ) of the location of the innermost stable orbit of the system , @xmath44 . in more than one - third of the simulations with @xmath79 0.1 au ( all of which use @xmath80 1/3 and have @xmath44 @xmath8 0.2 au )
, the innermost planetesimal in the initial disk survived the entire integration without a collision and remained close to its initial orbit at @xmath9 0.35 au ( fig . 7 and 8) . in eight of the ten simulations with @xmath6
= 0.2 and @xmath7 = 0 , the innermost embryo survived the integration without a collision .
binary systems with @xmath76 0.18 au have larger values of @xmath44 ( 0.32 au @xmath81 @xmath44 @xmath81 0.9 au ) that approach or exceed the initial orbit of the innermost body in the protoplanetary disk .
although the inner edge of the disk is truncated , the innermost planets formed in many of these simulations remain on orbits that are close to the system s @xmath44 . in the sjs
set of simulations and in the cb_.05_0_.5@xmath38 runs , an average of @xmath9 51% of the mass that composes the final planets remained in the largest planet formed , the same fraction of mass that composes the earth in the mercury - venus - earth - mars system .
the value of @xmath47 ( the fraction of final mass that composes the largest planet ) is generally higher in the close binary simulations that resulted in fewer planets . with the exception of two sets of runs ( the cb_.1_0_.5 set with an average value of @xmath47 = 0.35 , and the cb_.15_1/3_.5 set with an average of @xmath47 = 0.64 )
, all of the sets of simulations with @xmath79 0.2 au have values of @xmath47 that are statistically consistent with those in the sjs set of runs ( typically @xmath9 0.5 ) . the value of @xmath47 tends to be larger for planets that form around more eccentric binary stars , yet the effect of varying the stellar mass ratio for systems with common @xmath6 and @xmath7 appears to be statistically insignificant .
the statistic @xmath48 , which quantifies the orbital spacing of the final planets , ranged from 29 87 ( with an average of 44 ) in the sjs distribution , and is @xmath48 = 38 for the terrestrial planets in the solar system .
the cb_.1_0_.5 simulations resulted in a higher number of planets that were more closely spaced , with @xmath48 = 32 on average .
all other sets of close binary simulations with @xmath79 0.2 au resulted in planetary systems with @xmath48 values that are consistent with the @xmath48 distribution of the sjs set .
note that one - fourth of the simulations of binary stars with @xmath76 0.3 au produced only a single terrestrial planet , @xmath42 = 1 , in which case @xmath48 is omitted .
if we consider only sjs and close binary simulations which resulted in @xmath82 3 ( since two planet systems generally give high values of this statistic ) , then the sjs distribution has an average value of @xmath48 = 38 , the same as the solar system s terrestrial planets . comparing this limited distribution of @xmath48 to close binary simulations ( again omitting 2 planet systems ) with @xmath79 0.2 au ( nearly all simulations with larger @xmath4 resulted in 1 and 2 planet systems )
, we again find that only the cb_.1_0_.5 set results in a statistically different ( lower ) value of @xmath48 than the sjs @xmath82 3 distribution . the angular momentum deficit ( @xmath49 ) , which measures the orbital excitation of a planetary system , is an order of magnitude larger for the set of planet systems that formed in the sjs simulations ( @xmath49 = 0.02 , on average ) than for the solar system terrestrial planets ( @xmath49 = 0.002 ) .
most of the planetary systems that formed around close binaries with @xmath79 0.3 au ( in simulations in which the midplane of the disk began coplanar to the stellar orbit ) have values of @xmath49 that are on average comparable to the sjs distribution .
the exceptions are the two sets of simulations with @xmath6 = 0.1 and @xmath7 = 0 , both of which have smaller values of @xmath49 ( but even these sets have higher angular momentum deficit than do the actual terrestrial planets in our solar system ) .
the relatively high values of final planetary eccentricities and inclinations are a well known difficulty of models of the late stages of terrestrial planet growth within our solar system ( agnor et al . 1999 ,
chambers 2001 ) .
the radial distribution of mass in the final planetary systems is measured herein by the mass concentration statistic , @xmath50 . in the solar system , @xmath9 90% of the mass of the terrestrial planets
is concentrated in venus and in earth ( @xmath50 = 90 ) . with the exception of the simulations which resulted in a single planet ( for which this value is infinite ) ,
@xmath50 ranged from 21 81 ( with an average of @xmath50 = 40 ) in the sjs set of runs . in all of the close binary simulations that ended with @xmath82 3 , @xmath50 ranged from 25 85 . for most of the two - planet systems
, @xmath50 had much higher values , especially in systems with @xmath83 0.2 au .
the degree of radial mixing , @xmath51 , which sums the radial migrations of the bodies that form a final planet , is not known for our solar system , but has an average value of @xmath51 = 0.42 for the sjs planetary systems .
the values of @xmath51 are more widely varied among each close binary set . as @xmath4
is increased and more mass is lost , the degree of radial mixing is reduced , and the sets of simulations with @xmath84 0.2 au have smaller values of @xmath51 by a statistically significant amount than do simulations around a single star .
the percentage of mass that was lost in most of the close binary sets of simulations is statistically inconsistent with the average total mass loss in the sjs set of runs ( @xmath85 26% ) .
simulations with binary stars on circular orbits with @xmath75 0.1 au resulted in planetary systems that accreted , on average , more of the initial disk mass ( @xmath85 15% 18% ) than the sjs runs , which can be attributed to both the relatively weak stellar perturbations on the inner edge of the disk compared to the other binary star systems , and to the lack of secular resonances from the giant planets near the outer edge of the disk , which are an important source of perturbations in the sjs systems .
the amount of mass lost in the close binary simulations was typically much higher in systems with larger @xmath4 , and only the cb_.075_1/3_.5 set and the cb_.2_0_.5 set resulted in a comparable amount of mass loss as the sjs runs , as did the set which began with the midplane of the disk inclined by 30@xmath16 to the binary orbital plane , cb_.1_0_.5_30@xmath16 .
more mass was lost and fewer planets were formed ( on average ) when the stellar masses were unequal ( @xmath12 = 0.2 ) , since binaries with more extreme mass ratios ( smaller @xmath12 ) travel farther from the center of mass of the pair .
in addition to the differences in mass loss , all of the close binary simulations that began with the disk co - planar to the stellar orbit resulted ( on average ) in smaller values of the specific energy ( @xmath86 ) , and greater values of the specific angular momentum ( @xmath87 ) and specific @xmath57-component of angular momentum ( @xmath88 ) .
this is also probably related to the clearing of the region near 2 au by the @xmath89 resonance in the sjs system . in close binary systems with @xmath76 0.2 au
, the final specific energy was higher than ( or comparable to ) the system s initial specific energy ( @xmath90 1 ) , although still not as high as the average value of @xmath86 for the sjs runs ( @xmath9 1.26 ) . as @xmath6 and/or @xmath7 is increased , and planets form farther from the binary center of mass , the specific energy decreases ( even to as low as 50% of the system s initial specific energy ) while the changes in angular momenta increase by as much as 30% . the simulations that began with the midplane of the disk inclined relative to the stellar orbit , cb_.1_0_.5_30@xmath16 , were the only set that resulted in a greater average value of @xmath86 and a smaller average @xmath87 than the sjs runs .
these are both consequences of the inward drift of the disk resulting from the initial tilt of the disk to the binary orbital plane .
even larger manifestations of this effect are evident in the high - inclination simulations of terrestrial planet formation within the @xmath3 centauri ab system ( quintana et al .
the amount of mass loss ( an average of 25% ) , the average specific energy , and the average specific angular momentum are statistically consistent with the sjs simulations , but this is simply a result of the coincidental cancellation of greater inward motion and the absence of the @xmath89 resonance
. however , the @xmath57-component of angular momentum had values that were smaller than in the sjs simulations , because the final systems have similar total angular momentum yet larger inclinations ( which also lead to a statistically significant increase of the angular momentum deficit , @xmath49 ) .
in the present work , we have examined the effect of 14 different short - period binary star configurations ( each with a combined stellar mass of 1 m@xmath2 ) on the late stages of terrestrial planet formation within a circumbinary protoplanetary disk .
stellar mass ratios of 1:1 and 4:1 were examined , and the initial orbits of the stars were varied ( with semimajor axes between 0.05 au @xmath0 0.4 au and eccentricities @xmath80 0.8 ) such that the stellar apastron ranged from 0.05 au @xmath91 0.4 au .
the midplane of the disk began coplanar to the stellar orbit in all but one set of runs ; in that exceptional set , the initial inclination of the disk started at 30@xmath16 relative to the binary orbital plane .
giant planets analogous to jupiter ( at @xmath9 5.2 au ) and , in all but one set of runs , saturn ( at @xmath9 9.5 au ) were included .
the evolution of the protoplanets was followed using a symplectic ` close binary ' algorithm which was developed for this purpose ( chambers et al .
2002 ) , and 5 or 6 simulations were performed for each binary star system under study ( with small changes in the initial conditions of the disk ) to account for the chaotic nature of these @xmath26-body systems .
we statistically compared our results to a large set of simulations of the sun - jupiter - saturn system that began with virtually the same initial disk mass distribution ( initially performed by chambers ( 2001 ) , but also integrated herein ( appendix a ) ) .
the close binary stars with maximum separations @xmath4 @xmath5 @xmath6(1 + @xmath7 ) @xmath25 0.2 au and small @xmath7 had little effect on the accreting bodies , and in most of these simulations terrestrial planets formed over essentially the entire range of the initial disk mass distribution ( and even beyond 2 au in many cases ) . the stellar perturbations cause orbits to precess , thereby moving secular resonances out of the inner asteroid belt , allowing terrestrial planets to form from our initially compact disk and remain in stable orbits as far as 2.98 au from the center of mass of the binary stars .
the effects of the stellar perturbations on the inner edge of the planetesimal disk become evident in systems with larger @xmath6 ( and @xmath76 0.3 au ) and in most of the simulations with @xmath92 0 .
terrestrial - mass planets can still form around binary stars with nonzero eccentricity , but the planetary systems tend to be sparcer and more diverse . binary stars with @xmath76 0.3 au perturb the accreting disk such that the formation of earth - like planets near 1 au is unlikely . despite these constraints , at least one terrestrial planet ( at least as massive as the planet mercury ) formed in each of our simulations .
we thank john chambers for performing additional simulations presented in appendix a. fred adams , john chambers , and two anonymous referees provided informative suggestions which helped us improve the manuscript .
this research was supported in part by an astrobiology grant ( 21 - 344 - 53 - 1y ) and a grant from the nasa ames director s discretionary fund .
e. v. q. is a nas / nrc , and she was supported in the early stages of this research by a nasa graduate student researchers program fellowship and a michigan center for theoretical physics fellowship .
the initial conditions of the circumbinary disk used in our close binary simulations are taken from an earlier study of terrestrial planet formation within a disk around the sun with jupiter and saturn perturbing the system ( chambers 2001 ) . in a set of 16 accretion simulations , chambers ( 2001 )
varied the initial mass distribution of the circumstellar disk with the intent of determining which conditions best resulted in the formation of terrestrial planets with masses and orbits similar to those in the solar system . approximately 150 rocky bodies ( with a total disk mass @xmath93 2.5 m@xmath23 ) were placed between 0.3 au 2 au of the sun , with a surface density that is consistent with the minimum mass model of the solar nebula ( weidenschilling 1977 ) .
the masses of the bodies in the disk began with either a uniform or a ` bi - modal ' distribution ( in which half of the disk mass is divided into mars - sized planetary embryos that are embedded in a swarm of lunar - sized planetesimals ) .
the radius of each body was calculated assuming a material density of 3 g @xmath24 .
the disk was initially placed in the invariable plane of jupiter and saturn , and each system was evolved forward in time for @xmath9 150 300 myr .
two simulations , labeled ` j21 ' and ` j22 ' in chambers ( 2001 ) , were the most successful in reproducing terrestrial planets similar to those in the solar system .
each of these began with the bi - modal mass distribution , and differed only in their randomly chosen initial arguments of perihelion , longitudes of the ascending node , and mean anomalies of the embryos and planetesimals .
we chose to use the initial disk mass distribution from the j21 run for all of our close binary simulations in order to delineate the effects of an inner binary star system from the perturbations of a single star ( i.e. , varying the initial disk mass distribution in addition to examining the enormous binary star parameter space is computationally intensive ) . to examine the statistical differences between planets formed around close binaries and those formed around the sun
, we performed an additional 27 simulations of the sun - jupiter - saturn system ( using conditions nearly identical to the j21 and j22 integrations ) to provide a larger distribution of final planetary systems that form around the sun .
table 3 presents the planetary statistics ( as described in subsection 4.1 ) for all of the sjs simulations . in group 1 , the original @xmath94 hybrid symplectic algorithm ( chambers 1999 ) was used .
group 1a lists the planetary statistics for the original j21 and j22 simulations of chambers ( 2001 ) .
the next two groups each begin with the same initial conditions as j21 ( group 1b ) or j22 ( group 1c ) , with the exception of a small shift in the initial mean anomaly of one planetesimal by 1 9 meters . only the j22 simulation from group 1a was run for 150 myr , all other simulations of the sjs system listed into table 3 were integrated for 200 myr .
the @xmath94 algorithm was recently modified to model planetary accretion in binary star systems ( chambers et al .
2002 ) , and we used the hybrid symplectic algorithm that is built in this version to perform the integrations listed in group 2 . in theory
, this hybrid algorithm should produce results that are consistent with the hybrid algorithm in the original @xmath94 code .
the two simulations in group 2a , which had identical starting conditions as those in group 1a , were performed by j. e. chambers ( private communication , 2004 ) , and produced results statistically consistent with group 1a .
the nine simulations in group 2b ( which began with the same initial conditions as the corresponding runs from group 1b ) resulted in fewer planets and more mass loss , on average , than the final systems from group 1b .
table 3 gives the average values ( @xmath95 ) for the sjs runs ( group 1 and group 2 ) , and a statistical comparison ( using the kolmogorov - smirnov ( k - s ) test ) between group 1b and group 1c , and between group 1 and group 2 . the k - s value , d , and the associated probability , p , show that the planetary statistics of the final planets that form in the j21 set of simulations are consistent with those that formed in the j22 set of runs . comparing the distributions between group 1 and group 2 yield different results .
this may be due to the fact that six of the nine runs from group 2b resulted in two - planet systems , which may be a result of small number statistics . also note that the simulation which yielded a single planet does not have a finite value for @xmath48 and @xmath50 . in this case ,
the highest value for these statistics is taken when calculating the average , d , and p , in order to prevent biased results .
we include all of the runs from group 1 and group 2 when comparing the final systems that form around close binaries ( see table 2 ) .
finally , table 3 includes the planetary statistics for 3 simulations that began with the j21 initial disk around the sun , with neither giant planets nor a stellar companion perturing the system ( quintana et al .
as expected , the final planetary systems that form in these runs are qualitatively different than when massive companions are included .
in our nominal circumbinary disk model , the embryos and planetesimals begin on nearly circular and coplanar orbits about the center of mass of the pair of stars .
these initial conditions may not be the most appropriate for simulations in which the binary stars begin with larger stellar separations and higher eccentricities .
the least excited state orbiting highly eccentric binary star systems is one in which the bodies in the disk have higher initial eccentricities and aligned arguments of periastron . to examine the magnitude of this effect on the final outcome of the planets that form , we have performed an additional set of simulations of the @xmath6 = 0.2 au and @xmath7 = 0.5 binary star system ( with equal mass stars of 0.5 m@xmath2 ) , altering the initial disk conditions according to the following prescription .
we first performed a simulation of the @xmath6 = 0.2 au and @xmath7 = 0.5 system ( omitting the giant planets ) with massless test particles , beginning with the coordinates of the planetesimals and planetary embryos in our nominal disk .
the particles were followed for 100 years ( their eccentricity oscillations ranged from periods of years to decades ) , and the maximum eccentricity of each test particle , @xmath96 , was calculated .
we then set up grids with varying values of the eccentricity ( @xmath39/@xmath96 ) and argument of periastron ( @xmath97 ) for each of the 154 particles , and ran each system for 100 years in order to find the value of ( @xmath39 , @xmath97 ) that resulted in the smallest range in the magnitude of the eccentricity .
we varied @xmath39 and @xmath97 as follows , while keeping the remaining four orbital elements ( @xmath98 , @xmath15 , @xmath99 , and @xmath100 , which were randomly selected ) constant : @xmath101 0.21 @xmath102 0.7 ( with 0.01 intervals ) , with 72 values of @xmath97 ( at 5@xmath16 intervals ) examined for each @xmath39/@xmath96 .
@xmath101 0.01 @xmath102 0.2 ( with 0.01 intervals ) , with 36 values of @xmath97 ( at 10@xmath16 intervals ) examined for each @xmath103 . within each grid ,
the particle which had the smallest range of eccentricity ( @xmath104 - @xmath105 ) was found , and the initial coordinates of that particle were noted . excluding the values of semimajor axis at which all bodies were ejected from the system , the preferred value of @xmath106 ranged from 0.01 to 0.66 , with an average of 0.096 .
we then used these new values for the initial orbital elements of the bodies in the circumbinary disk , and performed five simulations with massive planetesimals and embryos , varying the position of one planetesimal in the disk , as done for our other planetary growth simulations .
figure 9 shows the bodies in the disk at the beginning of the simulation and also at time @xmath40 = 200,000 years for the cb_.2_.5_.5 ( a e ) set ( left column ) and the cbecc_.2_.5_.5 ( a e ) set ( right column ) . even in this early stage of the simulations ,
the evolution of the two circumbinary disks are comparable .
figure 10 presents the final planetary systems for each of these simulations .
the planetary statistics ( as described in section 4.1 ) for the final planetary systems are listed in table 4 and labelled cbecc_.2_.5_.5 ( a e ) .
table 4 also re - states the analogous statistics for the five original runs with @xmath6 = 0.2 au and @xmath7 = 0.5 that we performed , labelled as cb_.2_.5_.5 ( a e ) .
we used a k - s test to compare planetary systems resulting from the two sets of simulations , and found that the distributions for each of the statistics listed are statistically consistent .
we thus conclude that our choice of initial conditions for our nominal circumbinary disk are sufficient even for the highly eccentric binary star systems examined in this article .
one planetesimal in simulation cbecc_.2_.5_.5_e ended up stranded in a distant orbit , with periapse well beyond the orbits of the giant planets ( figure 11 ) .
this planetesimal was scattered outwards by the giant planets at @xmath40 = 23.6 yr , and became trapped in a 5:2 resonance with the outer giant planet at @xmath40 = 27.8 yr .
this resonance substantially reduced the planetesimal s eccentrity to @xmath39 = 0.4 , raising its periapse distance to above the apoapse of the orbit of the outermost giant planet ( the apoapse of this never exceeded @xmath107 = 10.84 au during the entire integration ) .
the planetesimal broke free of the resonance around @xmath40 = 47.8 yr , but its periapse distance remained high for the remainder of the simulation .
it is clearly in a chaotic orbit , capable of returning to the planetary region and being ejected from the system , but this may well not occur for many millions , if not billions , of years .
similar processes may have operated to remove trans - neptunian objects in our solar system from neptune s immediate control , and if they happened during the migration epoch of the giant planets , such objects could have been permanently stranded away from the planetary region ( gomes et al .
the ensemble of plausible initial conditions for simulations of planetary growth is immense .
fortunately , one single numerical simulation ( or a set thereof ) may be valid for a range of stellar and planetary masses , orbital periods , etc . for simulations to correspond exactly
, the following must apply : ( 1 ) : : the ratios of the masses of all of the bodies must be the same , i.e. , there is a uniform factor by which all masses must be multiplied , @xmath108 . ( 2 ) : : the ratios of the distances between objects must be multipled by an equal amount @xmath109 , and the physical sizes of the objects , as well as other factors involving removal of objects ( inner , outer boundaries in semimajor axis ) , must be multiplied by this same factor @xmath109 .
note that , in general , the densities of objects will change according to the following formula : + @xmath110 ( 3 ) : : initial ( and thus subsequent ) velocities must be scaled so that orbital geometries remain the same ( `` orbital elements '' ) .
for this to hold , the ratio of kinetic to potential energy must remain the same , so + @xmath111 + @xmath112 + therefore , + @xmath113 ( 4 ) : : time must scale in a manner such that the same processes take the same number of orbits , so : + @xmath114 one example of this scaling is that if the star s mass is increased , and those of the planetesimals / planetary embryos are increased by the same amount , then the simulations would be applicable either if planetesimal densities increase by this amount or if distances grow by the same factor as the physical radii of the planetesimals .
a less trivial example is that a simulation of the growth of rocky planetesimals with @xmath115 = 4 g @xmath24 at 2 au corresponds to rock - ice planetesimals with @xmath115 = 1.5 g @xmath24 at 2.77 au around the same single star or around another binary with the same masses but with a 38% larger semimajor axis .
note that we are free to select any positive values for two parameters ( @xmath108 and @xmath109 in the above discussion ) , but that these choices specify the values of the other three parameters .
collisional outcomes must also be scaled appropriately , although this is trivially satisfied by the `` perfect accretion '' assumption used herein . additionally , any luminosity - related items , temperature related items , radiation forces , etc .
( none of which were included in our simulations ) must be scaled appropriately .
correia , a. c. m. , udry , s. , mayor , m. , laskar , j. , naef , d. , pepe , f. , queloz , d. , santos , n. c. , 2005 .
the coralie survey for southern extra - solar planets .
xiii . a pair of planets around hd 202206 or a circumbinary planet ?
440 , 751 - 758 .
deeg , h. j. , doyle , l. r. , kozhevnikov , v. p. , blue , j. e. , martin , e. l. , schneider , j. , 2000 .
a search for jovian - mass planets around cm draconis using eclipse minima timing .
358 , l5-l8 .
gomes , r. s. , gallardo , t. , fernandez , j. a. , brunini , a. , 2006 .
on the origin of the high - perihelion scattered disk : the role of the kozai mechanism and mean motion resonances .
celestial mech .
91 , 109 - 129 .
kortenkamp , s. , wetherill , g. w. , 2000 .
terrestrial planet and asteroid formation in the presence of giant planets .
i. relative velocities of planetesimals subject to jupiter and saturn perturbations .
icarus 143 , 60 - 73 .
lissauer , j. j. , 2004 .
formation of giant planets and brown dwarfs . in : beaulieu ,
p . , des etangs , a. l. , and terquem , c. ( eds . ) , extrasolar planets : today and tomorrow , asp conference series 321 , san francisco , california , pp . 271 - 281 .
mathieu , r. d. , adams , f. c. , fuller , g. a. , jensen , e. l. n. , koerner , d. w. , sargent , a. i. , 1995 .
submillimeter continuum observations of the t tauri spectroscopic binary gw orionis .
j. 109 , 2655 - 2669 .
mathieu , r. d. , stassun , k. , basri , g. , jensen , e. l. n. , johns - krull , c. m. , valenti , j. a. , hartmann , l. w. , 1997 . the classical t tauri spectroscopic binary dq tau . i. orbital elements and light curves . astron .
j. 113 , 1841 - 1854 .
mathieu , r. d. , ghez , a. m. , jensen , e. l. n. , simon , m. , 2000 .
young binary stars and associated disks . in : mannings v. , boss , a. p. , russell , s. ( eds . ) , protostars and planets iv , univ . of arizona press , tucson , pp .
703 - 730 .
quintana , e. v. , 2003 .
terrestrial planet formation around alpha centauri b. in : deming , d. , and seager , s. ( eds . ) , scientific frontiers in research on extrasolar planets , asp conference series 294 , san francisco , california , pp .
319 - 322 .
udry , s. , mayor , m. , naef , d. , pepe , f. , queloz , d. , santos , n. c. , burnet , m. , 2002 .
the very low - mass companions of hd 141937 , hd 162020 , hd 168443 , and hd 202206 : brown dwarfs or `` superplanets '' ? .
390 , 267 - 279 .
weaver , h. a. , stern , s. a. , mutchler , m. j. , steffl , a. j. , buie , m. w. , merline , w. j. , spencer , j. r. , young , e. f. , young , l. a. , 2006 .
discovery of two new satellites of pluto .
nature 439 , 943 - 945 .
0 & & 0.2 & & & 2.2 & & 1.80 + 0 & & 0.5 & & & 2.3 & & 2.01 + 1/3 & & 0.2 & & & 3.5 [ 3.8 ] & & 3.15 + 1/3 & & 0.5 & & & 3.2 [ 3.5 ] & & 2.99 + 0.5 & & 0.2 & & & 3.9 & & 3.40 + 0.5 & & 0.5 & & & 3.6 & & 2.98 + 0.8 & & 0.2 & & & 4.3 & & 3.86 + 0.8 & & 0.5 & & & 3.9 & & 3.19 + lllllllllllllll [ tbl-2 ] & & & & & & & & & & & & & & + cb_.05_0_.5@xmath38_a & 500 & 6&1&3.53&7.12&0.288&27.1&0.0041&32.3&0.226&15.71&1.13&0.94&0.93 + cb_.05_0_.5@xmath38_b & 200 & 4&1&3.48&7.18&0.388&36.3&0.0063&29.5&0.343&13.57&1.14&0.93&0.93 + cb_.05_0_.5@xmath38_c & 500&4&1 & 3.48&7.28 & 0.515&34.4&0.0008&36.8&0.289&14.29 & 1.13&0.93&0.93 + cb_.05_0_.5@xmath38_d & 200 & 3&1&5.00&10.47&0.458&43.0&0.0047&34.7&0.410&18.93&1.18&0.91&0.91 + cb_.05_0_.5@xmath38_e & 200 & 2&1&7.47&16.02&0.905&40.0&0.0110&213.8&0.437&28.93&1.33&0.82&0.81 + * @xmath95&320 & 3.8 & 1 & 4.59 & 9.61 & 0.511 & 36.2 & 0.0054 & 69.4 & 0.341&18.29 & 1.18 & 0.91 & 0.90 + d & @xmath116&0.28&0.58 & @xmath116 & @xmath116&0.22&0.28&0.48&0.35&0.28&0.70&0.67&0.67&0.70 + p & @xmath116&0.832&0.066 & @xmath116 & @xmath116&0.968&0.832&0.200&0.572&0.810&0.014&0.021&0.021&0.014 + & & & & & & & & & & & & & & + cb_.05_0_.5_a & 500 & 3&1&3.42&7.52&0.500&45.3&0.0069&30.8&0.379&15.71&1.15&0.93&0.93 + cb_.05_0_.5_b & 200 & 3&1&3.46&7.29&0.561&48.4&0.0252&29.5&0.454&24.29&1.23&0.88&0.88 + cb_.05_0_.5_c & 500 & 5&1&5.00&10.95&0.355&30.3&0.0026&31.4&0.309&11.43&1.11&0.95&0.94 + cb_.05_0_.5_d & 200 & 3&1 & 5.12&10.50&0.473&42.4&0.0049&29.7&0.364&12.50&1.13&0.94&0.94 + cb_.05_0_.5_e & 500 & 5&1&3.43&7.17&0.294&27.8&0.0066&27.7&0.204&11.43&1.10&0.96&0.95 + * @xmath95&380&3.8 & 1 & 4.09 & 8.68 & 0.437 & 38.84 & 0.0092 & 29.8 & 0.342&15.07 & 1.14 & 0.93 & 0.93 + d & @xmath116&0.37&0.58&@xmath116&@xmath116&0.34&0.37&0.41&0.67&0.38&0.80&0.80&0.77&0.77 + p & @xmath116&0.501&0.066&@xmath116&@xmath116&0.620&0.501&0.355&0.021&0.457&0.003&0.003&0.005&0.005 + & & & & & & & & & & & & & & + cb_.075_1/3_.5_a & 200 & 3&1&1.56&3.66&0.436&40.2&0.0166&27.7&0.363&24.64&1.02&1.00&0.98 + cb_.075_1/3_.5_b & 200 & 2&0&5.89&16.47&0.910&54.1&0.0310&167.7&0.290&52.50&0.75&1.06&1.06 + cb_.075_1/3_.5_c & 200 & 3&0&2.21&5.96&0.566&41.3&0.0047&37.5&0.306&13.57&1.08&0.95&0.95 + cb_.075_1/3_.5_d & 200 & 2&3&5.50&13.39&0.487&42.9&0.0282&96.2&0.285&59.64 & 0.68&1.13&1.13 + cb_.075_1/3_.5_e & 200 & 3&1&1.54&3.58&0.489&42.9&0.0092&35.3&0.255&21.79&1.15&0.92&0.92 + * @xmath95&200 & 2.6 & 1 & 3.34 & 8.61 & 0.578 & 44.28 & 0.0179 & 72.9 & 0.300 & 34.43 & 0.94 & 1.01 & 1.01 + d & @xmath116 & 0.48&0.34 & @xmath116&@xmath116 & 0.42&0.55&0.37&0.37&0.64&0.37&0.80&0.935&0.968 + p & @xmath116&0.200&0.620 & @xmath116&@xmath116 & 0.337&0.096&0.479&0.501&0.032&0.501&0.003&0.000&0.000 + & & & & & & & & & & & & & & + cb_.1_0_.5_a & 200 & 4&1&2.40&5.70&0.478&36.1&0.0033&32.7&0.305&17.14&1.11&0.94&0.94 + cb_.1_0_.5_b & 200 & 5&1&1.80&3.74&0.313&29.8&0.0017&29.7&0.314 & 11.06&1.07&0.97&0.97 + cb_.1_0_.5_c & 200 & 4&1&1.88&4.22&0.335&33.2&0.0040&32.9&0.253&11.43&1.06&0.97&0.97 + cb_.1_0_.5_d & 200 & 4&1&2.97&6.44&0.297&30.7&0.0044&35.9&0.263&17.14&1.06&0.96&0.96 + cb_.1_0_.5_e & 200 & 4&1&2.62&5.84&0.345&30.1&0.0036&39.0&0.314&21.43&1.19&0.90&0.90 + * @xmath95&200&4.2 & 1 & 2.33 & 5.19 & 0.354 & 32.0 & 0.0034 & 34.0 & 0.290&15.64 & 1.10 & 0.95 & 0.95 + d & @xmath116&0.68&0.58&@xmath116&@xmath116&0.74&0.74&0.70&0.47&0.57&0.77&0.80&0.77&0.77 + p & @xmath116 & 0.019&0.066&@xmath116&@xmath116&0.008&0.008&0.014&0.212&0.071&0.005&0.003&0.005&0.005 + & & & & & & & & & & & & & + cb_.1_0_.2_a & 200 & 3&1&2.00&3.85&0.440&41.1&0.0038&35.6&0.314&19.29&1.11&0.94&0.94 + cb_.1_0_.2_b & 200 & 4&1&2.63&5.45&0.322&34.9&0.0053&29.7&0.422&16.79&1.14&0.94&0.94 + cb_.1_0_.2_c & 200 & 4&0&3.04&5.86&0.451&33.6&0.0019&33.4&0.317&11.79&1.07&0.96&0.96 + cb_.1_0_.2_d & 500&4&1 & 1.95 & 3.73 & 0.500&29.7&0.0041&36.9&0.393&20.00 & 1.17&0.91&0.91 + cb_.1_0_.2_e & 500&5&0&2.56&5.00&0.414&32.5&0.0031&31.2&0.305&18.93&1.13&0.94&0.94 + * @xmath95&320&4.0&0.6 & 2.44 & 4.78 & 0.425 & 34.36 & 0.0036 & 33.4 & 0.350 & 17.36 & 1.12 & 0.94 & 0.94 + d&@xmath116&0.48&0.34 & @xmath116 & @xmath116&0.38&0.54&0.67&0.45&0.37&0.67&0.80&0.87&0.90 + p&@xmath116&0.200&0.620 & @xmath116 & @xmath116&0.457&0.103&0.021&0.269&0.479&0.021&0.003&0.001&0.001 + & & & & & & & & & & & & & & + cb_.1_.8_.5_a & 200 & 3&2&2.29&4.31&0.440&35.3&0.0284&61.2&0.273&31.79&0.90&1.01&0.99 + cb_.1_.8_.5_b & 500&2&0 & 2.45 & 4.74&0.667&45.3&0.0100&78.8&0.305&38.93&0.95&0.98&0.98 + cb_.1_.8_.5_c & 200 & 3&2&1.84&3.33&0.634&43.4&0.0101&55.9&0.219&31.79&1.00&0.97&0.96 + cb_.1_.8_.5_d & 200 & 3&2&1.90 & 3.82&0.574&49.0&0.0276&37.2 & 0.274&39.64&1.06&0.95&0.93 + cb_.1_.8_.5_e & 200 & 4&1&1.52&2.88&0.328&36.5&0.0451&33.9&0.291&30.36&1.09&0.95&0.91 + * @xmath95 & 650&3.0&1.4&2.00&3.82 & 0.529 & 41.9 & 0.0242 & 53.4 & 0.272 & 34.50 & 1.00 & 0.97 & 0.96 + d&@xmath116&0.45&0.41 & @xmath116 & @xmath116&0.37&0.26&0.52&0.50&0.67 & 0.77 & 0.80&0.97&0.87 + p&@xmath116&0.269&0.374 & @xmath116 & @xmath116&0.479&0.889&0.135&0.155&0.021&0.005 & 0.003&0.000&0.001 + & & & & & & & & & & & & & & + cb_.1_.8_.2_a & 500 & 2&0&1.98&4.86&0.544&83.5&0.0841&19.7&0.527 & 46.79&0.95&1.03&0.97 + cb_.1_.8_.2_b & 500 & 3&0&1.93&4.50&0.468&38.7&0.0276&57.5&0.347&32.86&1.09&0.92&0.90 + cb_.1_.8_.2_c & 200 & 3&1&2.13&4.91&0.433&43.5&0.0060&39.5&0.208&35.71&0.93&1.02&1.02 + cb_.1_.8_.2_d & 165 & 1&0&2.82&6.78&1.000&@xmath116&0.0109&@xmath117&0.376&46.43&0.95&0.95&0.95 + cb_.1_.8_.2_e & 200 & 3&1&1.52&3.96&0.490&47.6&0.0378&30.5&0.368&30.00&1.02&0.98&0.95 + * @xmath95&313 & 2.4 & 0.4 & 2.07 & 5.00 & 0.587 & 59.4 & 0.0333 & 40.9 & 0.365 & 38.36 & 0.99 & 0.98 & 0.96
+ d & @xmath116&0.48&0.50&@xmath116&@xmath116&0.39&0.52&0.47&0.30&0.38&0.71&0.80&0.94&0.77 + p & @xmath116 & 0.200&0.155&@xmath116&@xmath116&0.435&0.135&0.212&0.742&0.457&0.012&0.003&0.000&0.005 + & & & & & & & & & & & & & & + cb_.1_0_.5_30@xmath16_a & 500 & 4&0&2.47&5.46&0.453&38.5&0.1182&29.8&0.387 & 28.21&1.42&0.83&0.74 + cb_.1_0_.5_30@xmath16_b & 500 & 2&1&2.18&5.06&0.724&50.5&0.0938&60.1&0.513&22.50&1.53&0.78&0.71 + cb_.1_0_.5_30@xmath16_c & 500 & 3&0&1.98&4.29&0.515&48.1&0.1473&25.5&0.467&18.93&1.48&0.82&0.71 + cb_.1_0_.5_30@xmath16_d & 200 & 4&1 & 1.79&3.69&0.421&36.5&0.0157&26.7&0.293&18.57&1.12&0.95&0.94 + cb_.1_0_.5_30@xmath16_e & 550&2&0&4.07&9.50&0.913&49.9&0.1610&186.1&0.361&42.14&1.21&0.85&0.72 + * @xmath95&450&3.0 & 0.4 & 2.50 & 5.60 & 0.605 & 44.7 & 0.1072 & 65.6 & 0.404 & 26.07 & 1.35 & 0.85 & 0.76 + d & @xmath116&0.31&0.50&@xmath116&@xmath116&0.39&0.32&0.77&0.47&0.19&0.30&0.54&0.57&0.77 + p & @xmath116&0.718&0.155&@xmath116&@xmath116&0.435&0.669&0.005&0.212&0.994&0.742&0.110&0.071&0.005 + & & & & & & & & & & & & & & + cb_.15_1/3_.5_a & 200 & 2&0&2.03&5.16&0.736&52.4&0.0203&79.5&0.174&56.79&0.77&1.09&1.08 + cb_.15_1/3_.5_b & 200 & 3&3&1.63&3.91&0.600&33.6&0.0070&83.7&0.313&44.64&1.09&0.92&0.91 + cb_.15_1/3_.5_c & 200 & 2&0&2.40&6.19&0.714&39.5&0.0135&120.6&0.275&45.00&0.86&1.02&1.01 + cb_.15_1/3_.5_d & 200 & 3&1 & 1.74&4.10&0.626&44.3&0.0134&60.4&0.270&37.86&0.93&1.00&0.99 + cb_.15_1/3_.5_e & 200 & 3&2&1.27&3.10&0.497&39.9&0.0066&57.7&0.260&39.64&1.00&0.97&0.96 + * @xmath95&200 & 2.6 & 1.2 & 1.82 & 4.49 & 0.635 & 41.9 & 0.0122 & 80.4 & 0.258 & 44.79 & 0.93 & 1.00 & 0.99 + d & @xmath116 & 0.48&0.34&@xmath116&@xmath116&0.61&0.32&0.39&0.90&0.74&0.97&0.80&0.94&0.97 + p & @xmath116 & 0.200&0.620&@xmath116&@xmath116&0.049&0.693&0.435&0.001&0.008&0.000&0.003&0.000&0.000 + & & & & & & & & & & & & & & + cb_.2_0_.5_a & 200 & 3&1&1.11&2.35&0.545&47.6&0.0064&40.0&0.182&23.93&1.02&0.97&0.97 + cb_.2_0_.5_b & 200 & 5&1&1.15&2.56&0.429&29.6&0.0058&45.8&0.248&20.00&1.08&0.94&0.94 + cb_.2_0_.5_c & 200 & 4&2&1.17&2.42&0.519&32.7&0.0075&58.4&0.405&32.50&1.12&0.91&0.91 + cb_.2_0_.5_d & 500 & 4&1 & 1.07&2.21 & 0.553&38.8&0.0053&34.2&0.298&32.14&1.12&0.94&0.93 + cb_.2_0_.5_e & 200 & 2&1&2.35&4.85&0.633&36.1&0.0052&109.4&0.242&43.57&0.95&0.98&0.97 + * @xmath95&260&3.6 & 1.2 & 1.37 & 2.88 & 0.536 & 37.0 & 0.0060 & 57.6 & 0.275 & 30.43 & 1.06 & 0.95 & 0.94 + d & @xmath116&0.28&0.58&@xmath116&@xmath116&0.41&0.34&0.45&0.41&0.64&0.44&0.80&0.87&0.87 + p & @xmath116&0.832&0.066&@xmath116&@xmath116&0.35&0.620&0.269&0.35&0.032&0.285&0.003&0.001&0.001 + & & & & & & & & & & & & & & + cb_.2_0_.2_a&200 & 3&2&1.24&2.30&0.659&59.0&0.0067&41.5&0.263&41.43&0.92&1.02&1.02 + cb_.2_0_.2_b&200 & 3&0&1.29&2.35&0.632&41.8&0.0041&84.2&0.253&48.57&1.08&0.93&0.92 + cb_.2_0_.2_c&500 & 3&3&1.28 & 2.51&0.661&48.1&0.0096&62.1&0.233&41.07&1.07&0.93&0.93 + cb_.2_0_.2_d&200 & 3&0&2.40&4.99&0.461&35.9&0.0073&60.4&0.220&45.00&0.85&1.05&1.05 + cb_.2_0_.2_e&200 & 3&0&1.27&2.43&0.487&49.0&0.0058&41.0&0.323&30.36&1.02&0.98&0.97 + * @xmath95&260&3.0&1&1.50&2.92&0.580&46.8&0.0067 & 57.8 & 0.258 & 41.29 & 0.99 & 0.98 & 0.98 + d & @xmath116&0.41&0.50&@xmath116&@xmath116&0.42&0.38&0.38&0.68&0.74 & 0.77&0.80&0.97&0.97 + p & @xmath116&0.355&0.155&@xmath116&@xmath116&0.337&0.457&0.457&0.019&0.008 & 0.005&0.003&0.000&0.000 + & & & & & & & & & & & & & & + cb_.2_.5_.5_a & 146 & 1&0&2.35&5.44&1.000&@xmath116&0.0193&@xmath117&0.147&73.58&0.74&1.07&1.07 + cb_.2_.5_.5_b & 200 & 2&0&2.25&4.65&0.890&57.2&0.0021&152.4&0.169&67.50&0.73&1.11&1.11 + cb_.2_.5_.5_c&154 & 1&0&3.56&8.54&1.000&@xmath116&0.0328&@xmath117&0.378&85.71&0.49&1.31&1.30 + cb_.2_.5_.5_d&115 & 1&0 & 2.70&5.92&1.000&@xmath116&0.0190&@xmath117&0.263&79.29&0.64&1.16&1.15 + cb_.2_.5_.5_e & 200 & 2&0&1.78&5.20&0.590&32.4&0.0104&185.7&0.155&62.50&0.67&1.15&1.15 + * @xmath95&163&1.4 & 0.0 & 2.53 & 5.95 & 0.896 & 52.2 & 0.0167&179.0 & 0.222&73.72 & 0.65 & 1.16 & 1.15 + d & @xmath116 & 0.74&0.67&@xmath116&@xmath116&0.77&0.70&0.41&0.97&0.74&0.97&0.80&0.97&0.97 + p & @xmath116 & 0.008&0.021&@xmath116&@xmath116&0.005&0.014&0.374&0.000&0.008&0.000&0.003&0.000&0.000 + & & & & & & & & & & & & & & + cb_.2_.5_.2_a & 200 & 2&1&1.15&2.95&0.547&36.9&0.0081&133.3&0.127&69.29&0.66&1.17&1.16 + cb_.2_.5_.2_b & 200 & 2&0&1.85&4.45&0.563&30.9&0.0126&203.1&0.179&63.21&0.69&1.13&1.12 + cb_.2_.5_.2_c & 500&2&1&1.82&4.62&0.691&67.6&0.0080&71.0&0.425&80.36&0.72&1.13&1.12 + cb_.2_.5_.2_d & 500 & 2&0&1.39&3.21&0.828&54.2&0.0244&141.1&0.083&77.15&0.73&1.09&1.09 + cb_.2_.5_.2_e & 200 & 2&1&1.36&3.34&0.860&46.5&0.0099&190.0&0.184&66.79&0.75&1.09&1.09 + * @xmath95&320&2.0 & 0.6 & 1.51 & 3.71 & 0.698 & 47.2 & 0.0126 & 147.7 & 0.200 & 71.36 & 0.71 & 1.12 & 1.12 + d&@xmath116&0.64&0.34&@xmath116&@xmath116&0.71&0.31&0.45&0.97&0.80&0.97&0.80&0.97&0.97 + p&@xmath116&0.032&0.620&@xmath116&@xmath116&0.012&0.718&0.254&0.000&0.003&0.000&0.003&0.000&0.000 + & & & & & & & & & & & & & & + cb_.3_0_.5_a&200 & 2&0&1.99&4.63&0.670&37.6&0.0088&153.9&0.166&64.28&0.67&1.15&1.15 + cb_.3_0_.5_b&200 & 3&0&1.32&2.90&0.667&41.8&0.0044&84.9&0.198&48.58&0.78&1.09&1.09 + cb_.3_0_.5_c&200 & 2&1&1.75&3.75&0.556&36.2&0.0056&126.2&0.176&55.00&0.80&1.07&1.06 + cb_.3_0_.5_d&200 & 2&0&2.28&4.96&0.734&51.8&0.0141&101.6&0.221&71.79&0.67&1.15&1.15 + cb_.3_0_.5_e&200 & 2&2&1.51&3.22&0.653&54.2&0.0051&65.9&0.197&56.79&0.80&1.08&1.08 + * @xmath95&200&2.2&0.6&1.77&3.89 & 0.656 & 44.3 & 0.0076 & 106.5 & 0.192 & 59.29&0.74 & 1.11&1.11 + d & @xmath116&0.64&0.50&@xmath116&@xmath116&0.74&0.32&0.34&0.97&0.80&0.97&0.80&0.97&0.97 + p&@xmath116 & 0.032&0.155&@xmath116&@xmath116&0.008&0.669&0.596&0.000&0.003&0.000&0.003&0.000&0.000 + & & & & & & & & & & & & & & + cb_.3_1/3_.5_a & 200 & 2&0&1.51&3.91&0.725&40.4&0.0043&235.5&0.245&85.71&0.61&1.21&1.21 + cb_.3_1/3_.5_b & 200 & 2&0&1.41&3.42&0.843&23.3&0.0017&811.3&0.185&75.00&0.68&1.14&1.14 + cb_.3_1/3_.5_c & 175 & 1&0&1.79&4.12&1.000&@xmath116&0.0011&@xmath117&0.181&86.07&0.64&1.17&1.17 + cb_.3_1/3_.5_d&200 & 1&1&2.17&4.87&0.960&@xmath116&0.0053&1839.4&0.167 & 91.07 & 0.53&1.29&1.29 + cb_.3_1/3_.5_e & 100 & 1&0&1.75&4.56&1.000&@xmath116&0.0040&@xmath117&0.105&78.57&0.62&1.19&1.19 + cb_.3_1/3_.5_f & 200 & 1&2&1.63&3.96&0.957&@xmath116&0.0069&541.8&0.208&83.57&0.65&1.17&1.16 + * @xmath95&179&1.3 & 0.5 & 1.71 & 4.14 & 0.914 & 37.6&0.0039 & 1184.5 & 0.182 & 83.33 & 0.62&1.19 & 1.19 + d & @xmath116&0.77&0.54&@xmath116&@xmath116&0.90&0.38&0.58&0.97&0.83&0.97&0.83&0.97&0.97 + p&@xmath116 & 0.002&0.069&@xmath116&@xmath116&0.000&0.365&0.043&0.000&0.001&0.000&0.001&0.000&0.000 + & & & & & & & & & & & & & & + cb_.4_0_.5_a & 500 & 2&1&1.56&3.24&0.598&34.1&0.0011&153.3&0.206&67.15&0.67&1.16&1.16 + cb_.4_0_.5_b & 1000 & 1&2&1.34&2.73&0.971&@xmath116&0.0006&2183.7&0.141&75.00&0.72&1.11&1.11 + cb_.4_0_.5_c & 929 & 1&0&2.27&5.35&1.000&@xmath116&0.0214&@xmath117&0.081&90.00&0.57&1.23&1.22 + cb_.4_0_.5_d & 500 & 2&2&1.10&2.40&0.595&25.6&0.0024&193.0&0.281&85.00&0.71&1.12&1.12 + cb_.4_0_.5_e & 200 & 1&1&1.88&4.47&0.986&@xmath116&0.0008&3356.0&0.282&75.00&0.60&1.22&1.22 + cb_.4_0_.5_f & 129 & 1&0&1.99&5.46&1.000&@xmath116&0.0193&@xmath117&0.222&85.35&0.57&1.22&1.22 + * @xmath95&543&1.3 & 1.0 & 1.69 & 3.94 & 0.858 & 32.7&0.0076 & 2099.7 & 0.202 & 79.58 & 0.64&1.18 & 1.17 + d & @xmath116&0.77&0.27&@xmath116&@xmath116&0.81&0.64&0.67&0.97&0.74&0.97&0.83&0.97&0.97 + p & @xmath116&0.002&0.795&@xmath116&@xmath116&0.001&0.017&0.011&0.000&0.004&0.000&0.001&0.000&0.000 + & & & & & & & & & & & & & & + mvem & @xmath116 & 4.0&0.0&@xmath116&@xmath116&0.509&37.7&0.0018&89.9&@xmath116&@xmath116&@xmath116&@xmath116&@xmath116 + sjs_ave & 200&3.0&0.7&@xmath116&@xmath116&0.514&44.4&0.0157&40.5&0.420&25.92&1.26&0.87&0.87 + sun_ave & 867 & 4.0&12.3&@xmath116&@xmath116&0.393&38.2&0.0318&15.9&0.360&0.48&1.06&1.01&1.00 + sun_ave ( @xmath61 2 au ) & 867 & 3.0&0.7&@xmath116&@xmath116&0.483&36.8&0.0061&34.3&@xmath116&18.66@xmath118 & 1.21&0.90&0.90 + @xmath3 cen a ( @xmath29 30@xmath16)&481&3.9&0.2&@xmath116&@xmath116&0.432&38.4&0.0098&30.9&0.354&15.69&1.24&0.90&0.94 + * * * * * * * * * * * * * * * * * group 1a & & & & & & & & & & & & + j21 & 4 & 0 & 0.335 & 31.3 & 0.0061 & 37.3 & 0.328 & 17.14 & 0.71 & 1.17 & 0.91 & 0.91 + j22 & 4 & 0 & 0.326 & 33.7 & 0.0105 & 35.0 & 0.281 & 18.57 & 1.43 & 1.19 & 0.91 & 0.90 + * group 1b & & & & & & & & & & & & + j21_1 & 3 & 2 & 0.532 & 43.1 & 0.0048 & 39.6 & 0.402 & 21.43&5.36 & 1.29 & 0.87 & 0.87 + j21_2 & 3 & 1 & 0.379 & 33.3 & 0.0041 & 49.3 & 0.508 & 18.93&0.00 & 1.24 & 0.87 & 0.87 + j21_3 & 3 & 1 & 0.394 & 36.9 & 0.0032 & 34.9 & 0.518 & 20.71&0.36 & 1.16 & 0.92 & 0.92 + j21_4 & 3 & 4 & 0.400 & 29.7 & 0.0132 & 61.5 & 0.339 & 26.79&7.14 & 1.33 & 0.84 & 0.83 + j21_5 & 4 & 0 & 0.496 & 34.0 & 0.0061 & 36.3 & 0.362 & 20.00&0.00 & 1.17 & 0.92 & 0.91 + j21_6 & 4 & 2 & 0.551 & 35.8 & 0.0083 & 34.8 & 0.296 & 23.57&0.00 & 1.23 & 0.89 & 0.89 + j21_7 & 2 & 0 & 0.571 & 55.0 & 0.0118 & 40.0 & 0.396 & 14.64&2.14 & 1.20 & 0.89 & 0.89 + j21_8 & 1 & 0 & 1.000 & @xmath116 & 0.0686 & @xmath117 & 0.498 & 34.29&1.79 & 1.48 & 0.73 & 0.72 + j21_9 & 3 & 0 & 0.491 & 52.8 & 0.0044 & 22.2 & 0.411 & 22.50&1.79 & 1.26 & 0.91 & 0.91 + * group 1c & & & & & & & & & & & & + j22_1 & 3 & 3 & 0.404 & 36.5 & 0.0044 & 34.6 & 0.414 & 20.36 & 0.00 & 1.20 & 0.91 & 0.90 + j22_2 & 3 & 0 & 0.518 & 40.9 & 0.0117 & 48.7 & 0.344 & 30.36 & 1.43 & 1.33 & 0.84 & 0.84 + j22_3 & 3 & 0 & 0.521 & 46.0 & 0.0188 & 37.5 & 0.580 & 22.86 & 0.36 & 1.24 & 0.88 & 0.87 + j22_4 & 4 & 0 & 0.418 & 33.4 & 0.0088 & 36.1 & 0.255 & 18.93 & 0.71 & 1.18 & 0.91 & 0.91 + j22_5 & 3 & 1 & 0.392 & 34.7 & 0.0069 & 42.3 & 0.463 & 29.64 & 1.07 & 1.31 & 0.86 & 0.85 + j22_6 & 4 & 0 & 0.377 & 36.0 & 0.0033 & 34.1 & 0.339 & 18.93 & 1.43 & 1.18 & 0.91 & 0.91 + j22_7 & 4 & 0 & 0.433 & 37.5 & 0.0066 & 33.1 & 0.311&25.00 & 0.71 & 1.25 & 0.89 & 0.88 + j22_8 & 3 & 1 & 0.381 & 45.8 & 0.0260 & 25.0 & 0.486 & 18.93 & 0.36 & 1.19 & 0.92 & 0.91 + j22_9 & 4 & 0 & 0.379 & 36.8 & 0.0063 & 40.2 & 0.257 & 25.36 & 1.43 & 1.21 & 0.89 & 0.89 + * @xmath95 ( group 1)&3.25 & 0.75 & 0.465&39.4 & 0.0117&39.2&0.389&22.45&1.41&1.24&0.88&0.88 + & & & & & & & & & & & & + * group 2a & & & & & & & & & & & & + j21@xmath119 & 4 & 1 & 0.488 & 34.5 & 0.0122 & 33.3 & 0.307 & 22.50 & 1.43 & 1.21&0.90&0.90 + j22@xmath119 & 3 & 1 & 0.733 & 42.1 & 0.0077 & 81.0 & 0.412 & 26.43 & 0.00 & 1.26&0.86&0.85 + * group 2b & & & & & & & & & & & & + j21_1@xmath119&2&0&0.871&87.0&0.0599 & 37.4&0.784 & 29.29&12.50&1.49&0.77&0.76 + j21_2@xmath119&2&0&0.513&65.9&0.0674 & 29.0&0.580&29.29&1.07&1.31&0.83&0.82 + j21_3@xmath119&2&0&0.651&54.2&0.0259 & 49.0&0.386&25.71&4.64&1.29&0.84&0.84 + j21_4@xmath119&2&0&0.684&80.8&0.0336 & 21.4&0.518&32.50&0.71&1.36&0.86&0.85 + j21_5@xmath119&3&0&0.527&39.8&0.0051 & 51.2&0.465&32.14&0.71&1.33&0.84&0.84 + j21_6@xmath119&2&1&0.592&54.6&0.0036 & 44.0&0.550&26.43&1.79&1.30&0.86&0.86 + j21_7@xmath119&2&0&0.684&54.3&0.0115 & 50.0&0.594&25.71&0.71&1.27&0.86&0.86 + j21_8@xmath119&4&2&0.397&37.2&0.0126 & 35.9&0.278&16.07&9.29&1.25&0.88&0.88 + j21_9@xmath119&3&3&0.488&38.3&0.0122 & 39.2&0.345&26.79&0.71&1.27&0.87&0.87 + * @xmath95 ( group 2)&2.64&0.73&0.603 & 53.5 & 0.0229 & 42.9 & 0.474 & 26.62 & 3.05 & 1.30 & 0.85 & 0.85 + & & & & & & & & & & & & + * @xmath95 ( all)&3.03&0.74&0.514&44.4&0.0157&40.5&0.420&23.93&1.99&1.26&0.87&0.87 + & & & & & & & & & & & & + * group 1b , group 1c & & & & & & & & & & & & + d & 0.22&0.22&0.44&0.22&0.33&0.33&0.33&0.22&0.44&0.22&0.22&0.22 + p & 0.958&0.958&0.250&0.958&0.603&0.603&0.603&0.958&0.250&0.958&0.958&0.958 + & & & & & & & & & & & & + * group 1 , group 2 & & & & & & & & & & & & + d & 0.57&0.25&0.51&0.51&0.44&0.25&0.40&0.62&0.26&0.56&0.57&0.57 + p & 0.012&0.722&0.032&0.032&0.096&0.679&0.147&0.005&0.658&0.014&0.012&0.012 + & & & & & & & & & & & & + & & & & & & & & & & & & + * sun - only & & & & & & & & & & & & + s_1&4&9&0.462&39.7&0.0136&19.0&0.335&0.000&0.357&1.06&1.00&1.00 + s_2&4&8&0.357&38.2&0.0342&16.0&0.359&0.000&1.071&1.05&1.00&0.99 + s_3&4&20&0.361&36.7&0.0476&12.8&0.385&0.000&0.000&1.07&1.02&1.00 + * @xmath95&4&12.3&0.393 & 38.2 & 0.0318 & 15.9 & 0.360 & 0.000 & 0.476 & 1.06 & 1.01 & 1.00 + & & & & & & & & & & & & + * sun - only , all & & & & & & & & & & & & + d & 0.68&0.97&0.52&0.39&0.77&0.75&0.36&0.75&0.49&0.75&0.97&0.97 + p & 0.041&0.001&0.194&0.548&0.012&0.017&0.631&0.017&0.256&0.017&0.001&0.001 + & & & & & & & & & & & & + * * * * * * * * * * * * * & & & & & & & & & & & & & & + cbecc_.2_.5_.5_a&200&3&0&1.89&4.22&0.481&33.0&0.0091&95.4&0.211&62.14&0.75&1.10&1.10 + cbecc_.2_.5_.5_b&100&1&0&3.17&8.87&1.000&@xmath116&0.0975&@xmath117&0.089&84.29&0.55&1.16&1.14 + cbecc_.2_.5_.5_c&200&2&2&2.06&4.38&0.610&40.1&0.0048&115.1&0.213&62.50&0.66&1.18&1.17 + cbecc_.2_.5_.5_d&200&2&0&1.76&3.82&0.652&48.2&0.0043&83.4&0.120&58.93&0.72&1.13&1.13 + cbecc_.2_.5_.5_e&200&1&1&3.68&9.46&0.976&@xmath116&0.0564&49.5&0.476&85.36&0.46&1.37&1.35 + * @xmath95&180 & 1.8 & 0.6 & 2.51 & 6.15 & 0.744 & 43.5 & 0.0344 & 91.7 & 0.222 & 70.64 & 0.63 & 1.19 & 1.18 + & & & & & & & & & & & & & & + cb_.2_.5_.5_a&146 & 1&0&2.35&5.44&1.000&@xmath116&0.0193&@xmath117&0.147&73.58&0.74&1.07&1.07 + cb_.2_.5_.5_b & 200 & 2&0&2.25&4.65&0.890&57.2&0.0021&152.4&0.169&67.50&0.73&1.11&1.11 + cb_.2_.5_.5_c&154 & 1&0&3.56&8.54&1.000&@xmath116&0.0328&@xmath117&0.378&85.71&0.49&1.31&1.30 + cb_.2_.5_.5_d & 115 & 1&0 & 2.70&5.92&1.000&@xmath116&0.0190&@xmath117&0.263&79.29&0.64&1.16&1.15 + cb_.2_.5_.5_e & 200 & 2&0&1.78&5.20&0.590&32.4&0.0104&185.7&0.155&62.50&0.67&1.15&1.15 + * @xmath95&163&1.4 & 0.0 & 2.53 & 5.95 & 0.896 & 52.2 & 0.0167 & 179.0 & 0.222 & 73.72 & 0.65 & 1.16 & 1.16 + & & & & & & & & & & & & & & + d & @xmath116&0.20&0.20&0.40&0.60&0.40&0.60&0.40&0.80&0.40&0.40&0.20&0.20&0.20 + prob&@xmath116&1.000&1.000&0.697&0.209&0.697&0.209&0.697&0.036&0.697&0.697&1.000&1.000&1.000 + & & & & & & & & & & & & & & + * *
|
most stars reside in binary / multiple star systems ; however , previous models of planet formation have studied growth of bodies orbiting an isolated single star . disk material has been observed around both components of some young close binary star systems .
additionally , it has been shown that if planets form at the right places within such disks , they can remain dynamically stable for very long times .
herein , we numerically simulate the late stages of terrestrial planet growth in circumbinary disks around ` close ' binary star systems with stellar separations 0.05 au @xmath0 0.4 au and binary eccentricities 0 @xmath1 0.8 . in each simulation ,
the sum of the masses of the two stars is 1 m@xmath2 , and giant planets are included .
the initial disk of planetary embryos is the same as that used for simulating the late stages of terrestrial planet formation within our solar system by chambers ( 2001 , making more terrestrial planets , icarus 152 , 205 - 224 ) , and around each individual component of the @xmath3 centauri ab binary star system by quintana et al .
( 2002 , terrestrial planet formation in the @xmath3 centauri system , astrophys .
j. 576 , 982 - 996 ) .
multiple simulations are performed for each binary star system under study , and our results are statistically compared to a set of planet formation simulations in the sun - jupiter - saturn system that begin with essentially the same initial disk of protoplanets .
the planetary systems formed around binaries with apastron distances @xmath4
@xmath5 @xmath6(1 + @xmath7 ) @xmath8 0.2 au are very similar to those around single stars , whereas those with larger maximum separations tend to be sparcer , with fewer planets , especially interior to 1 au .
we also provide formulae that can be used to scale results of planetary accretion simulations to various systems with different total stellar mass , disk sizes , and planetesimal masses and densities . _
keywords : _ planetary formation ; terrestrial planets ; extrasolar planets
|
at ultralow temperature phase separation of quantum gases has been realized experimentally , in particular in binary mixtures of bose - einstein condensates ( becs ) . for a survey ,
see , e.g. , refs .
early theoretical studies on trapped multicomponent becs @xcite appeared at the same time as experiments that witnessed weakly segregated dual becs @xcite .
ten more years were necessary to observe bec components that are strongly segregated @xcite but by now these are commonly studied experimentally @xcite by use of becs with different spin states , isotopes and species .
these experiments also revived the theoretical interest , much of which is focused on physical phenomena where the interface , separating the bec components , plays a key role .
different contributions in this context can be found in @xcite and a few very recent and relevant research activities concern the statics of interface characteristics @xcite , capillary wave dispersion relations @xcite , the kinetics @xcite and domain formation @xcite during phase segregation , and nambu - goldstone modes @xcite . in this paper
we build further upon earlier work @xcite devoted to the calculation of static interfacial properties of bec binary mixtures within gross - pitaevskii ( gp ) theory @xcite .
our main observation is that with a few new ideas and limited extra analytical work , a remarkable next step can be made towards physically transparent and powerful approximations that elucidate the physics of interfaces in soft condensed matter .
this paper is organized as follows .
section ii recapitulates concisely the frame - work of gp theory .
section iii defines the tpa model and illustrates its power in the context of the wall energy for a single condensate .
section iv presents the application of the tpa to the interfacial tension and discusses our main analytical expression , with emphasis on the strong segregation regime .
conclusions and an outlook are given in section v.
this section provides a summary of the theoretical frame - work given in more detail in section ii of @xcite .
consider the mean - field gross - pitaevskii lagrangian density @xcite @xmath3 with hamiltonian density @xmath4 + g_{12 } \lvert\psi_1\rvert^2 \lvert\psi_2\rvert^2.\ ] ] for atomic species @xmath5 , @xmath6 is the wave function of the condensate or order parameter " , @xmath7 the atomic mass , @xmath8 the repulsive intra - species interaction strength , @xmath9 the repulsive inter - species interaction strength and @xmath10 the @xmath11-wave scattering length . by introducing the dimensionless quantities , @xmath12 , with @xmath13 the healing length and @xmath14 the number density of condensate @xmath5 in bulk , @xmath15 , @xmath16 , and @xmath17 , where @xmath18 , and @xmath19 the chemical potential of condensate @xmath5 , we scale the lagrangian density in and hamiltonian density in to @xmath20 with @xmath21 + k \lvert\psi_1\rvert^2 \lvert\psi_2\rvert^2,\ ] ] where the pressure @xmath22 is given by @xmath23 , which is independent of the label @xmath5 provided the mixture is at bulk two - phase coexistence . we restrict our attention in this paper to this two - phase equilibrium .
next we make a transformation of the dimensionless lagrangian density by writing @xmath24 we then have a lagrangian density in terms of the new order parameters @xmath25 , @xmath26 - \hat{\mathcal{v } } ( \phi_1,\phi_2),\ ] ] in which the potential @xmath27 takes the form @xmath28 + k \lvert\phi_1\rvert^2 \lvert\phi_2\rvert^2.\ ] ] recall that for @xmath29 the two components are immiscible and a phase segregated bec forms @xcite .
it is well known that the euler - lagrange equations associated with @xmath30 yield the time - dependent gp equations .
these reduce to the time - independent gp equations ( tigpe ) when the order parameter @xmath31 is static , and these read @xmath32 which , for the given potential , implies @xmath33 \phi_{j } = 0 ; \;\ ; j=1,2 \ ; ( j \ne j').\ ] ] for describing a planar interface at @xmath34 , separating condensate 1 situated at @xmath35 from condensate 2 situated at @xmath36 , we consider order parameters that are translationally invariant in the @xmath37 and @xmath38 directions . for describing an interface the tigpe must be solved with the boundary conditions @xmath39 where @xmath40
. likewise , for describing a condensate , say 1 , adsorbed at an ( ideal ) optical wall @xcite at @xmath34 , we again consider an order parameter that is translationally invariant along @xmath37 and @xmath38 , and solve the tigpe with the boundary conditions @xmath41 finally , we recall the limit of strong segregation @xmath42 , which plays an important role in our work . in this limit
the segregation of the two condensates is complete .
the overlap of the order parameters becomes zero and it does so in such a way that the interaction term @xmath43 in the potential becomes negligible .
the gp equations decouple in this limit and the exact solution to the gp equations for the interface consists of two adjacent wall - like profiles @xcite @xmath44 ; \ ; j=1,2.\ ] ]
following up on the dpa introduced in @xcite we expand to second order ( harmonic approximation ) the quartic potential @xmath27 in about its two ( equal ) minima , which correspond to the bulk values for the order parameters .
moreover , and this defines the tpa , we _ also _ expand about its local maximum at the origin ( @xmath45 ) .
however , the height of the maximum , @xmath46 , is treated as an adjustable parameter , with respect to which the surface or interfacial energies can be minimized . in this section
we focus on a single - component bec with order parameter @xmath47 , and hence take @xmath48 . for simplicity of notation , we set @xmath49 , @xmath50 and @xmath51 . for obtaining the order parameter profile @xmath52 in the half - space @xmath53 , assuming a hard - wall boundary condition at @xmath34 , we make use of , on the one hand , the expansion ( to second order ) of the gp potential @xmath54 about bulk condensate 1 , which leads to @xcite @xmath55 and , on the other hand , the expansion about the origin in order parameter space , which leads to @xmath56 where @xmath57 is a matching value that is to be determined and @xmath46 is the adjustable height already discussed .
note that , for @xmath58 , we have @xmath59 @xcite .
furthermore , we require that @xmath60 be _ continuous _ at @xmath57 . with this requirement , and define the tpa _ model _ potential
. incidentally , it suffices for our purposes to work with a real order parameter , so the modulus signs can henceforth be dropped .
fig.1 shows the functions that constitute the three potentials , gp , dpa and tpa .
note that for a certain range of shifts " @xmath46 , there are _ two _ possible matching points for the tpa , where one can pass continuously from the inner ( tpa ) to the outer ( dpa ) branch ( fig.1a ) . for @xmath61
these two points merge into a single matching point , at which the branches are tangential , and @xmath62 ( fig.1b ) .
note that , as compared to the dpa potential , the tpa potential is much closer to the gp potential .
( color online ) shown are the dpa potential ( orange ; convex curve ) and the tpa branch ( blue ; concave curve ) for the choice @xmath63 for the shift . for this choice
two intersections appear .
a possible tpa potential then consists of the tpa branch from @xmath64 to either one of the two intersections and then switches to the dpa branch . , scaledwidth=80.0% ] ( color online ) shown are the dpa potential ( orange ; convex curve ) , the tpa branch ( blue ; concave curve ) and the gp potential ( green ) for the special choice @xmath65 , for which the tpa branch is tangent to the dpa curve in the degenerate intersection point ( black dot ) at @xmath66 .
the tpa potential follows the tpa branch from @xmath64 to @xmath66 and then switches ( smoothly ) to the dpa potential .
, scaledwidth=80.0% ] using the tpa potential(s ) and solving the el equations with the boundary conditions is a simple exercise , which leads to the following order parameter profile , which is physically acceptable provided @xmath67 , i.e. , @xmath68 , @xmath69 with @xmath70 a matching point defined through @xmath71 .
the order parameter @xmath72 is , of course , required to be continuous at @xmath70 .
interestingly , the continuity of @xmath73 at @xmath57 implies that also the first derivative of the order parameter is continuous .
this follows from insisting that a unique first integral of the el equations , or constant of the motion " describes the entire trajectory ( for more detail and a more general calculation , see section iv ) .
this is appropriate for equilibrium profiles , whereas for constrained ( non - equilibrium ) profiles , not considered in this work , the constant of the motion is in general discontinuous at the matching point @xcite .
there is still a free parameter in the problem , being @xmath70 , because any choice of @xmath67 provides at least one ( and at most two ) values for @xmath70 .
interestingly , an optimal matching point @xmath70 can be determined by minimizing the wall energy with respect to this parameter .
the wall energy ( or wall surface tension ) is defined as the excess energy per unit area relative to the energy of a spatially uniform ( bulk ) order parameter . within the tpa model
the wall energy can be obtained as follows .
starting from the grand potential for a condensate confined to the half - space @xmath74 , @xmath75 = 2 p_0 \xi a \int_0^{\infty } \mathrm{d } \rho \left [ \left(\partial_{\rho } \phi \right)^2 + \hat{\mathcal{v}}_{\mathrm{tpa } } ( \phi,0)\right],\ ] ] where @xmath76 is the wall area , and using the constant of the motion " @xcite , @xmath77 we substitute , for order parameter profiles that satisfy the tigpe , in , to obtain the wall energy within the tpa model , @xmath78 + p_0v}{a } = 4 p_0 \xi \left ( \int_0^{\lambda } + \int_{\lambda}^{\infty}\right ) \mathrm{d } \rho \left(\partial_{\rho } \phi \right)^2.\ ] ] the integrals are readily performed and the result is the following function of @xmath70 , @xmath79 with @xmath80 .
we can now optimize the parameter @xmath70 by minimizing the wall energy with respect to it .
this leads to the simple result @xmath81 this in turn implies a simple matching value of the order parameter , @xmath82 remarkably , this minimum is achieved when the two branches of the tpa potential become tangent in a degenerate point of intersection ( see fig.1b ) .
this happens for @xmath61 .
this tangency implies coincident derivatives , which in turn implies , through the el equations , that also _
the second derivative _ of the order parameter is continuous at the matching point .
the tpa order parameter profile is very smooth .
we observe that , in a mechanical analogy of our model @xcite , the order parameter is the position of a particle moving in a potential @xmath83 , so that at the matching time " its position , its velocity and its acceleration are continuous ( as is the force acting on it ) . for this optimal matching point , the reduced wall energy takes the value @xmath84 which is to be compared with the dpa approximation @xmath85 = 2.82842 ... @xcite , and the exact value in gp theory , @xmath86 = 1.88561 ... fig.2 allows one to compare qualitatively the order parameter profiles for a single condensate at a hard wall in the dpa and tpa models , against the exact solution of the full gp theory , which for this simple case is a tanh profile .
notice that the tpa leads to an overall improvement over the dpa , and is especially better near the wall , as expected , because the order parameter is small there .
its initial slope ( 0.816 ) is close to the exact one ( gp theory ; slope 0.707 ) , whereas the initial slope predicted by the dpa model ( 1.414 ) is ( much ) larger .
( color online ) order parameter profiles for a single condensate at a hard wall . shown
are the ( exact ) gp profile ( green ) given by the analytic solution for @xmath87 , the dpa ( orange curve ) and the tpa ( blue curve ) given by .
the tpa profile features a matching point ( black dot ) , at @xmath88 , corresponding to the degenerate intersection of the two potentials shown in fig.1b . at this matching point
the order parameter , its first and its second derivative are continuous . , scaledwidth=80.0% ]
in this section we apply the approach developed for a single condensate in the previous section , to the binary mixture with order parameters @xmath89 and @xmath90 .
we aim at deriving the profiles for the interface between the two becs and at calculating the interfacial tension . for obtaining the ( real ) order parameter profiles , subject to the boundary conditions , we make use of , on the one hand , the expansions ( to second order ) of the gp potential about bulk condensates 1 and 2 , for @xmath53 and @xmath91 , respectively .
this leads to @xcite , @xmath92 on the other hand , we also make use of the expansion about the origin in order parameter space , which leads to @xmath93 where @xmath94 and @xmath95 are two matching values that satisfy @xmath96 and are to be determined .
besides these three potential sheets , we need two more .
indeed , we also need mixed " potentials , to be used where one order parameter , but not the other , crosses over from one regime to the next . in this way we can guarantee continuity of @xmath47 , @xmath97 and @xmath73 everywhere , also
when as is generally the case the matching points for @xmath47 and @xmath97 lie at different positions , i.e. , different values of @xmath98 .
these mixed potentials read @xmath99 with the continuity and symmetry requirements @xmath100 and @xmath101 , respectively .
note that the potential sheet defined in coincides with that of the dpa model @xcite .
furthermore , we require that @xmath102 be _ continuous _ at @xmath103 and @xmath104 in both order parameters . with these requirements , , and define the tpa _ model _ potential for the binary mixture
. incidentally , it suffices for our purposes to work with real order parameters , so the modulus signs can henceforth be dropped .
the triple - parabola - approximated gp equations are obtained by replacing the potential @xmath27 in the gp equations by the potential @xmath73 , which leads to the following three sets of tigpe equations for the tpa model , @xmath105 @xmath106 @xmath107 it is instructive to give explicitly the first integrals of these equations of motion , the so - called constants of the motion .
these are , in the respective regimes , @xmath108 @xmath109 @xmath110 with , by symmetry , @xmath111 .
these constants of the motion can be complemented by an equation which expresses the analogue of the conservation of total mechanical energy @xmath112 .
we have @xmath113 , with @xmath114 the kinetic energy of the particle in the mechanical analogy .
the value of @xmath112 is obtained as minus the value that @xmath73 assumes in either one of the bulk phases , being minus @xmath115 .
we now assume , as is appropriate for equilibrium profiles ( but not for non - equilibrium ones @xcite ) , that one and the same constant @xmath112 applies to the entire trajectory . then the following useful identity holds , covering all the regimes of the tpa encountered in the half - space @xmath87 ( the half - space @xmath91 is covered by interchanging the subscripts 1 and 2 ) , @xmath116 with @xmath117 , where @xmath46 has been introduced in .
we now solve the tpa for the equations of motion in each regime and point out some properties of the solutions .
for concreteness and to alleviate the notation we focus on the half - space @xmath118 .
the solutions in the other half - space can be obtained by a judicious interchange of labels and/or signs , since the equations describing condensate 1 and 2 are the same when the scaled coordinates @xmath119 and @xmath120 are used .
close to the bulk phase ( 1,0 ) the solutions are the previously established dpa forms @xcite , @xmath121 where , in line with our previous considerations , the matching points satisfy @xmath122 and @xmath123 .
close to the origin ( 0,0 ) in order parameter space , the solutions within the tpa are , @xmath124 where we have used the symmetry of the solutions and the interface midpoint property @xmath125 .
we now turn to simple arguments for determining the matching values @xmath94 and @xmath95 for the order parameters . from the constants of the motion we infer , for the matching of @xmath47 , @xmath126 which has a degenerate solution @xmath127 at @xmath128 , while for @xmath129 there are two solutions
. we will be interested in the solution that minimizes the grand potential ( see further ) .
next , for the matching of @xmath97 we obtain , again by using a constant of the motion , the simple result @xmath130 expressing the continuity of the order parameters and their derivatives at the matching points provides us now with the following solutions for the amplitudes , @xmath131 @xmath132 with @xmath133 , and also @xmath134 @xmath135 with @xmath136 , so that @xmath137 furthermore , @xmath138
@xmath139 the parameter @xmath140 plays a special role .
it is a free parameter in the model . assuming @xmath141 , from the tpa solutions it is easily seen that ( for finite @xmath2 ) the second derivative of @xmath97 changes sign at @xmath142 .
likewise the second derivative of @xmath47 changes sign at @xmath143 . for comparison , in the gp theory the second derivatives of the profiles pass ( smoothly ) through zero at points that are in general different from the midpoint of the interface ( which lies at @xmath34 ) .
exceptions are @xmath144 ( solvable exactly @xcite ) and , evidently , @xmath145 , for which the two inflection points coincide with the midpoint ( where the profiles intersect ) .
in contrast , within the dpa the inflection points always coincide with the midpoint .
the tpa is thus qualitatively different from the dpa in that it allows for the possibility that the inflections points differ from the midpoint .
the tpa shares this property with the gp theory .
however , the value of @xmath140 is arbitrary .
we shall see ( in the next subsection ) that , since the order parameters are decoupled , the interfacial tension within the tpa is _ independent _ of the choice of @xmath140 . indeed , the grand potential only depends on the sum @xmath146 .
we consider it therefore most appropriate to choose @xmath147 such that _ the interface midpoints of the dpa and tpa coincide_. this allows one to compare the interfacial profiles in the two models without bias . in view of these considerations
we have , @xmath148 and @xmath149 where the latter is the result of the dpa @xcite .
we now proceed to study the interfacial tension variationally .
this will allow us to obtain unique values for @xmath150 , @xmath94 and @xmath95 . using the constants of the motion we can calculate the interfacial tension within the tpa in terms of the following contributions , @xmath151 which remain after the
bulk grand potential @xmath152 of a homogeneous phase has been subtracted from the grand potential @xmath153 of the configuration consisting of the two condensates and the interface in between them , i.e. , @xmath154 = 2 p_0 a \int_{-\infty}^{\infty } \mathrm{d } z \left [ \left(\partial_{\rho_1 } \phi_1 \right)^2 + \left(\partial_{\rho_2 } \phi_2 \right)^2 + \hat{\mathcal{v}}_{\mathrm{tpa } } ( \phi_1,\phi_2)\right],\ ] ] note that , at bulk two - phase coexistence , @xmath155 is the same for the two condensates . taking into account the different regimes encountered in the tpa and using the respective constants of the motion , we can rewrite as @xmath156 again using the constants of the motion this
can be converted into integrals over @xmath47 and @xmath97 , with the result @xmath157 + \sqrt{2 } ( 1-\phi_{+}^{\times } ) \right ) ( \xi_1+\xi_2),\ ] ] with @xmath158 .
it is easily verified that @xmath159 reaches a local minimum at @xmath127 , precisely where @xmath150 has its minimum ( and @xmath160 ) .
this local minimum _ defines _ the tpa .
( color online ) trajectory plots in the order parameter plane for @xmath161 and @xmath162 ( symmetric case ; solid blue line ) , and for @xmath161 and @xmath163 ( asymmetric case ; dashed blue line ) .
the trajectories represent an interface that connects the bulk phases at ( 1,0 ) and ( 0,1 ) .
the contours of constant potential @xmath27 ( grey lines ) are shown for the potential sheets that are relevant in the different sectors visited by the trajectory .
close to the bulk points the potential is fully determined by the dpa appropriate for the bulk phase in the immediate vicinity of the trajectory , dpa(1,0 ) or dpa(0,1 ) , and is given in .
when the trajectory traverses the black line to enter the square defined by @xmath164 , @xmath165 , the relevant potential is of mixed character , mix1 or mix2 , and is given in .
next , closer to the interface midpoint , when the trajectory enters the smallest square defined by @xmath166 , @xmath165 , the relevant potential is of pure tpa character , as defined in , _ provided _ the trajectory crosses the red ( south or west ) borders of the square ( as it does in the figure ) .
the contour lines of the pure tpa potential sheet are segments of circles ( green lines ) and are rather far apart . for hypothetical trajectories that would enter the small square through the black ( north or east ) borders ,
the relevant potential remains the mixed one .
the contour lines of the mixed potentials ( grey ) are more closely spaced .
note that the relevant potentials are continuous across the ( red or black ) borders , as can be seen by inspecting the contours .
, scaledwidth=65.0% ] we can now obtain the interface profiles as well as the interfacial tension uniquely .
we first discuss the interface profiles . to this end
we illustrate how the tpa works by showing examples of interface trajectories in the order parameter plane .
subsequently , we show the corresponding interface profiles themselves . fig.3 presents the interface trajectory in the space of the two order parameters for @xmath161 ( fairly strong segregation ) , for the symmetric case @xmath167 and for the asymmetric case @xmath168 .
also shown in the figure are the contours of constant potential and the various sectors ( triangles and squares ) corresponding to different potential sheets encountered by the trajectory .
fig.4 illustrates the interface trajectory for @xmath144 ( intermediate segregation ) , for the symmetric case @xmath167 and for the asymmetric case @xmath168 .
note that for the symmetric case the trajectory coincides with the antidiagonal of the quadrant .
this perfect symmetry ( @xmath169 ) is a property which the dpa shares with the exact solution of the gp equations for @xmath144 @xcite .
( color online ) trajectory plots in the order parameter plane for @xmath144 and @xmath162 ( symmetric case ; solid blue line ) , and for @xmath144 and @xmath163 ( asymmetric case ; dashed blue line ) .
the description of the figure is similar to that of fig.3 and we refer to that figure for details . , scaledwidth=70.0% ] fig.5a shows the interface profiles corresponding to @xmath161 and @xmath167 , while fig.5b presents the interface profiles corresponding to @xmath161 and @xmath168 . in each figure
, three profiles are displayed : the numerical solution to the gp equations , the dpa and the tpa .
as we described , the tpa profiles are characterized by weak singularities at the matching points ( black dots ) . at the upper matching points the order parameter and its first and second derivatives are continuous , whereas at the lower matching points the second derivative is discontinuous .
note that the tpa offers a modest improvement over the dpa .
recall that , by construction , the interface midpoints within the dpa and the tpa coincide .
likewise , fig.6a shows the interface profiles corresponding to @xmath144 and @xmath167 , while fig.6b presents the interface profiles corresponding to @xmath144 and @xmath168 . for this case of intermediate segregation the dpa and the tpa are almost equivalent .
recall that the exact solution to the gp equations , for @xmath144 and @xmath167 , takes a very simple tanh form @xcite and that the midpoint lies at @xmath170 , a property which is also respected by the dpa ( and tpa ) .
( color online ) interface profiles for @xmath161 and @xmath167 ( symmetric case ) .
the numerically exact solution within gp theory is shown ( solid green and dashed green lines ) , together with the dpa solutions ( solid orange and dashed orange lines ) and the tpa solutions ( solid blue and dashed blue lines ) . the latter feature ( weak )
singularities at the matching points .
, scaledwidth=80.0% ] ( color online ) interface profiles for @xmath161 and @xmath168 ( an asymmetric case ) . ,
scaledwidth=80.0% ] ( color online ) interface profiles for @xmath144 and @xmath167 ( symmetric case ) .
the analytically known exact solution within gp theory is shown ( solid green and dashed green lines ) , together with the dpa solutions ( solid orange and dashed orange lines ) and the tpa solutions ( solid blue and dashed blue lines ) .
the latter feature ( weak ) singularities at the matching points . , scaledwidth=80.0% ] ( color online ) interface profiles for @xmath144 and @xmath168 ( an asymmetric case ) .
, scaledwidth=80.0% ] next we return to the interfacial tension . substituting into the special parameter values we determined , we arrive at our main result @xmath171 this expression can be used for @xmath172 . indeed , for @xmath173 the interface midpoint value approaches the matching value @xmath127 and the matching position @xmath174 approaches the midpoint position @xmath34 .
fig.8 shows the interfacial tension versus @xmath175 in gp theory and in the dpa and tpa models .
we can obtain @xmath146 explicitly as a function of @xmath2 using , which leads to @xmath176 this function is displayed in fig.7 .
we recall that only the sum @xmath146 is relevant for the interfacial tension , while shifts of the individual matching point positions @xmath174 and @xmath140 affect the profiles but not the interfacial tension as long as their sum is invariant .
( color online ) dependence of the matching point position sum @xmath146 on the inverse of the interaction parameter @xmath2 .
for @xmath177 this sum takes the value @xmath178 and it reaches zero at @xmath179 , where the applicability of the tpa ends . the individual matching point positions @xmath174 and @xmath140 can both be determined numerically through the requirement that the interface midpoint value of the order parameters be identical in dpa and tpa . ,
scaledwidth=80.0% ] the vanishing of @xmath180 at @xmath179 signifies that in the interval @xmath181 the tpa reduces simply to the dpa .
recall that @xmath29 is a necessary condition for phase segregation of the condensates .
for @xmath182 the tpa model becomes relevant and leads to an interfacial tension that is lower than that in the dpa model .
note that in the strong segregation limit ( @xmath42 ) we recover , as we should , the sum of two wall tensions , @xmath183 with the wall tension as obtained in . for comparison
, we recall the result for the interfacial tension in the dpa model @xcite , @xmath184 ( color online ) reduced interfacial tension versus the inverse of the interaction parameter @xmath2 . shown
are the numerically exact result within gp theory ( lower curve ; black ) , the dpa result ( upper curve ; blue ) and the tpa refinement ( middle curve ; red ) applicable for @xmath185 .
the respective values in the strong segregation limit ( @xmath177 ) are indicated at the arrows
. also indicated is the point where the tpa curve intersects the dpa curve at @xmath179 ( with unequal slopes ) . for @xmath186 ( weak segregation regime ) the interface trajectory does not visit the pure tpa sector ( cf .
fig.3 or fig.4 ) .
, scaledwidth=80.0% ]
in this work we have introduced and applied an analytic approach , the tpa , that offers a useful improvement over the well - known dpa strategy . in the tpa , harmonic approximations to a potential are not only invoked near the minima or fixed points , but also near a local maximum , in as far as the smooth curvature near this maximum , but not its height , are respected .
this is a meaningful approach , because in contrast to the value in the minima ( equilibrium bulk solutions ) , the value in the maximum is in itself not of physical significance . while the dpa is a widely applied approximation in the context of variational problems in particle mechanics and statistical mechanics , we are not aware of any previous study introducing or employing a tpa .
quite generally , the tpa allows one to improve over the dpa while still obtaining analytical results with modest calculational effort .
* for the wall energy the tpa result constitutes a significant quantitative improvement over the dpa result .
moreover , the variational calculation intrinsic to the tpa leads to an order parameter profile with remarkable smoothness ( at the matching point it is continuous and possesses continuous first and second derivatives ) .
furthermore , the initial slope of the order parameter ( at the wall ) within the tpa is also much improved with respect to the dpa . * for the interfacial profiles and the interfacial tension the tpa offers a modest but interesting improvement over the dpa results . in the strong segregration regime
the improvement for the interfacial tension is significant , simply because the wall energies are ( much ) more accurate in the tpa . in the weak segregation limit the tpa leads to the same interface structure and interfacial tension as the dpa , simply because tpa and dpa coincide for @xmath187 . for intermediate and strong segregation ,
say @xmath188 , the tpa becomes worthwhile and offers a significant improvement over the dpa for profiles and interfacial tension .
furthermore , the tpa shares the qualitatively important feature with the ( exact ) gp theory that the inflection points of the profiles need not coincide with the midpoint of the interface . *
the analytic expression for the interfacial tension within tpa our main result is of a remarkable simplicity , and provides a useful and compact refinement of the dpa result , especially for strong segregation .
n.v.t is supported by the vietnam national foundation for science and technology development ( nafosted ) j.o.i . and c .- y.l . by fwo flanders under grant nr . fwo.103.2013.09 within the framework of the fwo - nafosted cooperation . j.o.i . and c .- y.l
. are furthermore supported by ku leuven grant ot/11/063 .
is also funded by ministry of education and training of vietnam .
the authors thank phat tran huu , xintian wu , wenan guo and jesper koning for discussions .
d. m. stamper - kurn and w. ketterle , spinor condensates and light scattering from bose - einstein condensates , in coherent atomic matter waves , edited by r. kaiser , c. westbrook , and f. david , proceedings of the les houches summer school of theoretical physics , lxxii , 1999 ( springer , newyork , 2001 ) , b. malomed , in emergent nonlinear phenomena in bose- einstein condensates , edited by p. g. kevrekidis , d. j. frantzeskakis , and r. carretero - gonzalez ( springer - verlag , berlin , 2008 ) , chap .
15 , p. 287 ; d. s. hall , in emergent nonlinear phenomena in bose - einstein condensates , edited by p. g.kevrekidis , d. j. frantzeskakis , and r. carretero - gonzalez ( springer - verlag , berlin , 2008 ) , chap .
16 , p. 307 .
|
accurate and useful analytic approximations are developed for order parameter profiles and interfacial tensions of phase - separated binary mixtures of bose - einstein condensates .
the pure condensates 1 and 2 , each of which contains a particular species of atoms , feature healing lengths @xmath0 and @xmath1 .
the inter - atomic interactions are repulsive .
in particular , the effective inter - species repulsive interaction strength is @xmath2 . a triple - parabola approximation ( tpa )
is proposed , to represent closely the energy density featured in gross - pitaevskii ( gp ) theory .
this tpa allows us to define a model , which is a handy alternative to the full gp theory , while still possessing a simple analytic solution .
the tpa offers a significant improvement over the recently introduced double - parabola approximation ( dpa ) . in particular , a more accurate amplitude for the wall energy ( of a single condensate ) is derived and , importantly , a more correct expression for the interfacial tension ( of two condensates ) is obtained , which describes better its dependence on @xmath2 in the strong segregation regime , while also the interface profiles undergo a qualitative improvement .
|
icecube is a kilometer - cube scale cherenkov detector at the geographical south pole , designed to search for muons from high energy neutrino interactions .
the arrival directions and energy information of these muons can be used to search for point sources of astrophysical neutrinos , one of the primary goals of icecube .
the main component of icecube is an array of optical sensors deployed in the glacial ice at depths between 1450 m and 2450 m. these digital optical modules ( doms ) , each containing a 25 cm diameter photomultiplier tube with accompanying electronics within a pressure housing , are lowered into the ice along `` strings . ''
there are currently 59 strings deployed of 86 planned ; the data analyzed here were taken in a 40 string configuration , which was in operation between april 2008 and april 2009 .
there are 13 lunar months of data within that time . in this analysis we present results from 8 lunar months of the 40 string configuration . for a muon with energy on the order of a tev
, icecube can reconstruct an arrival direction with order @xmath0 accuracy .
for down - going directions , the vast majority of the detected muons do not originate from neutrino interactions , but from high energy cosmic ray interactions in the atmosphere .
these cosmic ray muons are the dominant background in the search for astrophysical neutrinos .
they can also be used to study the performance of our detector . in particular
, we can verify the pointing capability by studying the shadow of the moon in cosmic ray muons . as the earth travels through the interstellar medium , the moon blocks some cosmic rays from reaching the earth .
thus , when other cosmic rays shower in the earth s atmosphere and create muons , there is a relative deficit of muons from the direction of the moon .
icecube detects these muons , not the primary cosmic rays .
since the position and size of the moon is so well known , the resulting deficit can be used for detector calibration .
the idea of a moon shadow was first proposed in 1957 @xcite , and has become an established observation for a number of astroparticle physics experiments ; some examples are given in references @xcite .
experiments have used the moon shadow to calibrate detector angular resolution and pointing accuracy @xcite .
they have also observed the shift of the moon shadow due to the earth s magnetic field @xcite .
the analysis described here is optimized for a first observation , and does not yet include detailed studies such as describing the shape of the observed deficit .
these will be addressed in future studies .
data transfer from the south pole is limited by the bandwidth of two satellites ; thus , not all downgoing muon events can be immediately transmitted .
this analysis uses a dedicated online event selection , choosing events with a minimum quality and a reconstructed direction within a window of acceptance around the direction of the moon .
the reconstruction used for the online event selection is a single ( i.e. , not iterated ) log - likelihood fit .
the online event selection is defined as follows , where @xmath1 denotes the moon declination : * the moon must be at least @xmath2 above the horizon . * at least 12 doms must register each event .
* at least 3 strings must contain hit doms .
* the reconstructed direction must be within 10@xmath3 of the moon in declination .
* the reconstructed direction must be within 40@xmath4 of the moon in right ascension ; the @xmath5 factor corrects for projection effects .
these events are then sent via satellite to the northern hemisphere for further processing , including running the higher - quality 32-iteration log - likelihood reconstruction used in further analysis .
the moon reached a maximum altitude of @xmath6 above the horizon ( @xmath7 ) in 2008 , when viewed from the icecube detector .
the trigger rate from cosmic ray muons is more than 1.2 khz in the 40 string configuration , but most of those muons travel nearly vertically , and thus they can not have come from directions near the moon .
only @xmath8 of all muons that trigger the detector come from angles less than @xmath9 above the horizon .
furthermore , muons which are closer to horizontal ( and thus closer to the moon ) must travel farther before reaching the detector .
they need a minimum energy to reach this far ( see fig . [ ecrm ] ) : the cosmic ray primaries which produce them must have energies of at least 2 tev .
.,width=288 ] three typical months of data are shown in fig .
[ livetime ] , along with the position of the moon above the horizon .
the dominant shape is from the strong increase in muon flux with increasing angle above the horizon : as the moon rises , so do the event rates near the moon .
this can be seen clearly in the correlation between the two sets of curves .
there is a secondary effect from the layout of the 40 strings .
one dimension of the detector layout has the full width ( approximately 1 km ) of the completed detector , while the other is only about half as long .
when the moon is aligned with the short axis , fewer events pass the filter requirements .
this causes the 12 hour modulation in the rate .
a simulated data sample of @xmath10 downgoing muon events was generated using corsika @xcite . between the true and reconstructed track .
the y - axis shows the fraction of events with this or lower angular error .
the blue curve shows the event sample after offline event selection , and the red curve shows the event sample after online event selection.,width=288 ] is plotted here .
the deviations of the central 9 bins are shown in red .
the surrounding bins are shown with a black line histogram , and fit with a gaussian curve.,width=288 ] a set of cuts was developed using the following estimated relation between the significance @xmath11 , the efficiency @xmath12 of events passing the cut , and the resulting median angular resolution @xmath13 of the sample : @xmath14 since the deficit is based on high statistics of events in the search bin , this function provides a good estimator for optimizing the significance .
the following cuts were chosen : * at least 6 doms are hit with light that hasnt been scattered in the ice , allowing a -15 nsec to + 75 nsec window from some minimal scattering .
* projected onto the reconstructed track , two of those hits at least 400 meters apart . *
the @xmath15 estimated error ellipse on the reconstructed direction has a mean radius less than @xmath16 .
the cumulative point spread function of the sample after the above quality cuts is shown as the blue line in fig .
[ psf ] .
the size @xmath17 of the search bin is optimized for a maximally significant observation using a similar @xmath18-error based argument and the resulting relation , which follows . using the cumulative point spread function of the sample after quality cuts
, we have : @xmath19 maximizing this significance estimator gives an optimal search bin radius of @xmath20 .
this analysis uses square bins with an area equal to that of the optimized round bin , with side length @xmath21 . to show that the data are stable in right ascension @xmath22
, we show , in fig .
[ rawdata ] , the number of events in the central declination band .
the errors shown are @xmath18 .
the average of all bins excluding the moon bin is 27747 , which is plotted as a line to guide the eye .
the moon bin has 852 events below this simple null estimate .
this represents a @xmath23 deficit using @xmath18 errors .
although this shows that the data are stable , this error system is vulnerable to variations in small data samples .
although we do nt see such variations here , we considered it prudent to consider an error system which takes into account the size of the background sample .
we used a standard formula from li and ma @xcite for calculating the significance of a point source : @xmath24 where @xmath25 is the number of events in the signal sample , @xmath26 is the number of events in the off - source region , and @xmath22 is the ratio between observing times on- to off - source .
we take @xmath22 instead as the ratio of on- to off - source areas observed , since the times are equal .
the above significance formula is applied to the moon data sample in the following way .
the data are first plotted in the standard moon - centered equatorial coordinates , correcting for projection effects with a factor of @xmath5 .
the plot is binned using the @xmath27 bin size optimized in the simulation study .
each bin successively is considered as an on - source region .
there is a very strong declination dependence in the downgoing muon flux , so variations of the order of the moon deficit are only detectable in right ascension .
thus , off - source regions are selected within the same zenith band as the on - source region .
twenty off - source bins are used for each calculation : ten to the right and ten to the left of the on - source region , starting at the third bin out from the on - source bin ( i.e. , skipping two bins in between ) .
for a region of 7 bins or @xmath28 in declination @xmath1 and 23 bins or @xmath29 in right ascension @xmath22 around the moon , the significance of the deviation of the count rate in each bin with respect to its off - source region was calculated , as described in section [ section : significance ] .
the result is plotted in fig .
[ signif ] .
the moon can be seen as the @xmath30 deficit in the central bin , at @xmath31 . to test the hypothesis
that the fluctuations in the background away from the moon are distributed randomly around 0 , we plot them in fig .
the central 9 bins , including the moon bin , are not included in the gaussian fit , but are plotted as the lower , shaded histogram .
the width of the gaussian fit is consistent with 1 ; therefore , the background is consistent with random fluctuations .
icecube has observed the shadow of the moon as a @xmath30 deviation from event counts in nearby regions , using data from 8 of the total 13 lunar months in the data taking period with the 40-string detector setup . from this , we can conclude that icecube has no systematic pointing error larger than the search bin , @xmath21 . in the future
, this analysis will be extended in many ways .
first , we will include all data from the 40 string detector configuration .
we hope to repeat this analysis using unbinned likelihood methods , and to describe the size , shape , and any offset of the moon shadow .
we will then use the results of these studies to comment in more detail on the angular resolution of various reconstruction algorithms within icecube .
this analysis is one of the only end - to - end checks of icecube systematics based only on experimental data .
the tibet as gamma collaboration , m. amenomori , , _ multi - tev gamma - ray observation from the crab nebula using the tibet - iii air shower array finely tuned by the cosmic - ray moon s shadow _ , arxiv : astro - ph/0810.3757v1
|
in the absence of an astrophysical standard candle , icecube can study the deficit of cosmic rays from the direction of the moon .
the observation of this `` moon shadow '' in the downgoing muon flux is an experimental verification of the absolute pointing accuracy and the angular resolution of the detector with respect to energetic muons passing through .
the moon shadow has been observed in the 40-string configuration of icecube .
this is the first stage of icecube in which a moon shadow analysis has been successful .
method , results , and some systematic error studies will be discussed .
icecube , moon shadow , pointing capability
|
this is the first of two lectures on a new technique for calculating the temperature equilibration rate between electrons and ions in a weakly to moderately coupled fully - ionized plasma , exact to leading and next - to - leading order in the plasma number density .
this calculation was first performed in section 12 of ref .
@xcite , a work whose primary focus was the charged particle stopping power in a plasma .
this paper assumed familiarity with a number of field theory concepts , and section 12 relied heavily on the charged particle stopping power results derived in previous sections of that work .
in contrast , these lectures will be self - contained .
i will either derive or motivate the requisite field theory background for a complete reading of ref .
@xcite , with an emphasis on some of the subtleties of the calculational techniques and the concepts behind them ( this lecture ) .
more to the point , since the rate calculation stands on its own , it should be presented on its own ( the following lecture ) . indeed ,
since the calculation of the rate is somewhat less involved than that of the stopping power , it more clearly illustrates the tools and concepts imported from field theory .
in addition to clarifying the method of dimensional continuation , this lecture will place ref .
@xcite in the context of more familiar and traditional approaches to the rate problem .
in particular , i will show that dimensional continuation can be viewed as a systematic implementation of the approach based on convergent kinetic equations .
finally , in in the next lecture , i will go on to derive the main result from section 12 of ref .
@xcite , the rate coefficient ( [ bpsrate ] ) of this lecture . by working in the born approximation , and adopting the methods of ref .
@xcite , i will derive this result in a much simpler manner than originally presented in ref .
@xcite . while lectures i and ii are self - contained , they are complementary and should be read as a unit . ) of this lecture .
this would provide two independent calculations of the rate ( [ bpsrate ] ) , but alas , time does not permit these last three supplements . ] the strategy employed by ref .
@xcite , hereafter referred to as bps , consists of two steps .
first , we will find a dimensionless parameter in which to perform a _ controlled _ perturbative expansion of the rate , expanding to leading and next - to - leading order in this parameter .
second , we will deploy a technique from quantum field theory that will allow us to calculate the coefficients of these leading and subleading terms _ exactly_. the exact leading order term is not very difficult to find , and has been known since the classic work of spitzer .
the next - to - leading order term , on the other hand , was not known exactly until the recent bps calculation .
the third - order term provides an estimate of the error of the calculation . when the plasma is weakly to moderately coupled
, the error will be small and the rate will be approximated quite accurately by the first two terms of this expansion . to calculate the expansion coefficients , we will exploit a field theory technique known as dimensional regularization ( or dimensional continuation , as i will call it here ) .
this application of dimensional continuation is quite different from its intended purpose in the renormalization procedure of quantum field theory .
dimensional continuation was originally developed as an elegant regularization scheme in which the fundamental symmetries of a field theory could be maintained while still rendering finite the otherwise infinite integrals that arise when calculating feynman diagrams .
i will show how this technique can be used in a novel fashion to extract next - to - leading order physics that has , until now , remained inaccessible .
in other words , i will show how dimensional continuation provides an exact result for the corresponding coulomb logarithm of the process in question .
i will also take the opportunity to correct a small algebra mistake for the electron - ion equilibration rate presented in section 12 of ref .
@xcite .
the general formalism starts with a plasma composed of multiple species labeled by an index @xmath5 , the various species being delineated by of a common electric charge @xmath6 and a common mass @xmath7 . with @xmath8 fixed .
while this simplifying assumption has its limitations , it facilitates the analytic calculation of the stopping power and temperature equilibration rate .
however , one can take the @xmath8 to be fractional to mock - up ionization in a simple manner . or better yet , it should be possible to combine the bps results with models of the ionization effects . for a hot low-@xmath9 plasmas , such as a deuterium - tritium plasma during icf ignition ,
the ions are likely to be fully ionized in any event , so this is not a serious restriction for clean thermonuclear burn with low-@xmath9 impurities . ]
each species is assumed to be in thermal equilibrium with itself at temperature @xmath10 with a spatially uniform number density @xmath11 .
i will drop the subscript on the charge of the electron and write @xmath12 ( with @xmath13 ) , although the electron mass will be denoted by @xmath14 , and the number density and temperature of the electron plasma component by @xmath15 and @xmath16 respectively .
i will use a lower case subscript @xmath17 to denote a single ion species of charge @xmath18 , mass @xmath19 , number density @xmath20 , and temperature @xmath21 .
i will employ a capital - i subscript to denote properties that correspond to the collective set of ions , such as the total ion number density @xmath22 or a common ion temperature @xmath23 .
when equilibrium distributions are required in calculations , they are assumed to be maxwell - boltzmann , although a generalization to fermi - dirac statistics can be accomplished with more effort @xcite .
for problems involving hot thermonuclear burn , however , the fugacity is small and maxwell - boltzmann statistics is an accurate approximation .
let @xmath24 denote rate at which the energy density of plasma species @xmath25 changes because of its coulomb interactions with another species @xmath5 ( the rate _ from _
the @xmath25-species _ to _ the @xmath5-species ) .
this rate is proportional to the temperature difference , and can be expressed by @xmath26 the sign convention in ( [ dedtei ] ) implies that when the rate coefficients @xmath27 are positive , then energy will flow from the hotter species to the cooler species , as it must . section 12 of bps used dimensional continuation to calculate the general rate coefficients @xmath27 in a weakly coupled , but otherwise arbitrary , plasma .
for simplicity , we will not perform the general calculation until lecture iii . in this and the following lecture ,
we will concentrate on the energy exchange between electrons and ions only .
since the electron mass @xmath14 is so much smaller than a typical ion mass @xmath28 , the electrons will come into equilibrium first with temperature @xmath16 on some time scale @xmath29 .
the energy transfer rate among ions is a factor @xmath30 slower than the corresponding rate for electrons , and therefore the ions will equilibrate to a common temperature @xmath23 in a time @xmath31 .
finally , as the electrons and ions exchange energy through coulomb interactions , these systems too will equilibrate on a time scale .
consequently , one finds a hierarchy of time scales , and it indeed makes sense to consider the electron and ion systems as having distinct temperatures @xmath16 and @xmath23 , with subsequent equilibration between them . taking @xmath32 and @xmath33 in ( [ dedtei ] ) , and since the ions have a common temperature @xmath34 , the rate equation of interest is obtained by summing over the ion components of ( [ dedtei ] ) to give @xmath35 where @xmath36 and @xmath37 . for reasons to be discussed shortly ,
rationalized units are preferred for dimensional continuation , and i will employ this choice in all that follows .
nonetheless , expression ( [ bpsrate ] ) is written in a manner that does not depend upon this choice : the debye wave number @xmath44 , and the plasma frequencies @xmath45 and @xmath50 can be calculated in your favorite units .
for example , in gaussian units where the electric potential takes the form @xmath51 , the debye wave number and the plasma frequency of species @xmath5 are given by and . in the rationalized units employed here , the electric potential is given by @xmath52 , and we have @xmath53 \omega_b^2 & = & \frac{e_b^2\ , n_b}{m_b } \ .
\label{wbdef}\end{aligned}\ ] ] the square of the total debye wave number is @xmath54 , and the total debye wave length is @xmath55 .
let us now consider an arbitrary plasma component of mass @xmath56 , which i will otherwise leave unspecified , and let @xmath57 denote the boltzmann distribution for this species .
then the average ( kinetic ) energy density of this component is @xmath58 if we work to leading and next - to - leading order in the number density , calculating the energy exchange between plasma components will then involve keeping a tally only of the kinetic energy , as in ( [ ebkin ] ) .
this is because the potential energy is higher order in the number density [ or more precisely , the potential energy is higher order in the plasma coupling @xmath59 , to be defined later in ( [ gdefa ] ) ] . as the system interacts with other plasma components through mutual coulomb interactions , it will loose or gain energy depending on the temperature gradients with other species , and the energy exchange rate
is given by @xmath60 0.1 cm in contrast to ( [ dedtei ] ) and ( [ dedtei ] ) , for ease of notation i have temporarily dropped the plasma component subscripts on the rate , and i will keep with this convention until the final calculation presented in section [ sec : calcleading ] .
we see that the entire problem is bound up in calculating the rate of change @xmath61 from an appropriate kinetic equation that captures the relevant physics .
as it turns out , however , there is a serious problem in performing all such calculation with the coulomb potential in three dimensions : the integrals in the kinetic equations diverge logarithmically , and they do so at both large and small distances . for processes in which large - angle scattering is important , such as the charged particle stopping power
, it is natural to use the boltzmann equation , which i will write in the abbreviated form @xmath62 \ , \label{besimp}\end{aligned}\ ] ] where @xmath63 is the gradient in position space , and @xmath64 $ ] is the scattering kernel , whose precise form will not concern us until the next lecture .
the gradient vanishes because of spatial uniformity , so we will set @xmath65 .
the boltzmann equation was designed to account for the statistical effects of short - distance collisions , and although its original context was classical , quantum two - body scattering effects can easily be incorporated .
in fact , since the scattering phase shifts @xmath66 are known analytically for the coulomb potential , ref .
@xcite used this to calculate the two - body quantum corrections to all orders [ in the quantum parameter @xmath67 to be defined in ( [ etadef ] ) ] .
the kernel @xmath64 $ ] therefore contains all short - distance or ultraviolet physics , for both classical and quantum scattering .
however , ( in three spatial dimensions ) the coulomb potential is long - range , and the integrals in @xmath64 $ ] diverge in the infrared ; or equivalently , if we write the scattering kernel @xmath64 $ ] in terms of momentum integrals , the divergence appears at small values of momentum .
this was not a problem in boltzmann s original formulation of ( [ besimp ] ) , since the coulomb potential was unknown at that time , and he modeled particle collisions in terms of hard - sphere scattering . in summary , the boltzmann equation gets the short - distance physics correct , including quantum two - body scattering , but it misses the infrared physics .
the fact that the boltzmann equation misses the long - distance physics manifests itself as an infrared divergence in the scattering kernel @xmath64 $ ] , thereby rendering calculations meaningless ( unless we tame , or regularize , this divergence ) . given that the boltzmann equation misses the long - distance or infrared ( ir ) physics ,
we might be tempted to try the lenard - balescu equation , which i will write in the abbreviated form @xmath68 \ , \label{lbesimp}\end{aligned}\ ] ] where @xmath69 $ ] is a scattering kernel whose exact form will be needed only in the next lecture .
again , the gradient term will be set to zero because of spatial uniformity .
the lenard - balescu equation takes the form of a fokker - plank equation , with the kernel @xmath69 $ ] chosen to capture the correct ir physics .
however , for the coulomb potential ( in three spatial dimensions ) , the lenard - balescu equation misses the short - distance or ultraviolet ( uv ) physics , and this is manifested by a uv divergence in @xmath69 $ ] .
the situation for the lenard - balescu equation is exactly reversed compared to that of the boltzmann equation .
this is what ref .
@xcite calls the `` complementarity '' of these two kinetic equations , and in the dimensional continuation procedure we will use this fact to our advantage .
the rate equation ( [ dedtbe ] ) reduces to a one - dimensional integral over the entire range of physical length scales ( or momentum scales , if one so chooses ) , from zero all the way to infinity .
trouble arises for the coulomb potential in three dimensions since the integral in question is logarithmically divergent at both integration limits .
we must therefore regulate the integral in some manner .
dimensional continuation is one such procedure , but there are others .
this divergence problem was first worked around by simply cutting off the divergent integrals by hand , with the cutoffs themselves being chosen by physical arguments ( rather than a calculation ) .
the energy exchange rate we are considering is but an example of a larger class of problems involving characteristic , but disparate , length or energy scales in which the measured quantity of interest is ( logarithmically ) insensitive to the physics above and below these scales . for these problems ,
the simplest and most intuitive regularization scheme is to replace the offending integration limits , _
i.e. _ infinity and zero , by the finite and non - zero physical scales of the problem .
these two scales then act as formal integration cutoffs , giving a finite logarithm of the ratio of the scales . furthermore , because the system is insensitive to the physics above and below the respective cutoffs , this procedure provides a physically meaningful result . expressed in terms of length , we will denote the long- and short - distance scales by @xmath70 and @xmath71 , and the integral over scales leads to a finite logarithm involving the ratio of the physical length scales , so that @xmath72 where @xmath73 is an easily determined prefactor with dimensions of energy density per unit time . as we shall see , a calculation to leading order in the number density is sufficient to provide the coefficient @xmath73 , while a next - to - leading order calculation is required to find the exact terms under the logarithm . the problem with this regularization prescription , which i will call the _ heuristic scheme _
, is that we can only estimate the values of the physical scales @xmath70 and @xmath71 to within factors of order one or so .
for example , it is physically reasonable that the scale of the long distance cutoff in a plasma is set by a debye length , so that @xmath74 with @xmath75 being a dimensionless constant of order unity ; but what determines the exact value of @xmath75 ?
in fact , how does arise naturally from the kinetic equations themselves , rather than simply being chosen by hand ? and should one use the total debye wave number @xmath1 , or just the contribution from the electrons @xmath44 ? the origin of the short distance cutoff @xmath71 is even less clear . in the extreme classical limit
, we expect this scale to be set by the classical distance of closest approach @xmath76 between two colliding particles , so that .
, as this involves a somewhat arbitrary choice in the thermal averaging procedure .
] on the other hand , we expect quantum effects to dominate when @xmath76 becomes smaller than some thermal compton wavelength @xmath77 , with @xmath78 being a typical thermally averaged momentum transfer . in this case
we expect @xmath79 .
worse yet , the intermediate region in which neither classical nor quantum processes dominate is often realized in a weakly coupled plasma , and in this case it is even more obscure how one should choose @xmath71 .
we must interpolate between the extreme classical and the extreme quantum scales , but in the literature the exact procedure is always somewhat _
ad hoc_. for example , a common choice is to define @xmath80 to be the sum of the squares of the classical and quantum scales , so that @xmath81 i will have more to say about this in section [ sec : ginside ] .
the heuristic scheme forces us to choose the specific forms of @xmath70 and @xmath71 motivated by imprecise physical arguments or heuristic exercises , which leads us into the art of model building rather than systematic calculation .
indeed , the very notation that we must _ choose _ a cutoff is misleading , since the physics itself must conspire to render all integrals finite .
consequently , the heuristic method suffers from an unknown coefficient under the logarithm , and only the approximate value of the ratio @xmath82 can be determined with this method ( in fact , this ratio varies across an order of magnitude over models in the literature , rather than factors of two or three ) .
as we shall see , determining the constant under the logarithm exactly is equivalent to determining the next - to - leading order term exactly ; therefore , models of the form ( [ ratemodel ] ) are accurate only to leading order , and no better . rather than merely regulating the _ integrals _ in a rate derived from the kinetic equations , as with ( [ ratemodel ] ) , a more sophisticated approach involves regularizing the divergences in the
_ kinetic equations _ themselves . in other words ,
the _ theory _ itself is regularized , rather than a particular quantity being calculated within the theory .
the method of dimensional continuation falls into this category , albeit with somewhat more subtle mathematical machinery than traditional approaches .
these approaches , of which refs
. @xcite are good examples , are summarized and placed into a common framework by aono in ref .
@xcite . as discussed in section [ sec : problemwithproblem ] of this lecture , one can view the boltzmann and leonard - balescu equations as providing complementary physics since they both succeed and fail in complementary regimes .
the boltzmann equation gets the short - distance physics correct , while the leonard - balescu equation captures the long - distance physics ; conversely , boltzmann and leonard - balescu miss the long- and short - distance physics , respectively .
this complementarity motivates a class of kinetic equations of the form @xcite @xmath83 + l[f ] - r[f ] \ , \label{convke}\end{aligned}\ ] ] where @xmath84 $ ] is a carefully chosen `` regulating kernel '' designed to subtract the long - distance divergence of the boltzmann equation and the short - distance divergence of the lenard - balescu equation . at the same time
, the kernel @xmath84 $ ] must preserve the correct short - distance physics of the boltzmann equation and the correct long - distance physics of the lenard - balescu equation ( a minimal requirement of the regulating kernel @xmath84 $ ] is that it take a hippocratic oath to _ do no harm _ , at least to subleading order in the plasma coupling ) .
each term on the right - hand - side of ( [ convke ] ) is separately divergent , but collectively they lead to a finite collision kernel if properly interpreted . ) should be separately regularized in some manner , rendering the individual kernels @xmath64 $ ] , @xmath69 $ ] , and @xmath84 $ ] separately finite : @xmath64 $ ] must be regulated at long - distances , @xmath69 $ ] at short - distances , and @xmath84 $ ] at both long- and short - distances .
if this procedure is executed correctly , so that the long - distance divergences of @xmath84 $ ] and @xmath64 $ ] cancel , as do the short - distance divergences of @xmath84 $ ] and @xmath69 $ ] , both in a consistent fashion , then the entire right - hand - side of ( [ convke ] ) remains finite as the cutoffs are removed .
if @xmath84 $ ] does not disturb the @xmath85 physics , then the convergent kinetic equation will be accurate to @xmath85 .
i will have more to say about how one regulates long- and short - distances _ consistently _ in section [ sec : lamb ] , and how this can be a quite non - trivial process .
] as an example , gould and dewitt @xcite regulated the long - distance divergence of the boltzmann equation by simply replacing the coulomb potential with a debye screened yukawa - like potential , @xmath86 where , as not to confuse symbols , i write @xmath87 as the base of the natural logarithm and @xmath88 as the electric charge . while ref .
@xcite performed this operation by hand using physical arguments , one could easily introduce a kernel @xmath84 $ ] to do the same . while the approach to convergent kinetic equations described by aono might appear to be more rigorous than the aforementioned model building approach of ( [ ratemodel ] ) , it is no more systematic : methods based on ( [ convke ] ) or its equivalent do not contain the ability to estimate their own error , _
i.e. _ they can not determine their domain of applicability .
there is nothing in the formalism of ( [ convke ] ) that keeps track of the plasma coupling constant , or the order to which we are working in this constant .
indeed , one does not generally think of ( [ convke ] ) in terms of a perturbation theory .
in contrast , ref .
@xcite calculates the rate using a systematic expansion in the plasma number density , or more precisely , in a dimensionless plasma coupling parameter @xmath59 [ to be defined by ( [ gdefa ] ) and discussed at length in section [ sec : eerate ] ] .
although written in a disguised form , the bps rate coefficient ( [ bpsrate ] ) is an expansion to leading and next - to - leading order in the coupling parameter : the leading order term goes like @xmath89 , the next - to - leading order is proportional to @xmath90 , and the @xmath91 term provides an estimate of the error . translating the work of gould and dewitt @xcite into the language of a perturbative expansion in a plasma coupling constant , it turns out that their result is valid to order @xmath90 and is in agreement with bps to this order .
however , in their final result , gould and dewitt retain spurious higher order terms in @xmath92 .
i call these terms `` spurious '' because ref
. @xcite did not calculate the full set of order @xmath92 terms , but only some of these terms .
indeed , the notion of a systematic expansion in a small dimensionless parameter does not enter their calculational framework . as this example shows , convergent kinetic equations can be more accurate than the heuristic model building technique of the previous section , but one can not be sure of their accuracy until a comparison with a systematic calculation has been made , as with bps and ref .
i will illustrate the main points of dimensional continuation with an example involving the modified bessel function @xmath93 , with an emphasis on analytic continuation and how this can be used to extract leading and next - to - leading order behavior .
this example was first presented in ref .
@xcite , and for pedagogical purposes it was also included in appendix a of ref .
this example contains all the essential features of dimensional continuation , but in a mathematically simple form , and while it is an imperfect analogy , as all analogies are , it is explicit in all its details .
we will show that the modified bessel function @xmath93 has the expansion @xmath94 to leading order ( lo ) and next - to - leading order ( nlo ) in @xmath95 , with @xmath96 denoting euler s constant .
this expansion is quite accurate for small values @xmath95 , with an error of order @xmath97 rather than @xmath95 for symmetry reasons .
the asymptotic expansion ( [ kzerolonlo ] ) is a well known result @xcite , but it is rather difficult to prove by conventional methods because of the non - analytic leading - log behavior .
however , the method of dimensional continuation allows us to derive this result rather easily .
the price one pays for this ease of derivation is that one must learn ( or recall ) a bit of mathematical machinery which , at first sight , seems unrelated to the problem at hand .
we start with the general integral representation of the modified bessel functions @xcite , @xmath98 as the notation in ( [ knuint ] ) suggests , we can think of @xmath99 as the dimension of space and the integration variable @xmath100 as the wave number . in this analogy
, the argument @xmath95 corresponds to the dimensionless coupling parameter @xmath59 of the plasma .
the following dictionary provides a useful mnemonic in relating this mathematical example to the plasma physics problem of real interest : @xmath101 \nonumber \nu & \rightarrow & \text{spatial dimension } \\[-6pt ] \nonumber k & \rightarrow & \text{wave number } \\[-6pt ] dk\,k^{\nu-1 } & \rightarrow & \text{integration measure } d^\nu k \ . \label{physanalogy}\end{aligned}\ ] ] pushing our physical analogy further , if we think of @xmath99 as being the dimension of space , then it should always be a positive integer , which i will express by the conventional set theory notation @xmath102 ; however , nothing in the integral representation ( [ knuint ] ) requires that @xmath103 .
we can therefore think of @xmath99 in expression ( [ knuint ] ) as being a continuous real variable , or indeed , a complex variable if circumstances warrant .
similarly , for a real physical system written in the appropriate integral form , there is nothing in any law of physics that prevents us from interpreting the dimension of space @xmath99 as being a complex number .
continuing @xmath99 from the positive integers into the complex plane is what i mean by `` dimensional continuation .
'' there will be times when we restrict our attention to the real numbers only , rather than the complex numbers in general , and i will refer to this as dimensional continuation as well . as we shall see , this procedure of taking @xmath104 or @xmath105 will allow us to regulate otherwise infinite integrals in a systematic and perturbative fashion .
finite manipulations can then be performed , the divergent poles will cancel from physically measurable quantities , and afterward we can take @xmath99 to the appropriate integer dimension ( in this analogy we take @xmath106 , rather than @xmath107 as we do for the physics problem ) .
let us first calculate the leading order in @xmath95 behavior of @xmath108 for positive and negative values of @xmath99 . for small positive values of @xmath95 , the leading order @xmath95-behavior can be obtained by replacing the exponential in ( [ knuint ] ) by one , _ except _ in the regions @xmath109 and @xmath110 , where the exponential is required for convergence .
taking @xmath111 first , note that the integral ( [ knuint ] ) is dominated by small values of @xmath100 near the lower limit of integration . in terms of the analogy ( [ physanalogy ] ) where @xmath100 is a wave number , this corresponds to the situation in which long - distance ir physics is dominant .
therefore , when @xmath111 and the integral is dominated by small values of @xmath100 , the leading order contribution to ( [ knuint ] ) can be obtained from the leading order behavior of the exponential , that is to say , the replacement @xmath112 will capture the entire leading order in @xmath95 behavior for negative values of @xmath99 .
note that ( [ expa ] ) provides convergence as @xmath113 , while large-@xmath100 convergence is provided by the prefactor @xmath114 since @xmath111 .
we will denote this leading order contribution by @xmath115 , and using the substitution ( [ expa ] ) we write @xmath116 in the last equality of ( [ knult ] ) , we have made the variable change @xmath117 , and we have used the standard integral representation for the gamma function , @xmath118 this representation of the gamma function only converges when @xmath119 ( actually , ( [ gammaone ] ) is valid over the complex @xmath120-plane with @xmath121 ) . when @xmath122 and @xmath123 , we may indeed use ( [ gammaone ] ) .
we can find the leading order in @xmath95 contribution when @xmath124 in a similar manner . in this case , the integral is dominated by large values of @xmath100 at the upper limit of integration , and we make the substitution @xmath125 thereby giving the leading order result @xmath126 note that the exponential provides convergence as @xmath110 , while the integrand possesses an integrable singularity at @xmath127 for @xmath128 ( the integrand is non - singular at @xmath127 when @xmath129 ) .
we will eventually take the @xmath130 limits of ( [ knult ] ) and ( [ knugt ] ) , since we are interested in @xmath93 and not @xmath108 .
while we could take @xmath99 to be a general real or complex number until the limit is required , it is easier to consider only small values of @xmath99 from the start ( we need work no higher than linear order , since this and higher orders vanish when @xmath131 ) . to do this
, we expand the gamma function @xmath132 to linear - order in its argument using @xmath133 . taking @xmath134 in this expansion
gives @xmath135 \hskip1 cm : ~\text{lo in $ x$ when } \nu > 0 \label{knua } \\[5pt ] k_\nu^{{\scriptscriptstyle < } } ( x ) & = & -\frac{1}{2\nu}\ , \left(\frac{x}{2}\right)^\nu~ \big[1 + \nu \gamma \big ] \hskip1.05 cm : ~\text{lo in $ x$ when } \nu < 0 \ .
\label{knub}\end{aligned}\ ] ] expressions ( [ knua ] ) and ( [ knub ] ) are accurate to linear order in the dimension @xmath99 ; on the other hand , ( [ knua ] ) gives the leading order in @xmath95 contribution to @xmath108 as defined by ( [ knuint ] ) when @xmath124 , and ( [ knub ] ) gives the leading order in @xmath95 contribution to @xmath108 when @xmath123 .
to compare these two expressions , we must analytically continue them to a common dimension .
we will discuss this further in the next section . in terms of our physics analogy , expression ( [ knua ] )
captures the leading order short - distance physics in the @xmath136 regime ; the pole at @xmath137 corresponds to a small-@xmath100 divergence , which , pushing our physical analogy again , would reflect missing or incomplete long - distance physics ( as with the boltzmann equation ) .
the situation is completely reversed for ( [ knub ] ) , which captures the leading order long - distance physics for @xmath136 , with the pole at @xmath138 corresponding to a large-@xmath100 divergence arising from missing short - distance physics ( like the lenard - balescu equation ) . as functions of @xmath99 ,
we see from ( [ knua ] ) and ( [ knub ] ) that @xmath139 and @xmath115 are analytic in @xmath99 , except for the simple pole at @xmath138 .
as we shall see , the analytic continuation to complex @xmath99 takes the same functional form as the individual expressions ( [ knua ] ) and ( [ knub ] ) , each defined separately for @xmath124 and @xmath123 , respectively .
since analytic continuation plays such a central role in dimensional continuation , at least mathematically , i would like to briefly discuss the conditions under which a function can be analytically continued from one region of the complex plane to another . recall that a function @xmath140 is said to be _ analytic at a point _ @xmath141 in the complex plane @xmath142 , if and only if its derivative exists not only at @xmath141 , but also at every point within some open neighborhood of @xmath141 .
a function @xmath140 is _
analytic on a domain _ @xmath143 in the complex plane if it is analytic at each point in @xmath143 .
analyticity is a very stringent condition on a function , since the existence of the derivative of a complex function is a much more robust constraint than the corresponding existence of the derivative of a function on a real domain .
this is because in the two - dimensional complex plane , the limiting procedure defining the derivative must exist regardless of the direction used in taking the limit .
in fact , analyticity at a point @xmath141 is so strong that it implies the existence and continuity of all derivatives @xmath144 for any order @xmath145 @xcite . in other words ,
an analytic function on @xmath143 can be thought of as being infinitely smooth on @xmath143 , even though the definition of analyticity itself invokes only the existence of the first derivative , albeit on a neighborhood .
analyticity is such a stringent condition , that the behavior of an analytic function in a small domain is enough to determine its behavior in a larger region .
even if the function is only known along a one - dimensional curve in the complex plane , such as a portion of the real axis , this is enough to uniquely determine the function in the complex plane .
, in practice we can analytically continue a function defined on the integers without this theorem .
we appeal to carlson s theorem only to guarantee the uniqueness of this procedure .
] this is because the derivatives of an analytic function exist to all orders , and therefore a taylor series expansion about a point of analyticity always exists with some non - zero radius of convergence .
for example , one of the powers of analytic continuation is that the original function may be defined in any manner over a subregion , and this can generate a unique function over a larger region . as an application of this ,
take the analytic function defined by the infinite geometric series @xmath146 a series that converges only for @xmath147 . upon defining @xmath140 by ( [ fseriesdef ] )
, we therefore take the domain @xmath148 to be the unit disk about the origin , excluding the unit circle itself . in this domain
, the geometric series converges to @xmath149 note , however , that the function @xmath150 is defined over the entire complex plane except @xmath151 , a region i will call @xmath152 . since the function @xmath140 is only defined within the unit circle , and since @xmath140 and @xmath59 agree within the unit circle , the function @xmath153 is the unique analytic continuation of @xmath154 . as a more relevant example , consider @xmath115 with @xmath111 as given by ( [ knub ] ) . to compare this with @xmath139 , which is determined by ( [ knua ] ) for @xmath124 , we must analytically continue @xmath115 to the positive real @xmath99-axis .
we can think of @xmath115 as a sequence of functions of an independent variable @xmath99 indexed by a continuous label @xmath95 ; therefore , in a more suggestive notation , we temporarily write @xmath155 . while the collection of functions @xmath156 are defined in ( [ knub ] ) on the negative @xmath99-axis @xmath157 ( excluding the simple pole at zero ) , they can be analytically continued to the complex @xmath99-plane @xmath158 . furthermore , the functions take the same algebraic form on the complex plane , namely , @xmath159 ~~\text{for}~~
\nu \in \mathbb{c } \ .
\label{fxnu}\end{aligned}\ ] ] we can now restrict our attention from @xmath158 in general to the positive @xmath99-axis @xmath160 ( excluding zero ) .
this allows us to directly compare @xmath115 and @xmath139 at @xmath161 using the same algebraic forms as given by ( [ knub ] ) and ( [ knua ] ) . in the next section
, we discuss the implications of analytically continuing @xmath115 from the negative axis @xmath123 to the positive axis @xmath124 .
alternately , we could continue @xmath162 to the region @xmath163 using the same functional form as ( [ knua ] ) , and compare this with @xmath115 .
we now illustrate the key mathematical result that allows dimensional continuation to extract not only the leading , but the next - to - leading order terms .
recall from ( [ knua ] ) and ( [ knub ] ) that @xmath139 and @xmath115 are both leading order in @xmath95 for @xmath124 and @xmath123 , respectively .
since these functions were calculated for mutually exclusive values of @xmath99 , one might think that they can not be compared . when viewed as an analytic function in the complex
, however , we have seen that @xmath115 is also a function over the domain @xmath124 , in which case both @xmath115 and @xmath139 can be compared at the same values of @xmath99 and @xmath95 .
since the algebraic form is so simple , @xmath115 takes the same functional form when analytically continued to @xmath124 as it did for @xmath123 . as illustrated in fig .
[ fig : knucont ] , this means that @xmath115 becomes next - to - leading order in @xmath95 along the positive real axis : @xmath135 \hskip1 cm : ~\text{lo in $ x$ when } \nu > 0 \label{knuaplus } \\[10pt ] k_\nu^{{\scriptscriptstyle < } } ( x ) & = & -\frac{1}{2\nu}\ , \left(\frac{x}{2}\right)^\nu~ \big[1 + \nu \gamma \big ] \hskip1.04 cm : ~\text{nlo in $ x$ when } \nu > 0 \ . \label{knubplus}\end{aligned}\ ] ] the expressions ( [ knuaplus ] ) and ( [ knubplus ] ) remain accurate to linear - order in @xmath99 .
( 120,120)(-75,0 ) ( -101,60)(1,0)220 ( 124,58)@xmath99 ( 0,56)(0,1)8 ( -2.65,57.3)@xmath164 ( -2.00,49)@xmath165 ( -60,66)@xmath115 ( -100,44)@xmath166 $ ] ( -60,30)@xmath167 ( -60,19)@xmath168 ( -120,8)lo : large when @xmath169 ( 30,66)@xmath115 ( 20,44)@xmath166 $ ] ( 40,30)@xmath167 ( 40,19)@xmath170 ( 20,8)nlo : small when @xmath169 ( 0,75)(45,45)[t ] ( 22,72)(0,-1)1 ( -40,115)analytically continue ( -42,105)around the @xmath171 pole to see that ( [ knubplus ] ) is indeed next - to - leading order in @xmath95 relative to ( [ knuaplus ] ) , note that the of the leading order contribution for @xmath124 can be written . i have used the absolute value @xmath172 to emphasize that the power of @xmath95 in ( [ knuaplus ] ) is strictly negative when @xmath124 .
similarly , along the positive @xmath99-axis we find the behavior @xmath173 for @xmath124 , and we see that @xmath174 for @xmath175 .
this means @xmath176 for @xmath124 and @xmath177 , and we are therefore justified in calling @xmath139 leading order in @xmath95 and @xmath115 next - to - leading order . strictly speaking
, we have only shown that @xmath115 is subleading relative to @xmath139 when @xmath124 . to conclude that @xmath178 is indeed next - to - leading order relative to @xmath179 , it is important to establish that there are no powers of @xmath95 between @xmath180 and @xmath181 in the expansion of @xmath108 . for
@xmath124 , one simply subtracts ( [ knua ] ) from ( [ knuint ] ) , and it becomes clear that this error is higher order in @xmath95 than @xmath180 . for a more detailed proof of this ,
see footnote 2 of ref .
a similar statement holds for @xmath139 , namely , as we analytically continue from @xmath124 to @xmath123 , the quantity @xmath139 switches from leading order to next - to - leading order in @xmath95 relative to @xmath115 .
we have now assembled enough results to find @xmath93 to leading and next - to - leading order in @xmath95 : we simply add the expressions ( [ knuaplus ] ) and ( [ knubplus ] ) and take the limit of vanishing @xmath99 .
note that this does not lead to any form of `` double counting . ''
instead , we are simply adding the next - to - leading order term ( [ knubplus ] ) to the leading order term ( [ knuaplus ] ) at a common value of . upon taking the limit of vanishing @xmath99 , or more precisely @xmath182
since @xmath99 is always positive in ( [ knuaplus ] ) and ( [ knubplus ] ) , we obtain @xmath93 to leading and next - to - leading order in @xmath95 .
we now calculate this limit , proving that @xmath183 = - \ln x + \ln 2 - \gamma \ . \label{knuaa}\end{aligned}\ ] ] let us first expand @xmath184 in powers of @xmath99 .
we will denote @xmath185 , from which we find @xmath186 , or in summary : @xmath187 this allows us to express ( [ knuaplus ] ) and ( [ knubplus ] ) as @xmath188 \bigg[1 - \nu \gamma \bigg ] + { \cal o}(\nu ) = \phantom{- } \frac{1}{2\nu } - \frac{1}{2}\ln\!\left(\frac{x}{2}\right ) - \frac{\gamma}{2 } + { \cal o}(\nu ) \ , \end{aligned}\ ] ] and @xmath189 \bigg[1 + \nu \gamma \bigg ] + { \cal o}(\nu ) = -\frac{1}{2\nu } - \frac{1}{2}\ln\!\left(\frac{x}{2}\right ) - \frac{\gamma}{2 } + { \cal o}(\nu ) .\end{aligned}\ ] ] when we divide ( [ zpmnu ] ) by a factor of @xmath99 , as required by ( [ knuaplus ] ) and ( [ knubplus ] ) , note that we find : ( i ) a pole @xmath190 from the first term in ( [ zpmnu ] ) , ( ii ) a non - analytic finite contribution @xmath191 from the second term , and ( iii ) the error in @xmath99 becomes @xmath192 , which is the same order in @xmath99 that we are neglecting in ( [ knuaplus ] ) and ( [ knubplus ] ) . the error in @xmath99 , which vanishes in the limit @xmath193 , should not be confused with the error in @xmath95 , the latter being @xmath194 for vanishing @xmath99 .
note that the pole terms cancel upon adding ( [ knugt ] ) and ( [ knult ] ) , so that @xmath195 thereby giving ( [ knuaa ] ) as @xmath182 .
as we have discussed , there are no @xmath95-dependent terms that lie between @xmath196 and @xmath178 , so this procedure has captured the leading and next - to - leading order behavior in @xmath95 . in exactly the same way , we can also calculate the leading order and next - to - leading order contribution to @xmath93 by taking the limit from the left , @xmath197 = - \ln x + \ln 2 - \gamma \ . \label{knubb}\end{aligned}\ ] ] i should point out a potential notational problem in ( [ knuaa ] ) and ( [ knubb ] ) . concentrating on ( [ knuaa ] ) for the moment , the limit @xmath182 indicates that both terms @xmath198 and @xmath199 are understood to live in dimensions @xmath124 , with the second term @xmath199 having been analytically continued from .
the notation with which the term @xmath199 is written in ( [ knuaa ] ) , however , does not indicate that it has been analytically continued
. this should be no cause for confusion , however , since @xmath199 takes the same functional form in any dimension @xmath99 ; therefore , a separate notation indicating that the @xmath199 in ( [ knuaa ] ) has been analytically continued is unnecessary .
we can simply add @xmath198 and @xmath199 as calculated in @xmath124 and @xmath123 respectively .
before moving on to the details of dimensional continuation , we must first discuss the plasma expansion parameter @xmath59 .
since the problems in plasma physics are usually so complicated as to preclude a perturbative approach , most plasma physicists do not usually think in terms of expanding systematically in a small dimensionless parameter .
however , for a weakly to moderately coupled plasma , it is a quite fruitful approach to perturbatively expand the rate in a small dimensionless coupling constant . that such a universal parameter for a plasma exists was discussed at length in ref .
@xcite , where it was shown that any physical quantity associated with a plasma whose species are in equilibrium with themselves ( such as the plasma we are studying ) can be expanded in _
powers of a dimensionless coupling constant @xmath59 defined by @xmath200 \label{gdef } g & \equiv & \frac{\text{coulomb energy for two charges at separation } \lambda_{{\rm\scriptscriptstyle d } } } { \text{temperature in energy units } } \ , \\[-10pt]\nonumber\end{aligned}\ ] ] with @xmath201 being the debye length of the plasma .
since the potential energy between two like charges is given by @xmath202 in rationalized units , and writing the debye wave number as @xmath203 , the coupling parameter is therefore @xmath204 for a multicomponent plasma , there is actually a coupling constant for each pair of components , @xmath205 with @xmath206 defined by ( [ kbdef ] ) .
however , when the pairs have approximately the same coupling strength , then the single parameter ( [ gdefa ] ) adequately characterizes the entire plasma . expressing the charges as @xmath207 and @xmath208 ,
we can write @xmath209 as @xmath210 and we see that the coupling constant is proportional to the cube of the electric charge , the square root of the density , and the inverse ( 3/2)-power of the temperature .
recall that the usual plasma parameter @xmath211 is defined in a similar manner to ( [ gdef ] ) , except that the charge separation is determined not by @xmath201 , but by the inter - particle spacing in the plasma , @xmath200 \label{gammadef } \gamma & \equiv & \frac{\text{coulomb energy for two charges at inter - particle separation } d}{\text{temperature in energy units } } \ .
\\[-10pt]\nonumber\end{aligned}\ ] ] the inter - particle spacing @xmath212 is defined in several ways in the literature , but the idea is to transform the plasma number density @xmath213 into a length scale , so that @xmath214 .
the most common convention is to define @xmath212 to be the radius of a sphere containing , on average , a single plasma particle , so that @xmath215 , and therefore @xmath216 with this convention , the relation between the two plasma coupling parameters for a single plasma species is @xmath217 , and for an arbitrary number of plasma species we always find @xmath218 .
we can therefore use either @xmath59 or @xmath211 to characterize the strength of the plasma , as @xmath59 and @xmath211 become large or small together .
the dimensional continuation method requires that we perform calculations in an arbitrary spatial dimension @xmath99 ; however , the parameter @xmath219 as given by ( [ gdefa ] ) is _ dimensionless _ ( in the engineering sense ) only in three spatial dimensions . as such
, it is meaningless to call @xmath59 large or small in any dimension other than @xmath220 .
this is no cause for alarm , however , since in footnote [ foot : gnudef ] , i will construct a dimensionless coupling parameter @xmath221 in arbitrary spatial dimensions @xmath99 .
the parameter @xmath221 can then be used to characterize the plasma strength in @xmath99 dimensions , and it will have the properties that ( i ) @xmath222 and ( ii ) @xmath223 as @xmath107 . property ( i ) implies that a @xmath221-expansion is the same as a @xmath59-expansion , and ( ii ) implies this correspondence continues down to three dimensions , where @xmath59 becomes the relevant expansion parameter . ] for our purposes , however , there is an an advantage to using @xmath59 rather than @xmath211 since physical quantities are expanded as _
powers of @xmath59 , while they expand in fractional powers of @xmath211 .
as i have said , any plasma quantity can be written as a power series expansion in integer power of @xmath59 , with the possible exception of non - analytic terms involving @xmath224 .
for the process of energy exchange via coulomb interactions , this non - analyticity arises from the competition between disparate physical length scales . as an expansion in @xmath59 , the rate of energy exchange takes the form @xmath225 in ( [ dedtnlo ] ) ,
i have indicated the leading order in @xmath59 ( lo ) and the next - to - leading order in @xmath59 ( nlo ) terms in the : the first term is leading order relative to the second because @xmath226 for small @xmath59
. the minus sign on the leading order term of ( [ dedtnlo ] ) is a matter of convention .
since the logarithm @xmath224 will be negative in a weakly coupled plasma ( recall @xmath227 ) , the minus sign renders the coefficient @xmath3 positive when the energy exchange is positive .
the coefficient @xmath3 was first calculated by spitzer .
the coefficient @xmath228 , however , is very difficult to calculate , and this was the main purpose of bps @xcite .
it is convenient to define the dimensionless coefficient @xmath4 by @xmath229 , in which case we can write @xmath230 we see , then , that knowing the next - to - leading order term is equivalent to knowing the exact coefficient @xmath4 under the logarithm .
note that the minus sign renders the coulomb logarithm positive when @xmath59 is very small , in keeping with convention . for the heuristic model building of section
[ sec : heuristic ] , let us pause for a moment and show that the argument of the coulomb logarithm in ( [ ratemodel ] ) is indeed proportional to @xmath59 , as required by ( [ lngsqu ] ) . on physical grounds we saw that the long - distance scale @xmath70 is set by a debye length , and therefore we nominally set .
in the extreme classical limit , the short - distance cutoff @xmath71 is set by the classical distance of closest approach @xmath76 , so that .
for simplicity , we will choose the coefficient @xmath75 such that @xmath231 between ( [ ratemodel ] ) and ( [ lngsqu ] ) , in which case @xmath232 let us consider two unit charges of mass @xmath56 approaching one another with zero impact parameter .
the rms speed of each particle is determined by @xmath233 while energy conservation @xmath234 gives the distance of closest approach , @xmath235 in the extreme classical regime , we see that the argument of the coulomb logarithm in ( [ lnmodel ] ) is indeed proportional to the plasma coupling constant , @xmath236 let us now look at the _ ad hoc _ interpolation ( [ bminint ] ) between the classical and quantum regimes . up to this point
i have said very little about quantum mechanics .
while i will not dwell on quantum corrections , i will briefly discuss a dimensionless expansion parameter that characterizes the strength of the quantum two - body scattering correction .
there are many ways of defining such a parameter , but i will follow ref .
@xcite , taking @xmath200 \eta & \equiv & \frac{\text{classical distance of closest approach } } { \text{thermal wavelength } } \ .
\label{etadefwords } \\[-10pt]\nonumber\end{aligned}\ ] ] with this definition , quantum corrections are large when @xmath237 . motivated by the de broglie wavelength of a particle , the thermal wavelength of a plasma species is given by @xmath238 , where @xmath239 is a typical momentum transfer suffered during a collision .
definition ( [ etadefwords ] ) yields @xmath240 , or more succinctly @xmath241 from which ( [ bminint ] ) gives @xmath242 in the extreme quantum limit in which @xmath237 , this becomes @xmath243 .
finally , note that the factors inside the bps coulomb logarithm ( [ bpsrate ] ) are also proportional to the coupling constant @xmath59 , and upon dropping the electron subscripts for convenience , we find @xmath244 we see that the @xmath59-dependence of the coulomb logarithm arises quite naturally . however , the accompanying coefficient under the logarithm might also possess @xmath67-dependence , thereby obscuring the @xmath59-dependence unless we are careful . before describing what dimensional continuation is ,
allow me to first state what it is not .
dimensional continuation is _ not _ performing an integral to a fractional power of the spatial dimension , as with the meaningless expression @xmath245 instead , dimensional continuation is the following .
suppose some physical quantity of interest can be written as an integral over a kernel @xmath246 where @xmath140 is determined by the physical equations of motion , whether classical or quantum .
the integrand is of course a function of the physical parameters , such as the masses and charges of the fundamental particles , and i have abbreviated this dependence by the parameter @xmath56 . for definiteness , we will think of @xmath100 as a wave number with dimensions of an inverse length .
the laws of physics , from which ( [ intthree ] ) follows , are usually written in three dimensional space .
thus , we usually take @xmath247 to be a three - dimensional vector , and we integrate @xmath247 over the entire three - dimensional euclidean space @xmath248 .
the known fundamental laws of physics themselves , however , do not specify a particular spatial dimension in which they hold .
in fact , as far as the known laws of physics are concerned , the actual value @xmath99 of the spatial dimension can be viewed as a free _ integer _ parameter : it is simply an unexplained empirical fact that we live in three dimensions .
we can therefore express any three - dimensional physical quantity or law , such as newton s equation of motion or gauss law , in any number of integer dimensions @xmath103 .
for example , we can write down a corresponding quantity to ( [ intthree ] ) in an arbitrary number of dimensions , @xmath249 where the wave vector @xmath247 is now a @xmath99-dimensional vector , and we integrate over the entire @xmath99-dimensional euclidean space @xmath250 .
i have placed a subscript on the integrand @xmath251 to indicate that it is determined by the theory expressed in @xmath99 dimensions . at this point ,
the spatial dimension is a non - negative _ integer _ , so that @xmath252 . since the integer @xmath99 is arbitrary in ( [ intnu ] )
, i have indicated that the corresponding quantity contains @xmath99-dependence by writing @xmath253 ; however , for notational simplicity i will drop the parametric dependence of quantities such as mass and simply write @xmath254 .
a quantity that diverges in three - dimensions will be finite in arbitrary @xmath99 , but it will typically exhibit a simple pole of the form @xmath255 where @xmath256 is finite at @xmath257 . of the pole .
] dimensional continuation is simply the act of treating @xmath254 as a function of a complex argument @xmath99 , _ after _ the integral ( [ intnu ] ) has been performed for _ all _ positive integer values of @xmath99 .
this is not an unfamiliar procedure , as the factorial function @xmath258 with @xmath259 can be generalized to the gamma function @xmath260 in which @xmath261 .
indeed , given any function @xmath254 defined on the integers , with only mild restrictions placed on the function at large values of the argument , carlson s theorem @xcite allows us to continue this function uniquely to the complex plane . how does dimensional continuation work in practice ?
we will look at a few specific examples in sections [ sec : physdim ] and [ sec : solution ] , but for now let us consider a general physical quantity @xmath262 in which the integrand in ( [ intthree ] ) depends solely upon the modulus of @xmath247 , so that @xmath263 in such a case it is not uncommon that the integrand in the generalization ( [ intnu ] ) is only a function of the modulus of the @xmath99-dimensional wave vector @xmath247 , with the same functional form as the integrand in ( [ qthree ] ) . in other words , in ( [ intnu ] ) we have @xmath264 with @xmath265 , thereby allowing us to write @xmath266 since the integrand is a function only of @xmath100 , we can extract the angular integrals and write @xmath267 at this point , the dimension @xmath99 is simply an arbitrary positive integer , @xmath259 . as we will show in the next paragraph
, the integration over all angles gives @xmath268 this leaves a one - dimensional integral to perform , in which @xmath99 simply acts as a parameter , @xmath269 the physical quantity now becomes the product of ( [ intomega ] ) and ( [ intk ] ) with @xmath259 . in the case of ( [ intomega ] ) , we already know how to analytically continue @xmath99 to complex values . on the other hand , for ( [ intk ] ) we can think of @xmath99 as being an arbitrary complex number when performing the one - dimensional integral over @xmath100 , and
therefore we can regard @xmath270 as a function over the complex @xmath99-plane , thereby giving @xmath271 in this manner , we can regard @xmath254 as a function of a complex argument @xmath99 , and by carlson s theorem @xcite , this is the unique continuation from positive integer values of @xmath99 to complex values of @xmath99 .
as an example of this procedure , let us prove ( [ intomega ] ) .
first , consider the one - dimensional gaussian integral @xmath272 if we multiply both sides together @xmath99 times ( with @xmath259 ) , we find @xmath273 where the wave vector @xmath247 in the exponential of the last expression is the @xmath99-dimensional vector @xmath274 , and @xmath275 is the @xmath99-dimensional inner product @xmath276 .
as in ( [ qfactor ] ) , we can factor the angular integrals out of the right - hand - side of ( [ gaussnu ] ) , and the remaining one - dimensional integral can be converted to a gamma function with the change of variables @xmath277 : @xmath278 solving for @xmath279 in ( [ pinuomega ] ) gives ( [ intomega ] ) .
as an aside , it is interesting to note that we have also found the hyper - area of a sphere of radius @xmath280 in @xmath250 , @xmath281 and integrating ( [ areanu ] ) gives the hyper - volume of a @xmath99-dimensional ball of radius @xmath280 , @xmath282 so far we have only introduced dimensional continuation as a regularization prescription , a means by which infinite theories can be rendered temporarily finite .
this is the use to which dimensional continuation is commonly employed in field theory ; however , this alone is not sufficient to render the method suitable for our purposes . instead , there are also physics reasons that make dimensional continuation applicable to the problem at hand .
i will now show that the coulomb potential , and indeed physics in general , behaves differently in different spatial dimensions .
in particular , short - distance or ultraviolet ( uv ) physics dominates when @xmath283 ; while long - distance or infrared ( ir ) physics dominates when @xmath284 . in @xmath257 dimensions , both uv and ir physics are equally important .
this means that dimensional continuation acts as a `` physics sieve , '' allowing us to capture the leading uv and ir physics simply by performing the relevant integrals in dimensions greater than or less than the traditional @xmath257 . in the next few paragraphs
, we will discuss why changing the dimension of space emphasizes either long- or short - distant physics .
understanding this point is crucial for all that follows .
we now turn to finding the @xmath99-dimensional coulomb potential , an exercise that succinctly illustrates how the physics of a system changes with the dimension of space . besides illustrating that uv physics dominates in higher dimensions ( and conversely )
, this example will also clarify the manner by which one performs physical calculations in arbitrary dimensions .
let us consider poisson s equation @xmath285 there is nothing in this equation _ per se _ that restricts us to three dimensions .
, where the electric and magnetic fields have been expressed in terms of the anti - symmetric field tensor @xmath286 .
] we _ choose _ to describe the electric field @xmath287 and the spatial coordinate @xmath288 as three - dimensional vectors because we live in three dimensions . however
, from the mathematics alone , we could equally well consider these vectors as living in an arbitrary @xmath99-dimensional space @xmath250 .
the rectilinear coordinates would then become @xmath289 , with a similar expression for the electric field , while the gradient would be @xmath290 .
it is convenient to write poisson s equation in its integral representation , which will allow us to calculate the electric field of a point charge with relative ease .
consider a point - charge at the origin given by @xmath291 , and let @xmath292 be any volume containing the charge @xmath88 .
then , in a general number of dimensions , we can integrate ( [ peq ] ) to obtain @xmath293 note that the dimensionality of space is now explicitly indicated by the integration measure . for our purposes ,
the advantage of the integral representation of poisson s equation is that the electric field of a point charge in the @xmath99-dimensional space can easily be calculated by the same symmetry principles that hold in three dimensions .
suppose the volume @xmath292 is a spherical ball @xmath294 of radius @xmath280 centered on the charge .
the boundary of @xmath294 is a sphere of dimension @xmath295 and will be denoted by @xmath296 ( i am using the common notation @xmath297 in differential geometry for the boundary of a manifold ) . from rotational symmetry , the electric field @xmath287 of a point charge
is directed radially outward and lies normal to the surface @xmath298 at each point .
we will denote the magnitude of the electric field at radius @xmath280 by @xmath299 .
recall that in ( [ areanu ] ) , we calculated the hyper - area of the @xmath300-sphere @xmath296 to be @xmath301 with @xmath302 . since the divergence theorem holds in an arbitrary euclidean space ( like the laws of physics , there is nothing in the divergence theorem to restrict the dimensionality of space to three ) , we can write @xmath303 at position @xmath288 , the electric field therefore takes the form @xmath304 where i am using the notation @xmath305 , with @xmath306 being a unit vector pointing in the direction of @xmath307 and @xmath308 being the magnitude .
it is often more convenient to work with the electric potential , a scalar quantity @xmath309 defined by @xmath310 .
in fact , we need the potential energy , so that @xmath311 with @xmath312 for @xmath220 we have @xmath313 , which is the origin of the @xmath314 of rationalized units . note from ( [ vnu ] ) that the engineering unit of electric charge is a function of the dimension @xmath99 .
this is because the @xmath99-dimensional coulomb potential ( [ vnu ] ) must have units of energy , and consequently the engineering unit of @xmath315 is @xmath316 .
it is quite natural that a composite quantity , made from the fundamental units of mass , length and time , change its engineering dimension with the dimension of space . .
long - distance or infrared ( ir ) physics dominates when @xmath284 .
uv and ir physics are equally important in @xmath257 .
, title="fig : " ] -0.8 cm figure [ fig : coulomb ] shows the coulomb potential for @xmath220 , along with two representative dimensions on either side of @xmath220 .
as the figure illustrates , the short - distance or uv behavior of the coulomb potential becomes more severe as the dimension increases above @xmath220 , while the long - distance or ir behavior dominates for dimensions below @xmath257 .
the arbitrary integration constant for the potential energy has been adjusted in each case so that all three graphs intersect at a single point . this was purely for aesthetics , as it renders the differences between the potentials more apparent . despite the trouble we went through in the previous paragraph to find the coefficients @xmath317 , in this paragraph ( and only in this paragraph )
i have temporarily set @xmath318 .
this will make it easier to compare the @xmath280-dependence of various potentials , and i would rather opt for clarity over notational consistency . for the representative potential with dimension below @xmath257 , i chose to graph the one - dimensional potential rather than the two - dimensional potential , where @xmath319 is an arbitrary integration constant with units of length . in both cases the potential increases without bound at long distances , thereby illustrating the dominance of long - distance or ir physics in dimensions less than three
however , i chose to graph @xmath320 rather than @xmath321 because the latter possesses a potentially misleading divergence as @xmath322 : unlike the short - distance or uv divergence associated with the potential in dimensions greater than three , the @xmath323 divergence of @xmath321 is integrable , and consequently harmless
. therefore , for purposes of illustration , the potential @xmath324 makes the point better than @xmath325 .
remarkably , we now see that by simply selecting the dimension @xmath99 , we can dial a potential @xmath326 that filters either long - distance or short - distance physics .
( 120,120)(-75,0 ) ( -101,60)(1,0)220 ( 0,0.5)(0,1)60 ( -2.3,0)@xmath164 ( -2.3,-8)@xmath6 ( -10,30)@xmath327 ( 0,60.5)(1,0)30 ( 10,65)@xmath328 ( -90,65)projectile ( 0,0.5)(1,2)30 ( 31,52)@xmath329 ( 1.9,15)@xmath330 ( 20,25)@xmath331 i would now like to show how the pole at @xmath257 arises from the @xmath99-dimensional coulomb potential . in the temperature equilibration process , individual plasma species exchange energy by mutual coulomb interactions .
for example , consider a particle in the plasma with charge @xmath329 , mass @xmath332 , and velocity @xmath333 , and suppose it passes another charge @xmath6 with an impact parameter @xmath327 .
this is illustrated in fig .
[ fig : scat ] , where , to zeroth order , the projectile follows the straight line @xmath334 as a function of the time @xmath335 , with @xmath336 .. for this , we can work with a hot dilute plasma where a linear trajectory will suffice .
] in @xmath99-dimensional space , the coulomb potential is given by ( [ vnu ] ) , and the corresponding electric field @xmath287 by ( [ e ] ) , with @xmath88 replaced by @xmath6 .
the projectile therefore acquires a momentum transfer @xmath337 and it suffers a corresponding change in energy @xmath338 the component of the electric field along the direction of motion @xmath333 integrates to zero in ( [ delp ] ) , while the component normal to the trajectory gives the impulse @xmath339 & \sim & \frac{1}{b^{\nu-2}}\ , \hat{\bf b } \ . \label{delp}\end{aligned}\ ] ] the temperature equilibration rate contains a factor involving the cross - section weighted energy transfer , and we see that the energy exchange between @xmath340 and @xmath25 can be written @xmath341 where we have used the fact that @xmath342 .
i will elaborate further on this example in the next lecture , but for now one should simply note that the rate ( [ dedtsim ] ) implies that large @xmath99 is dominated by short - distance physics and small @xmath99 is dominated by long - distance physics .
moreover , expression ( [ dedtsim ] ) gives @xmath343 as the dividing line between these two regions . to see this
, note that for @xmath344 the impact parameter integral is not sensitive to the large distance cut off , and we may simply take the limit @xmath345 to obtain @xmath346 conversely , for @xmath347 , we can set @xmath348 , with @xmath349 the results displayed are the dominant forms in the two regions of spatial dimensionality about @xmath257 . in either case , the stopping power contains a pole @xmath350 .
there are a number of important consequences arising from the uv and ir behaviors of the coulomb potential ( [ vnu ] ) .
i will discuss this more fully in section [ sec : dimred ] , but for now recall that the derivation of the boltzmann equation , as presented in ref .
@xcite for example , breaks down for the coulomb potential in three spatial dimensions because of the aforementioned infrared singularity .
however , in @xmath283 the `` textbook derivation '' of the boltzmann equation with the coulomb potential ( [ vnu ] ) is finite and completely rigorous .
the simple pole at @xmath257 in the scattering kernel corresponds to an ir divergence because of the long - range nature of the coulomb force in three dimensions .
furthermore , because dimensions greater than three enhance the uv physics , the classical born , bogoliubov , green , kirkwood , and yvon ( bbgky ) hierarchy reduces to the boltzmann equation to leading order in @xmath59 ( or to leading order in the number density ) when @xmath283 .
a similar reduction from bbgky holds for the lenard - balescu equation in @xmath347 , and the `` textbook derivation '' @xcite is also rigorous in these dimensions . in @xmath257 ,
the derivations of the boltzmann and lenard - balescu equations break down for the coulomb potential .
this is not because the @xmath257 version of the bbgky hierarchy is divergent , but because the truncation procedure that leads to the boltzmann and lenard - balescu equations breaks down for the three dimensional coulomb potential . indeed , bbgky is completely finite for the coulomb potential in @xmath257 , albeit completely useless for our purposes .
section ii of bps @xcite provides more details , especially the two paragraphs between eqs .
( 2.7 ) and ( 2.8 ) .
the lamb shift is interesting for us because it provides another connection with the coulomb logarithm , both in the historical details and in much of the physics .
the lamb shift is a small energy split in the otherwise degenerate @xmath351 and @xmath352 states of total angular momentum @xmath353 .
the measured value is about @xmath354 , with the @xmath355 state lying above the @xmath356 state .
therefore , when an electron makes a transition from the @xmath357-state to the @xmath358-state , it emits a microwave photon of frequency @xmath359 . calculating the observed value of the lamb shift was the first great success of quantum electrodynamics ( qed ) , the relativistic quantum theory of light and matter . in terms of the bohr magneton @xmath360 by introducing the dimensionless @xmath59-factor : @xmath361 . using relativistic single - particle quantum mechanics , in 1928
dirac predicted @xmath362 ( exactly ) . in 1948
schwinger used qed to calculate the radiative corrections , and he found @xmath363 , which was in excellent agreement with experiment .
today , the electron s magnetic dipole moment has been calculated to include @xmath364 terms , and is in agreement with experiment to 10 significant figures , the most accurately verified quantity in the history of physics . ]
dirac s relativistic model of the hydrogen atom , _
i.e. _ his relativistic theory of the electron in a coulomb potential , made the prediction that the energy levels of the hydrogen atom ( neglecting the hyperfine structure ) depend only upon the principal quantum number @xmath213 and the total angular momentum @xmath365 ( the sum of the orbital momentum @xmath366 and the spin @xmath367 of the electron ) .
in particular , dirac predicted that the @xmath355 and @xmath356 states should be degenerate . at the 1947 shelter island conference in new york
, w. lamb and r. retherford announced the results of their highly sensitive experiment measuring the emission frequency of photons in a @xmath351-@xmath352 transition , thereby establishing the unequivocal experimental existence of the lamb shift . while almost degenerate compared to the binding energy of the hydrogen atom , today s accepted experimental splitting
is @xcite @xmath368 \delta \nu_\text{\,lamb } & \equiv & \frac{\delta e_\text{lamb}}{h } = 1057.845(9)\,{\rm mhz } \ , \end{aligned}\ ] ] where plank s constant in the form @xmath369 is the conversion factor between energy and frequency .
hans bethe had attended the shelter island conference , and on the train ride back he performed the first calculation of the lamb shift , finding the value .
bethe s calculation was recognized as being only a rough approximation , neglecting high energy relativistic effects , but its close agreement with experiment was cause for optimism . the dominant contribution to
the lamb shift comes from the bound @xmath351 and @xmath352 electrons exchanging virtual photons with the atomic nucleus .
there are also several other subdominant mechanisms , such as the vacuum polarization of the photon and the anomalous magnetic dipole moment of the electron .
these give , respectively , the contributions @xmath370 and @xmath371 . ]
these radiative corrections effectively smear the point - like nature of the nucleus , thereby altering the atomic energy levels ( the electron no longer `` sees '' a pure coulomb potential ) .
any possible number of photon exchanges with the nucleus are permitted , from a single high - energy photon ( hard / uv physics ) to many low - energy photons ( soft / ir physics ) , and this means there are two disparate but competing energy scales in the problem . in a manner similar to the coulomb energy - loss rate ( [ ratemodel ] ) , the lamb shift takes the form @xmath372 the uv scale is set by @xmath373 and the ir scale @xmath374 is set by the binding energy of the hydrogen atom @xmath375 to understand the origin of these scales , note that @xmath376 takes its value from the energy at which relativistic effects for the electron become important ( the rest - mass energy of the electron ) , while @xmath374 takes its value from the only low - energy scale in the problem ( namely , the binding energy of the atom ) .
the soft photon ir interactions can be handled by nonrelativistic means , while the hard uv photons require a more complicated relativistic treatment .
for this reason , bethe used the simpler nonrelativistic formalism , cutting his calculation off at the relativistic energy scale @xmath377 at which his formalism broke down .
this is akin to cutting the calculation of the energy exchange rate off at some small distance scale @xmath71 .
bethe then concentrated on the low energy theory , which is analogous to looking at the lenard - balescu equation in the rate calculation .
bethe found a logarithmic uv divergence , in the same way the lenard - balescu equation has a uv divergence , but he was able to regularize the infinity by applying a technique known as mass renormalization , thereby rendering his calculation finite .
his calculation was completely rigorous at low energies , and bethe found the coefficient @xmath73 and the low energy cutoff @xmath374 , both _ exactly _
@xcite : @xmath378 the binding energy @xmath379 of the hydrogen atom is given by ( [ e0 ] ) , and while i will not write it down , the coefficient @xmath380 is rigorously defined in terms of sums of matrix elements of the various intermediate states .
these matrix elements are sufficiently complicated that one can only calculate them numerically , and to three significant figures bethe found @xmath381 .
bethe s exact calculation of @xmath374 is akin to an exact calculation of @xmath70 in the rate problem .
this , however , is a point where the analogy is not precise : bethe was able to exactly calculate @xmath374 through the regularization procedure of mass renormalization , which has no counterpart in plasma physics . on the other hand ,
lyman spitzer was only able to estimate the value of the maximum impact parameter in the coulomb logarithm to be of order @xmath382 .
bethe was able to perform this feat because qed is a fundamental theory of nature , while the boltzmann equation is not . using ( [ e0 ] )
, we can express ( [ lambscematic ] ) as @xmath383 where @xmath384 is the fine structure constant .
bethe s calculation therefore gives @xmath385 or @xmath386 .
bethe did not , however , calculate the _ exact _ coefficient under the logarithm . since his calculation broke down at relativistic energies , he used the somewhat _ ad hoc _ value ( rather than some multiple thereof ) for the uv cutoff .
it turns out that the high energy corrections to ( [ delebethe ] ) are rather small , so bethe s result was perhaps more accurate than warranted .
since bethe only estimated the maximum energy cutoff , his result ( [ delebethe ] ) is only accurate to leading order in @xmath387 , and the constant under the logarithm required a more accurate treatment .
the analogy further continues : bethe calculated the lamb shift to leading order , just as spitzer calculated the coulomb logarithm to leading order .
both calculations found the correct coefficient in front of the logarithm , and the order of magnitude of the argument of the logarithm .
these calculations failed to extract the exact constant under the logarithm , although bethe managed to find the exact expression for the low - energy cutoff . in this sense , bethe is the spitzer of the lamb shift . shortly after this , r. feynman and j. schwinger independently calculated the high energy contribution using their respective relativistically covariant formalisms , and their calculations were exact to leading and next - to - leading order in @xmath387 . after adding their contribution of the high energy corrections to bethe s low energy form , the calculation of the lamb shift was complete and the constant under the logarithm was fully determined .
or so it would seem .
simultaneously , j. french and v. weisskopf completed an independent calculation of the lamb shift that disagreed with feynman - schwinger , albeit only slightly in the coefficient under the logarithm .
the plot thickened as yet another calculation of the lamb shift was completed that agreed with french - weisskopf , this time from n. kroll and lamb himself .
so now we seem to have a real problem : the schwinger - feynman calculations agree with each other , but disagree with the french - weisskopf and kroll - lamb calculations .
schwinger and feynman had developed independent but equivalent formalisms that manifestly exhibited the relativistic covariance of the theory , while the other four had used a cumbersome formalism developed in the 1930 s ( now called _ old - fashioned perturbation theory _ ) . as it turns out ,
schwinger and feynman had made the same subtle mistake , and weisskopf s calculation was correct .
the contribution to the lamb shift we have been considering has the form @xmath388 where @xmath389 is bethe s low energy result ( [ kbethe ] ) . as with the bps coulomb logarithm
, we can bring the additive constant in ( [ lambcorrect ] ) inside the logarithm , and using ( [ e0 ] ) we can express ( [ lambcorrect ] ) as @xmath390 in terms of the bethe s coefficient @xmath391 of ( [ delebethe ] ) , the complete lamb shift coefficient is ( apart from the photon vacuum polarization and the electron dipole moment corrections , which were mentioned in footnote [ foot : othercorrections ] , but will not concern us further ) .
french spent the next year tracing down the origins of the discrepancy between the calculations .
he could not find an error in the feynman - schwinger high energy calculation , nor was there an error in bethe s low energy calculation .
instead , the error was in the way the high - energy calculation of feynman - schwinger was `` joined '' onto the low - energy result of bethe .
the high - energy calculation is also infinite and in need of regularization , but this time it contains an ir divergence .
this is analogous to the boltzmann equation containing the correct short - distance physics , but nonetheless diverging at long distances .
therefore , feynman - schwinger had to introduce an intermediate step in which they regulated their high energy theories in the infrared .
they chose to do this by assigning a small mass to the photon , and then taking this mass to zero at the end of the calculation . while this is a common ir regularization scheme in qed , it is incommensurate with bethe s regularization scheme in his low energy calculation , where he used a simple cutoff procedure in a high - energy integral .
french - weisskopf and kroll - lamb had gotten the correct result because they calculated both the low - energy and the high - energy contributions using the _ same _ formalism , and therefore with consistent regularization schemes for both short and long distances . for
more on the consequences of regulating large and small scales in an incommensurate manner , see feynamn s footnote 13 in ref .
@xcite of this lecture , which i have quoted in the bibliography in its entirety .
that one must calculate the large and small scales in exactly the same manner is not a minor point , as underscored by the stature of the physicists who failed to realize its importance when they first calculated the lamb shift from the then fledgling theory of quantum electrodynamics ( qed ) .
there are a number of consistent regularization schemes in use today in qed and other field theories of nature , with dimensional regularization being one of the most popular and easy to use .
these regularization and renormalization schemes have allowed us to calculate a great many experimentally verified quantities , to extremely high precision , and there is no longer any doubt in their correctness .
while extending the dimension of space to complex values might at first seem unsettling , this procedure works . when dimensional continuation was first introduced into quantum field theory , there were strong reactions against it .
however , calculations performed with this method agree with calculations using other regularization schemes , and more important , with experiment . in time , particle physicists learned to accept the notion that one can perform correct three dimensional calculations by working in arbitrary complex dimensions . for our concerns , we note that refs .
@xcite and @xcite have indeed calculated the lamb shift using the method of dimensional continuation .
the problem of feynman - schwinger is avoided , and dimensional continuation gives the correct experimentally observed result with much less effort than more traditional methods . as the above discussion illustrates , this is no small achievement . in summary
, dimensional continuation is powerful because ( i ) it is a consistent regularization scheme that ( ii ) lends itself to a perturbative analysis , and ( iii ) requires relatively simple ( or at least easily learned ) calculational tools .
dewitt s calculation @xcite was certainly consistent , in that it matched the long and short - distance physics commensurately [ to order @xmath392 .
this is because he starts with a finite regulated theory that treats long- and short - physics together within a single framework , albeit with a regularization scheme that does not lend itself to a systematic perturbative analysis . by comparison with the bps result , we know dewitt was accurate to order @xmath90 , inclusive .
but , as illustrated by feynman s footnote 13 of ref .
@xcite , and by the story of the lamb shift above , any attempt at treating the long- and short - distance physics by separate regularization schemes ( as much of the plasma literature currently does ) , will likely miss the very constants they are trying to calculate .
recall , it took two correct calculations to find the error in feynman s single calculation .
dimensional continuation performs all necessary book keeping , at both small and large scales , and in a manner that affords simple calculations and perturbative expansions .
( 120,120)(-75,0 ) ( -20,80)(100,50 ) ( -40,90 ) ( -55,70 ) ( @xmath99 ) ( -50,55)(-1,-1)30 ( 14,55)(1,-1)30 ( -100,40)@xmath410(38,40)@xmath411 ( 50,5 ) ( 43,-10 ) ( 78,0)(100,50 ) ( -150,7 ) ( -150,-8 ) ( -113,0)(100,50 ) ( -20,55)(0,-1)80 ( -32,-35)@xmath220 ( -45,-60 ) ( -17,-60)(80,35 ) note that this is a first - principles derivation of the rates ( [ dedtgtthree ] ) and ( [ dedtltthree ] ) in their respective dimensions @xmath99 .
let me reiterate the argument once again , although with a slightly different emphasis . for simplicity ,
we consider the purely classical regime first , adding quantum mechanics in a moment .
the classical bbgky hierarchy for the coulomb potential in three spatial dimensions is well defined and finite .
we run into trouble only when we attempt to truncate the hierarchy and derive lower - order kinetic equations , such as the boltzmann and the lenard - balescu equations .
the necessity of truncated equations is of course clear ( even the three - body problem can not be solved analytically ) .
unfortunately , however , the coulomb potential in three dimensions produces divergent scattering kernels in these truncated equations . rather than creating a model of the ostensibly divergent scattering kernel , we shall instead systematically regulate the divergences by letting the spatial dimension depart from its empirically measured value of three .
logarithmic divergences in three dimensions then become simple poles of the form @xmath350 in arbitrary dimensions . as with the regularization procedure of quantum field theory ,
our starting point here is a _
well - defined and finite theory _ , albeit in @xmath99 dimensions ,
_ regularized in a consistent fashion at all length and energy scales_. let us now return to classical bbgky , but this time in a spatial dimension @xmath99 of arbitrary positive _ integer _ value ( we are not yet considering continuous values of @xmath99 ) . at first sight , the hierarchy equations in @xmath99 dimensions are just as useless as those in three are simply too many of them to solve .
however , if we are willing to work to leading order accuracy in the plasma coupling @xmath59 , which is quite accurate for a weakly to moderately coupled plasma , then : ( i ) in @xmath283 we can truncate bbgky to the @xmath99-dimensional boltzmann equation , and ( ii ) in @xmath284 we can truncate bbgky to the @xmath99-dimensional lenard - balescu equation .
quantum scattering effects in the plasma will not modify the @xmath99-dimensional lenard - balescu equation , but they will modify the @xmath99-dimensional boltzmann equation . since we require the leading order term in the rate to be exact , we must include two - body quantum scattering effects exactly ( but no more than two - body effects , since these are subleading fact , three - body and higher correlations and scattering effects , both classical and quantum , can and _ should _ be neglected to leading order in @xmath59 ) .
two - body quantum effects can be accounted for by replacing the classical cross section in the boltzmann scattering kernel with the the two - body quantum cross section ( this can be performed _
exactly _ since all scattering phase shifts @xmath66 are known for the coulomb potential in three dimensions ) .
these ideas will be illustrated in complete detail in the next lecture for a particularly useful but simple case .
we now return specifically to the electron - ion temperature equilibration rate of ( [ dedtei ] ) . to obtain the leading order in @xmath59 behavior when
@xmath283 , we calculate the rate using ( [ dedtgtthree ] ) . as our calculation in the next lecture
will reveal , this rate is proportional to @xmath90 ( or the number density ) and takes the form as we have already pointed out in footnote [ foot : smallg ] , the parameter @xmath219 given by ( [ gdefa ] ) is dimensionless only for @xmath257 . from definition ( [ gdef ] ) and the coulomb potential ( [ vnu ] ) , the proper _ dimensionless _ expansion parameter should be @xmath412 .
however , since @xmath413 , we can absorb factors of @xmath414 into any accompanying coefficient , such as @xmath415 in ( [ dedtonecal ] ) .
therefore , powers of the dimensionless coupling @xmath221 also count powers of the three dimensional coupling @xmath59 .
we are therefore free to think of the expansion , even in @xmath99 dimensions , in terms of the three dimensional parameter @xmath59 . ]
@xmath416 where , for definiteness , we have restored the electron - ion subscript to the rate , as in ( [ dedtei ] ) . in the next lecture
we shall calculate @xmath417 explicitly , but for now it will suffice to note that @xmath418 depends upon the spatial dimension @xmath99 and the quantum parameter @xmath67 .
the simple pole at @xmath257 reflects the long - distance or infrared divergence of the coulomb potential in three spatial dimensions , and it arises from an integral over the radial coordinate of the coulomb potential ( [ vnu ] ) . in a similar manner , the leading order behavior in dimensions @xmath284 is given by ( [ dedtltthree ] ) . in the next lecture , we will see that this takes the form @xmath419 there is no in @xmath420 since the leading order long - distance physics is purely classical .
in the next lecture , we will preform the integrals to establish ( [ dedtonecal ] ) and ( [ dedttwocal ] ) , thereby calculating the coefficients @xmath421 and @xmath420 exactly ( for notational simplicity , i will hereafter drop the from @xmath418 ) .
note that ( [ dedttwocal ] ) also contains a simple pole at @xmath257 arising from an integration of ( [ vnu ] ) , but this time the pole corresponds to missing short - distance physics or the ultraviolet divergence of the lenard - balescu equation in three dimensions . in general ,
if a three - dimensional integral diverges , then the corresponding integral will typically contain a simple pole of the form @xmath350 . in this way we can transform a divergent integral into a convergent quantity that we can manipulate , and
this is how we shall regularize the divergent boltzmann and lenard - balescu equations in three dimensions .
since the rates @xmath422 of ( [ dedtonecal ] ) and @xmath423 of ( [ dedttwocal ] ) were calculated in mutually exclusive dimensional regimes , one might think that they can not be compared .
however , even though ( [ dedttwocal ] ) was originally calculated in @xmath284 for integer values of @xmath99 , we can analytically continue the result to complex values of @xmath99 ( in the same way that the factorial function , which operates on positive integers , can be generalized to the gamma function , which operates on the whole complex plane ) .
in fact , if we continue @xmath424 to real values of @xmath99 with @xmath283 , then we can directly compare ( [ dedtonecal ] ) and ( [ dedttwocal ] ) .
upon writing the @xmath59-dependence of ( [ dedttwocal ] ) as @xmath425 , when @xmath283 we see that ( [ dedttwocal ] ) is higher order in @xmath59 than ( [ dedtonecal ] ) . by power
counting arguments , there are no powers of @xmath59 in ( [ dedtonecal ] ) between @xmath90 and @xmath426 for @xmath283 , and therefore ( [ dedttwocal ] ) indeed provides the correct next - to - leading order term in @xmath59 when the dimension is analytically continued to @xmath283 , @xmath427 where i have written the exponent of @xmath59 as @xmath428 rather than @xmath429 .
this is illustrated in fig .
[ fig : nlodedt ] . as we shall see in the next section
, the individual pole - terms in ( [ dedtonecal ] ) and ( [ nlogterm ] ) will cancel giving a finite result when the leading and next - to - leading order terms are added .
the resulting finite quantity will therefore be accurate to leading and next - to - leading order in @xmath59 as the @xmath107 limit is taken .
alternately , we could have analytically continued ( [ dedtonecal ] ) to dimensions @xmath347 , where it would become next - to - leading order relative to ( [ dedttwocal ] ) . in either case , the leading and next - to - leading order contribution can be found by simply adding ( [ dedtonecal ] ) and ( [ dedttwocal ] ) and then taking @xmath107 .
( 120,120)(-75,0 ) ( -101,60)(1,0)230 ( 134,58)@xmath99 ( 0,56)(0,1)8 ( -2.3,57.3)@xmath164 ( -2.0,49)@xmath430 ( -60,66)@xmath423 ( -80,44)@xmath431 ( -60,30)@xmath167 ( -80,15)@xmath432 ( -120,0)lo : large when @xmath433 ( 30,66)@xmath423 ( 30,44)@xmath434 ( 40,30)@xmath167 ( 40,15)@xmath435 ( 20,0)nlo : small when @xmath433 ( 0,75)(45,45)[t ] ( 22,72)(0,-1)1 ( -40,115)analytically continue ( -42,105)around the @xmath220 pole to find the three dimensional rate @xmath436 , accurate to leading and next - to - leading order in the plasma coupling @xmath59 , we add the leading order expression ( [ dedtonecal ] ) for @xmath437 to the next - to - leading order expression ( [ nlogterm ] ) for @xmath423 , and then take the limit @xmath438 : @xmath439 + { \cal o}(g^3 ) \ . \label{dedtorderg3}\end{aligned}\ ] ] for the same reasons as given in section [ sec : assembly ] , this does not lead to any form of `` double counting . ''
instead , we are simply adding the next - to - leading order term ( [ nlogterm ] ) to the leading order term ( [ dedtonecal ] ) at a common value of .
since ( [ dedtorderg3 ] ) lies at the heart of dimensional continuation , allow me review the reasoning behind this expression one final time .
recall from ( [ dedtonecal ] ) and ( [ dedttwocal ] ) that @xmath422 and @xmath423 are both leading order in @xmath59 for @xmath283 and @xmath284 , respectively .
since these functions were calculated for mutually exclusive values of @xmath99 , they must be analytically continued to the same value of @xmath99 for purposes of comparison . in equation ( [ dedtorderg3 ] ) ,
we have chosen to continue @xmath423 to @xmath283 , which takes the same algebraic form as it did for @xmath284 .
the analytic continuation has the effect of rendering @xmath423 subleading in @xmath59 relative to @xmath422 .
there are no powers of @xmath59 between these two terms for any value of @xmath99 , and therefore the limiting procedure @xmath438 gives the three dimensional rate _
exactly _ to leading and next - to - leading order accuracy in @xmath59 .
we have therefore found the leading order and next - to - leading order in @xmath59 contributions for @xmath283 , @xmath440 + { \cal o}(\nu-3 ) \ . \label{nlosum}\end{aligned}\
] ] i have not indicated the higher order error in @xmath59 , but instead i am emphasizing here the error associated with the @xmath99-expansion .
that is to say , ( [ nlosum ] ) contains an implicit error that approaches @xmath441 as @xmath438 , while i have chosen to explicitly display the first - order error in @xmath442 . in the sum ( [ nlosum ] ) , the error in @xmath59 remains nonzero in the three dimensional limit , while the @xmath99-error in ( [ nlosum ] ) vanishes as @xmath107 . in taking the @xmath107 limit
, we must keep in mind that terms proportional to @xmath443 will give a non - vanishing result when multiplied by the pole @xmath350 .
upon expanding the to linear order in @xmath444 we find @xmath445 which can be written as @xmath446 we must now expand the coefficients @xmath421 and @xmath420 in powers of @xmath447 . as we shall see in the next lecture , and this is a _ crucial _
point , the leading order in @xmath448 terms are identical , and so the expansions take the form @xmath449
g(\nu ) & = & -a + \epsilon \,g_1 + { \cal o}(\epsilon^2 ) \ . \label{gexp}\end{aligned}\ ] ] in the next lecture , we will perform the integrals in ( [ dedtgtthree ] ) and ( [ dedtltthree ] ) , thereby allowing us to exactly compute @xmath450 and the linear terms @xmath451 and @xmath452 .
the remaining procedure is now straightforward . upon using ( [ gexp])([gexp ] ) in ( [ dedtonecal ] ) and ( [ nlogterm ] ) , the leading and next - to - leading order terms in @xmath59
now take the form @xmath453
\frac{d{\cal e}_{e{{\rm\scriptscriptstyle i}}}^{{\scriptscriptstyle < } } } { dt } & = & \phantom{- } \frac{a}{\nu-3 } \ , g^2 - a\ , g^2 \ln g - g_1 \,g^2 + { \cal o}(\nu-3;g^3 ) \ . \label{localcb}\end{aligned}\ ] ]
i have now indicated here the error associated with the @xmath59-expansion .
because the leading order terms in ( [ hexp ] ) and ( [ gexp ] ) are equal , the simple poles in ( [ localca ] ) and ( [ localcb ] ) cancel .
therefore , the limit @xmath438 of expression ( [ nlosum ] ) gives the rate @xmath454 with @xmath455 , in agreement with ( [ dedtnlo ] ) . in this way
, bps has calculated the energy exchange accurate to leading order and next - to - leading order in @xmath59 .
i hope this account of the simple field theoretic apparatus necessary for a reading of bps has been helpful .
having understood the reasoning behind bps , especially the manner by which analytic continuation turns a leading order result into a next - to - leading order result , we can summarize the bps procedure by the following prescription : 1 .
calculate the rate @xmath456 in the regime @xmath283 using the boltzmann equation , including quantum corrections to all orders in @xmath67 .
this captures the leading order physics in dimensions greater than three . 2 .
calculate the corresponding rate @xmath457 in the regime @xmath284 using the lenard - balescu equation .
this physics is classical , and captures the leading order behavior in dimensions less than three .
add the terms @xmath456 and @xmath457 and take the @xmath107 limit .
this gives the energy exchange rate in three dimensions accurate to leading and next - to - leading order in @xmath59 , @xmath458 .
that is to say , this gives the coefficients @xmath3 and @xmath4 exactly .
this prescription provides complete analytic expressions for the coefficients @xmath3 and @xmath459 , with @xmath459 being of particular interest . in the language of the coulomb logarithm , we write @xmath460 .
the fundamental interactions of nature can be expressed as quantum field theories ( except for gravity , which to date has only been expressed as a classical field theory , _
i.e. _ general relativity ) .
for this reason , quantum field theory reveals something very deep about the structure of nature .
however , for our purposes , we only need to think of quantum field theory as an elegant collection of tools for packaging and solving many - body problems , and plasma physics concerns itself with the many - body problem par excellence .
it should therefore come as little surprise that quantum field theory can be useful for plasma physics .
reference @xcite is a nice example of cross fertilization between two quite different branches of physics , and it is gratifying for someone who has worked in both particle and plasma physics that such seemingly different subjects can inform one another . perhaps the most dramatic difference between particle and plasma physics lies with their respective methods and outlooks , and not so much the subject matter itself .
a number of plasma physicist have explained to me that that plasma physics is not an _ exact _ science . while the nature of the subject might render a first - principles approach limited , particularly for subtopics like magneto - hydrodynamics in tokamaks or self - organized behavior in strongly coupled plasmas , i believe the plasma physicist could still benefit from the more rigorous outlook of particle physics .
conversely , the particle physicist could benefit from plasma physics .
in fact , until new high energy experiments come on - line , plasma physics may have more to offer particle physics than the reverse . until recently ,
high energy physics was a spectacular field in which to work , but , with a few exceptions such as lattice gauge theory and rhic physics at bnl , the field seems to have been milked dry .
high energy physics is in dire need of direct experimental input .
plasma physics , on the other hand , still has many interesting and quite challenging problems of experimental and astrophysical relevance , and a number of these seem tailor - made for the field theorist .
the particle physicist could therefore benefit by an excursion ( or even a longer stay ) in plasma physics .
hence , my hidden agenda in these lectures : it would be quite nice to entice a few particle theorists to work on some of the interesting problems of plasma physics
. it would also be nice to convince a few plasma physicists that particle theory has something to offer them as well .
this manuscript arose from a series of lectures presented to the applied physics division s plasma working group in august 2006 at the los alamos national laboratory .
i would kindly like to thank guy dimonte for providing the opportunity to speak , and the other members of the plasma working group for their questions and interest during the talks .
i would particularly like to thank jerome daligault for reading the manuscript for clarity and accuracy , and lowell brown for a number of useful discussions .
this work was funded by the icf codes projects ja1k and jl1c , and the thermonuclear burn project javd .
l. s. brown , d. l. preston , and r. l. singleton jr .
, _ charged particle motion in a highly ionized plasma _ , phys .
* 410 * ( 2005 ) 237 - 333 , e - print archive physics/0501084 , los alamos preprint la - ur-042713 .
l. s. brown , d. l. preston , and r. l. singleton jr . , _ plasma stopping power including subleading order _ , proceedings of international conference on strongly coupled coulomb systems ( sccs 2005 ) , moscow , russia , june 2005 , physics/0509168 , la - ur-05 - 6527 .
g. n. watson , _ theory of bessel functions _ , @xmath462 ed .
, cambridge univ .
press , 1952 , section 3.71 ; i. s. gradshteyn and i. m. ryzhik , _ tables of integrals , series , and products _ , corrected and enlarged ed . , academic press , 1980 , formula 8.447.3 .
f. carlson , _ sur une classe de sries de taylor _ , ( 1914 ) dissertation , uppsala , sweden .
the theorem is proved , for example , in e. c. titchmarsh , _ the theory of functions _
, 2nd ed . , oxford univ
. press ( london , 1960 ) , sections * 5.8 * and * 5.81*. j. j. sakurai , _ advanced quantum mechanics _
, addison - wesley , @xmath465 printing , 1982 , p 71 ; see also pp 280 , 293 .
another classic treatment is j. d. bjorken and s. d. drell , _ relativistic quantum mechanics _
, mcgraw - hill , 1964 , section * 8.7*. r. p. feynman , _ space - time approach to quantum electrodynamics _ , phys . rev . * 76 * ( 1949 ) 769 , footnote 13 , which i will quote in it entirety : 0.1 cm that the result given in b [ r.p .
feynman , phys .
74 , 1430 ( 1948 ) ] in eq .
( 19 ) was in error was repeatedly pointed out to the author , in private communication , by v.f .
weisskopf and j.b .
french , as their calculation completed simultaneously with the author s early in 1948 gave a different result .
french has finally shown that although the expression for the radiationless scattering in b , eq . ( 18 ) or ( 24 ) above is correct , it was incorrectly joined onto bethe s non - relativistic result .
he shows that the relation @xmath466 used by the author should have been @xmath467 .
this results in adding a term @xmath468 to the logarithm in b , eq .
( 19 ) so that the result now agrees with that of j.b .
french and v.w .
weisskopf , phys .
75 , 1240 ( 1949 ) and n.h .
kroll and w.e .
lamb , phys .
75 , 388 ( 1949 ) .
the author feels unhappily responsible for the very considerable delay in the publication of french s result occasioned by this error .
this footnote is appropriately numbered .
|
plus 1pt minus 1pt0.3 cm this is the first of two lectures on the technique of dimensional continuation employed by brown , preston , and singleton ( bps ) to calculate such quantities as the charged particle stopping power and the temperature equilibration rate in a plasma . in this exposition
we will examine some of the more basic points of dimensional continuation , with an emphasis on the coulomb logarithm for electron - ion temperature equilibration .
dimensional continuation , or dimensional regularization as it is more properly known in quantum field theory , was originally developed as part of the renormalization procedure for the theories of the electroweak and other fundamental interactions in particle physics .
dimensional continuation is so general , in fact , that any _ theory _ can be unambiguously lifted to dimensions beyond three , and therefore the technique is powerful enough to apply in many other settings .
the technique , however , is not well known outside the field theory and particle physics communities .
this exposition will therefore be self - contained , intended for those who are not specialists in quantum field theory , and i will either derive or motivate any requisite field theory results or concepts . of particular relevance is the analogy between the coulomb logarithm as calculated by lyman spitzer on the one hand , and the lamb shift as calculated by hans bethe on the other . while dimensional continuation is a well developed and a thoroughly tested method for regularizing any quantum field theory ,
bps employs the method in a novel way that provides the leading and subleading behavior for processes that involve competing disparate energy or length scales .
in particular , bps calculated the temperature equilibration rate to leading and next - to - leading order in the plasma number density for any two species in a plasma that are in thermal equilibrium with themselves , but not necessarily with each other .
no restriction is made on the charge , mass , or the temperature of the plasma species .
it is , however , assumed that the plasma is not strongly coupled in the sense that the dimensionless plasma coupling parameter @xmath0 is small , where @xmath1 is the debye wave number of the plasma . to leading and next - to - leading order in this coupling ,
the temperature equilibration rate is of the generic form @xmath2 . the precise numerical coefficient @xmath3 in front of the logarithm has been known for some time , while bps have recently computed the constant @xmath4 under the logarithm .
it should be emphasized that the bps result is not a model , but rather it is an exact calculation of the leading terms in a well - defined perturbation theory .
this exact result differs from approximations and models given in the literature . 0.2 cm plus 1pt minus 1pt plus 1pt minus 1pt
|
the recent cleo experiments related with and supply possibilities to extract the information about the structures of these two pseudoscalar mesons . for example , the cleo collaboration @xcite has reported very large branching ratios for inclusive production @xmath3 and for the exclusive decay @xmath4 @xmath5 to explain the abnormally large production of in the standard model , either large portion of intrinsic @xmath6 component in @xcite or large coupling of to @xmath7 @xcite , even both of them , must be concluded . at the same time , the data of the @xmath8 , @xmath9 and @xmath10 transition form factors at higher energies reported by cleo @xcite suggest that may have very different non - perturbative properties from @xmath11 and , while the latter two have similar wave functions .
it is reasonable because the physical and states consist dominantly of flavor su(3 ) octet @xmath12 and singlet @xmath13 , respectively .
the usual mixing scheme reads [ mixing ] latexmath:[\ ] ] from the bhl model .
the transition form factor of su(3 ) singlet or octet is @xmath92 where @xmath93 or 8 presents the singlet or octet .
the charge factors are @xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 , and @xmath99 .
the hard scattering amplitude @xmath100 is the same as eq .
( [ hard ] ) and @xmath101 is the bhl wave function , eq .
( [ bhlw ] ) , for each flavor quark with the corresponding @xmath102 , @xmath103 and @xmath104 as input parameters .
the quark mass @xmath103 in the factor from the melosh rotation will be adopted as @xmath105 mev or @xmath106 mev , depending on the flavor in @xmath101 .
summing over all flavors , we obtain the transition form factors of physical @xmath0 and @xmath1 states from eq .
( [ form ] ) [ lcphy ] @xmath107 which , to ensure the feasibility of perturbation theory , should be valid only for large @xmath25 . while fitting the data , we will choose only the data points with @xmath108 gev@xmath109 .
it is interesting to note that the form factors @xmath110 and @xmath111 follow the same pattern of mixing as the state , eq .
( [ mixing ] ) .
as @xmath112 , the gauge invariance requires that eq .
( [ lcphy ] ) contributes exact one half to @xmath113 .
inputing a mixing angle @xmath79 , @xmath114 and @xmath115 can be fixed by @xmath113 .
the theoretical predictions obtained from eq .
( [ corrected ] ) are shown in fig . 2 .
the best fit value of @xmath79 is @xmath116 .
however , we find that the differences between different choices of @xmath79 are so small that it is in fact unreliable to determine the mixing angle , e.g. @xmath117 can not be excluded by the data .
the reason is that @xmath113 , which we use as the normalization condition , shares the same mixing mechanism as the form factor @xmath118 .
the only difference comes from the different contributions between the @xmath90 quark and the @xmath119(@xmath120 ) quark at different @xmath25 .
if @xmath121 , the mixing will be meaningless in this approach .
the same conclusion also holds as @xmath48 , while all quarks can be treated as massless .
since the perturbation calculations are only valid for large @xmath25 , this minor difference is negligible . at present
, neither the data nor the theoretical approach has reached the accuracy to distinguish the @xmath90 or @xmath119(@xmath120 ) quark in the transition form factors of pseudoscalar mesons . therefore , unless other normalization conditions ( e.g. @xmath122 , @xmath123 and @xmath124 , and so on . )
are involved , it is unreliable to determine the mixing angle from the data of the transition form factors .
inclusion of other normalization conditions brings additional uncertainties and goes beyond the present work .
the intrinsic quark " is one of the novel properties for hadrons and there have been many examples where the non - valence intrinsic quark " components seem important @xcite .
the flavor singlet meson @xmath13 may have a large portion of @xmath2 component and strong coupling to gluons through qcd axial anomaly .
using operator product expansion and qcd low energy theorems , the non - perturbative intrinsic charm content of the meson was evaluated semi - quantitatively in ref .
@xcite to be @xmath125 mev , which suffices to explain the large production in @xmath126 decay reported by cleo .
furthermore , ref .
@xcite suggested @xmath127 mev to account for the data if @xmath2 pair is in color octet .
especially , the sign of @xmath128 was fixed as minus in their work . a similar size , @xmath129 mev ,
was argued in ref .
@xcite , too .
the gluonic components of @xmath13 play no role in electromagnetic transition form factors because the coupling of two photons to two gluons is very small .
they are invisible here .
however , the @xmath2 pair has distinct behavior from the light quark pairs . unlike the @xmath130 pair
, different sizes of @xmath2 component change the form factor a lot . in the following ,
the current mass and constituent mass of @xmath131 quark are treated as the same and we adopt @xmath132 gev .
the hard scattering amplitude for the heavy quark pair is @xmath133 it is noted that the helicity - flip amplitude can not be ignored since it is proportional to current quark mass .
a direct calculation gives the higher helicity contributions as @xmath134 the total contributions of all helicity components can be expressed as eq .
( [ lcsu3 ] ) by changing the hard scattering amplitude as @xmath135 the spacial wave function is written as @xmath136\ .\ ] ] the @xmath2 contributions to the transition form factor have very different @xmath25 behavior from light quark ones because of the higher helicity contributions , the heavy quark propagator , and the different wave function .
comparison with light quark contributions is shown in fig .
3 with their relative sizes at zero momentum transfer normalized to unity . this difference may result in observable effects in the transition form factors .
now we explore the form factor with the interested value @xmath137 mev .
the decay constant is connected with the wave function as the pionic case @xcite as @xmath138 the above decay constant is multiplied by a factor of @xmath139 comparing with the previous sections to be consistent with refs .
@xcite .
@xmath140 is easily obtained by the definition of the decay constant , eq .
( [ dc ] ) , with the input @xmath141 and the mixing angle @xmath79 .
the mixing angle is fixed at @xmath142 because it is , by our numerical results , still not sensitive even when @xmath2 is included .
at the same time , we neglect the difference between @xmath90 quarks and @xmath119(@xmath120 ) quarks in order to concentrate on @xmath131 quarks , i.e. , let @xmath143 mev .
this approximation is valid since the difference is minor , which can be seen in last section .
the obtained form factor , comparing with @xmath144 , is shown in fig .
the corresponding decay constants are @xmath145 these values for light quarks are similar to that with @xmath144 because the @xmath2 contributions at zero momentum transfer are very small and change the normalization condition only slightly .
they are not far from the naive expectation in su@xmath74(3 ) limit at @xmath142 : @xmath146 since the flavor singlet has different non - perturbative properties from that of the octet , the parameter @xmath86 is not necessarily the same as @xmath147 .
in fact , with @xmath137 mev , a very good fit can be obtained by adjusting @xmath148 to 0.3 gev@xmath68 rather than @xmath149 gev@xmath68 .
however , the decay constants will increase apparently for light quarks ( about 30% for @xmath119 and @xmath120 quark ) : @xmath150 this difference will result in apparent increasement to the radiative decay of mesons , such as @xmath151 , @xmath152 and @xmath153 .
choi and ji @xcite have found that the same value @xmath154 for all of these flavor nonet mesons can produce satisfied decay widths for these decay channels .
an increasement of 30% , if @xmath155 gev@xmath68 rather than @xmath85 , may exceed too much to fit the data of their radiative decays .
therefore the above numerical results disfavor such a choice and thus disfavor a large portion of @xmath2 component as @xmath156 .
there have been recently a number of investigations @xcite which support a small @xmath157 than that was estimated in ref .
our above conclusion is in agreement with these results from other considerations .
for example , in ref .
@xcite , @xmath158 was suggested . to check the consistency of such a possibility in our approach , we adjust the input to @xmath159 and present the result in fig . 4 .
the decay constants for different flavor quarks are @xmath160 with @xmath149 gev@xmath68 .
we notice that the obtained results are very close to the case of @xmath161 and our analysis allow a small @xmath157 around @xmath162 .
this is consistent with the results in refs .
the electromagnetic transition form factors of @xmath0 and @xmath21 are presented comparing with @xmath20 in this paper .
the numerical results show that there still exists a gap between the data and the light - cone perturbation calculation with the brodsky - huang - lepage wave function as the input for the non - perturbative aspects of the mesons . to model the higher fock state and higher order contributions
, we assume that the fraction size of these contributions are similar to all three mesons : pion , @xmath0 , and @xmath21 .
the ratio of the data of the @xmath8 transition form factor to the theoretical calculation is multiplied to the light - cone results of the @xmath9 and @xmath37 form factors .
with such a correction , the obtained transition form factors can be compared with the data . in su@xmath74(3 ) , it is in fact unreliable to determine the mixing angle at present unless additional normalization conditions other than the @xmath163 decay widths are included .
the reason is that the same mixing mechanism occurs in both the transition form factors and the normalization conditions .
both the experimental ( especially for @xmath1 ) and theoretical accuracies are not high enough to distinguish the @xmath119(@xmath120 ) and @xmath90 quark pairs in these pseudoscalar mesons at high @xmath25 .
the heavy quark pair has different @xmath25 behavior from the light ones .
it is possible to explore the size of the @xmath2 component in the flavor singlet .
our results disfavor a large portion of @xmath2 component in @xmath1 .
such a conclusion is in agreement with a number of recent investigations with a small portion of intrinsic charm in from other considerations .
b. behrens , talk presented at _ the second international conference on b physics and cp violation _ , march 24 - 27 , 1997 , honolulu , hawaii .
f. wrthwein , hep - ex/9706010 .
i. halperin , a. zhitnitsky , phys .
d * 56 * , 7247 ( 1997 ) .
for further development , see , e.v .
shuryak and a. zhitnitsky , phys . rev . *
d 57 * , 2001 ( 1998 ) .
cheng and b. tseng , phys .
b415 * , 263 ( 1997 ) .
d. atwood and a. soni , phys .
b405 * , 150 ( 1997 ) ; w.s .
hou and b. tseng , phys .
. lett . * 80 * , 434 ( 1998 ) . the cleo collaboration , j. gronbery _ et al .
_ , phys . rev . *
d 57 * , 33 ( 1998 ) .
venugopal , barry r. holstein , hep - ph/9710382 .
r. jakob , p. kroll , and m. raulfs , j. phys .
g * 22 * , 45 ( 1996 ) . h.n .
li and g. sterman , nucl . phys .
* b381 * , 129 ( 1992 ) . t. feldmann and p. kroll , hep - ph/9711231 .
choi and c .- r .
ji , nucl . phys .
* a618 * , 291 ( 1997 ) ; phys . rev . *
d 56 * , 6010 ( 1997 ) ; s.j .
brodsky , t. huang , g.p .
lepage , in _ particle and fields 2 _ , edited by a.z .
capri and a.n .
kamal ( plenum publishing corporation , new york , 1983 ) , p. 143 ; t. huang , _ proceedings of xx - th international conference on high energy physics _ ,
madison , wisconsin , 1980 , edited by l. durand and l.g .
pondrom , aip con .
69 ( aip , new york , 1981 ) , p. 1000 . v.v .
anisovich , d.i .
melikhov , and v.a .
nikonov , phys .
d * 55 * , 2918 ( 1997 ) .
cao , t. huang , and b .- q .
ma , phys .
d * 53 * , 6582 ( 1996 ) .
lepage and s.j .
brodsky , phys .
d * 22 * , 2157 ( 1980 ) ; * 24 * , 1808 ( 1981 ) . s. j. brodsky and g. p. lepage , phys .
rev . lett . * 53 * , 545 ( 1979 ) ; phys . lett . * 87b * , 359 ( 1979 ) . s. j. brodsky and g. p. lepage , _ perturbative quantum chromodynamics _ , edited by a. h. mueller ( singapore , world scientific , 1989 ) , p. 93 .
a. ali and c. greub , phys . rev .
* d 57 * , 2996 ( 1998 ) ; a. ali , j. chay , c. greub , and p. ko , hep - ph/9712372 .
t. feldmann , p. kroll , and b. stech , hep - ph/9802409 .
petrov , hep - ph/9712497 .
f. araki , m. musakhanov , and h. toki , hep - ph/9803356 .
e. wigner , ann . math . * 40 * , 149 ( 1939 ) ; h.j .
melosh , phys .
d * 9 * , 1095 ( 1974 ) .
chung , f. coester , b.d .
keister , and w.n .
polyzou , phys .
c * 37 * , 2000 ( 1988 ) .
t. huang , b .- q .
ma , and q .- x .
shen , phys .
d * 49 * , 1490 ( 1994 ) .
ma , z. phys .
* a 345 * , 321 ( 1993 ) ; b .- q .
ma and t. huang , j. phys .
* g 21 * , 765 ( 1995 ) .
cao , j. cao , t. huang , and b .- q .
ma , phys .
d * 55 * , 7107 ( 1997 ) .
ma , j. phys . * g 17 * , l53 ( 1991 ) ; b .- q . ma and q .- r .
zhang , z. phys .
* c 58 * , 479 ( 1993 ) ; s.j . brodsky and f. schlumpf , phys .
* b 329 * , 111 ( 1994 ) ; b .- q . ma , phys . lett . *
b 375 * , 320 ( 1996 ) ; p. kroll and m. raulfs , phys .
* b387 * , 848 ( 1996 ) ; s. ong , phys .
d * 52 * , 3111 ( 1995 ) ; a.v .
radyushkin and r.t .
ruskov , phys . lett . *
b374 * , 173 ( 1996 ) ; v. braun and i. halperin , phys . lett . *
b328 * , 457 ( 1994 ) ; a.p .
bakulev and s.v .
mikhailov , mod .
a11 , 1611 ( 1996 ) .
musatov , a.v .
radyushkin , phys .
d * 56 * , 2713 ( 1997 ) .
v.l . chernyak and a.r .
zhitnitsky , phys .
rep . * 112 * , 173 ( 1984 ) .
f. del aguila and m.k .
chase , nucl .
* b193 * , 517 ( 1981 ) ; e. braaten , phys .
d * 28 * , 24 ( 1983 ) ; e.p .
kadantseva , s.v . mikhailov and a.v .
radyushkin , sov .
. phys . * 44 * , 326 ( 1986 ) .
the l3 collaboration , cern - ppe/97 - 110 .
see , e.g. , s. j. brodsky , p. hoyer , c. peterson , and n. sakai , phys .
* b 93 * , 451 ( 1980 ) ; s.j . brodsky and b .- q .
ma , phys . lett . *
b 381 * , 317 ( 1996 ) ; s.j . brodsky and m. karliner , phys .
lett . * 78 * , 4682 ( 1997 ) ; and references therein .
f. yuan and k.t .
chao , phys .
d * 56 * , 2495 ( 1997 ) .
1 . the pion - photon transition form factor with different parameters . the dashed line is the form factor without melosh rotation .
the solid line is with melosh rotation .
the dashed - dotted line is the result as @xmath164 , while the evolution of the bhl wave function is neglected .
|
the electromagnetic transition form factors of @xmath0 and @xmath1 are calculated in the light - cone perturbation theory .
we show that it is unreliable to determine the -mixing angle without any additional normalization conditions other than their decay widthes to two photons .
the possible intrinsic @xmath2 component in the flavor singlet is investigated .
the heavy quark pair has distinct properties from the light ones in electromagnetic transition processes of pseudoscalar mesons .
it is possible to explore the size of @xmath2 component and our numerical results disfavor a large portion of @xmath2 component . /#1#1/
# 1#1/
|
cluster algebras have been introduced by fomin and zelevinsky around 2000 and have enjoyed a remarkable success story in recent years .
they attractively link various areas of mathematics , like combinatorics , algebraic lie theory , representation theory , algebraic geometry and integrable systems and have applications to mathematical physics . in an attempt to categorify cluster algebras ( without coefficients ) , cluster categories have been introduced by buan , marsh , reiten , reineke , todorov @xcite .
more precisely , these are orbit categories of the form @xmath6 $ ] where @xmath7 is a quiver without oriented cycles , @xmath8 is the bounded derived category of the path algebra @xmath9 ( over an algebraically closed field @xmath10 ) and @xmath11 and @xmath12 $ ] are the auslander - reiten translation and shift functor on @xmath8 , respectively .
remarkably , these cluster categories are again triangulated categories by a result of keller @xcite .
quivers of dynkin types @xmath13 play a special role in the theory of cluster algebras since they parametrize cluster - finite cluster algebras , by a seminal result of fomin and zelevinsky @xcite .
the corresponding cluster categories @xmath14 where @xmath7 is a dynkin quiver are triangulated categories with finitely many indecomposable objects and their structure is well understood by work of amiot @xcite .
important objects in cluster categories are the cluster - tilting objects .
a cluster - tilted algebra of type @xmath7 is by definition the endomorphism algebra of a cluster - tilting object in the cluster category @xmath15 .
the corresponding cluster - tilted algebras of dynkin types @xmath16 , @xmath17 and @xmath0 are of finite representation type and they can be constructed explicitly by quivers and relations .
namely , the quivers of the cluster - tilted algebras of dynkin type @xmath7 are precisely the ones obtained from @xmath7 by performing finitely many quiver mutations . moreover , in this case the quiver of a cluster - tilted algebra uniquely determines the relations @xcite ; we shall review the corresponding algorithm in section 2 below . in this paper
we address the question of when two cluster - tilted algebras of dynkin type @xmath1 , @xmath2 or @xmath3 have equivalent derived categories .
the analogous question has been settled for cluster - tilted algebras of type @xmath18 by buan and vatne @xcite ( see also work of murphy on the more general case of @xmath19-cluster tilted algebras of type @xmath18 @xcite ) and by the first author @xcite for type @xmath20 . note that the cluster - tilted algebras in these cases are gentle algebras @xcite .
it turns out that two cluster - tilted algebras of type @xmath18 are derived equivalent if and only if their quivers have the same number of 3-cycles . for distinguishing such algebras up to derived equivalence one uses the determinants of the cartan matrices ;
these have been determined explicitly for arbitrary gentle algebras by the second author in @xcite . a derived equivalence classification of cluster - tilted algebras of other dynkin types @xmath17 and @xmath0 has been open . in this paper
we settle this question for type @xmath0 , i.e. we obtain a complete derived equivalence classification for cluster - tilted algebras of types @xmath1 , @xmath2 and @xmath3 .
there are two natural approaches to address derived equivalence classification problems of a given collection of algebras arising from some combinatorial data .
the top - to - bottom approach is to divide these algebras into equivalence classes according to some invariants of derived equivalence , so that algebras belonging to different classes are not derived equivalent .
the bottom - to - top approach is to systematically construct , based on the combinatorial data , tilting complexes yielding derived equivalences between pairs of these algebras and then to arrange these algebras into groups where any two algebras are related by a sequence of such derived equivalences . to obtain a complete derived equivalence classification one has to combine these approaches and hope that the two resulting partitions of the entire collection of algebras coincide .
the invariant of derived equivalence we use in this paper is the integer equivalence class of the bilinear form represented by the cartan matrix of an algebra @xmath16 . as
this invariant is sometimes arithmetically subtle to compute directly , we instead compute the determinant of the cartan matrix @xmath21 and the characteristic polynomial of its asymmetry matrix @xmath22 , defined whenever @xmath21 is invertible over @xmath23 , and encode them conveniently in a single polynomial that we call the _ polynomial associated with @xmath21_. this quantity is generally a weaker invariant of derived equivalence , but in our case it will turn out to be enough for the classification .
note that unlike as in type @xmath16 , the determinant itself is not sufficient for distinguishing the algebras up to derived equivalence .
we stress that the asymmetry matrix and its characteristic polynomial are well defined whenever the cartan matrix is invertible over @xmath23 , even without having any categorical meaning , as follows from ( * ? ? ?
* section 3.3 ) . in the special case
when @xmath16 has finite global dimension , the asymmetry matrix @xmath24 , or better minus its transpose @xmath25 , is related to the coxeter transformation which does carry categorical meaning , and its characteristic polynomial is known as the coxeter polynomial of the algebra .
the tilting complexes we use are inspired by quiver mutations in the following sense . for a vertex ,
we consider all the incoming arrows and build , based on this combinatorial data , a two - term complex of projective modules , see also similar constructions in @xcite , @xcite .
we call a mutation at a vertex `` good '' if the corresponding complex is a tilting complex and moreover its endomorphism algebra is the cluster - tilted algebra of the mutated quiver . in other words , a `` good '' mutation produces a derived equivalence between the corresponding cluster - tilted algebras .
of course , there are also `` bad '' mutations , for two reasons : the complex might not be a tilting complex or even if it is , its endomorphism algebra might not be a cluster - tilted algebra .
it turns out that for cluster - tilted algebras of type @xmath0 the two approaches can be successfully combined to give a complete derived equivalence classification .
more precisely , our main result is the following . [ mainthm ] the following conditions are equivalent for two cluster - tilted algebras @xmath16 and @xmath26 of dynkin type @xmath0 : 1 .
@xmath16 and @xmath26 have the same associated polynomial ; 2 .
the cartan matrices of @xmath16 and @xmath26 represent equivalent bilinear forms over @xmath27 ; 3 .
@xmath16 and @xmath26 are derived equivalent ; 4 .
@xmath16 and @xmath26 are connected via a sequence of `` good '' mutations .
in addition to the above general statement we make the derived equivalence classification explicit by providing complete lists of the algebras contained in each derived equivalence class ( up to sink / source equivalence ) .
note that the implication ( c ) @xmath28 ( b ) holds in general for any two ( finite - dimensional ) algebras @xmath16 and @xmath26 , and that the implication ( b ) @xmath29 ( a ) holds whenever the associated polynomials are defined , i.e. when the cartan matrices are invertible over @xmath30 .
moreover , for cluster - tilted algebras the implication ( d ) @xmath28 ( c ) is evident from the definition .
we also note that since the cluster - tilted algebras we consider involve only zero- and commutativity - relations ,
they can in fact be defined over any commutative ring @xmath10 .
the combinatorial nature of our construction of tilting complexes via `` good '' mutations will then imply that such algebras corresponding to quivers in the same class will be derived equivalent for any @xmath10 .
therefore one may view the derived equivalences arising from `` good '' mutations as `` universal '' , being independent on the auxiliary algebraic data ( specified by @xmath10 ) .
let us briefly describe the above main result in some more detail . for precise definitions of the cluster - tilted algebras involved
we refer to sections 3 ( type @xmath1 ) , a / b ( type @xmath2 ) , and c / d ( type @xmath3 ) below . in the following tables
we list the associated polynomials of the cluster - tilted algebras , and also the total number of algebras in each derived equivalence class . for type
@xmath1 the mutation class consists of 67 quivers .
the corresponding cluster - tilted algebras turn out to fall into six derived equivalence classes as follows . & & & + & & & + @xmath31 & 20 & @xmath32 & 19 + & & & + @xmath33 & 16 & @xmath34 & 7 + & & & + @xmath35 & 3 & @xmath36 & 2 + for type @xmath2 the mutation class consists of 416 quivers .
the derived equivalence classes of the corresponding cluster - tilted algebras are again characterized by the associated polynomials ; there are 14 classes in total , given as follows . & & & + & & & + @xmath37 & 64 & @xmath38 & 2 + & & & + @xmath39 & 32 & @xmath40 & 56 + & & & + @xmath41 & 72 & @xmath42 & 8 + & & & + @xmath43 & 8 & @xmath44 & 17 + & & & + @xmath45 & 124 & @xmath46 & 11 + & & & + @xmath47 & 16 & @xmath48 & 1 + & & & + @xmath49 & 4 & @xmath50 & 1 + for type @xmath3 the mutation class consists of 1574 quivers .
the corresponding cluster - tilted algebras turn out to fall into 15 different derived equivalence classes which are characterized as follows . & & & + & & & + @xmath51 & 128 & @xmath52 & 221 + & & & + @xmath53 & 64 & @xmath54 & 22 + & & & + @xmath55 & 256 & @xmath56 & 167 + & & & + @xmath57 & 16 & @xmath58 & 38 + & & & + @xmath59 & 384 & @xmath60 & 118 + & & & + @xmath61 & 72 & @xmath62 & 4 + & & & + @xmath63 & 48 & @xmath64 & 24 + & & & + @xmath65 & 12 & & + the paper is organized as follows . in section 2
we collect some background material ; in particular we recall the notion of quiver mutation , describe the results of buan , marsh and reiten on cluster - tilted algebras of finite representation type , review the fundamental results on derived equivalences and then discuss invariants of derived equivalence such as the equivalence class of the euler form , in particular leading to the determinant of the cartan matrix and the characteristic polynomial of its asymmetry matrix as derived invariants . in section 3
we discuss derived equivalences for cluster - tilted algebras of dynkin type @xmath1 in detail .
the quivers of these algebras are given by those in the mutation class of type @xmath1 ; this mutation class can easily be reproduced by the reader using keller s software @xcite .
we first give in section 3 a list of the derived equivalence classes , sorted by the associated polynomial which is the crucial invariant for our purposes .
we also give the cartan matrix of one representative in each class which we shall need later in our computations . as main result of this section
we prove the main theorem [ mainthm ] for type @xmath1 . to this end
we have to find explicit tilting complexes for the cluster - tilted algebras of type @xmath1 and we have to determine their endomorphism rings .
the necessary calculations are carried out in detail in sections [ sec : det2 ] - [ sec : det4 ] .
the tilting complexes we use are closely related to quiver mutations at a single vertex , and we explain this construction in section [ sec : vertex - to - complex ] . for types
@xmath2 and @xmath3 we have followed a different strategy of presentation since the number of algebras involved becomes very large .
we first list the algebras but without drawing the quivers ; again , the quivers can be found using keller s software .
we then present the results on derived equivalences for cluster - tilted algebras of types @xmath2 and @xmath3 in a very concise form which is explained at the beginning of the respective sections . for each group of algebras with the same associated polynomial we then provide tilting complexes and list their endomorphism rings , but without giving any details on the calculations .
however , we hope that we have provided enough information so that interested readers should easily be able to check our findings
. the first two authors are grateful to bernhard keller for suggesting to contact the third author on questions left open in a first version of this paper which could then be completely settled in the present joint version .
a _ quiver _ is a finite directed graph @xmath7 , consisting of a finite set of vertices @xmath66 and a finite set of arrows @xmath67 between them .
a fundamental concept in the theory of fomin and zelevinsky s cluster algebras is mutation ; for quivers this takes the following shape .
let @xmath7 be a quiver without loops and oriented @xmath68-cycles .
for vertices @xmath69 , let @xmath70 denote the number of arrows from @xmath71 to @xmath72 , where @xmath73 means that there are @xmath74 arrows from @xmath72 to @xmath71 .
the _ mutation of @xmath7 at the vertex @xmath75 _ yields a new quiver @xmath76 obtained from @xmath7 by the following procedure : 1 .
add a new vertex @xmath77 .
2 . for all vertices
@xmath78 , different from @xmath75 , such that @xmath79 , set the number of arrows @xmath80 from @xmath71 to @xmath72 in @xmath76 as follows : + if @xmath81 and @xmath82 , then @xmath83 ; + if @xmath84 and @xmath85 , then @xmath86 .
3 . for any vertex @xmath71 , replace all arrows from @xmath71 to @xmath75 with arrows from @xmath77 to @xmath71 , and replace all arrows from @xmath75 to @xmath71 with arrows from @xmath71 to @xmath77 .
4 . remove the vertex @xmath75 .
two quivers are called _ mutation equivalent _ if one can be obtained from the other by a finite sequence of mutations .
the _ mutation class _ of a quiver @xmath7 is the class of all quivers mutation equivalent to @xmath7 .
it is known from the seminal results of fomin and zelevinsky @xcite that the mutation class of a dynkin quiver @xmath7 is finite .
cluster - tilted algebras arise as endomorphism algebras of cluster - tilting objects in a cluster category , see @xcite . for the special case of dynkin
quivers the cluster - tilted algebras are known to be of finite representation type . moreover , by a result of buan , marsh and reiten @xcite they can be described as quivers with relations by a simple combinatorial recipe to be recalled below . as a consequence , a cluster - tilted algebra of dynkin type
is uniquely determined by its quiver .
let @xmath7 be a quiver and throughout this paper let @xmath10 be an algebraically closed field .
we can form the _ path algebra _ @xmath9 , where the basis of @xmath9 is given by all paths in @xmath7 , including trivial paths @xmath87 of length zero at each vertex @xmath71 of @xmath7 .
multiplication in @xmath9 is defined by concatenation of paths .
our convention is to compose paths from right to left . for any path @xmath88 in @xmath7 let @xmath89 denote its start vertex and @xmath90 its end vertex
then the product of two paths @xmath88 and @xmath91 is defined to be the concatenated path @xmath92 if @xmath93 . the unit element of @xmath9 is the sum of all trivial paths , i.e. , @xmath94 .
we recall some background from @xcite .
an oriented cycle in a quiver is called _ full _ if it does not contain any repeated vertices and if the subquiver generated by the cycle contains no other arrows . if there is an arrow @xmath95 in a quiver @xmath7 then a path from @xmath72 to @xmath71
is called _ shortest path _ if the induced subquiver is a full cycle .
we now describe cluster - tilted algebras of dynkin type by a quiver with relations , i.e. in the form @xmath96 where @xmath7 is a finite quiver and @xmath97 is some admissible ideal in the path algebra @xmath9 .
recall that the quivers associated with cluster - tilted algebras of dynkin type are precisely the quivers in the mutation class of the corresponding dynkin quiver .
relations are linear combinations @xmath98 of paths @xmath99 in @xmath7 , all starting in the same vertex and ending in the same vertex , and with each @xmath100 non - zero in @xmath10 .
if @xmath101 , we call the relation a
_ zero - relation_. if @xmath102 and @xmath103 , and we call it a _ commutativity - relation _ ( and say that the paths @xmath104 and @xmath105 commute ) .
it will turn out that for cluster - tilted algebras of dynkin type the ideal @xmath97 can be generated by only using zero - relations and commutativity - relations .
finally , a relation @xmath106 is called _ minimal _ if whenever @xmath107 , where @xmath108 is a relation for every @xmath71 , then there is an index @xmath72 such that both @xmath109 and @xmath110 are scalars .
[ buan , marsh and reiten @xcite ] a cluster - tilted algebra @xmath16 of finite representation type is of the form @xmath111 , where @xmath7 is mutation equivalent to a dynkin quiver and where the ideal @xmath97 can be described as follows
. let @xmath71 and @xmath72 be vertices in @xmath7 . 1 .
the ideal @xmath97 is generated by minimal zero - relations and minimal commutativity - relations .
assume there is an arrow @xmath95 .
then there are at most two shortest paths from @xmath72 to @xmath71 . 1 . if there is exactly one , then this is a minimal zero - relation .
2 . if there are two , @xmath112 and @xmath113 , then @xmath112 and @xmath113 are not zero in @xmath16 and there is a minimal relation @xmath114 .
3 . up to multiplication by non - zero elements of @xmath10
there are no other minimal zero - relations or commutativity - relations than the ones coming from 2 .
[ mutate_examples ] we consider the following quiver @xmath7 of type @xmath1 if we mutate at vertex @xmath68 , we get the following quiver @xmath115 the corresponding cluster - tilted algebra is of the form @xmath116 where @xmath97 is generated by the zero - relations @xmath117 and @xmath118 ( and there are no commutativity - relations ) . mutating the latter quiver at the vertex @xmath119 leads to the quiver @xmath120 here , the ideal of relations of the corresponding cluster - tilted algebra is generated by the zero - relations @xmath121 , @xmath122 , @xmath123 and @xmath124 and the commutativity - relation @xmath125 . in this section
we briefly review the fundamental results on derived equivalences .
all algebras are assumed to be finite - dimensional @xmath10-algebras . for a @xmath10-algebra @xmath16 the bounded derived category of @xmath16-modules
is denoted by @xmath126 .
recall that two algebras @xmath127 are called derived equivalent if @xmath126 and @xmath128 are equivalent as triangulated categories . by a famous theorem of rickard @xcite derived equivalences can be found using the concept of tilting complexes . [ tilting - complex ] a _ tilting complex _ @xmath129 over @xmath16 is a bounded complex of finitely generated projective @xmath16-modules satisfying the following conditions : 1 .
@xmath130)=0 $ ] for all @xmath131 , where @xmath12 $ ] denotes the shift functor in @xmath126 ; 2 . the category @xmath132 ( i.e. the full subcategory consisting of direct summands of direct sums of @xmath129 ) generates the homotopy category @xmath133 of projective @xmath16-modules as a triangulated category .
we can now formulate rickard s seminal result .
[ rickard @xcite ] two algebras @xmath16 and @xmath134 are derived equivalent if and only if there exists a tilting complex @xmath129 for @xmath16 such that the endomorphism algebra @xmath135 .
let @xmath16 be a finite - dimensional algebra over a field @xmath10 and let @xmath136 be a complete collection of non - isomorphic indecomposable projective @xmath16-modules ( finite - dimensional over @xmath10 ) .
the _ cartan matrix _ of @xmath16 is then the @xmath137 matrix @xmath21 defined by @xmath138 .
denote by @xmath139 the triangulated category of _ perfect _ complexes of @xmath16-modules inside the derived category of @xmath16 , that is , complexes ( quasi - isomorphic ) to finite complexes of finitely generated projective @xmath16-modules .
the grothendieck group @xmath140 is a free abelian group on the generators @xmath141 , \dots , [ p_n]$ ] , and the expression @xmath142)\ ] ] is well defined for any @xmath143 and induces a bilinear form on @xmath140 , known as the _ euler form _ , whose matrix with respect to the basis of projectives is @xmath144 .
the following proposition is well known . for
the convenience of the reader , we give the short proof , see also the _ proof _ of proposition 1.5 in @xcite .
let @xmath16 and @xmath134 be two finite - dimensional , derived equivalent algebras .
let @xmath145 denote by number of their non - isomorphic indecomposable projectives .
then the matrices @xmath21 and @xmath146 represent equivalent bilinear forms over @xmath147 , that is , there exists @xmath148 such that @xmath149 . indeed , by @xcite , if @xmath16 and @xmath134 are derived equivalent , then @xmath139 and @xmath150 are equivalent as triangulated categories .
now any triangulated functor @xmath151 induces a linear map from @xmath140 to @xmath152 .
when @xmath153 is also an equivalence , this map is an isomorphism of the grothendieck groups preserving the euler forms .
thus , if @xmath154 $ ] denotes the matrix of this map with respect to the bases of indecomposable projectives , then @xmath154^t c_b [ f ] = c_a$ ] . in general , to decide whether two integral bilinear forms are equivalent is a very subtle arithmetical problem .
therefore , it is useful to introduce somewhat weaker invariants that are computationally easier to handle . in order to do this ,
assume further that @xmath21 is invertible over @xmath23 .
in this case one can consider the rational matrix @xmath22 ( here @xmath155 denotes the inverse of the transpose of @xmath21 ) , known in the theory of non - symmetric bilinear forms as the _ asymmetry _ of @xmath21 .
let @xmath16 and @xmath134 be two finite - dimensional , derived equivalent algebras with invertible ( over @xmath23 ) cartan matrices .
then we have the following assertions , each implied by the preceding one : 1 .
there exists @xmath148 such that @xmath149 .
there exists @xmath148 such that @xmath156 .
there exists @xmath157 such that @xmath156 .
4 . the matrices @xmath24 and
@xmath158 have the same characteristic polynomial . for proofs and discussion ,
see for example ( * ? ? ?
* section 3.3 ) . since the determinant of an integral bilinear form is invariant under equivalence , we can combine it with the characteristic polynomial @xmath159 of the asymmetry matrix @xmath24 to obtain a discrete invariant of derived equivalence , namely @xmath160 .
we call this invariant the _ polynomial associated with @xmath21_. the matrix @xmath22 ( or better , minus its transpose @xmath161 ) is related to the _ coxeter transformation _ which has been widely studied in the case when @xmath16 has finite global dimension ( so that @xmath21 is invertible over @xmath147 )
. it is the @xmath10-theoretic shadow of the serre functor and the related auslander - reiten translation in the derived category .
the characteristic polynomial is then known as the _ coxeter polynomial _ of the algebra . in general , @xmath24 might have non - integral entries .
however , when the algebra @xmath16 is _ gorenstein _ , the matrix @xmath24 is integral , which is an incarnation of the fact that the injective modules have finite projective resolutions . by a result of keller and reiten @xcite ,
this is the case for the cluster - tilted algebras in question .
let @xmath162 be an algebra given by a quiver @xmath163 with relations .
since @xmath164 is the unit element in @xmath16 we get a decomposition @xmath165 , hence the ( left ) @xmath16-modules @xmath166 are the indecomposable projective @xmath16-modules , and the cartan matrix @xmath167 of @xmath16 is the @xmath145-by-@xmath145 matrix whose entries are @xmath168 , where @xmath169 .
any homomorphism @xmath170 of left @xmath16-modules is uniquely determined by @xmath171 , the @xmath10-vector space generated by all paths in @xmath7 from vertex @xmath71 to vertex @xmath72 that are non - zero in @xmath16 .
in particular , we have @xmath172 , i.e. , computing entries of the cartan matrix for @xmath16 reduces to counting paths in @xmath7 . for cluster - tilted algebras of dynkin type the entries of the cartan matrix
can only be @xmath173 or @xmath174 , as the following result shows .
[ buan , marsh , reiten @xcite ] [ dimension_hom_space ] let @xmath16 be a cluster - tilted algebra of finite representation type .
then @xmath175 for any two indecomposable projective @xmath16-modules @xmath176 and @xmath177 .
[ ex : cartan ] we have a look at the quivers in example [ mutate_examples ] again , and compute the cartan matrices of the corresponding cluster - tilted algebras . for the dynkin quiver @xmath7 of type @xmath1 with the above orientation
we get the following cartan matrix @xmath178{cccccc } 1&1&1&0&0&0\\ 0&1&1&0&0&0\\ 0&0&1&0&0&0\\ 0&0&1&1&0&0\\ 0&0&1&1&1&0\\ 0&0&1&0&0&1
\end{array } \right)$ ] since there are no zero- or commutativity - relations . for the quiver @xmath179 obtained by mutation from @xmath7 at vertex 2 , the corresponding cartan matrix @xmath180 has the form @xmath181 @xmath178{cccccc } 1&0&1&0&0&0\\ 1&1&0&0&0&0\\ 0&1&1&0&0&0\\ 0&1&1&1&0&0\\
0&1&1&1&1&0\\ 0&1&1&0&0&1 \end{array } \right)$ ] for @xmath96 since the paths from vertex @xmath174 to @xmath68 , from @xmath68 to @xmath119 and from @xmath119 to @xmath174 are zero .
finally , for the quiver @xmath182 obtained from @xmath179 by mutating at vertex 3 , the cluster - tilted algebra has cartan matrix @xmath183 @xmath178{cccccc } 1&0&0&0&0&0\\ 1&1&1&0&0&0\\ 1&1&1&1&0&1\\ 0&1&0&1&0&0\\
0&1&0&1&1&0\\ 0&1&0&0&0&1 \end{array } \right)$ ] .
note that the two paths from vertex @xmath119 to vertex @xmath68 ( over @xmath184 or @xmath185 ) are the same since we have the commutativity - relation @xmath186 . for calculating the endomorphism ring @xmath187 of a tilting complex @xmath129 over the algebra @xmath16 , we can use the following statement which explicitly gives the cartan matrix of the endomorphism ring in terms of the tilting complex and the cartan matrix of @xmath16 . let @xmath129 be a tilting complex over @xmath16 with endomorphism algebra @xmath188 , and
let @xmath189 be the indecomposable direct summands of @xmath129 .
then the cartan matrix @xmath146 of @xmath134 is given by @xmath190 , where @xmath191 is the matrix defined by @xmath192 = \sum_{j=1}^n p_{ij } [ p_j]\ ] ] ( that is , its @xmath71-th row is the class of the summand @xmath193 in @xmath140 written in the basis @xmath141 , \dots , [ p_n]$ ] ) . continuing example [ ex : cartan ] , let @xmath194 be the complex over the cluster - tilted algebra corresponding to @xmath179 defined by @xmath195 where the @xmath176 are in degree @xmath173 for @xmath196 and @xmath197 is in degree @xmath198 .
then @xmath129 is a tilting complex and the corresponding matrix @xmath199 is given by @xmath200 @xmath178{cccccc } 1&0&0&0&0&0\\ 0&1&0&0&0&0\\ 1&0&-1&1&0&1\\ 0&0&0&1&0&0\\
0&0&0&0&1&0\\ 0&0&0&0&0&1 \end{array } \right)$ ] , so that @xmath201 .
in fact , @xmath202 is isomorphic to the cluster - tilted algebra corresponding to @xmath182 , see section [ sec : a2a7 ] .
it is sometimes convenient to use the following alternating sum formula , arising from the fact that for a bounded complex @xmath203 of projective modules , we have @xmath204 = \sum ( -1)^r [ x^r]$ ] in @xmath140 .
[ happel @xcite ] [ p : happel ] for an algebra @xmath16 let @xmath205 and @xmath206 be bounded complexes of projective @xmath16-modules .
then @xmath207)= \sum\limits_{r , s } ( -1)^{r - s}\dim\mathrm{hom}_a(x^r , y^s).\ ] ] in particular , if @xmath208 and @xmath209 are direct summands of the same tilting complex , then @xmath210
for the mutation class of @xmath1 we start with the following quiver and determine all quivers which can be obtained from it by a finite number of mutations . for this , one can use the software of b. keller @xcite .
the mutation class of @xmath1 consists of 67 quivers . for the purpose of derived equivalence classifications of the corresponding cluster - tilted algebras it suffices to consider the quivers up to sink / source equivalence , and there are @xmath211 quivers up to sink / source equivalence . we can divide the corresponding cluster - tilted algebras into six groups by computing the polynomials associated with their cartan matrices .
recall from the introduction that these associated polynomials are obtained by multiplying the determinant of the cartan matrix by the characteristic polynomial of its asymmetry matrix .
it will turn out that these six groups form the six derived equivalence classes of the cluster - tilted algebras of type @xmath1 .
we list in the table below all quivers in the mutation class of type @xmath1 . for each group of sink / source equivalent quivers we give only one picture where certain arrows are replaced by undirected lines ; this has to be read that these lines can take any orientation .
we also give the cartan matrix of the corresponding cluster - tilted algebra of one particular representative in each group of sink / source equivalent quivers . from this
cartan matrices one can easily read off to which orientation of the undirected lines it corresponds ; in fact , for a line between vertices @xmath71 and @xmath72 the arrow is going from @xmath71 to @xmath72 if the @xmath212 entry in the cartan matrix is non - zero , and from @xmath72 to @xmath71 otherwise .
we sort the @xmath211 classes of cluster - tilted algebras of type @xmath1 ( up to sink / source equivalence ) according to their associated polynomials , and number them according to the output of b. keller s software @xcite , i.e. the cluster - tilted algebras are denoted by @xmath213 .
|
we address the question of when cluster - tilted algebras of dynkin type @xmath0 are derived equivalent and as main result obtain a complete derived equivalence classification .
it turns out that two cluster - tilted algebras of type @xmath0 are derived equivalent if and only if their cartan matrices represent equivalent bilinear forms over the integers which in turn happens if and only if the two algebras are connected by a sequence of `` good '' mutations . for type @xmath1
all details are given in the paper , for types @xmath2 and @xmath3 we present the results in a concise form from which our findings should easily be verifiable .
derived equivalence classification of cluster - tilted algebras of dynkin type @xmath0 janine bastian@xmath4 , thorsten holm@xmath4 , and sefi ladkani@xmath5 @xmath4 leibniz universitt hannover , institut fr algebra , zahlentheorie und diskrete mathematik , welfengarten 1 , d-30167 hannover , germany + bastian,[email protected] @xmath5 max - planck - institut fr mathematik , vivatsgasse 7 , d-53111 bonn , germany +
|
the bipolar homunculus nebula surrounding @xmath0 carinae offers an unusually well - constrained laboratory to study a wide array of physical processes in the interstellar medium .
the distance , size , shape , orientation , and structure of the nebula are all known ( smith 2006 ; davidson et al .
2001 ) , as are its expansion velocity and age ( morse et al . 2001 ; currie & dowling 1999 ; smith & gehrz 1998 ) .
ionized ejecta outside the homunculus indicate nitrogen - rich gas ( davidson et al . 1986
; dufour et al . 1997 ; smith & morse 2004 ) , and the recent detection of ammonia in the homunculus suggests that it too is n - rich ( smith et al .
the homunculus itself is mostly neutral , since the star s dense wind extinguishes nearly all of the lyman continuum produced by the system .
this means that familiar temperature and density diagnostics of low - density ionized gas in h ii regions ( e.g. , osterbrock & ferland 2006 ) can not be used here . at uv
/ visual wavelengths the homunculus is seen mainly in reflection . the dust that absorbs and scatters this light acts as a calorimeter by re - emitting the absorbed luminosity in the infrared ( ir ) , providing us with a practical and accurate measure of the bolometric luminosity of the central source ( e.g. , smith et al.2003b ; cox et al . 1995 ; hackwell et al . 1986 ; davidson 1971 ; pagel 1969 ; westphal & neugebauer 1969 ) . despite these constraints
, @xmath0 car s nebula is such a bright and complex object that it leaves no shortage of observational puzzles .
these can sometimes make it difficult to see the trees for the forest .
relevant physical conditions in the homunculus nebula can be constrained if we understand the origin of the distinct double - shell structure in the walls of the nebula .
these layers manifest themselves in three independent ways that have been clearly observed .
first , high resolution thermal - ir images reveal a stratification in the dust color temperature , with a thin outer shell at @xmath9140 k , and a warmer , thicker inner zone with a color temperature of @xmath9200 k ( smith et al .
second , near - ir spectra show an outer skin of h@xmath1 , and an inner shell in tracers of dense , low - ionization atomic gas like collisionally - excited lines of [ fe ii ] ( smith 2006 , 2002a ) .
these two layers are segregated from one another .
third , the two distinct shells can be seen at different radial velocities in uv absorption profiles along our line - of - sight to the star ( gull et al .
2006 , 2005 ; nielsen et al .
the outer molecular layer is also detected in uv absorption lines of ch and oh ( verner et al .
2005 ) , and the inner layer is also seen in visual - wavelength emission lines like [ ni ii ] and [ fe ii ] , plus faint emission of h@xmath10 and [ n ii ] ( davidson et al.2001 ; hillier & allen 1992 ; allen & hillier 1993 ; smith et al.2003a ) .
a detailed account of this structure was given recently in paper i ( smith 2006 ) .
it is this pronounced double - shell structure that we aim to reproduce in the model presented here .
we are particularly interested in the physical conditions that allow h@xmath1 to form and survive in such a thin layer .
so far , @xmath0 carinae is the only luminous blue variable ( lbv ) known to be surrounded by molecular gas in its own ejecta , while nearly all lbv nebulae exhibit bright near - ir emission of [ fe ii ] ( smith 2002b ; smith & hartigan 2006 ) .
the h@xmath1 is probably a consequence of the extreme youth and high density of the nebula .
new observations presented in paper i , obtained with the phoenix spectrograph on the gemini south telescope , have shown the h@xmath1 layer to be surprisingly thin ( only a few per cent of the radius ) and uniformly so , despite the fact that it traces a wide range of different radii at various latitudes in the bipolar nebula .
this hints that the near - ir h@xmath1 emission arises at the inside surface of an optically - thick shell , which absorbs the remainder of whatever fuv radiation penetrates the inner [ fe ii]-emitting zone .
by varying the density in this outer shell , we can test what physical density permits h@xmath1 to survive .
this density constraint , in turn , provides an independent measure of the mass of the homunculus because the geometric thickness of the shell is known .
the total mass is thought to be at least 12 m@xmath8 derived from ir dust observations ( smith et al .
car s grains may have unusual composition ( chesneau et al .
here we are fine - tuning an earlier uniform - density shell model for the physical conditions in the homunculus ( ferland et al .
2005 ) in order to account for the observed thin structure of the h@xmath1 layer .
we simulate the conditions within the nebula using the development version of cloudy , last described by ferland et al .
the incident stellar continuum is represented by an interpolated 20,000 k costar atmosphere with a total luminosity of 5@xmath310@xmath11 . to account for the hard x - rays from the colliding wind binary system , we add a high - energy component corresponding to a 3@xmath310@xmath12 k blackbody with a luminosity of 30 @xmath13 ( e.g. , corcoran et al . 2001 ) .
the lack of a prominent h ii region shows that few hydrogen - ionizing photons strike the inner edge of the nebula , most likely due to absorption by the stellar wind .
therefore , we extinguish the net continuum by photoelectric absorption from a neutral hydrogen layer of 10@xmath14 @xmath15 to account for this . some high - energy photons are transmitted and they help drive the chemistry .
we also include the galactic background cosmic ray ionization rate , although the actual ionization rate may be higher if radiative nuclei are present .
cosmic rays have effects that are similar to x - rays , providing ionization that helps drive the chemistry .
we adopt solar gas phase abundances except for the following : he / h=0.4 , [ c / h]=4.4 , [ n / h]=2.82 , and [ o / h]=4.12 . the gas : dust mass ratio was set to the normal ism value of 100 , and we adopted silicate grains rather than graphitic grains because the 2175 absorption feature is absent in observations ( e.g. , hillier et al .
2006 ) and the 10 @xmath16 silicate emission feature is strong ( see , however , chesneau et al .
2005 ) .
we chose the nebular geometry to be a thick shell separated into two radial zones , where the concentric shells are permitted to have different densities .
table 1 gives our adopted geometric and physical properties of the double - shell structure .
the inner shell ( zone 1 ) has lower density and is geometrically thicker , corresponding to the gas traced by infrared [ fe ii ] lines and warm @xmath9200 k dust .
the outer shell ( zone 2 ) is very thin , and corresponds to the material traced by h@xmath1 emission and cooler @xmath9140 k dust .
the outer edge of the inner shell is directly in contact with the inner surface of the outer shell .
we calculated the physical conditions in these two zones separately in cloudy , using the _ transmitted _ continuum at the outside of zone 1 as the _ incident _ continuum for zone 2 .
although the shells were calculated using spherical symmetry in cloudy , they are meant to approximate the observed properties of the nebula roughly along our sight line through the polar region of the bipolar lobes .
the equatorial region of the nebula with much smaller radii was modeled separately ( see below ) .
the inner and outer radii for the two zones were taken directly from observations ( paper i ) and are not free parameters .
our goal is to investigate the densities in these two zones , to answer the following questions : 1 ) what is the density in zone 1 that is consistent with the observed @xmath17 value of @xmath1810@xmath19 @xmath5 ( smith 2002a ) and permits significant fuv radiation to fully penetrate the shell , causing iron to remain as fe@xmath6 throughout ?
2 ) what is the minimum density in zone 2 that allows hydrogen to become predominantly molecular , and iron to recombine to fe@xmath7 ?
the most important free parameters here are therefore the value of @xmath20 chosen for each zone , but other factors like the gas : dust mass ratio and the filling factor play roles as well
. the main determining factor in whether or not h@xmath1 can survive in zone 2 is the attenuation of fuv radiation in the lyman and werner bands by the total column density of gas and dust in both zones . based on spatially - resolved infrared [ fe ii ] spectra ( paper i ) we adopt a volume filling factor of 0.5 in zone 1 . from similar h@xmath1 spectra and the size scales of structures seen in _ hst
_ images ( morse et al .
1998 ) we adopt a filling factor of 0.9 for the thin outer shell in zone 2 .
a fraction of 0.9 also matches the fraction of the total stellar luminosity absorbed by dust grains that is reemitted in the ir , while @xmath910% of the starlight escapes at near - uv and visual wavelengths .
finally , we note that we have ignored the possible influence of the little homunculus ( ishibashi et al .
this inner nebula has a much lower mass a few percent of the homunculus ( smith 2005 ) and is apparently clumpy on large scales , allowing both radiation and wind to leak through .
thus , its influence should be minimal , absorbing whatever small amount of lyman continuum radiation may escape the dense stellar wind , causing strong variability in the radio continuum ( duncan et al .
figure 1 shows the results of a cloudy calculation matched to observations of the polar lobes of the homunculus roughly along our line - of - sight ( paper i ) .
the transition between zones 1 and 2 is sudden at a depth of a little more than 7@xmath310@xmath21 cm , showing a thick inner zone traced by [ fe ii ] and a very thin outer shell seen in h@xmath1 emission . in order to reproduce this sharp transition ,
the simulation required a strong density increase of a factor of 1530 from zone 1 to 2 . throughout zone 1 , hydrogen remains fully atomic , while fe is singly ionized ( representative values about halfway through zone 1 give an h@xmath1 fraction of 5@xmath310@xmath22 and a neutral fe fraction of 2@xmath310@xmath23 ) .
the electron density from infrared [ fe ii ] line ratios in zone 1 is n@xmath2410@xmath19 @xmath5 ( smith 2002a ) . to match this constraint ,
the hydrogen density in zone 1 was set to 10@xmath25 @xmath5 , since n@xmath26/n@xmath27 was only about 3% at this distance from the star .
only 12% of the hydrogen was ionized at the innermost edge of zone 1 , probably through balmer continuum ionization from the 2s level .
this may account for the faint h@xmath10 emission there ( smith et al .
2003a ; davidson et al .
2001 ) , as well as faint radio continuum ( duncan et al .
1997 ) . in both zone 1 and 2 ,
heating of the gas was dominated ( @xmath2890% ) by grain - gas photoelectric heating , whereas direct heating by line absorption was minimal . to investigate the formation of h@xmath1 and the recombination of fe in zone 2 we considered several different values for the hydrogen density , and discuss a few representative values here :
n@xmath27=10@xmath29 , 10@xmath30 , 10@xmath31 , 10@xmath32 , and 10@xmath4 @xmath5 .
table 1 lists the properties in zone 2 at these various densities , and figure 2 gives a detailed view of the transition to molecular gas for each of these adopted densities .
with n@xmath27=10@xmath29 @xmath5 , we found that h@xmath1 did not form at all , with h remaining fully atomic and fe remaining singly ionized throughout zone 2 . when the density was raised to 2@xmath310@xmath29 @xmath5
, h@xmath1 did begin to form at the outer edge of zone 2 ; the molecular fraction reached a maximum of about 40% at the outermost edge ( fig . 2 ) .
however , fe remained singly ionized throughout zone 2 at this density .
this is not in agreement with observations , since fe i absorption is seen from zone 2 in uv spectra ( nielsen et al .
2005 ) , so the density must be higher than 2@xmath310@xmath29 @xmath5 . at a density of 3@xmath310@xmath29 @xmath5 , hydrogen became 99% molecular at the outer edge of the shell ( fig .
2 ) , but fe@xmath6 still did not recombine to fe@xmath33 .
observations suggest that h@xmath1 should form and the recombination of fe should occur _ early _ in zone 2 , especially at the remote polar regions where the radiation field is weaker than at other latitudes .
this condition is met only for higher hydrogen densities around 10@xmath32 to 10@xmath4 @xmath5 ( fig . 2 ) .
at a depth of only about 10@xmath34 cm into zone 2 ( 1020% of the way through the layer ) , the h@xmath1 fraction begins to rise sharply .
this is because its destruction rate collapses as uv radiation in the lyman and werner bands becomes optically thick and is extinguished by the much higher column density of gas and dust encountered here .
once molecules begin to form , the transition is rapid .
hydrogen remains fully molecular throughout the rest of zone 2 , while fe@xmath6 does finally recombine .
the fe@xmath6/fe@xmath7 transition occurs at a slightly larger depth than the atomic / molecular hydrogen transition , because fe@xmath7 can still be ionized by balmer continuum photons with wavelengths longer than the lyman - werner bands .
the hydrogen density is probably not much above 10@xmath4 @xmath5 , because then the h@xmath1 shell would be thicker than observed , and the extinction to the star would be unreasonably high .
the bottom panel of figure 1 shows the dust temperature for various grain radii at a range of depths through the shell . in zone 1 , we see that only the largest grains with radii @xmath350.2 @xmath16 can match the observed dust color temperatures .
the transition to higher densities at the beginning of zone 2 is accompanied by a sharp drop in the grain temperature , as observed in mid - ir images ( smith et all.2003b ) .
this sharp drop to lower grain temperatures only occurred for the highest - density models with n@xmath27=10@xmath32 to 10@xmath36 @xmath5 in zone 2 .
the grain temperature in zone 2 appears to be independent of grain size .
this is perhaps because much of the short - wavelength fuv radiation has been extinguished and the grains are heated predominantly by radiation at near - uv , visual , and ir wavelengths instead , where the absorption efficiency is less sensitive to grain size ( e.g. , draine 2003 ) . in this case
, grain - gas photoelectric heating will drop as well .
in fact , we see that the grain temperature eventually drops to within 20 k of the equilibrium blackbody temperature t@xmath37 ( dashed line ) , independent of grain radius .
the fact that the temperature in zone 2 is nearly independent of grain size is interesting , because it contradicts long - held conventional wisdom .
the observed near - blackbody color temperature has often been taken as an indication that the grains have radii of 12 @xmath16 ( smith et al . 2003b , 1998 ; polomski et al .
1999 ; hackwell et al . 1986
; mitchel et al . 1983 ) .
however , figure 1 indicates that grains this large may not be necessary to explain the grain temperature in zone 2 after all ( although some large grains are still needed for the nearly gray extinction ; rodgers 1971 ; hillier 2006 ; whitelock et al .
in particular , a population that includes _ both _ large and small grains is evidently allowed in zone 2 .
the visual extinction through zone 1 is only @xmath91 mag , compared to about 4.59 magnitudes for zone 2 ( table 1 ) .
thus , the thin outer h@xmath1 shell dominates the visual extinction , the total infrared luminosity , and also the scattered light seen in visual - wavelength images and hard x - rays ( corcoran et al .
this is why model shapes for the homunculus derived from [ fe ii ] or [ ni ii ] emission from zone 1 ( davidson et al .
2001 ; allen & hillier 1991 ) are too small compared to images , whereas model shapes derived from h@xmath1 emission have the correct apparent size ( paper i ) .
the stellar continuum that illuminates the nebula is strongly reprocessed by the inner region before it reaches the denser outer zone .
figure 3 shows the stellar continuum ( the smoother dotted line ) that is incident upon the inner nebula .
the gray solid line with many spectral emission features is the net emission that emerges from the inner region and goes on to irradiate the molecular gas in zone 2 .
this continuum is a combination of the attenuated incident continuum plus reprocessed emission from the inner nebula .
the uv continuum has been heavily absorbed by dust within the inner region .
thermal dust emission dominates the infrared continuum , and the dust is warm enough for the 10 @xmath16 silicate feature to come into emission .
many emission lines , mainly atomic and singly ionized species , are formed in the inner region .
the major effect of this reprocessed continuum will be upon dust in the outer nebula .
much of the incident uv continuum has been converted into thermal - ir emission already .
the uv is the most effective in photoelectrically heating gas , so the outer nebula is cooler as a result of the absence of this light , as noted earlier .
the very strong infrared continuum heats the outer nebula indirectly , by first heating the dust , which then shares its energy with the gas as the result of gas - dust collisions .
the emergent spectrum from zone 2 with @xmath20=10@xmath32 @xmath5 is shown with the solid black line in figure 3 .
as expected , the thermal - ir continuum is shifted toward longer wavelengths because of the cooler dust , and the uv / visual light is more severely attenuated than for zone 1 .
interestingly , when we compare the spectrum from zone 1 alone to the emergent spectrum from zone 2 ( which includes the emergent spectrum of zone 1 that passes through zone 2 ) , we see that the 10 @xmath16 silicate feature has a different shape the silicate profile from zone 2 is somewhat muffled , indicating that the emission is beginning to turn optically thick .
in fact , when the density is increased to @xmath20=10@xmath4 @xmath5 in zone 2 ( dashed gray line ) the silicate feature turns clearly to self - absorption .
this trend in the silicate profile shape has two important implications .
first , when the emission begins to turn optically thick , the silicate profile appears wider and flatter , and the peak moves toward longer wavelengths .
this effect will modify the required contribution from other grain species like corrundum ( see chesneau et al .
second , the strong silicate feature observed in @xmath0 car s mid - ir spectrum may indicate that densities of @xmath20=10@xmath4 @xmath5 are too high ( although the effects of clumping are an issue ) . if densities near our lower bound @xmath20=10@xmath32 @xmath5 are preffered , then a total mass of the homunculus near 16 m@xmath8 is favored over a much larger value of @xmath930 m@xmath8 ( see 5.1 ) .
this is , perhaps , a relief .
high spatial resolution spectroscopy of the silicate profile in the homunculus could provide a powerful and independent constraint on the density and total mass of the nebula .
near - infrared h@xmath1 emission is seen at all latitudes in the homunculus ( paper i ) , even though the distance from the star changes by an order of magnitude . in the side walls of the polar lobes at mid - latitudes ,
the oblique incident angle of the radiation ( i.e. `` seasons '' ) will mitigate the stronger radiation field at closer radii to the central star , but this geometric advantage disappears for the molecular gas at the pinched - waist of the nebula .
thus , we also investigate the physical conditions in the equatorial plane of the homunculus , where h@xmath1 is seen nearest to the star .
the equatorial model was identical to the polar lobe / line - of - sight model , except that the radii and thicknesses of zones 1 and 2 were adjusted to match observations ( paper i ) , and the density in zone 1 was increased .
the equatorial model is summarized in table 2 .
figure 4 shows the results of a cloudy calculation matched to observations of the radial stratification in the equatorial plane of the homunculus ( paper i ) .
the inner radius is set approximately equal to the distance between the star and the weigelt knots , which is 0@xmath382 to 0@xmath383 ( smith et al . 2004 ; dorland et al .
2004 ) , or roughly 10@xmath21 cm if we correct for the 41 inclination angle ( paper i ; davidson et al .
2001 ) . in order to match observed electron densities of @xmath3910@xmath29 @xmath5 in the weigelt knots and other bright equatorial ejecta ( smith 2002 ; hamann et al . 1999 , 1994 ; davidson et al .
1995 ) , the simulations required a hydrogen density of about n@xmath27=2@xmath310@xmath29 @xmath5 or more .
to 10@xmath40 @xmath5 ( hamann et al .
] this was also the highest density that allowed h to remain predominantly atomic and fe to remain as fe@xmath6 throughout zone 1 .
even though the material is much closer to the star and feels a much stronger radiation field than in the polar model , @xmath210@xmath4 @xmath5 in zone 2 was still sufficient to allow a sharp transition to molecular hydrogen and neutral fe . unlike the polar model
, the inner edge of zone 1 experienced a more substantial h@xmath6 fraction of @xmath975% , while @xmath920% of the iron was ionized to fe@xmath41 .
this ionization occurs despite the fact that , aside from hard x - rays ( corcoran et al .
2001 ) , we did not include radiation from a hot companion star in our calculation . throughout most of zone 1 ,
the typical molecular hydrogen fraction was of order 10@xmath42 , while the neutral fe fraction was typically 10@xmath23 .
the bottom panel of figure 4 shows the dust temperature for grains in the equatorial model . in zone 1
, the predicted dust temperatures greatly exceed observational constraints on the color temperature for most grain sizes .
the hottest extended dust in the core of the homunculus ( associated with the weigelt knots and more distant ejecta ) is about 550 k ( smith et al .
2003b ; chesneau et al .
this is marginally consistent with @xmath43=0.2 @xmath16 grains within observational uncertainty , but the observed tempertures would be more consistent with significantly larger grains .
the transition to lower dust temperature at the boundary between zones 1 and 2 is not as stark as in the polar model , but like the polar model , the grain temperature in zone 2 drops and becomes insensitive to grain size ( fig . 4 ) . here
it fully reaches temperatures as low as the equilibrium blackbody temperature t@xmath37 .
figure 5 shows the incident ( dotted ) and transmitted energy distributions for the equatorial model .
the transmitted radiation is that which escapes from the outer boundary of zone 2 .
this emission at the equator is of particular interest , as it illuminates material in the equatorial skirt at larger radii .
some of that equatorial gas has pecular spectral properties , like the so - called `` strontium filament '' ( hartman et al .
2004 ; zethson et al . 2001 ; bautista et al . 2006 , 2002 ) , the very bright and possibly fluorescent [ ni ii ] @xmath447379 emission ( davidson et al . 2001 ) , and he i @xmath4410830 emission ( smith 2002 ) . in figure 5
we see that radiation shortward of 2000 is severely attenuated , while near - uv radiation beyond this limit remains strong .
our most important observational constraint on the density in the polar lobes is that molecular hydrogen is able to survive and fe becomes neutral , even though they are within only 0.1 pc of the most luminous hot supergiant in our galaxy .
furthermore , both of these conditions should be met in the polar model of zone 2 with some room to spare , since h@xmath1 emission is seen to be fairly uniform at all latitudes , including lower latitudes closer to the equator ( and closer to the star ) where the radiation field is stronger .
as we have discussed above , these constraints are only met for high densities of n@xmath27=10@xmath45 @xmath5 in zone 2 . from the thermal - ir spectral energy distribution and conservative assumptions , smith
( 2003b ) derived a lower limit to the total mass of the homunculus of about 12.5 m@xmath8 ( about 11 m@xmath8 in the cool 140 k shell , and 1.5 m@xmath8 in the warmer inner shell ) .
if this mass of cool dust were spread evenly over the volume of the h@xmath1 shell inferred observationally , the density would be at least n@xmath27=10@xmath31 @xmath5 in zone 2 ( paper i ) .
however , our cloudy calculations show that even higher densities are likely , as noted above .
these higher average densities distributed over the same volume would imply a total gas mass for the homunculus in the range of roughly 1535 m@xmath8 .
thus , our study confirms the large mass of the homunculus deduced from ir observations of dust , but uses an independent method sensitive to the gas density .
it also confirms that previous ir estimates were indeed lower limits to the total mass .
we also find that 35 m@xmath8 is an upper limit to the mass , since for densities much higher than 10@xmath4 @xmath5 , the transition to molecular gas would occur earlier and the h@xmath1 shell would be thicker than observed .
caveats : although our mass estimate is sensitive to the gas density , it is not completely independent of the gas : dust ratio , since dust helps attenuate the uv radiation in the lyman - werner bands that destroys h@xmath1 , and grains act as a catalyst by providing a surface for h@xmath1 formation .
however , like mass estimates from thermal - ir dust emission , these are probably underestimates if they are wrong .
our cloudy calculations were tuned to match the observed structure in ir lines of h@xmath1 and [ fe ii ] , but they naturally reproduce the bimodal stratification of the observed dust color temperature as well .
high - resolution thermal - ir images show a thick inner shell at @xmath9200 k and a thin outer shell at @xmath9140 k ( smith et al .
2003b ) , where the cooler dust color temperatures are coincident with the outer h@xmath1 shell .
this is clearly reproduced in our model ( fig . 1 ) .
our earlier simulations with cloudy used a uniform density of 10@xmath29 @xmath5 throughout a thick shell ( ferland et al .
2005 ) and were able to reproduce a transition from h to h@xmath1 and fe@xmath6 to fe@xmath7 at some point within the shell , _ but in that model a sharp drop in dust temperture did not occur at the same position as the onset of h@xmath1_. instead , in the constant - density thick - shell model , the dust temperature began to drop immediately at the inner radius of the shell and continued smoothly thereafter .
therefore , the observed sharp transition from a geometrically thick zone of @xmath9200 k to a very thin layer of @xmath9140 k dust coincident with the onset of the thin h@xmath1 shell ( fig
. 1 ) gives a robust and independent confirmation that a sharp density contrast between the two zones is indeed required . because of the strong density increase of a factor of 15 - 30 from zone 1 to zone 2
, the model also successfully reproduced the relative contributions to the total mass observed in each component .
fits to the ir spectral energy distribution ( smith et al .
2003b ) show that the 200 k dust in the inner shell only contains about 10% of the total mass in the nebula , while the rest of the mass resides in the thin outer shell at 140 k. because the outer shell is so thin , it must be much denser in order to have a higher total mass .
the relative masses for the inner ( @xmath91 m@xmath8 ) and outer ( 16 - 33 m@xmath8 ) shells in table 1 agree with these observational constraints . by stark contrast
, a uniform - density thick shell would have much more mass in the [ fe ii ] zone .
interestingly , the radial behavior of the dust temperature brings into question long - held notions about the grain properties .
several authors have noted that the low dust color temperatures at thermal - ir wavelengths are close to the equilibrium blackbody temperatures at those radii , suggesting large grain sizes above @xmath461 @xmath16 ( mitchell et al .
1983 ; mitchell & robinson 1986 ; hackwell et al.1986 ; robinson et al .
1987 ; apruzese 1975 ; smith et al . 1998 , 2003b ; polomski et al .
silicate emission feature , but chesneau et al .
( 2005 ) , mitchell & robinson ( 1978 ) , and hyland et al .
( 1979 ) have suggested that this may be due to a partial contribution from other grain emission features like corrundum .
] in contrast , we find that the dust temperature in the outer layer is _ independent of grain size _ ( figs . 1 and 4 ) , and approaches the equilibrium blackbody temperature even for small grains .
the material becomes optically thick and severely attenuates the uv radiation , leaving the burden of energy balance to heating by longer wavelength radiation . at longer wavelengths
, the difference between absorption efficiency and the thermal - ir emissivity is smaller , allowing small and large grains to maintain similar temperatures .
the dust temperature never gets below the equilibrium blackbody temperature in our calculations .
therefore , the very low dust temperature of 110 k that morris et al .
( 1999 ) deduced from fits to the far - ir spectral energy distribution is probably too low , especially at the small radii in the equator suggested by those authors .
although the temperature is apparently independent of grain size in zone 2 , very small grains are not permitted in zone 1 because corresponding high temperatures are not seen there .
the color temperature in zone 1 is about @xmath9200 k , which would be consistent with grains of radius @xmath350.2 @xmath16 .
thus , our calculations relax but do not eliminate the constraints on large grain size posed by the low observed dust temperatures in the homunculus . allowing the presence of somewhat smaller grains may be relevant to the high degree of polarization ( thackeray 1961 ; visvanathan 1967 ; warren - smith et al . 1979 ; meaburn et al . 1987 ; 1993 ;
walsh & ageorges 2000 ) and the unusual scattering properties of the homunculus seen in high - resolution polarization imaging ( schulte - ladbeck et al .
1999 ) .
an accurate model for zone 1 determines the radiation field that penetrates it to illuminate the outer h@xmath1 shell ( fig .
3 ) , but the inner shell is interesting in its own right because of the physical conditions that give rise to the extraordinarily bright infrared [ fe ii ] emission seen there ( paper i ) .
these bright infrared [ fe ii ] emission lines are a common feature of other lbv nebulae ( smith 2002b ; smith & hartigan 2006 ) , which is in contrast to the outer h@xmath1 shell that seems to make @xmath0 car unique among hot luminous stars . to investigate this bright [ fe ii ] emission in detail
, we ran cloudy using the more sophisticated large fe atom .
this predicts the strength of numerous fe lines , including the distinct [ fe ii ] ( @xmath47 ) and ( @xmath48 ) transitions in the 12 @xmath16 region .
for example , with solar fe abundance , our cloudy model for zone 1 predicts a total luminosity for [ fe ii ] @xmath4416435 of about 65 l@xmath8 . this line has been observed extensively in @xmath0 car in both moderate and high resolution spectra ( smith 2002a ; paper i ) , and exhibits an average brightness in the polar lobes of about 2@xmath310@xmath49 ergs s@xmath50 @xmath15 arcsec@xmath51 ( although the local value is highly spatially dependent ) .
integrating this over the roughly 140 - 150 arcsec@xmath52 projected area of the homunculus would indicate a total flux in this line of 3@xmath310@xmath22 ergs s@xmath50 @xmath15 ( excluding the bright [ fe ii ] contribution from the little homunculus ; smith 2005 ) .
this corresponds to a total observed [ fe ii ] @xmath4416435 luminosity of about 4050 l@xmath8 .
the agreement between our model s predicted luminosity of 65 l@xmath8 and the observed value of @xmath945 l@xmath8 in this same line is remarkable , considering that the bipolar homunculus has a somewhat smaller projected area on the sky than the equivalent - radius spherical shell in cloudy .
this match indicates that the gas - phase fe abundance is indeed roughly solar in zone 1 .
a solar gas - phase fe abundance here is actually quite astonishing . in dusty regions of the ism ,
the abundance of fe is normally depleted by two orders of magnitude due to the formation of fe - bearing grains ( e.g. , shields 1970 ) .
we do nt expect the overall dust - to - gas ratio in zone 1 to be unusually low , since the mass inferred for the gas from the observed electron density agrees with the mass measured from the 200 k dust component that occupies the same region ( uncertainties of a factor of 2 are likely , but not factors of 100 ) .
therefore , some process has selectively destroyed fe - bearing grains _ without _ destroying the remaining grain population , or some mechanism has specifically inhibited the formation of fe - bearing grains .
shock speeds @xmath2880 km
s@xmath50 can destroy iron grains and release the corresponding atomic fe into the gas phase .
while we see no spectroscopic evidence for such fast shocks in the polar lobes today , shocks of this speed may have been active earlier in the life of the homunculus .
if such shocks destroyed the fe - bearing grains early - on , they may have been released into an environment where the density was too low to reform the grains , locking fe in the gas phase .
alternatively , the ejecta in the homunculus are c- and o - poor , and n - rich .
the low oxygen abundance may have inhibited the formation of iron oxides , which can be an important repository for storing fe in the solid phase of the ism . in the absence of c , o , and fe
, grains might form preferentially with al , mg , or ca instead . interestingly
, mid - ir spectroscopy of the homunculus strongly suggests the presence of al - bearing grains like corrundum ( mitchell & robinson 1978 ; hyland et al .
1979 ; chesneau et al .
we also know that fe is present in the gas phase in the outer h@xmath1 shell , since fe@xmath7 is seen in absorption there in uv spectra ( gull et al . 2005 ; nielsen et al.2005 ) .
the existence of a region of the ism with a high dust content and where fe is predominantly in the gas phase has important ramifications beyond the study of @xmath0 carinae , and will be discussed elsewhere .
we have shown that the observed double - shell structure of the homunculus requires a strong density jump in the walls of the nebula .
the observed transition from atomic to molecular gas ( paper i ) and the corresponding drop in dust temperature ( smith et al .
2003b ) result when fuv radiation propagating through the low - density inner zone suddenly encounters the high densities in the thin outer shell .
the absorption of this uv radiation in the lyman - werner bands by gas and dust in the dense outer shell regulates the destruction rate of h@xmath1 , allowing the h@xmath1 to survive .
the required density contrast between these two zones is a factor of roughly 1530 .
while our cloudy calculations explain how the observed properties of the double - shell structure result from a strong density stratification , they do not explain the origin of the density structure itself .
since the homunculus appears to be expanding ballistically ( smith & gehrz 1998 ) and because the current stellar wind is too weak to shape the massive nebula , it is likely that this density structure was determined early . the observed
stratification is reminiscent of the layered structure seen in cooling zones behind shock fronts ; although this is unlikely to be occuring at the present time ( paper i ) , the observed density structure may be frozen - in to the ejecta from shocks that occurred during the event itself as the dense material cooled .
as noted in the previous section , shocks present during an earlier epoch would also be interesting from the point of view of the gas - phase fe abundance .
the rapid cooling of this dense layer may have triggered severe thermal instabilities , which in turn , could explain the currently - observed clumping and fragmentation seen in images of the homunculus ( morse et al .
1998 ) . alternatively
, the low - density inner zone may result from a photoevaporative flow off the inside face of the much denser outer h@xmath1 shell , as is commonly seen in photodissociation regions at the surfaces of giant molecular clouds .
finally , an obvious possibility is that the two shells simply result from separate ejection events during the @xmath920 yr timespan of the great eruption . we know that this is at least plausible from the existence of the outer ejecta , the homunculus , and the little homunculus , indicating that @xmath0 car has suffered multiple mass ejections ( walborn 1976 ; walborn et al .
1978 ; ishibashi et al . 2003 ; smith 2005 ) .
our models that reproduce the fundamental observed structure have constrained the most basic physical properties of the homunculus : namely , its density and temperature .
these will allow further refinements to models trying to understand emission line spectra , unusual grain properties , molecular chemistry , abundances , and radiative transfer models of the appearance of @xmath0 car s nebula in images . the physical explanation for this distinct double - shell structure may have wider applications as well , especially to objects like planetary nebulae .
for example , the bipolar nebula m2 - 9 has a similar and pronounced double - shell structure , with a thin outer h@xmath1 shell and an inner [ fe ii ] shell ( smith et al .
2005 ; hora & latter 1994 ) . the uv radiation field of m2 - 9 is known to be highly asymmteric and time variable because of a central binary system ( e.g. , doyle et al .
2000 ) , so it is likely that a density jump controls the shape of the h@xmath1 lobes , as in @xmath0 car . our cloudy models for @xmath0 car would then suggest that the transition from atomic to molecular gas in such pne would be accompanied by a similar sharp drop in observed dust color temperature at mid - ir wavelengths , possibly insensitive to grain size as well . since @xmath0 car is the only lbv nebula known to have a dense h@xmath1 shell , we must ask if this is a truly unique property of @xmath0 car itself , or if it is simply a consequence of @xmath0 car being observed at a special time so soon after its mass ejection ; the homunculus is also the _ youngest _ lbv nebula . we expect that in the decades immediately following the outburst , the ejecta around @xmath0 car were extremely dense and self - shielding , permitting the formation of not only dust , but also h@xmath1 and even more complex polyatomic molecules like ammonia ( smith et al .
the physical conditions in this early high - density epoch are of great interest .
this material has since been expanding and thinning , as the observed extinction drops and the star brightens ( whitelock et al .
1983 ; davidson et al .
1999 ; smith et al .
2000 ) .
due to the high mass of the homunculus and its relatively empty surroundings , the shell will continue to expand quasi - ballistically .
thus , the hydrogen density will drop as @xmath53 , and because of ballistic motion , roughly as @xmath54 as well . in another couple
hundred years , the density in the nebula will drop below the threshold at which h@xmath1 can survive ( this may happen even sooner , since @xmath0 car resides in a giant h ii region with a strong ambient uv field ) .
eventually , the homunculus will become an atomic shell resembling the massive shells seen around other lbvs and lbv candidates , like the more evolved nebulae around the pistol star ( figer et al .
1999 ) , ag car ( voors et al .
2000 ) , and p cygni ( smith & hartigan 2006 ) . at these later stages ,
the homunculus may even overtake and mix with more distant outer ejecta ( walborn 1976 ) , appearing as a single ring - like lbv nebula , thereby erasing the details of its eruptive history .
we have modeled the basic double - shell structure in the homunculus , as traced most vividly by near - ir emisison from [ fe ii ] and h@xmath1 ( paper i ) .
our aim was to constrain the density which permits the survival of h@xmath55 , and thus , to provide an independent constraint on the mass of the homunculus and the physical conditions in the nebula .
the main observational constraints on our cloudy models were the geometric thickness of each zone , the electron density in the inner [ fe ii ] zone , and the survival of both h@xmath1 and fe@xmath7 in the thin outer zone .
our primary conclusions are as follows : \3 .
the relative mass of gas in the two zones closely matches that derived for warm and cool dust components in the same two zones ( smith et al .
this suggests that previous assumptions of a normal gas : dust mass ratio of 100 are not wildly in error .
we predict a sharp drop in the dust temperature between the inner and outer zones of the double shell , as observed ( smith et al .
the temperature in the outer zone approache the equilibrium blackbody temperature independent of grain size , due to shielding of uv radiation .
this relaxes constraints on the unusually large grains inferred from observed dust temperatures .
even though dust formation was efficient in the ejecta of @xmath0 car , we find that the gas phase fe abundance is still roughly solar .
one normally expects fe to depleted by two orders of magnitude in dusty environments .
this has occurred despite the severe depletion of c and o in the ejecta .
n.s . was supported by nasa through grant hf-01166.01a from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
research into the physical processes in the ism by g.j.f . is supported by nsf ( ast0307720 ) and nasa ( nag5 - 12020 ) .
log r@xmath56 & cm & 17.255 & 17.398 & 17.398 & 17.398 & 17.398 & 17.398 + log r@xmath57 & cm & 17.398 & 17.415 & 17.415 & 17.415 & 17.415 & 17.415 + log @xmath58r & cm & 16.845 & 16.0 & 16.0 & 16.0 & 16.0 & 16.0 + log n@xmath27 & @xmath5 & 5.5 & 7.0 & 6.7 & 6.5 & 6.3 & 6.0 + log n@xmath59 & @xmath15 & 22.35 & 23.0 & 22.7 & 22.5 & 22.3 & 22.0 + m@xmath60 & m@xmath8 & 1 & 33 & 16 & 11 & 7.5 & 3.7 + filling factor & & 0.5 & 0.9 & 0.9 & 0.9 & 0.9 & 0.9 + log n@xmath26 & @xmath5 & 3.7 & 4.5 & 4.35 & 4.26 & 4.2 & 4.0 + t@xmath61 & k & 170 & 118 & 142 & 169 & 190 & 191 + fe@xmath6/fe & & 1.0 & 2(-4 ) & 0.97 & 1.0 & 1.0 & 1.0 + n(h@xmath6)/n@xmath27 & & 0.016 & 2(-12 ) & 2(-3 ) & 3(-3 ) & 3(-3 ) & 1(-4 ) + max 2n(h@xmath1)/n@xmath27 & & 5(-10 ) & 1.0 & 1.0 & 0.99 & 0.42 & 2(-4 ) + d 2n(h@xmath1)/n@xmath27=0.5 & ... & ... & 0.15 & 0.36 & 0.61 & ... & ... + d 2n(h@xmath1)/n@xmath27=0.99 & ... & ... & 0.41 & 0.88 & ... & ... & ... + @xmath62 & mag & 1.1 & 8.9 & 4.5 & 2.8 & 1.8 & 0.9 + log r@xmath56 & cm & 16.0 & 16.498 + log r@xmath57 & cm & 16.498 & 16.618 + log @xmath58r & cm & 16.332 & 16.0 + log n@xmath27 & @xmath5 & 6.3 & 7.0 + log n@xmath59 & @xmath15 & 22.63 & 23.0 + filling factor & & 0.5 & 0.9 +
|
we present models that reproduce the observed double - shell structure of the homunculus nebula around @xmath0 carinae , including the stratification of infrared h@xmath1 and [ fe ii ] emission seen in data obtained with the phoenix spectrograph on gemini south , as well as the corresponding stratified grain temperature seen in thermal - infrared data .
tuning the model to match the observed shell thickness allows us to determine the threshold density which permits survival of h@xmath1 .
an average density of @xmath2(0.51)@xmath310@xmath4 @xmath5 in the outer zone is required to allow h@xmath1 to exist at all latitudes in the nebula , and for fe@xmath6 to recombine to fe@xmath7 .
this gives independent confirmation of the very large mass of the homunculus , indicating a total of roughly 1535 m@xmath8 ( although we note reasons why the lower end of this range is favored ) . at the interface between the atomic and molecular zones , we predict a sharp drop in the dust temperature , in agreement with the bimodal dust color temperatures observed in the two zones . in the outer molecular shell
, the dust temperature drops to nearly the blackbody temperature , and becomes independent of grain size because of self - shielding at shorter uv wavelengths and increased heating at longer wavelengths .
this relaxes constraints on large grain sizes suggested by near - blackbody color temperatures .
finally , from the strength of infrared [ fe ii ] emission in the inner shell we find that the gas - phase fe abundance is roughly solar .
this is astonishing in such a dusty object , where one normally expects gaseous iron to be depleted by two orders of magnitude .
|
the ever - increasing amount of information that is produced in the age of big data calls for efficient techniques for storage and processing of a large number of high - dimensional signals .
compact representations can be obtained with different methods depending whether the particular task requires signal reconstruction ( e.g. , image and video compression for delivery and visualization ) or the goal is to infer some information from the signals ( e.g. , in classification , regression , information retrieval , etc . ) .
embeddings provide compact representations of signals for the latter tasks .
formally , an embedding is a transformation that maps a set of signals in a high dimensional space to a lower dimensional space , in such a way that the geometry of the set is approximately preserved .
the concept of embedding has been successfully used in the context of information retrieval @xcite , where it is usually called `` hashing '' .
an important class of signal embeddings are those preserving the distances among pairs of signals . johnson and lindenstrauss @xcite famously stated that an embedding can be realized with a lipschitz mapping to approximately preserve euclidean distances with a dimension of the embedding space that only depends on the desired distortion and logarithmically in the number of signals to be embedded .
random projections have been shown to implement such embedding with high probability .
several extensions have later been proposed , allowing one to approximately preserve the angle between signals @xcite , control the maximum distance that is embedded @xcite , or preserve the jaccard distance @xcite .
recently , some works have studied learning embeddings @xcite from training data to derive compact codes by exploiting the particular geometry of the dataset ( e.g. , signals living close to a manifold ) .
in this paper we propose an approach to construct an embedding that is not based on learning and does not require a training set of data , but rather is adapted to a single reference signal .
this choice maintains to some degree the universality of random projections and it is useful when the data present no particular structure .
jegou et al .
@xcite empirically explored a similar idea by proposing to choose hash functions with a robustness criterion , that essentially measures how far a random projection falls from the edges of a quantization interval .
our work presents a rigorous analytical treatment of a binary embedding obtained from the selection of the random projections of a reference signal with largest magnitude .
the analysis of the embedding provides insights on its advantages , particularly in mitigating the difficulty of low - contrast nearest neighbor problems and superior performance on classification tasks , e.g. in a neural network .
[ def : embedding ] a mapping @xmath0 of metric spaces , endowed with distances @xmath1 and @xmath2 is called an embedding with distortion @xmath3 if @xmath4 for some constant @xmath5 and for all @xmath6 .
a well - known binary embedding is the sign random projections @xcite where a random matrix @xmath7 made of independent and identically distributed ( i.i.d . )
gaussian entries is used to compute some random projections , which are then quantized to a binary representation by keeping their sign .
the hamming distance between the binary vectors approximately preserves the angle between the signals in the original space @xcite , i.e. , @xmath8 , being @xmath9 and @xmath10 the @xmath11-th row of @xmath7 .
a reference signal @xmath12 is used to generate the adaptive embedding in the following way . a number @xmath13 of random projections
is computed by means of an i.i.d .
gaussian matrix @xmath14 as @xmath15 .
the @xmath16 entries with largest magnitude are identified and their locations stored in vector @xmath17 .
the @xmath16-bit resulting binary code is : @xmath18 where @xmath19 is the matrix @xmath7 restricted to the rows indexed by @xmath17 .
the locations vector @xmath17 is saved as side information of the embedding and used as in whenever a new signal is to be embedded , i.e. , @xmath20 , where @xmath21 is a generic signal and @xmath22 its embedding .
the first theorem we prove confirms that the hamming distance @xmath23 , i.e. the number of differing entries , between binary codes obtained with the adaptive embedding concentrates around its expected value .
[ thm : expected ] let @xmath24 be a set of @xmath25 signals and @xmath6 .
let @xmath14 with @xmath26 , @xmath15 and the locations @xmath17 of the @xmath27 entries in @xmath28 with largest magnitude be known .
let @xmath29 and @xmath30 .
then for @xmath31 , @xmath32 with @xmath33 = \frac{1}{m}\sum_{i=1}^m p_i \\ p_i & = \frac{1}{2 } + \frac{1}{2}\mathrm{erf}\left ( -y_{l_i } \frac{{{\ensuremath{\bm{\mathrm{u}}}}}^t { { \ensuremath{\bm{\mathrm{v}}}}}}{\sqrt{2}\sigma\vert { { \ensuremath{\bm{\mathrm{u}}}}}\vert \sqrt{\vert { { \ensuremath{\bm{\mathrm{u}}}}}\vert^2 \vert { { \ensuremath{\bm{\mathrm{v}}}}}\vert^2 - ( { { \ensuremath{\bm{\mathrm{u}}}}}^t { { \ensuremath{\bm{\mathrm{v}}}}})^2 } } \right ) \end{aligned}\ ] ] let us consider a single measurement in the @xmath34 location @xmath35 and @xmath36 .
then @xmath37 $ ] is a bivariate gaussian with zero mean and covariance @xmath38 .
suppose that @xmath39 is observed to be @xmath40 , then the conditional distribution of @xmath41 given @xmath40 is @xmath42 .
after quantization of the measurements , the probability of mismatching bits in position @xmath34 is @xmath43 define the following random variable @xmath44 then , @xmath45 is a poisson binomial random variable measuring the hamming distance between @xmath46 and @xmath22 .
. can be readily obtained using chernoff bounds @xcite for the tails of @xmath47 .
the previous theorem holds for a fixed pair of signals .
it is customary to derive an asymptotic result on the number of measurements needed to provide a distortion @xmath48 around the expectation when signals are drawn from a finite set of cardinality @xmath25 .
standard derivation using a union bound on eq .
yields @xmath49 , which is exactly the same as classic results on non - adaptive random projections @xcite .
the advantages of the proposed method are , in fact , due to the modified expected value rather than the variance . moreover , the previous theorem supposed we knew the values of the projections of the reference signal at the locations kept as side information .
this allows us to compute the exact expected value of the hamming distance in the embedded space as function of the inner product ( or correlation coefficient ) between the original signals
. however , it might be useful to have some a - priori knowledge about the embedding without the need to know the reference signal .
the following theorem approximately bounds the expected value of the embedding by characterizing the order statistics of @xmath50 .
the @xmath51-th order statistic of a statistical sample is equal to its @xmath51th - smallest value .
let us call @xmath52 the probability density function of the @xmath51-th order statistic of @xmath50 .
we could then in principle compute the probability of bit mismatch as @xmath53 and then repeat the same poisson binomial argument as before .
however , this is cumbersome to compute and we instead derive some bounds . under the same assumptions as theorem 1 , and being @xmath54 the expected value of the @xmath55-th order statistic of a sample of @xmath56 of size @xmath57 : @xmath58 \leq \frac{1}{2}+ \frac{1}{2}\mathrm{erf}\left ( \frac{-e_{2(m_\text{pool}-m+1);2m_\text{pool } } { { \ensuremath{\bm{\mathrm{u}}}}}^t { { \ensuremath{\bm{\mathrm{v}}}}}}{\sqrt{2}\sigma\vert { { \ensuremath{\bm{\mathrm{u}}}}}\vert \sqrt{\vert { { \ensuremath{\bm{\mathrm{u}}}}}\vert^2 \vert { { \ensuremath{\bm{\mathrm{v}}}}}\vert^2 - ( { { \ensuremath{\bm{\mathrm{u}}}}}^t { { \ensuremath{\bm{\mathrm{v}}}}})^2 } } \right)\end{aligned}\ ] ] we first notice that @xmath59 , @xmath60 so that @xmath61 = \frac{1}{m}\sum_{i=1}^m p_i \leq p_{m}$ ] . then , @xmath62 \leq g ( { \mathbb{e}}[\tau ] ) $ ] by applying jensen s inequality to @xmath63 , i.e. , the same gaussian tail probability as before .
also notice that the convexity of @xmath64 allows us to use jensen s inequality .
@xmath65 = \tilde{e}_{(m_\text{pool}-m+1);m_\text{pool}}$ ] is the expected value of the @xmath66-th order statistic of a sample of size @xmath13 from a half gaussian ( since we consider @xmath50 ) .
we then notice that that is equivalent @xcite to @xmath54 , i.e. , the @xmath55-th order statistic of a sample of size @xmath57 of a full gaussian with zero mean and @xmath67 variance . as a further remark ,
according to @xcite an approximation of the expected value of the desired order statistic is : @xmath68 being @xmath69 and @xmath70 the inverse cdf of a normal distribution with zero mean and variance @xmath67 .
so far we considered distances between a test signal and the reference used to adapt the embedding .
we will now consider what happens to the distance between any arbitrary pair of signals @xmath21 and @xmath71 .
qualitatively , we can say that the curve of the expected value of the hamming distance in the embedded space as function of the original distance between @xmath21 and @xmath71 will be somewhere between the one predicted by theorem [ thm : expected ] and the one of non - adaptive sign random projections depending on how much @xmath71 is close to the reference @xmath12 .
the following theorem formalizes this concept .
[ thm : expected_3party ] let @xmath24 be a set of @xmath25 signals and @xmath72 .
let @xmath14 with @xmath26 , @xmath15 and the locations @xmath17 of the @xmath27 entries in @xmath28 with largest magnitude be known .
let @xmath29 , @xmath30 , and @xmath73 .
then for @xmath31 , @xmath74 with @xmath75 = \frac{1}{m}\sum_{i=1}^m p_i \\ p_i & = \left [ f\left ( \begin{bmatrix } 0\\ + \infty \end{bmatrix } \right ) - f\left ( \begin{bmatrix } 0\\ 0 \end{bmatrix } \right ) \right ] \left [ 1 - f\left ( \begin{bmatrix } + \infty\\ 0 \end{bmatrix } \right ) \right ] \nonumber \\ & + \left [ f\left ( \begin{bmatrix } + \infty\\ 0 \end{bmatrix } \right ) - f\left ( \begin{bmatrix } 0\\ 0 \end{bmatrix } \right ) \right ] f\left ( \begin{bmatrix } + \infty\\ 0 \end{bmatrix } \right),\end{aligned}\ ] ] being @xmath76 the cdf of a bivariate gaussian with mean @xmath77 and covariance @xmath78 .
the proof is similar to the proof of theorem [ thm : expected ] .
let us consider a single measurement in the @xmath34 location @xmath35 , @xmath36 , @xmath79 .
then @xmath80 $ ] is gaussian with zero mean and covariance @xmath81 .
suppose that @xmath39 is observed to be @xmath40 , then the conditional distribution of @xmath82 $ ] given @xmath40 is @xmath83 \vert y_i=\tau_i ) \sim \mathcal{n}\left ( \mathbf{\mu } ' , \mathbf{\sigma } ' \right)$ ] with @xmath77 and covariance @xmath78 .
after quantization of the measurements , the probability of mismatching bits in position @xmath34 is @xmath84 define the random variable @xmath85 as in , then @xmath45 is a poisson binomial random variable measuring the hamming distance between @xmath22 and @xmath86 .
eq . can be readily obtained using chernoff bounds @xcite for the tails of @xmath47 . the key distinction between sign random projections @xcite and
the method presented in this paper is that the former provides a linear relationship between the hamming distance in the embedded space and the angle in the original space . on the other hand
, the binary adaptive embedding provides a nonlinear relationship between the two distances , with the important property that the hamming distances observed with the adaptive embedding are always smaller than those observed with sign random projections .
this property is at the core of the improved performance of the embedding for tasks such as binary classification , as discussed in sec .
[ sec : experimental ] .
[ fig:2party ] shows the hamming distance in the embedded space as function of the inner product between the test signal and the reference signal in the original space .
it can be noticed how the curve of the adaptive embedding always lies below the one for the non - adaptive embedding . for a fixed value of @xmath13 increasing the umber of measurements @xmath16 will reduce the variance and move the expected value towards that of the non - adaptive embedding .
viceversa , for a fixed @xmath16 increasing @xmath13 will lower the curve .
[ fig:3party ] shows the hamming distance between the first test signal and the second test signal in the embedded space as function of the inner product between the first test signal and the second test signal ( @xmath87 ) and between the second test signal and the reference signal ( @xmath88 ) in the original space .
notice how for decreasing @xmath88 the shape of the embedding tends to the one of a non - adaptive embedding .
, @xmath89 .
non - adaptive curve uses sign random projections @xcite . ]
[ fig:2party ] , @xmath89 . ] [ fig:3party ]
a measure of difficulty @xcite of a nearest neighbor search problem is the contrast @xmath90 , which is defined as the ratio between the distance of the closest false neighbor and that of the farthest true neighbor .
locality sensitive hashing @xcite is able to find approximate nearest neighbors in a time @xmath91 that is sublinear in the database size @xmath25 but that degenerates to linear search as the contrast approaches 1 .
the curse of dimensionality makes low contrast more probable when high - dimensional spaces are considered .
the adaptive embedding presented in this paper can be used as a dimensionality reduction technique to solve approximate nearest neighbor search more efficiently in low - contrast scenarios . in order to show this we develop an experiment in a high - dimensional space where the goal is to find the nearest neighbors of a given signal within a certain radius .
true neighbors are generated as standard gaussian vectors with @xmath92 i.i.d .
entries with an expected correlation coefficient equal to 0.07 to a reference that is also used as query .
disturbing signals are i.i.d .
gaussian with zero expected correlation .
notice that they are almost orthogonal to each other but the contrast is low because the true neighbours are weakly correlated with our query .
this problem is not unrealistic and , in fact , as an example , it occurs in the detection of photo - response non - uniformity artifacts from camera sensors @xcite@xcite , used to attribute a given picture to a given camera sensor . in this experiment
all the signals in the database are adaptively embedded , i.e. , the locations of the @xmath16 entries of largest magnitude are identified and stored with the binary code .
the random projections of the query are computed and appropriately subsampled according to the locations stored for each database signal under test .
the storage requirement for each database entry is the sum of the bits needed by the binary measurements and the overhead due to the adaptively chosen locations and it amounts to @xmath93 $ ] bits . fig .
[ fig : roc ] shows the receiver operating characteristic ( roc ) showing the probability of detection of a true neighbor against the probability of false alarm .
we notice that the adaptive embedding provides a performance closer to the uncompressed case .
two non - adaptive strategies are presented for a fair comparison . a non - adaptive method using the same storage as the adaptive method would use @xmath94 $ ] binary random projections , with the drawback of increased computational complexity in the hamming distance evaluation .
the second strategy equalizes computational complexity , thus using @xmath95 non - adaptive measurements .
this is advantageous in terms of storage but it performs significantly worse .
finally , we compared the proposed method with the universal embedding of boufounos et al .
@xcite which is a kind of adaptive embedding where the quantizer can be parametrized in order to distort the expected value of signal distances , similarly to the embedding proposed in this paper . the universal embedding bounds the maximum distance that is embedded , beyond which points become indistinguishable .
it is therefore expected to have poor performance in low - contrast , low - correlation scenarios , as it appears from fig .
[ fig : roc ] . , where the quantization step is @xmath96 . , @xmath97 , @xmath98
the query is used as reference signal . ]
[ fig : roc ] in this section we apply the adaptive embedding to a multiclass linear classifier in order to improve its storage and computational efficiency .
linear classifiers are widely used in the context of deep neural networks , where the layers of the network are trained to disentangle the features of each class and a simple linear classification layer provides the class labels .
a @xmath51-class linear classifier can be written as @xmath99 , being @xmath100 the class label , @xmath101 the weights vectors and @xmath102 a feature vector .
the weights are learned during the training phase using a suitable loss function such as the hinge loss @xcite for support vector machines or the softmax cross - entropy @xcite for multinomial logistic regression which is more popular in deep neural networks . since
the feature vectors may be high dimensional and the number of classes large , this operation may require significant storage space for the real - valued weights as well as computational resources to compute all the inner products .
this can be overcome using an embedding such as sign random projections .
after the training phase is completed , the weights are embedded in a compact binary code for each class . during predictions the feature vectors
are also embedded , the hamming distance with the weights is computed and the class label corresponding to the minimum distance is selected , i.e. @xmath103 , being @xmath104 and @xmath105 . replacing sign random projections with the proposed adaptive embedding can improve the classification performance of the compressed system .
overall , there are as many adaptive embeddings as the number of classes .
the random projections of each @xmath101 are used to compute a different set of locations @xmath106 for each class . at test time , the feature vector is embedded @xmath51 times to generate @xmath107 ( this amounts to generating @xmath13 non - adaptive projections and then subsampling according to the corresponding @xmath106 ) .
hence , the class label is given by @xmath108 , being @xmath109 and @xmath110 .
the following experiment is a classification problem on the cifar-10 dataset @xcite comprising 10 classes .
we implemented the same convolutional neural network architecture presented in @xcite .
this network is composed of 8 convolutional layers followed by 2 fully connected layers all with relu activation units @xcite and a final linear layer .
the last linear layer outputs one of the @xmath111 class labels from a @xmath112-dimensional input feature vector . after conventional training of the network
, we replaced the layer weights with its embedded codes as explained above . for the adaptive method we used @xmath113 .
table [ table : cnn ] shows the classification accuracy as function of the number of measurements used by the embedding .
it can be noticed that the adaptive embedding allows to achieve a significant dimensionality reduction at a negligible loss in terms of classification accuracy , with respect to both sign random projections @xcite and the universal embedding @xcite .
the quantization step size of the universal embedding has been optimized via cross validation to value @xmath114 .
.classification accuracy [ cols="^,^,^,^,^",options="header " , ]
this paper presented a technique to generate compact binary codes from high - dimensional signals adapting them to a reference signal .
the resulting embedding displays interesting properties that allow to improve performance in classification tasks when those are performed in the reduced - dimensionality domain . future work will focus on generalizing the approach to sub - gaussian and structured sensing matrices . m. s. charikar , `` similarity estimation techniques from rounding algorithms , '' in _ proceedings of the thiry - fourth annual acm symposium on theory of computing _ , ser .
stoc 02.1em plus 0.5em minus 0.4emnew york , ny ,
usa : acm , 2002 , pp . 380388 .
[ online ] .
available : http://doi.acm.org/10.1145/509907.509965 l. jacques , j. n. laska , p. t. boufounos , and r. g. baraniuk , `` robust 1-bit compressive sensing via binary stable embeddings of sparse vectors , '' _ ieee trans .
inf . theory _ ,
59 , no . 4 , pp . 20822102 , april 2013 .
a. broder , `` on the resemblance and containment of documents , '' in _ proceedings of the compression and complexity of sequences 1997_. 1em plus 0.5em minus 0.4emwashington , dc , usa : ieee computer society , 1997 , pp .
[ online ] .
available : http://dl.acm.org/citation.cfm?id=829502.830043 y. weiss , a. torralba , and r. fergus , `` spectral hashing , '' in _ advances in neural information processing systems 21 _ , d. koller , d. schuurmans , y. bengio , and l. bottou , eds.1em plus 0.5em minus 0.4emcurran associates , inc .
, 2009 , pp
. 17531760 .
[ online ] .
available : http://papers.nips.cc/paper/3383-spectral-hashing.pdf k. he , f. wen , and j. sun , `` k - means hashing : an affinity - preserving quantization method for learning binary compact codes , '' in _ the ieee conference on computer vision and pattern recognition ( cvpr ) _ , june 2013 .
h. jegou , l. amsaleg , c. schmid , and p. gros , `` query adaptative locality sensitive hashing , '' in _ 2008 ieee international conference on acoustics , speech and signal processing _ ,
march 2008 , pp .
825828 .
k. beyer , j. goldstein , r. ramakrishnan , and u. shaft , _ _ when is `` nearest neighbor '' meaningful?__1em plus 0.5em minus 0.4emberlin , heidelberg : springer berlin heidelberg , 1999 , pp . 217235 . [ online ] .
available : http://dx.doi.org/10.1007/3-540-49257-7_15 d. valsesia , g. coluccia , t. bianchi , and e. magli , `` compressed fingerprint matching and camera identification via random projections , '' _ ieee transactions on information forensics and security _
10 , no .
14721485 , july 2015 .
|
we use some of the largest order statistics of the random projections of a reference signal to construct a binary embedding that is adapted to signals correlated with such signal .
the embedding is characterized from the analytical standpoint and shown to provide improved performance on tasks such as classification in a reduced - dimensionality space .
binary embeddings , random projections
|
in the paper @xcite soergel gave a combinatorial description of a certain category of harish - chandra bimodules over a simple lie algebra @xmath0 .
this category was and continues to be of primary interest in the infinite - dimensional representation theory of simple lie algebras .
soergel described the category anew as a full subcategory of bimodules over a certain ring @xmath1 . the objects in this subcategory are now commonly called soergel bimodules .
the category of soergel bimodules is additive and monoidal , and soergel constructed an isomorphism between the grothendieck ring of his category and the integral form of the hecke algebra of the weyl group @xmath2 of @xmath0 .
hence , soergel s construction provides a categorification of the hecke algebra .
given a @xmath3-dimensional @xmath4-vector space @xmath5 and a generic @xmath6 , there are commuting actions of the quantum group @xmath7 and the hecke algebra @xmath8 of the symmetric group @xmath9 on @xmath10 .
these actions turn the quotient @xmath11 of the quantum group and @xmath12 of the hecke algebra by the kernels of these action into a dual pair .
a categorical realization of the triple @xmath13 was given by grojnowski and lusztig @xcite via categories of perverse sheaves on products of flag and partial flag varieties , also see @xcite .
many foundational ideas about categorification were put forward by igor frenkel in the early 90 s ( a small fraction of these ideas formed a part of the paper @xcite ) . in particular , frenkel conjectured @xcite that quantum groups and not just their finite - dimensional quotients @xmath11 can be categorified .
these conjectures remained open until recently , when categorifications of quantum @xmath14 and @xmath15 were discovered in @xcite and @xcite , with a related but different approach developed in @xcite . in the categorifications
@xcite of quantum groups , 2-morphisms are given by linear combinations of planar diagrams , modulo local relations .
the parallel objective would be to categorify the hecke algebra in the same spirit , using planar diagrams .
soergel had already provided a categorification , so it remains to ask whether his category can be rephrased diagrammatically .
diagrammatics should also provide a presentation of the category by generators and relations . a similar question was recently posed by libedinsky @xcite , who essentially produced such a description for categorifications of hecke algebras associated to `` right - angled '' coxeter systems . here
we answer this question positively in the case of the hecke algebra associated to the symmetric group .
this is , of course , the hecke algebra that appears in the schur - weyl duality for @xmath16 . for notational convenience , we use @xmath17 , not @xmath18 , as our parameter and define a diagrammatical version @xmath19 of the category of soergel bimodules that categorifies the hecke algebra of the symmetric group @xmath20 . in some sense ,
diagrammatic categorifications are very `` low - tech , '' in that they can be described easily and do not rely on heavy machinery .
while the proof that soergel bimodules categorify the hecke algebra is based on elaborate commutative algebra , showing that indecomposable bimodules descend to the kazhdan - lusztig basis of the hecke algebra utilizes kazhdan - lusztig theory @xcite .
this , in turn , is intrinsically linked to fundamental developments in representation theory that relate highest weight categories with d - modules on flag manifolds @xcite , as well as the theory of perverse sheaves @xcite and etale cohomology .
one hopes that a diagrammatic approach will help to visualize and work with these sophisticated constructions .
we start with an intermediate category @xmath21 whose objects are finite sequences @xmath22 of numbers between @xmath23 and @xmath18 .
an object is represented graphically by marking @xmath24 points in the standard position ( say , having coordinates @xmath25 ) on the @xmath26-axis and assigning labels @xmath27 to marked points from left to right .
morphisms between @xmath28 and @xmath29 are given by linear combinations ( with coefficients in a ground field @xmath30 ) of planar diagrams embedded in the strip @xmath31 $ ] .
these diagrams are decorated planar graphs , where edges may extend to the boundary @xmath32 .
each edge carries a label between @xmath23 and @xmath18 , so that the induced labellings of the lower and upper boundaries are @xmath28 and @xmath29 , respectively . in the interior of the strip , we allow * vertices of valence 1 , * vertices of valence 3 with all 3 edges carrying the same label , * vertices of valence 4 seen as intersections of @xmath33 and @xmath34-labelled lines with @xmath35 , * vertices of valence 6 with the edge labelling @xmath36 , reading clockwise around the vertex .
furthermore , boxes labelled by numbers between @xmath23 and @xmath17 are allowed to float in the regions of the graph .
we impose a set of local relations on linear combinations of these diagrams , including invariance of diagrams under all isotopies .
a subset of the relations says that @xmath33 is a frobenius object in the category @xmath21 .
these diagrammatic generators and relations were guessed by studying generating morphisms between tensor products of basic soergel bimodules @xmath37 and writing down relations between these morphisms . the space of morphisms in @xmath21 between @xmath28 and @xmath29 is naturally a graded vector space .
allowing grading shifts and direct sums of objects , then restricting to grading - preserving morphisms , and finally passing to the karoubian closure of the category results in a graded @xmath30-linear additive monoidal category @xmath19 .
our main result ( theorem [ mainthm ] in section [ sec - conseq ] ) is an explicit equivalence between this category and the category @xmath38 of soergel bimodules .
the category @xmath19 is monoidal , and can be viewed as a 2-category with a single object ( where in a diagrammatic 2-category , regions are labelled by objects , and edges by 1-morphisms ) .
this may make it easier to explore similarities with the categorifications of quantum groups in @xcite , where regions of diagrams are labelled by integers in @xcite and integral weights of @xmath39 in @xcite .
boxes floating in regions are superficially analogous to floating bubbles of @xcite .
unlike the diagrammatic categorifications in @xcite , our lines do nt carry dots and are not oriented .
there is another way to view our diagrammatics , which is not developed in this paper .
rouquier @xcite defined an action of the coxeter braid group associated to @xmath2 on the category of complexes of soergel bimodules up to homotopies .
this action was later used in an alternative construction @xcite of a triply - graded link homology theory @xcite categorifying the homfly - pt polynomial @xcite . in this approach ,
a product soergel bimodule @xmath40 is depicted by a planar diagram given by concatening elementary planar diagrams lying in the @xmath41-plane that consist of @xmath17 strands going up , with @xmath33 and @xmath42-st strands merging and splitting , see ( * ? ?
? * figures 2,3 ) .
morphisms between product bimodules can be realized by linear combinations of foams decorated two - dimensional cw - complexes embedded in @xmath43 with suitable boundary conditions .
foams have been implicit throughout papers on triply - graded link homology ( see @xcite for instance , where various arrows between planar diagrams can be implemented by foams ) , and explicitly appear in the papers on their doubly - graded cousins , see @xcite and references therein .
foams are 3-dimensional objects they are two - dimensional cw - complexes embedded in @xmath43 that produce cobordisms along the @xmath44-axis direction between planar objects corresponding to product soergel bimodules .
the planar diagrams of our paper are two - dimensional encodings of these foams , essentially projections of the foams onto the @xmath45-plane along the @xmath26-axis .
it was shown in @xcite that the action of the braid group on the homotopy category of soergel bimodules extends to a ( projective ) action of the category of braid cobordisms .
thus , the homotopy category of @xmath19 produces invariants of braid cobordisms , so that our planar diagrammatics carry information about four - dimensional objects .
this informational density indicates efficiency of such encodings .
* acknowledgments . *
the authors were supported by the nsf grants dms-0706924 and dms-0739392 .
they are grateful to geordie williamson for remarks and suggestions on an earlier version of this paper .
henceforth we will fix a positive integer @xmath18 . indices
@xmath33 , @xmath34 , and @xmath3 will range over @xmath46 if not otherwise specified .
finite ordered sequences of such indices ( allowing repetition ) will be denoted @xmath47 , as well as @xmath29 and @xmath48 . for sequences of length @xmath49 where the single entry is @xmath33 , we use @xmath33 and @xmath28 interchangeably . occassionally @xmath42 will also be used as an index , and whenever this occurs we make the tacit assumption that @xmath50 so that all indices used remain between 1 and @xmath18 .
the same goes for @xmath51 , @xmath52 and the like .
we denote the length 0 sequence by the empty set symbol @xmath53 .
we work over a field @xmath30 , usually assuming that char @xmath54 , and sometimes specializing it to @xmath4 .
given a noetherian ring @xmath1 , the category @xmath55 is the full subcategory of @xmath1-bimodules consisting of objects which are finitely generated as left @xmath1-modules . if @xmath1 is graded , the category @xmath56 is the analogous subcategory of graded @xmath1-bimodules and grading - preserving homomorphisms .
let @xmath57 be a coxeter system of a finite weyl group @xmath2 , with length function @xmath58 , and @xmath59 the identity .
the hecke algebra @xmath8 is an algebra over @xmath60}}$ ] ( we follow soergel s use @xcite of the variable @xmath61 ; related variables are denoted in the literature by @xmath62 and @xmath63 ) , which is free as a module with basis @xmath64 .
multiplication in this basis is given by @xmath65 when @xmath66 , and @xmath67 for @xmath68 .
@xmath69 is the identity element in @xmath8 and will often be written as @xmath23 . in the case
we are interested in presently , @xmath70 , and @xmath71 consists of the transpositions @xmath72 for @xmath73 .
the element @xmath74 will be denoted @xmath75 .
the hecke algebra has a presentation over @xmath60}}$ ] , being generated by @xmath75 subject to the relations @xmath76 clearly then , @xmath8 is also generated as an algebra by @xmath77 , @xmath78 , and the relations above transform into @xmath79 we often write the monomial @xmath80 as @xmath81 where @xmath47 .
notice that @xmath82 .
let @xmath83 be the involution of @xmath60}}$ ] determined by @xmath84 .
it extends to an involution of @xmath8 given by @xmath85 in particular , @xmath86 kazhdan and lusztig @xcite defined a pair of bases @xmath87 and @xmath88 for @xmath8 , which immediately proved to be of fundamental importance for representation theory and combinatorics .
the two bases are related via a suitable involution of @xmath8 , and the elements of the second kazhdan - lusztig basis @xmath89 are determined by the two properties : @xmath90 where @xmath91 $ ] has @xmath61-degree strictly less than @xmath92 for @xmath93 and @xmath94 .
there is no simple formula expressing @xmath95 in terms of @xmath96 , but observe that @xmath97 and @xmath98 .
let @xmath99}}$ ] be the @xmath60}}$]-linear map given by @xmath100 and @xmath101 if @xmath102 .
thus , @xmath103 simply picks up the coefficient of @xmath69 in @xmath26 .
the easily checked property @xmath104 for any @xmath105 implies that @xmath106 , @xmath107 . the map @xmath103 turns @xmath8 into a symmetric frobenius @xmath60}}$]-algebra .
denote by @xmath108 a @xmath61-antilinear antiinvolution @xmath109 defined uniquely by @xmath110 .
the antiinvolution and @xmath61-antilinearity conditions say that @xmath111 and @xmath112 , for @xmath113 and @xmath114}}$ ] .
consider the pairing @xmath115}}$ ] of @xmath116modules given by @xmath117 it satisfies the following properties : 1 .
the pairing is _ semi - linear _ , i.e. @xmath118 while @xmath119 , for @xmath114}}$ ] .
2 . @xmath120 is self - adjoint , i.e. @xmath121 and @xmath122 .
if @xmath123 with @xmath124 then @xmath125 . such a monomial @xmath81 is called an _ increasing monomial _ , and @xmath28 an _ increasing sequence_. when @xmath126 , the sequence @xmath28 is empty and @xmath127 .
[ unicity ] it is not difficult to observe that @xmath128 is the unique form satisfying these three properties . this is because the hecke algebra has a spanning set over @xmath60}}$ ] consisting of monomials @xmath81 , and every monomial may be reduced , by cycling the last @xmath120 to the beginning and by applying the hecke algebra relations , to an increasing monomial .
this is a simple combinatorial argument that we leave to the reader .
let @xmath129}}$ ] be given by @xmath130 this is a @xmath60}}$]-linear map , equipping @xmath8 with a symmetric frobenius @xmath60}}$]-algebra structure , just like @xmath103 did .
we have @xmath131 and @xmath132 .
in @xcite soergel introduced a category of bimodules which categorified the hecke algebra , and later generalized his construction to any coxeter group @xmath2 @xcite . within the category @xmath56 , for @xmath1 a certain graded @xmath30-algebra ( @xmath30 an infinite field of characteristic @xmath133 ) , he identified indecomposable modules @xmath134 for @xmath135 , such that the only indecomposable summands of tensor products of @xmath134 s are @xmath136 for @xmath137 .
thus , the subcategory of @xmath56 generated additively by the @xmath134 has a tensor product , and its grothendieck ring is isomorphic to @xmath8 , under the isomorphism sending @xmath138 to @xmath139 $ ] .
moreover , every @xmath134 shows up as a summand of some tensor product of various @xmath140 for @xmath68 .
while the general @xmath134 may be difficult to describe , @xmath140 has an easy description .
it is conjectured in @xcite that this isomorphism sends @xmath141 $ ] to @xmath95 for all @xmath142 , and it is proven for @xmath143 and @xmath2 a weyl group in @xcite .
henceforth we specialize to the case where @xmath70 and @xmath144 . here
one may also define the soergel bimodule @xmath145 instead of @xmath134 , as discussed in @xcite .
the @xmath134 are obtained by taking the fundamental @xmath18-dimensional representation @xmath5 of @xmath20 and looking at the coordinate rings of twisted diagonals in @xmath146 .
the @xmath145 are defined analogously with the standard ( @xmath17)-dimensional representation , which is @xmath5 plus the trivial representation .
effectively , @xmath147 $ ] and the entire story of @xmath134 translates to @xmath145 by base extension .
both versions will categorify the hecke algebra , and we will interest ourselves with the @xmath145 story below ; to deal with @xmath134 instead , see section [ subsec - quotient ] .
since we only use @xmath145 below , we will denote them as @xmath134 instead to avoid an apostrophe catastrophe . with these conventions
, we now make the story explicit .
let @xmath148 $ ] be the ring of polynomials in @xmath17 variables , with the natural action of @xmath20 .
the ring @xmath1 is graded , with @xmath149 .
if @xmath150 is the subgroup of @xmath20 generated by transpositions @xmath151 , then we denote the ring of invariants under @xmath150 as @xmath152 or @xmath153 .
thus @xmath154 are the invariants under the transposition @xmath155 .
since @xmath1 is an @xmath153-algebra , @xmath156 is a bifunctor sending two @xmath1-bimodules to an @xmath1-bimodule .
henceforth @xmath157 with no subscript denotes tensoring over @xmath1 , while @xmath158 denotes tensoring over the subring @xmath152 .
most commonly we will just use @xmath159 for various indices @xmath33 .
the soergel bimodule @xmath37 is @xmath160 , where @xmath161 denotes a grading shift .
we denote by @xmath162 the tensor product @xmath163 .
note that @xmath164 and @xmath165 an element of this module can be written as a sum of terms given by @xmath166 polynomials , one in each slot separated by the tensors .
multiplication by an element of @xmath1 in any particular slot is an @xmath1-bimodule endomorphism . for each @xmath33
there is a map of graded vector spaces @xmath167 sending @xmath168 which lowers the degree by two .
since @xmath169 is @xmath170-invariant , it is not hard to see that @xmath171 is also @xmath170-invariant . since @xmath172 , this implies that @xmath1 is a free graded @xmath154-module of rank two , with homogeneous generators 1 and @xmath173 .
likewise , @xmath37 is a free graded @xmath1-module with a basis @xmath174 , under the action of @xmath1 by left multiplication .
a product element decomposes in this basis as @xmath175 in particular , this gives an isomorphism @xmath176 as graded left @xmath1-modules . iterating and ignoring the grading
, we see that @xmath162 is a free left @xmath1-module of rank @xmath177 , when @xmath28 has length @xmath24 .
rewriting a term @xmath178 as a sum of terms like above will happen often , and we refer to it as `` sliding '' the polynomial @xmath179 across the tensor . if @xmath179 is @xmath170-invariant then it may be slid across leaving nothing behind , while an arbitrary @xmath179 leaves terms with at either 1 or @xmath173 behind ( alternatively , we may choose to leave 1 and @xmath180 behind , if it is more convenient ) .
we also remark on spanning sets for @xmath162 .
for instance , @xmath181 has a spanning set @xmath182 where @xmath183 range over all monomials in @xmath1 .
the bimodule @xmath184 for @xmath185 has a spanning set @xmath186 , over all monomials @xmath183 , since any polynomial in the middle can be slid to the left leaving at most @xmath173 behind ( or @xmath180 , which we choose when @xmath187 ) , and that can be slid to the right ; thus @xmath188 generates it as a @xmath1-bimodule .
the bimodule @xmath189 has a spanning set @xmath190 , over all monomials @xmath183 .
this is because all polynomials anywhere between the two @xmath33 tensors may be slid out , leaving @xmath173 somewhere in - between . as an exercise , the reader may generalize this argument to an arbitrary @xmath162 and find a spanning set as a @xmath1-bimodule , consisting of @xmath191 terms , where @xmath192 is the number of pairs @xmath193 such that @xmath194 and @xmath195 for @xmath61 between @xmath196 and @xmath197 , which is equal to @xmath24 minus the number of distinct indices in @xmath28 . for more information on soergel bimodules and their applications
we refer the reader to @xcite and references therein .
several subcategories of @xmath55 and @xmath56 will play a role in what follows .
let @xmath198 be the full subcategory of @xmath55 whose objects consist of @xmath162 for all sequences of indices @xmath28 . since @xmath1 is a commutative ring , the hom spaces in @xmath198 are in fact enriched in @xmath56 .
let @xmath199 be the subcategory of @xmath56 whose objects are finite direct sums of various graded shifts of objects in @xmath198 and the morphisms are all grading - preserving bimodule homomorphisms .
finally , let @xmath38 be the karoubi envelope of @xmath199 , a category equivalent to the full subcategory of @xmath56 which contains all summands of objects of @xmath199 : @xmath200 in general , the karoubi envelope is the category which formally includes all `` summands '' , where a summand of an object @xmath201 is identified by an idempotent @xmath202 corresponding to projection to that summand .
since @xmath56 is abelian , it is idempotent - closed , and the karoubi envelope of @xmath199 can be realized as a subcategory in @xmath56 .
we refer the reader to wikipedia and other easily found online sources for basic information about karoubi envelopes .
this final category @xmath38 ( the soergel category ) is a @xmath30-linear additive monoidal category with the krull - schmidt property .
soergel showed that , when @xmath30 is an infinite field of characteristic other than @xmath203 , the indecomposable bimodules in this category are enumerated by elements of the weyl group and grading shifts ( theorem 6.16 in @xcite ) .
they are denoted by @xmath204 for @xmath205 and @xmath206 .
an indecomposable @xmath134 is determined by the condition that it appears as a direct summand of @xmath162 , where @xmath207 and @xmath208 is a reduced presentation of @xmath209 , and does not appear as direct summand of any @xmath162 , for sequences @xmath28 of length less than @xmath210 .
the grothendieck ring of @xmath38 is isomorphic to @xmath8 , under an isomorphism sending product bimodule @xmath162 to product element @xmath81 and @xmath211 to @xmath61 .
relations ( [ eqn - bisq])([eqn - bibpbi ] ) on @xmath120 s lift to isomorphisms of graded bimodules @xmath212 the first isomorphism is given by sending @xmath213 in @xmath37 to either @xmath214 or @xmath215 in @xmath216 .
these two tensors were observed in the previous section to generate @xmath216 as an @xmath1-bimodule .
the second and third isomorphisms come from the following isomorphisms in @xmath56 : @xmath217 in the category @xmath198 , given any two sequences @xmath28 and @xmath29 , the space @xmath218 is a graded free left @xmath1-module .
hence we can define its _ graded rank _
, an element of @xmath60}}$ ] , by choosing a homogeneous @xmath1-basis @xmath219 of @xmath220 and letting @xmath221 .
we can extend this assignment all the way to @xmath38 , where we get a map @xmath222}}\ ] ] by taking @xmath223 to the graded rank of @xmath224 where the latter obviously has the structure of a graded free left @xmath1-module . shifting the grading of @xmath225 will shift the grading of this hom space in the same direction , while shifting the grading of @xmath201 will shift the grading of the hom space in the opposite direction .
hence this map descends to a @xmath61-semilinear product on the grothendieck ring of @xmath38 , @xmath226 } } , \ \ \ ( [ m],[n]){\stackrel{\mbox{\scriptsize{def}}}{=}}{{\rm grk}}\big ( \oplus_{j\in \z}{{\rm hom}}_{{\mathcal{sc}}}(m\{j\},n)\big ) .\ ] ] unsurprisingly , this is the same semilinear product as defined in section [ subsec - hecke ] . to check this , following remark [ unicity ] , we only need to show two things .
the first is that @xmath37 self - biadjoint , i.e. that @xmath227 and @xmath228 via some canonical adjointness maps .
this is known , and will become evident in the sequel . the second is that when @xmath28 is an increasing sequence , @xmath229 is a free left @xmath1-module of rank 1 , generated by a morphism of degree @xmath24 , the length of @xmath28 .
this is a simple argument , presented below .
an @xmath1-bimodule map from @xmath1 to @xmath37 is clearly determined by an element of @xmath37 on which right and left multiplication by polynomials in @xmath1 are identical .
any element of @xmath37 is of the form @xmath230 , and clearly @xmath231 for @xmath232 , and @xmath233 .
hence @xmath234 can be the image of 1 under a bimodule map from @xmath1 if and only if @xmath235 .
@xmath236 thus @xmath234 can be an image iff @xmath237 and @xmath238 , iff @xmath239 .
such elements form a left @xmath1-module generated by the case @xmath240 , @xmath241 , or in other words by @xmath242 .
the element @xmath234 has degree 1 in @xmath37 , so we deduce that @xmath243 .
let us call @xmath244 the corresponding map @xmath245 , @xmath246 .
one may use the same argument for the general case .
suppose @xmath28 is increasing and has length @xmath24 .
@xmath162 is a free left @xmath1-module of degree @xmath177 , with generators @xmath247 ranging over terms where either @xmath248 or @xmath249 .
as an exercise , the reader may find the criteria for a general element to be ad - invariant under @xmath173 , and verify that the only possible bimodule maps @xmath250 are @xmath1-multiples of the following iterated version of @xmath244 : @xmath251 the first map is @xmath252 , the second map is @xmath253 , and so forth .
this generator is a map of degree @xmath24 , so that @xmath254 and @xmath255,[b_{{\underline{\textbf{\textit{i}}}}}])=t^d$ ] .
it is worth reiterating here the dependence of the above story on the choice of @xmath30 .
it is clear that when @xmath70 , our definitions of @xmath37 above will make sense over any commutative ring @xmath30 , so we can define the analog of category @xmath198 for any @xmath30 .
one can still define the isomorphisms that descend to the hecke relations and the maps which correspond to the self - adjunction of @xmath37 over @xmath30 , and so we can calculate hom spaces as above . in particular , the graded rank of @xmath256 will be @xmath257 .
however , the size of the grothendieck group of the karoubi envelope of @xmath198 in not known for an arbitrary commutative ring @xmath30 . specializing to @xmath30 an infinite field of characteristic @xmath133 , we may use soergel s result ( * ? ? ?
* sazt 6.16 ) to state that the indecomposables in the karoubi envelope correspond to elements of @xmath2 , and that @xmath38 categorifies the hecke algebra .
when @xmath143 we can go even further and match @xmath141 $ ] with the kazhdan - lusztig basis element @xmath95 .
let us summarize the soergel categorification of the hecke algebra @xmath8 of the symmetric group @xmath20 and the various structures on it , following @xcite and assuming @xmath143 .
the ring @xmath60}}$ ] is canonically isomorphic to the grothendieck group of the category @xmath258 of finitely - generated graded free @xmath1-modules and grading - preserving homomorphism . under this isomorphism @xmath259\ ] ] @xmath260 $ ] goes to @xmath23 and @xmath261 $ ] to @xmath262 .
the bar involution on @xmath60}}$ ] lifts to the contravariant equivalence that takes @xmath263 to @xmath264 the latter naturally viewed as an @xmath1-module .
the hecke algebra @xmath8 is canonically isomorphic to @xmath265 , the grothendieck group of @xmath38 .
multiplication by @xmath61 operator corresponds to the grading shift : @xmath266 = t^d [ m]$ ] .
multiplication in the hecke algebra corresponds to the tensor product of bimodules : @xmath267 \cdot [ n ] \ : = \ [ m\otimes_r n].\ ] ] the isomorphism takes @xmath268 $ ] to @xmath120 and @xmath269 $ ] to @xmath81 .
the kazhdan - lusztig basis @xmath270 is the basis of balanced indecomposable objects @xmath271\}$ ] of the grothendieck group , @xmath272 $ ] .
the pairing @xmath273 becomes the graded rank of @xmath274 , viewed as ( either left or right ) free @xmath1-module : @xmath275,[n ] ) = { { \rm grk}}\big ( { { \mathop{\oplus}\limits_{j\in\z}}}{{\rm hom}}(m\{j\},n)\big ) .\ ] ] for a soergel bimodule @xmath201 the space of bimodule homomorphisms @xmath276 is just the @xmath277-th hochschild cohomology @xmath278 of @xmath201 .
thus , @xmath279 becomes the graded rank of @xmath277-th hochschild cohomology of @xmath201 , viewed as a free graded @xmath1-module : @xmath280 ) : = { { \rm grk}}\big ( \mathrm{hh^0(r , m)}\big ) .\ ] ] the bar involution on @xmath8 corresponds to the contravariant duality on the soergel category that takes @xmath201 to @xmath281 : @xmath282 } = [ { { \rm hom}}_{r'}(m , r')\{n(n+1)\}]=q^{n(n+1)}[{{\rm hom}}_{r'}(m , r')].\ ] ] here @xmath283 is the subring of @xmath20-invariants in @xmath1 , and the expression inside the square brackets on the right hand side is naturally an @xmath1-bimodule .
antiinvolution @xmath108 lifts to the antiequivalence @xmath284 of @xmath38 that takes a bimodule to the biadjoint bimodule . for each bimodule @xmath201 in @xmath38
there exists a bimodule @xmath285 such that the functor of tensoring with @xmath285 is left and right adjoint ( biadjoint ) to the functor of tensoring with @xmath201 .
bimodule @xmath285 is well - defined up to isomorphism .
a homomorphism @xmath286 of bimodules dualizes to the homomorphism @xmath287 , resulting in an antiequivalence of @xmath38 .
notice that @xmath284 takes @xmath37 to itself and @xmath162 to @xmath288 where @xmath29 is given by reading @xmath28 from right to left .
the categorification of map @xmath103 is the functor @xmath289 of taking the 0-th hochschild homology @xmath290 of a soergel bimodule .
in particular , @xmath291 ) = { { \rm grk}}\big ( \mathrm{hh}_0(m)\big ) , $ ] and the equation @xmath292 categorifies to an isomorphism @xmath293 of @xmath294-modules . hecke algebra @xmath8 has a trace more sophisticated than @xmath295 or @xmath103 , called the ocneanu trace @xcite , which describes the homfly - pt polynomial .
the categorification of the ocneanu trace utilizes all hochschild homology groups rather than just @xmath296 , see @xcite .
invertible elements @xmath75 that satisfy the braid relations become @xcite invertible complexes @xmath297 in the homotopy category of the soergel category ( bimodule @xmath1 sits in cohomological degree 1 ) .
their inverses @xmath298 become inverse complexes @xmath299 with @xmath1 in cohomological degree @xmath300 .
homomorphism of the braid group into the hecke algebra is categorified by a projective functor from the category of braid cobordisms between @xmath301-stranded braids to the category of endofunctors of the homotopy category of the soergel category @xcite .
we follow the standard rules for the diagrammatic calculus of bimodules , or more generally for the diagrammatic calculus of a monoidal category .
the @xmath1-bimodule @xmath37 is denoted by a point labelled @xmath33 placed on a horizontal line .
the tensor product of bimodules is depicted by a sequence of labelled points on a horizontal line , so that tensor products are formed `` horizontally '' .
a vertical line labelled @xmath33 denotes the identity endomorphism of @xmath37 .
labelled lines placed side by side denote the identity endomorphism of the tensor product of the corresponding bimodules .
more general bimodule maps are represented by some symbols connecting the appropriate lines , and are composed `` vertically '' , and tensored `` horizontally '' .
all diagrams are read from bottom to top , so that the following diagram may represents a bimodule map from @xmath302 to @xmath303 .
the horizontal line without marked points represents @xmath1 as a bimodule over itself .
the empty planar diagram represent the identity endomorphism of this bimodule .
planar diagrams without top and bottom endpoints ( without boundary ) represent more general endomorphisms of @xmath1 .
we generally do not draw the horizontal line at all .
we wo nt draw horizontal lines , only the marked points on them , which will appear as endpoints of planar diagrams .
the structure of bimodule categories ( or more generally strict 2-categories ) guarantees that a planar diagram will unambiguously denote a morphism of bimodules .
we will be using so many such pictures that it will become cumbersome to continuously label each line by an index .
generally , the calculations we do will work independently for each @xmath33 , and can be expressed with diagrams that use lines labelled @xmath33 , @xmath42 and the like . in these circumstances ,
when there is no ambiguity , we will fix an index @xmath33 and draw a line labelled @xmath33 with one style , a line labelled @xmath42 with a different style , and so forth , maintaining the same conventions throughout the paper .
we use different styles of lines because most printers are black and white , but we recommend that you do your calculations at home in colored pen or pencils instead ; we even refer to the labels as `` colors '' throughout this paper .
we use the styles above when referring to indices @xmath33 , @xmath42 , @xmath51 , and @xmath34 , where @xmath34 will be used unambiguously for any index which is `` far away '' from any other indices in the picture ( in other words , when drawing a picture only involving @xmath33-colored strands , we require @xmath35 , while for a picture involving both @xmath33 and @xmath42 we require @xmath304 or @xmath305 ) . [ prop - full ] suppose we choose the subset of the morphisms in @xmath198 , including the identity morphism of each object , as well as the following morphisms : 1 . the generating morphism from @xmath1 to @xmath37 .
2 . the isomorphisms that yield the hecke algebra relations , as well as the respective projections to and inclusions from each summand in ( [ eqn - ipibimod ] ) .
3 . the unit and counit of adjunction that make @xmath37 into a self - adjoint bimodule . consider @xmath306 the subcategory generated monoidally over the left action of @xmath1 by these morphisms , i.e. it includes left @xmath1-linear combinations , compositions , and tensors of all its morphisms .
then @xmath306 is a full subcategory , and thus it is actually @xmath198 . for any objects @xmath307 in @xmath198
, there is an inclusion @xmath308 of graded left @xmath1-modules ( since it is clearly an inclusion of @xmath1-modules , and all generating morphisms are homogeneous ) .
one can define @xmath309 for any graded left @xmath1-module @xmath201 by choosing generators of @xmath310 , where @xmath311 is the ideal of positively graded elements , and it is a simple argument that a submodule of a free graded @xmath1-module with the same graded rank is in fact the entire module .
so we need only show that hom spaces in @xmath306 have the same graded rank .
we can define a bilinear form on the free algebra generated by @xmath120 by the formula @xmath312 .
our direct sum decompositions imply that the hecke algebra relations are in the kernel of this bilinear form , so it descends to a bilinear form on the hecke algebra .
each @xmath120 will be self - adjoint .
when @xmath28 is increasing , @xmath313 contains the generator of the free rank one @xmath1-module @xmath229 , since that generator is the tensor of the generating morphisms from @xmath1 to various @xmath37 .
hence it is in fact the entire module , so @xmath314 , and @xmath125 . by unicity
, this inner product agrees with our earlier inner product on the hecke algebra .
in particular , the graded ranks agree , and the inclusion is full . below we will construct a category @xmath21 of diagrams via generators and local relations , where the hom spaces will be graded @xmath1-bimodules .
we will construct a functor @xmath315 from @xmath21 to @xmath198 , showing that our diagrams give graphical presentation of morphisms in @xmath198 .
the morphisms in the image of @xmath315 will include all the morphisms enumerated in proposition [ prop - full ] , hence the functor will be full . calculating the hom spaces in @xmath21 between certain objects ( corresponding to @xmath1 , @xmath162 for @xmath28 increasing ) , we may use a similar argument to the above proposition to show that they are they are free @xmath1-modules of the same graded rank as the hom spaces in @xmath198 .
then the functor @xmath315 will be faithful , and an equivalence of categories .
this describes @xmath198 in terms of generators and relations .
let @xmath316 be the category whose objects are finite direct sums of formal grading shifts of objects in @xmath21 , but whose morphisms only include degree 0 maps .
finally , let @xmath317 be the karoubi envelope of @xmath316 .
the functor @xmath315 lifts to functors @xmath318 and @xmath319 , as in the picture below , with all three horizontal arrows being equivalences of categories .
@xmath320 finally , if one draws the generating morphisms in @xmath21 using the appropriate pictures , the relations in @xmath21 enable us to view all morphisms unambiguously as graphs , instead of diagrams .
thinking of morphisms as graphs will be more convenient when we calculate hom spaces . to be very explicit , the diagrammatic definition of the category @xmath21 makes it clear how to define the functor @xmath315 , since it is constructed out of local diagrams which are just depictions of morphisms in @xmath198 .
the category would be entirely unchanged if one used different pictures for each generating morphism .
however , when the `` correct '' pictures are chosen for the generators , the picture of any morphism turns out to be an isotopy invariant , and morphisms connecting multiple lines can be seen as vertices in a graph , so that with this choice we may define @xmath21 with graphs instead
. it would be more difficult ( a priori ) to define the functor @xmath315 directly from the graphical presentation , but that presentation is easier to examine , without relation to @xmath198 .
this section and the next several will hold the definition of the category @xmath21 , which will be a @xmath30-linear additive monoidal category , with @xmath321-graded hom spaces .
later it will become clear that hom spaces are actually graded @xmath1-bimodules .
it is generated monoidally by @xmath18 objects @xmath73 , whose tensor products will be denoted @xmath47 .
morphisms will be given by ( linear combinations of ) diagrams inside the strip @xmath322 $ ] , constructed out of lines colored by an index @xmath33 , and certain other planar diagrams , modulo local relations .
the intersection of the diagram with @xmath323 , called the _ lower boundary _ , will be a sequence @xmath28 of colored endpoints , the source of the map , and the _ upper boundary _
@xmath29 will be the target .
a vertical line colored @xmath33 represents the identity map from @xmath33 to @xmath33 .
the monoidal structure consists of placing diagrams side by side , and composition consists of placing diagrams one above the other , in the standard fashion for diagrammatic categories .
we simultaneously define the functor @xmath324 .
it is defined on objects by @xmath325 , and on morphisms as below . to show that @xmath315 is a functor one must check that the relations hold in @xmath198 , a series of simple but tedious calculations that is postponed to section [ sec - proofs ] .
all morphisms defined below will be homogeneous , with a specified degree , and @xmath315 will preserve degrees of morphisms . instead of displaying all the generators and then all the relations , we will present them in an order based on the colors they use , for clarity of exposition .
we will only display each generator with one color @xmath33 , using the conventions described in section [ subsec - diagintro ] , but the generator exists for each index @xmath33 .
in fact , it will turn out that the set of all relations is invariant under all color changes that preserve adjacency , under isotopy and rotation , and under horizontal and vertical flips ! the generating morphisms which use only one color ( or fewer :
the final generator is a map from @xmath53 to @xmath53 ) are : @xmath326{dot.eps } } } \endxy } } & 1 & \mathrm{enddot } & { \xy ( 0,-5)*+{f { \otimes}g}="1 " ; ( 0,5)*+{fg}="2 " ; { \ar@{|- > } " 1";"2"};\endxy}\\ \\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{dot.eps } } \endxy } } & 1 & \mathrm{startdot } & { \xy ( 0,-5)*+{1}="1 " ; ( 0,5)*+{{x_i}{\otimes}1 - 1 { \otimes}{x_{i+1}}}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } \\ \\ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.3 , angle=180]{merge.eps } } } \endxy } } & -1 & \mathrm{merge } & \begin{array}{ccc } { \xy ( 0,-5)*+{f { \otimes}1 { \otimes}g}="1 " ; ( 0,5)*+{0}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } & \quad & { \xy ( 0,-5)*+{f { \otimes}{x_i}{\otimes}g}="1 " ; ( 0,5)*+{f { \otimes}g}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } \end{array } \\ \\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{merge.eps } } \endxy } } & -1 & \mathrm{split } & { \xy ( 0,-5)*+{f { \otimes}g}="1 " ; ( 0,5)*+{f { \otimes}1 { \otimes}g}="2 " ; { \ar@{|- > } " 1";"2"};\endxy}\\ \\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxi.eps } } \endxy } } & 2 & \mathrm{mult\ by\ } { x_i } & { \xy ( 0,-5)*+{1}="1 " ; ( 0,5)*+{{x_i}}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } \end{array } $ ] we give these maps names , but the names are very temporary . once we explore the meaning of isotopy invariance , we will stop distinguishing between merge and split , and call them both _
trivalent vertices_. similarly we will stop distinguishing between startdot and enddot , and call them both _
dots_. we also use a shorthand for the following compositions : @xmath327{cap.eps } } } \endxy } } { \stackrel{\mbox{\scriptsize{def}}}{=}}{\vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.4 , angle=180]{dotsplit.eps } } } \endxy } } & 0 & \mathrm{cup } & { \xy ( 0,-5)*+{1}="1 " ; ( 0,5)*+{{x_i}{\otimes}1 { \otimes}1 - 1 { \otimes}1 { \otimes}{x_{i+1}}}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } \\ \\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{cap.eps } } \endxy } } { \stackrel{\mbox{\scriptsize{def}}}{=}}{\vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{dotsplit.eps } } \endxy } } & 0 &
\mathrm{cap } & \begin{array}{ccc } { \xy ( 0,-5)*+{f { \otimes}1 { \otimes}g}="1 " ; ( 0,5)*+{0}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } & \quad & { \xy ( 0,-5)*+{f { \otimes}{x_i}{\otimes}g}="1 " ; ( 0,5)*+{fg}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } \end{array } \end{array } $ ] it is worth noting here that the maps indicated as the image of these generators under @xmath315 are in fact @xmath1-bimodule maps .
this is obvious for enddot and split . for merge ,
one need only confirm that the only polynomials that can move across the tensors , on both top and bottom , are @xmath170-invariant .
startdot is precisely the generator @xmath244 of @xmath328 discussed in section [ subsec - sbimcat ] .
we now list a series of relations using only one color , the _ one - color relations _ , dividing them into 4 sets for future reference .
the first set are known as _ polynomial slides _ , which are obviously analogous to the definitions of the modules @xmath37 .
@xmath329{multxi.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{line.eps } } \endxy}}\ + \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxp.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{line.eps } } \endxy}}\ & \
= \ & \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{line.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxi.eps } } \endxy}}\ + \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{line.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxp.eps } } \endxy } } \label{eq - slide1 } \\ \ \ & & \
\ \nonumber \\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxi.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxp.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{line.eps } } \endxy}}\ & \ = \ & \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{line.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxi.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxp.eps } } \endxy } } \label{eq - slide2 } \\ \ \ & & \ \ \nonumber \\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxj.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{line.eps } } \endxy}}\ & \ = \ & \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{line.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxj.eps } } \endxy } } \label{eq - slide3}\end{aligned}\ ] ] multiplication by @xmath173 maps , for any @xmath33 , are called _
boxes_. a morphism from @xmath53 to @xmath53 consisting of a sum of products of boxes will be called a _ polynomial _ , and will obviously correspond to an element @xmath330 . as a shorthand , we may also draw such a morphism as a box with the corresponding element @xmath331 inside .
as a map from @xmath53 to @xmath53 , and thus a closed diagram , a polynomial may be placed in any region of another diagram . the same obvious corollary as for @xmath37 is that any polynomial can be `` slid '' across a line colored @xmath33 , yielding a linear combination of morphisms with the lines unchanged , but with different configurations of boxes , and at most one box labelled @xmath33 left behind .
@xmath329{boxf.eps } } \endxy}}\ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.8]{line.eps } } \endxy } } \ & \ = \ & \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.8]{line.eps } } \endxy}}\ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{boxpf.eps } } \endxy}}\ \ + \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.5]{multxi.eps } } \endxy } } \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.8]{line.eps } } \endxy}}\ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{boxdf.eps } } \endxy } } \label{eq - slide5 } \\ \ \ & & \ \
\nonumber \\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.8]{line.eps } } \endxy}}\ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{boxf.eps } } \endxy } } \ & \ = \ & \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{boxpf.eps } } \endxy } } \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.8]{line.eps } } \endxy}}\ \ + \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{boxdf.eps } } \endxy } } \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.8]{line.eps } } \endxy}}\ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.5]{multxi.eps } } \endxy } } \label{eq - slide6 } \end{aligned}\ ] ] relations ( [ assoc])-([twistdot2 ] ) below state that @xmath33 is a frobenius object in the category @xmath21
. existence of the functor @xmath319 implies that @xmath37 is a frobenius object in the category @xmath198 .
@xmath332{doublemerge.eps } } \endxy } } = { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.2]{doublemerge.eps } } } \endxy } } \label{assoc}\ ] ] @xmath333{doublemerge.eps } } } \endxy } } = { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.2 , angle=180]{doublemerge.eps } } \endxy } } \label{coassoc}\ ] ] @xmath332{splitdot.eps } } \endxy } } = { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.5]{line.eps } } \endxy } } = { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.2]{splitdot.eps } } } \endxy}}\label{counit}\ ] ] @xmath333{splitdot.eps } } } \endxy } } = { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.5]{line.eps } } \endxy } } = { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.2 , angle=180]{splitdot.eps } } \endxy}}\label{unit}\ ] ] @xmath334{cupcap.eps } } \endxy } } \;\ ; = \;\ ; { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.2]{tallline.eps } } \endxy } } \;\ ; = \;\ ; { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.25]{cupcap.eps } } } \endxy}}\label{biadjoint}\ ] ] @xmath335{splitcup.eps } } \endxy } } \ =
\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{tallmerge.eps } } \endxy } } \ =
\ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.3]{splitcup.eps } } } \endxy } } \label{twistmerge}\ ] ] @xmath336{splitcup.eps } } } \endxy } } \
= \ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.3 , angle=180]{tallmerge.eps } } } \endxy } } \
= \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3 , angle=180]{splitcup.eps } } \endxy } } \label{twistsplit}\ ] ] @xmath335{capdot.eps } } \endxy } } \
= \ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.4 , angle=180]{dot.eps } } } \endxy } } \ = \ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.3]{capdot.eps } } } \endxy } } \label{twistdot1}\ ] ] @xmath336{capdot.eps } } } \endxy } } \ = \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{dot.eps } } \endxy } } \
= \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3 , angle=180]{capdot.eps } } \endxy } } \label{twistdot2}\ ] ] relations ( [ assoc ] ) and ( [ coassoc ] ) are associativity of the merge and coassociativity of the split .
the next two relations say that dot represents unit and counit morphisms relative to merge and split .
relation ( [ biadjoint ] ) says that the object @xmath33 is self - adjoint , with cup and cap being biadjointness morphisms .
biadjunctions in all six categories in the commutative diagram ( [ sixcategories ] ) satisfy cyclicity conditions . for more on diagrammatics of biadjointness and the cyclicity property
we refer the reader to @xcite . in this language ,
notice that ( merge , split ) and ( startdot , enddot ) are mates under this adjunction , and are cyclic . for simplicity of discussion , and ease of understanding for readers not versed in mates under adjunction , we will use the easily visualized notion of a _
twist_. given a morphism , one can twist it by taking a line which goes to the upper boundary and adding a cap , letting the line go to the other boundary instead .
an example is given below .
@xmath337{biadj1.eps } } \endxy } } \ \ \ $ ] twists to @xmath338{biadj2.eps } } \endxy}}$ ] one can also twist a downward line back up , or twist lines on the left as well .
two morphisms are twists of each other if they are related by a series of these simple twists , using cups and caps on the right and left side . for instance , relations ( [ twistmerge ] ) and ( [ twistsplit ] ) state that the merge is a simple twist of the split , for a variety of twists .
if one applies the same twist to every term in a relation , one gets a twist of that relation .
because of biadjointness ( [ biadjoint ] ) , twisting a line down and then back up will do nothing to the morphism .
once biadjointness is shown , all twists of a relation are equivalent , because twisting in the reverse direction we get the original relation back .
when a morphism has a total of @xmath339 inputs and outputs , twisting a single strand will often be referred to as rotation by @xmath340 degrees .
the above relations imply that twisting any of the above generators by 360 degrees will do nothing .
this is the notion of _ cyclicity _ , and it is precisely cyclicity in the presence of biadjoints which allows an isotopy class of a diagram to unambiguously represent a morphism .
moreover , merge and split are 60 rotations of each other , and each is invariant under 120 degree rotation , so we may represent them isotopy - unambiguously with pictures that satisfy the same properties .
a similar statement holds for startdot and enddot , which is why we will refer to these morphisms as dots and trivalent vertices from now on .
we also may allow `` isotopies '' that retract dots back into lines , as in the definition of the cap or the cup .
this particular `` isotopy '' always makes sense for diagrams representing frobenius objects .
henceforth , we do nt need to rigidly draw our diagrams with vertical lines only , but can draw lines freely , with the obvious convention that lines having boundary on the side of a diagram ( instead of the top or bottom ) are meant to be twisted upwards or downwards , and horizontal lines are meant to be turned into caps or cups in order that the picture represent an actual morphism in @xmath21 . in the diagrams above , we already took the liberty of bending vertical lines .
associativity and coassociativity are twists of each other .
this relation is written in a rotation - invariant form below , and shall be crucial in the sequel .
we refer to all such relations , which permit one to `` slide '' a line over a trivalent vertex , as _
associativity_. @xmath341{xdiagram.eps } } \endxy } } \ \
= \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.5]{xdiagram.eps } } \endxy } } \label{eqn - associativity}\ ] ] one could draw the above diagram unambiguously ( i.e. isotopy invariantly ) as a 4-valent vertex ( which is to say , the map is invariant under 90 degree twisting ) .
we refer to either picture above as an `` h '' .
similarly , by repeated use of associativity , there is only one way for @xmath18 strands to meet with @xmath342 trivalent vertices , and it is invariant under the appropriate rotation , so we could replace it with an @xmath18-valent vertex .
however , we will _ not _ make use of this simplification of drawings , for reasons that will become obvious in the sequel , and will use the term `` 4-valent vertex '' for other purposes .
note that relations ( [ assoc ] ) , ( [ biadjoint ] ) , ( [ twistmerge ] ) , and ( [ twistdot1 ] ) are sufficient to imply the other frobenius relations , because of the remarks about twisting made above .
here is the proof of half of ( [ twistsplit ] ) using ( [ twistmerge ] ) , as an illustrative example .
@xmath336{tallmerge.eps } } } \endxy } } \ = \ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.3 , angle=180]{twistcalc2.eps } } } \endxy } } \ \
= \ \ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.3 , angle=180]{splitcup.eps } } } \endxy}}\ ] ] now for the final set of one - color relations .
@xmath343{dotspacedot.eps } } \endxy } } \ = \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxi.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.75]{line.eps } } \endxy } } \ - \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.75]{line.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxp.eps } } \endxy } } \ = \ -\
{ \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxp.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.75]{line.eps } } \endxy } } \ + \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.75]{line.eps } } \endxy}}\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxi.eps } } \endxy}}\label{dotspacedot}\ ] ] @xmath343{doubledot.eps } } \endxy } } \
= \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxi.eps } } \endxy } } \ - \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{multxp.eps } } \endxy } } \label{doubledot}\ ] ] @xmath344{needle2.eps } } \endxy } } \ \ = \ \ 0 \ \ \ \ \ \ \ \ \ \ \ \ \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.35]{needle2witheye.eps } } \endxy } } \ \ = \ \ \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.65]{dot2.eps } } \endxy } } \label{needle}\ ] ] the second equality in ( [ dotspacedot ] ) is just the relation ( [ eq - slide1 ] ) .
relations ( [ needle ] ) together with the earlier ones imply @xmath344{needle.eps } } \endxy } } \ \ = \ 0 \ \ \ \ \ \ \ \ \ \ \ \ \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.35]{needlewitheye.eps } } \endxy } } \ \ \ = \ \ \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.29]{tallline.eps } } \endxy } } \label{needlewitheye}\ ] ] we refer to the relations ( [ dotspacedot])([needle ] ) as _ dot and needle relations_. they allow us to simplify diagrams with closed loops or paired dots . for example , @xmath345{example1.eps } } \endxy } } & \quad = \quad { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{example2.eps } } \endxy } } \quad = \quad { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{example3.eps } } \endxy}}\\ & \quad = \quad { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{example4.eps } } \endxy } } \quad = \quad { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{example5.eps } } \endxy } } \end{aligned}\ ] ] more generally @xmath344{polygoni3.eps } } \endxy } } \
= \ \ \ 0\ ] ] @xmath344{polygoni.eps } } \endxy } } \
= \ \ \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.35]{polygoni2.eps } } \endxy}}\ ] ] we see that a region disjoint from the boundary with a polynomial @xmath331 written in it simplifies as follows .
@xmath346{hole.eps } } \endxy } } \ \ \ = \ \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{hole2.eps } } \endxy}}\ ] ] therefore , any one - color diagram reduces to a planar forest ( planar graph without cycles ) , with polynomials written in regions .
there is another relation which is equivalent ( given the others ) to the first equality in ( [ dotspacedot ] ) .
@xmath347{twolines.eps } } \endxy } } & \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{twolinescalc2.eps } } \endxy } } \ \ = \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{twolinescalc3.eps } } \endxy } } \ - \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{twolinescalc4.eps } } \endxy } } \label{twolines}\\ \ \nonumber \\ & = \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{twolinesfinal1.eps } } \endxy } } \ -\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{twolinesfinal2.eps } } \endxy } } \nonumber \end{aligned}\ ] ] this relation quickly leads to the decomposition @xmath348 , see section [ subsec - functor ] . for a single color and two variables @xmath349 , the category above , modulo the relation @xmath350 , is equivalent to the category considered by libedinsky @xcite in the case of a single label @xmath196 .
morphisms given by dot , merge , split , and cap correspond to morphisms @xmath351 , @xmath352 , @xmath353 , @xmath354 and @xmath355 in ( * ? ? ?
* section 2.4 ) .
planar graphical notation , of paramount importance to us , is implicit in @xcite . from here on , we diverge from libedinsky s work , by generalizing to the case of the weyl group @xmath20 , while libedinsky @xcite investigates the right - angled case .
we now add some generators which mix adjacent colors , which we call _
6-valent vertices_. remember that the thick lines represent @xmath42 , and the thin lines represent @xmath33 .
@xmath356{6valent.eps } } } \endxy } } & 0 & \begin{array}{cc } { \xy ( 0,-5)*+{1 { \otimes}1 { \otimes}1 { \otimes}1}="1 " ; ( 0,5)*+{1 { \otimes}1 { \otimes}1 { \otimes}1}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } & { \xy ( 0,-5)*+{1 { \otimes}{x_i}{\otimes}1 { \otimes}1}="1 " ; ( 0,5)*+{({x_i}+ { x_{i+1 } } ) { \otimes}1 { \otimes}1 { \otimes}1 - 1 { \otimes}1 { \otimes}1 { \otimes}{x_{i+2}}}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } \end{array } \\ \\
{ \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{6valent.eps } } \endxy } } & 0 & \begin{array}{cc } { \xy ( 0,-5)*+{1 { \otimes}1 { \otimes}1 { \otimes}1}="1 " ; ( 0,5)*+{1 { \otimes}1 { \otimes}1 { \otimes}1}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } & { \xy ( 0,-5)*+{1 { \otimes}{x_{i+2}}{\otimes}1 { \otimes}1}="1 " ; ( 0,5)*+{1 { \otimes}1 { \otimes}1 { \otimes}({x_{i+1}}+ { x_{i+2 } } ) - { x_i}{\otimes}1 { \otimes}1 { \otimes}1}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } \end{array } \end{array } $ ] let us look at the first map above .
the generating set @xmath357 was chosen because @xmath173 can be slid freely between the second and third slots .
the image of the second generator can be written more symmetrically as @xmath358 , where the first term is a polynomial symmetric in all the relevant variables , and thus can be slid anywhere . in the other two terms , @xmath359 can not be slid freely under a line labelled @xmath42 . in contrast ,
@xmath360 and @xmath361 are not equal , since @xmath180 can not be slid over a line labelled @xmath42 , but the images of both these elements are easier to remember , and are shown below .
@xmath362{6valent.eps } } } \endxy } } & \begin{array}{cc } { \xy ( 0,-5)*+{1 { \otimes}{x_{i+1}}{\otimes}1 { \otimes}1}="1 " ; ( 0,5)*+ { 1 { \otimes}1 { \otimes}1 { \otimes}{x_{i+2}}}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } & { \xy ( 0,-5)*+{1 { \otimes}1 { \otimes}{x_{i+1}}{\otimes}1}="1 " ; ( 0,5)*+{{x_{i+2}}{\otimes}1 { \otimes}1 { \otimes}1}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } \end{array } \\ \\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{6valent.eps } } \endxy } } & \begin{array}{cc } { \xy ( 0,-5)*+{1 { \otimes}{x_{i+1}}{\otimes}1 { \otimes}1}="1 " ; ( 0,5)*+ { 1 { \otimes}1 { \otimes}1 { \otimes}{x_i}}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } & { \xy ( 0,-5)*+{1 { \otimes}1 { \otimes}{x_{i+1}}{\otimes}1}="1 " ; ( 0,5)*+{{x_i}{\otimes}1 { \otimes}1
{ \otimes}1}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } \end{array}\end{aligned}\ ] ] the way to remember this is that the variable which ca nt be slid is sent to the variable which ca nt be slid , from the middle on one side to the exterior on the other .
deriving this from the above definition is an exercise for the reader .
it will be shown in section [ subsec - functor ] that these are in fact @xmath1-bimodule maps .
we list more relations below . for brevity
we have only listed one color variant of the relations , but _ we also impose the same relations with the colors switched _
! the two color variants in general do not imply each other , and need to be checked separately .
@xmath335{threelines1.eps } } \endxy}}\ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{threelines4.eps } } \endxy } } \ \ - \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{threelines3.eps } } \endxy } } \label{eqn - threelines}\ ] ] @xmath363{ipipiprot1.eps } } \endxy } } \ \ \ = \ \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{6valent.eps } } \endxy } } \ \ \ = \ \ \ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.32]{ipipiprot1.eps } } } \endxy } } \label{eqn - ipipiprot}\ ] ] @xmath335{ipipipdot1.eps } } \endxy } } \ \ = \ \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{ipipipdot2.eps } } \endxy } } \ \ + \ \ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.3]{ipipipdot3.eps } } } \endxy } } \label{eqn - ipipipdot}\ ] ] @xmath335{ipipipass.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3 , angle=90]{ipipipass.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{ipipipass2.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.3]{ipipipass2.eps } } } \endxy } } \label{eqn - ipipipass}\ ] ] @xmath335{ipipipasswdot1.eps } } \endxy } } \ \ \ \ = \ \ \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{ipipipasswdot2.eps } } \endxy } } \label{eqn - ipipipasswdot}\ ] ] it will be shown in section [ subsec - functor ] that the first relation is related to the isomorphism @xmath364 .
the second relation shows that our drawing of the 6-valent vertex is isotopy invariant , and isotopy classes of diagrams built out of our local generators will still unambiguously designate a morphism .
the relation ( [ eqn - ipipipass ] ) contains a number of equalities , and it is clear that the last equality is merely a rotation of the color switch of the first equality . as an exercise , one can show that any pair of equalities from ( [ eqn - ipipipass ] ) will imply both color variants of ( [ eqn - ipipipasswdot ] ) as well as the remainder of the equalities from ( [ eqn - ipipipass ] ) ( hint : add a dot below the first equality in ( [ eqn - ipipipass ] ) ; use ( [ eqn - ipipipasswdot ] ) twice inside the third picture in ( [ eqn - ipipipass ] ) ) .
one can observe that , together with the earlier relations , the relations ( [ eqn - threelines ] ) for both colors , ( [ eqn - ipipiprot ] ) for one color , ( [ eqn - ipipipdot ] ) for both colors , and the first equality in ( [ eqn - ipipipass ] ) for both colors together imply the remainder of the relations .
the important feature to notice is that the 6-valent vertex can be visualized as two trivalent vertices , one of each color , that overlap .
a diagram as above with colors restricted to @xmath33 and @xmath42 can be thought of as a planar graph with external ( boundary ) vertices of valence @xmath23 and internal vertices of valence @xmath23 ( dots ) , @xmath365 and @xmath366 .
there are two types of vertices of valence @xmath23 and @xmath365
one for each color , and one type of valence @xmath366 vertex .
each of the two colors within forms a graph with univalent and trivalent internal vertices only .
we call ( [ eqn - ipipipasswdot ] ) and the first equality of ( [ eqn - ipipipass ] ) `` double overlap associativity '' relative to the thick subgraph ( labelled @xmath42 ) , because when viewed as an operation on the `` thick''-colored graph , they are nothing more than the associativity relation for one color . under the same transformations the `` thin''-colored graph ( labelled @xmath33 ) transforms differently , though . the last equality in ( [ eqn - ipipipass ] ) can be called `` double overlap associativity '' relative to the thin subgraph .
notice that in the above relations internal regions of diagrams are free of boxes .
fix @xmath34 , an index which is not adjacent to @xmath33 . in pictures involving both @xmath33 and @xmath42
, we also assume @xmath34 is not adjacent to @xmath42 .
remember that @xmath34 is represented by a dashed line .
this new generator is called a _
4-valent vertex _ , or a _
crossing_. @xmath367{4valent.eps } } } \endxy } } & 0 & { \xy ( 0,-5)*+{1 { \otimes}1 { \otimes}1}="1 " ; ( 0,5)*+{1 { \otimes}1 { \otimes}1}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } \end{array}\ ] ] note that this definition also covers the same picture with the colors reversed . the colors @xmath33 and @xmath34 can be switched freely since the only requirement was that they were distant from each other .
the fifth relation below holds when you switch @xmath33 and @xmath42 , which does need to be checked separately ( however , one color variant will follow quickly from the other by twisting and applying the first relation below ) . in the final relation ,
the new color represents an index @xmath3 which is far away from both @xmath33 and @xmath34 ; this is the reidemeister iii move .
@xmath346{ijijrot1.eps } } \endxy}}\ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{4valent2.eps } } \endxy}}\ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{ijijrot2.eps } } \endxy } } \label{eqn - ijijrot}\ ] ] @xmath343{farlines.eps } } \endxy}}\ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.5]{r2move.eps } } \endxy } } \label{eqn - r2move}\ ] ] @xmath368{ijijdot1.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.6]{ijijdot2.eps } } \endxy } } \label{eqn - ijijdot}\ ] ] @xmath346{pullfarthrutrivalent1.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{pullfarthrutrivalent2.eps } } \endxy } } \label{eqn - pullfarthrutrivalent}\ ] ] @xmath346{pullfarthru6valent1.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{pullfarthru6valent2.eps } } \endxy } } \label{eqn - pullfarthru6valent}\ ] ] @xmath346{r3move.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{r3move2.eps } } \endxy } } \label{eqn - r3move}\ ] ] all of the above relations are required ( although only one color variant of ( [ eqn - pullfarthru6valent ] ) is needed to show the other ) .
other relations similar to ( [ eqn - pullfarthrutrivalent ] ) and ( [ eqn - pullfarthru6valent ] ) , such as when @xmath34 only goes through one or two lines of the 6-valent vertex , are easy to derive using those relations and ( [ eqn - r2move ] )
. we will refer to ( [ eqn - r2move ] ) and ( [ eqn - r3move ] ) as the r2 and r3 moves respectively , because of the obvious analogy to knot theory .
the r2 relation is essentially the isomorphism @xmath369 , see section [ subsec - functor ] .
the relations ( [ eqn - r2move])-([eqn - r3move ] ) imply that a @xmath34-colored strand can just be pulled underneath any morphism only using colors distant from @xmath34 , whether it be a line , a dot , a trivalent vertex , or a 6-valent vertex .
once again , the 4-valent vertices are drawn so that morphisms are isotopy invariant .
we have now listed all the generators of our subcategory : trivalent , 4-valent , and 6-valent vertices , and dots .
there is one further relation , coming from the fact that @xmath42 and @xmath51 may not interact , but they do jointly interact with @xmath33 . the final relation will be called `` triple overlap associativity '' : @xmath346{threecolorass1.eps } } \endxy } } = { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{threecolorass2.eps } } \endxy } } \label{eqn - threecolorass}\ ] ] in the above diagram , dotted lines carry label @xmath51 , thick lines @xmath42 and thin solid lines @xmath33 . rotating this relation by 90 degrees , we get the same relation except with @xmath42 and @xmath51 switched , so that only one variant is needed to imply both .
the `` thick''-colored graph undergoes the associativity transformation .
the same is true ( symmetrically ) with the `` dotted''-colored graph .
we are effectively performing double overlap associativity on colors @xmath33 and @xmath42 , except with our third color @xmath51 interfering in the only place it can .
this concludes the definition of the category @xmath21 and the functor @xmath315 . because of the relations in @xmath21 , the morphism associated to a diagram is an isotopy invariant of the diagram , and the generators all satisfy some cyclicity relations .
moreover , the planar diagram generators can all be viewed as vertices in a graph with colored edges . henceforth , whenever we use the term _ graph _
, we are only considering a very special class of graphs .
graphs will all be planar graphs with boundary ( that is , edges which go to the boundary of the plane ) , and each edge of the graph will be labelled with an integer from @xmath23 to @xmath18 , referred to as its `` color . ''
we only allow univalent vertices ( dots ) , trivalent vertices where each edge has the same color , 4-valent vertices where the colors entering alternate between @xmath33 and @xmath34 for @xmath370 , and 6-valent vertices where the colors entering alternate between @xmath33 and @xmath42 .
finally , we allow edges which are not attached to the boundary or any vertex , which must necessarily form a circle .
graphs will be often represented by the letter @xmath371 .
we may as well consider a graph to be embedded in the strip @xmath322 $ ] , just like our diagrams above , instead of the plane . by an _
isotopy _ of graphs , we mean an isotopy of the graph as embedded in the strip , where we also allow isotopies to contract a dot into a line , or turn a line into a trivalent vertex attached to a dot , as in the definition of the cup and cap .
these procedures are called _ dot contraction _ and _ reverse dot contraction_. a graph in the strip splits the strip into regions .
the region on the far left will be called the _ lefthand region_. the _ righthand region _ is defined similarly .
the empty graph has a single region , which is both the lefthand and righthand region .
we may think of a graph as a representative ( of an isotopy class of graphs modulo relations ) of a morphism in @xmath21 , with all boxes ignored ( i.e. ignoring the data of polynomials in each region ) .
in particular , we could have defined the category @xmath21 to have morphisms which are isotopy classes of graphs with polynomials in each region , modulo the local relations ( we no longer need the relations which merely state isotopy invariance ) .
this is an equivalent definition .
we will think of morphisms in @xmath21 as being graphs with polynomials henceforth .
graphical notation is nice because cyclicity of our generating morphisms becomes inherent , and because planar graph theory can be used to aid some arguments . with this graphical terminology ,
hom spaces in @xmath21 are enriched in @xmath56 , where the action of @xmath1 on the left ( resp .
right ) consists of placing a polynomial in the lefthand ( resp .
righthand ) region , and the degree of a morphism is -1 for each trivalent vertex , 1 for each dot , 2 for each box , and 0 for each 6-valent vertex ( this is preserved by dot contraction ) .
we now proceed to investigate the category @xmath21 with the ultimate goal of proving the following theorem .
[ mainthm ] the functor @xmath315 from @xmath21 to @xmath198 is an equivalence of @xmath30-linear monoidal categories with hom spaces enriched in @xmath56 .
this theorem holds over any commutative ring @xmath30 . assuming that @xmath315 is in fact a functor , which will be verified in section [ sec - proofs ] , we know as per the discussion in section [ subsec - moral ] that @xmath315 is full , and need only verify that the hom spaces in @xmath21 are free left ( or right ) graded @xmath1-modules of the appropriate rank .
this chapter will focus on the consequences of our relations towards classifying all morphisms .
first , some terminology : given a set @xmath372 of graphs and a morphism @xmath373 in @xmath21 , we say that @xmath372 _ underlies _
@xmath374 if @xmath373 can be written as a linear combination of morphisms , each of which is given by a graph @xmath375 with polynomials in regions .
we say that a graph @xmath371 _ reduces _ to @xmath372 if @xmath372 underlies every morphism obtained from @xmath371 by placing arbitrary polynomials in each region .
clearly reduction is transitive , in that if @xmath371 reduces to @xmath372 , and every graph in @xmath372 reduces to @xmath376 , then @xmath371 reduces to @xmath376 . for instance , the relation ( [ twolines ] ) implies that if @xmath371 is a graph with two adjacent edges of the same color , we may reduce it to the same graph with those two edges replaced by an `` h '' . let @xmath377 be a subset of @xmath378 .
the @xmath377-_graph _ of a graph will be the subgraph consisting of all edges colored @xmath33 for @xmath379 .
in particular , a vertex which is 6-valent with colors @xmath379 and @xmath380 becomes a trivalent vertex for @xmath33 , and a vertex which is 4-valent with colors @xmath379 and @xmath381 is ignored , leaving a single @xmath33-colored edge in its wake .
the @xmath377-graph is itself a graph by our above definition .
more commonly , we will just consider the @xmath33-graph for a single color ( i.e. @xmath382 ) .
our general method will be to derive _ reduction moves _ , which replace a graph with a set of graphs it reduces to , in order to determine a small set of graphs that every morphism reduces to .
these reduction moves will often be determined by their action on the @xmath33-graph , since if we can simplify the @xmath33-graph for each color @xmath33 , one at a time , then we can simplify the graph .
the notion of reduction essentially ignores polynomials , so once we have found the graphs which underlie every morphism , we will investigate what happens when polynomials are placed in various regions .
we say a soergel graph @xmath371 is _ simpler _ than @xmath383 if one of the following conditions holds : 1 .
@xmath371 has fewer connected components . 2 .
@xmath371 has fewer 6-valent vertices , but the same number of connected components .
@xmath371 has fewer trivalent vertices , but the same number of 6-valent vertices and connected components .
this is a partial order , clearly satisfying the descending chain condition , that will serve often as a base of induction .
we call a _ boundary dot _ any connected component of a graph which consists entirely of an edge starting at the boundary and ending in a dot .
we call a _ double dot _ any connected component of a graph which consists entirely of a edge with a dot on both ends . in this section , we assume all graphs consist of a single color .
first we discuss certain kinds of reduction moves on one color graphs , which always reduce a graph @xmath371 to a single other graph @xmath384 .
they take a subdiagram looking like start , and replace that subdiagram with finish .
we allow dot contraction to work in both directions .
@xmath385{xdiagram.eps } } \endxy } } & { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{xdiagram.eps } } \endxy}}\\ \\ \mathrm{needle } & { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{needle2.eps } } \endxy } } & { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.4 , angle=180]{dot.eps } } } \endxy}}\\ \\ \mathrm{double dot removal } & { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{doubledot.eps } } \endxy } } & \quad\\ \\ \mathrm{dot contraction } & { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{dotsplit.eps } } \endxy } } & { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{cap.eps } } \endxy}}\\ \\ \mathrm{connecting } & { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{twolines.eps } } \endxy } } & { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{xdiagram.eps } } \endxy } } \end{array } $ ] remember , these are moves on graphs , not graphs with polynomials .
dot contraction is already included in the definition of isotopy , but it shows up here for a reason which shall be obvious soon . note that the needle move , by adding a dot on the bottom , yields a reduction from the circle to a double dot .
the only moves which change the connectivity of a graph are double dot removal , which deletes a connected component , and the connecting move , which has the potential to link two components into one .
an example of a reduction using these moves is shown below .
@xmath347{reduction1.eps } } \endxy } } & \rightarrow { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{reduction2.eps } } \endxy } } \rightarrow { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{reduction3.eps } } \endxy}}\\ & \rightarrow { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{reduction4.eps } } \endxy } } \rightarrow { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{reduction5.eps } } \endxy } } \end{aligned}\ ] ] in fact , an inductive argument shows the following reduction , which should be familiar from the previous chapter , and only uses the associativity and needle moves .
@xmath386{polygoni3.eps } } \endxy } } \rightarrow { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.35]{polygoni2.eps } } \endxy } } \end{aligned}\ ] ] this reduction immediately implies that any one - color graph reduces to a forest ( a graph with no cycles or circles ) .
the number of connected components is unchanged ! a forest is composed out of a disjoint union of connected trees .
each dot present must be connected either to the boundary , another dot , or a trivalent vertex .
hence it is immediately obvious , using dot contraction and double dot removal , that we may assume each tree either has no dots , or is a boundary dot .
in particular , when we make this assumption , we have removed each tree which had no boundary , and not changed the connectivity of trees touching the boundary .
just to be very precise , we use the following redundant definition . a _ simple tree _ @xmath377 with @xmath339 boundary lines
is the following ( connected ) graph with boundary : 1 . if @xmath387 then @xmath377 is a trivalent tree with @xmath388 vertices connecting all the boundary lines . 2 . if @xmath389 then @xmath377 is a single boundary dot .
3 . if @xmath390 then @xmath377 is the empty graph .
a _ simple forest _ is a forest , each of whose components is a simple tree .
note that any two simple trees with @xmath339 boundary lines are equivalent under the associativity move , so there is effectively one such tree .
it is also easy to observe that when one applies the connecting move to edges from two disjoint simple trees , one gets a connected graph which is in fact a simple tree .
so we have proven the following useful proposition .
[ prop - onecolor ] using the associativity , needle , dot removal and dot contraction moves , one can reduce any one - color graph to a simple forest .
all connected components of the original graph which did not have a boundary are removed , and after this is done , the connectedness is unchanged . using the connecting move in addition , we can reduce the graph to a simple tree .
in particular , a connected graph always reduces to a simple tree ( which may be empty ) , without needing the use of the connecting move .
there are two useful sets of one - color graphs with @xmath339 boundary lines , to which all others reduce .
the first is just the simple tree , and the latter is the collection of all simple forests .
the former is useful because there is a single graph , so we have fewer cases to deal with . the latter is useful because it does nt require the connecting move .
more importantly , these sets behave differently when we introduce polynomials into the equation .
the following result will not be used in this paper , and is an immediate consequence of our calculation of hom spaces , but is also an easy exercise .
there are two useful spanning sets for the space of graphs with polynomials with @xmath339 boundary lines .
the first consists of the simple tree , with an arbitrary polynomial in the lefthand region ( or rather , a basis for polynomials ) , and with either @xmath23 or @xmath173 in each other region .
this spanning set is actually a basis .
the second consists of the set of all forests , with an arbitrary polynomial in the lefthand region , and _ no other polynomials _ !
effectively , we may use relation ( [ dotspacedot ] ) to replace @xmath173 on one side of a line with @xmath180 on the other side , at the cost of replacing the line with two dots in a third diagram .
this lets us get from the first spanning set to the second .
relation ( [ twolines ] ) lets us go the other way . to show that the latter is not a basis , consider the three different ways to take a trivalent vertex and turn one edge into two dots : there is a relation between these and the trivalent vertex with a polynomial in the lefthand region .
now we demonstrate the procedure which allows us to simplify multicolor graphs .
let @xmath371 be a graph , and fix a neighborhood in @xmath371 whose @xmath33-colored graph is an `` h '' , that is , one of the two pictures below @xmath391{xdiagram.eps } } \endxy}}$ ] or @xmath392{xdiagram.eps } } \endxy}}$ ] let @xmath393 be the set of graphs @xmath394 which are identical to @xmath371 outside the chosen neighborhood , and whose @xmath33-colored graph in that neighborhood is an `` h '' with the other picture . in other words , @xmath393 is the set of all graphs where the associativity move has been locally applied to a specific `` h '' .
let @xmath372 be the set of graphs simpler than @xmath371 , using the definition of simpler found at the beginning of this section .
if @xmath395 is a set of relations or reduction moves , we say that @xmath395 allows the @xmath33-_colored associativity move _ to be applied to our chosen `` h '' , if we may reduce @xmath371 to @xmath396 using @xmath395 .
we say that @xmath395 allows the _ strict _ @xmath33-_colored associativity move _ if we may reduce @xmath371 to @xmath393 .
we say that @xmath395 allows one to apply the @xmath33-_colored associativity move _ to @xmath371 if it can be done for every such neighborhood , i.e. for every instance of an @xmath33-colored `` h '' in the graph .
similarly for the strict version .
similar definitions can be made for the strict and non - strict @xmath33-_colored double dot removal _
, @xmath33-_colored dot contraction _
, @xmath33-_colored needle move _ , and @xmath33-_colored connecting move_. one chooses a neighborhood whose @xmath33-colored graph looks like the first column in the table of section [ subsec - onecolorred ] , and asks if one can reduce the graph to pictures which locally look like the second column , possibly modulo simpler graphs .
all of the above are collectively called the @xmath33-_colored basic moves_. [ prop - imoves ] the relations defining @xmath21 allow one to apply the @xmath33-colored basic moves to any graph @xmath371 , so long as @xmath371 does not contain either @xmath51 or @xmath42 .
the proof of this proposition is found in section [ subsec - graphproof ] .
one may have wondered earlier why dot contraction was viewed as a reduction move , since it was built into the notion of isotopy .
however , one can see that the @xmath33-colored dot contraction move does not merely follow from isotopy , since the @xmath33-colored trivalent vertex could actually be a 6-valent vertex in the full graph , and distant lines may cross over the picture .
the power of the proposition can be seen immediately : [ cor - empty ] any graph without boundary @xmath371 can be reduced to the empty graph .
we induct on the simplicity of the graph and on the total number of colors present in the graph . by proposition [ prop - imoves ]
we may apply the basic colored moves to the lowest color @xmath33
. we may apply proposition [ prop - onecolor ] to replace every connected component of the @xmath33-graph with the empty set .
hence , applying our reductions to @xmath371 , we get a set of graphs which are either simpler than @xmath371 , or do not include the color @xmath33 . by the inductive hypothesis , each of these graphs reduces to the empty graph .
note that we did not use the @xmath33-colored connecting move in this argument , since we dealt with each connected component separately . in general , it is unneccessary for any of the proofs in this paper .
one can apply a similar procedure to a graph with boundary .
namely , if @xmath33 is the lowest ( resp .
highest ) color , we can reduce the @xmath33-graph to a simple forest ( or a simple tree ) .
then , within each region delimited by the @xmath33-graph , we can reduce the @xmath42-graph ( resp .
@xmath51-graph ) to a simple forest ( the number of boundary lines in each region will be determined by how many of the trivalent @xmath33-graph vertices actually come from 6-valent vertices ) .
one can iterate the process .
[ cor - oneofeach ] any graph whose boundary has at most one line of each color can be reduced to a disjoint union of boundary dots .
we know we can reduce the @xmath33-graph , for @xmath33 the lowest color present in the graph , to either the empty set or a boundary dot , depending on whether or not @xmath33 appears in the boundary . the dot need not be a boundary dot in the entire graph @xmath371 , but it can encounter only 4-valent vertices on the way to the boundary . since
a dot can be slid under a 4-valent vertex , we may turn the dot into a boundary dot .
now , the remainder of the graph does not intersect the @xmath33-graph at all , so it can be reduced as though @xmath33 were not present at all .
induction for the win .
now , adding back polynomials into the equation , we have a classification of all morphisms , at least in the easy case .
[ cor - grrnk ] the space @xmath397 , where @xmath28 is a length @xmath24 increasing sequence , is a free left ( or right ) @xmath1-module of rank 1 , generated by a homogeneous morphism of degree @xmath24 .
by corollary [ cor - oneofeach ] , the disjoint union of boundary dots underlies every morphism .
this graph has a single region , so the only way to construct a morphism from it is to place an arbitrary polynomial in that region . the easiest proof that there are no further relations , i.e. that all such morphisms are generated freely over @xmath1 by the morphism with the polynomial 1 in that region , is that this is easily observed to be true after application of the functor @xmath315 , since adding a polynomial below the dots corresponds to left multiplication by @xmath1 on @xmath162 , and @xmath162 is a free left @xmath1-module .
this generating morphism , which is nothing but @xmath24 boundary dots , has degree @xmath24 .
moreover , as we have seen , @xmath315 sends the boundary dot to the map @xmath244 from @xmath1 to @xmath37 , the free generator of @xmath398 , so that @xmath315 yields an isomorphism @xmath399 .
[ cor - biadjoint ] the object @xmath33 in @xmath21 is self - biadjoint , so that there are natural isomorphisms @xmath400 and @xmath401 .
these compose when @xmath402 to yield an isomorphism @xmath403 .
the first isomorphism and its inverse are shown below .
@xmath346{biadj1.eps } } \endxy } } \ \ \ \mapsto \ \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{biadj2.eps } } \endxy } } \quad \quad \quad \quad { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{biadj3.eps } } \endxy}}\ \ \ \mapsto \ \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{biadj4.eps } } \endxy}}\ ] ] that these maps compose to be the identity is exactly the relation ( [ biadjoint ] ) .
the second isomorphism is effectively the mirror of this one .
here is a drawing of the third isomorphism .
@xmath346{biadj5.eps } } \endxy } } \mapsto { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{biadj6.eps } } \endxy}}\ ] ] at this point , one could construct a @xmath61-semilinear product on the free algebra generated by @xmath120 , @xmath404 , via @xmath405 , and it would satisfy the three properties specified in section [ subsec - hecke ] . if we show that the hecke algebra relations are in the kernel of this semilinear product then it will descend to the hecke algebra , and by unicity will be exactly the semilinear product defined earlier .
this will then tell us that the hom spaces in @xmath21 are exactly what we wish them to be , and @xmath315 is faithful and full .
in fact , we will even get for free that @xmath406 is a free left @xmath1-module ! essentially , what we want to do is look in @xmath316 , where we have direct sums and grading shifts , and prove the isomorphisms ( [ eqn - auxbimod])-([eqn - auxbimod2 ] ) .
alternatively , we could look in the karoubi envelope @xmath19 , find idempotents corresponding to the auxilliary modules in ( [ eqn - auxbimod ] ) and friends , and show those isomorphisms .
finally , we could work entirely within @xmath21 and merely show the isomorphisms ( [ eqn - ipibimod ] ) after applying the hom functor .
for instance , showing that @xmath407 will be sufficient .
all these tactics are roughly the same .
the relation ( [ twolines ] ) precisely descends to @xmath408 .
we decompose the identity of @xmath409 into the sum of two idempotents , and obtain orthogonal projections from @xmath409 to @xmath33 of degrees 1 and -1 respectively .
@xmath346{twolines.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{twolinesfinal1v2.eps } } \endxy } } \ -\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{twolinesfinal2v2.eps } } \endxy}}\ ] ] stated very explicitly , we have two projection maps @xmath410 , and two inclusion maps @xmath411 , as indicated in the diagram above , which is just relation ( [ twolines ] ) divided in half . include the minus sign on the right picture into the map @xmath412
. then one can quickly check @xmath413 , @xmath414 , @xmath415 , @xmath416 , and @xmath417 .
so these must be projections to and inclusions of summands , and they are maps of the correct degree .
now we look at @xmath33 and @xmath34 which are not adjacent , which is the easy case .
we have the isomorphism @xmath418 .
it is clear that we may construct isomorphisms @xmath419 merely by precomposing with the appropriate 4-valent vertex , and ( [ eqn - r2move ] ) shows that these maps are inverses .
we may also extend our diagrammatic calculus to include additional modules from @xmath55 .
if we allow a new color of line , call it @xmath209 , to represent the bimodule @xmath420 , then we may define the following maps : @xmath421{ijdown.eps } } } \endxy } } & 0 & { \xy ( 0,-5)*+{1 { \otimes}1 { \otimes}1}="1 " ; ( 0,5)*+{1 { \otimes}1}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } \\ \\
{ \vcenter{\xy ( 0,0)*{\includegraphics[scale=.5]{ijdown.eps } } \endxy } } & 0 & { \xy ( 0,-5)*+{1 { \otimes}1}="1 " ; ( 0,5)*+{1 { \otimes}1 { \otimes}1}="2 " ; { \ar@{|- > } " 1";"2"};\endxy}\end{aligned}\ ] ] @xmath343{4valent.eps } } \endxy } } { \stackrel{\mbox{\scriptsize{def}}}{=}}{\vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{4valentdef.eps } } \endxy}}\ ] ] it is clear that composing these morphisms to get an endomorphism of @xmath422 will yield the identity map , and composing them to get an endomorphism of @xmath209 will also yield the identity map .
thus we get isomorphisms @xmath423 in @xmath19 .
the bimodule that @xmath209 represents is isomorphic to @xmath424 .
we now combine these techniques to deal with adjacent colors .
we have isomorphisms @xmath425 and @xmath426 .
if we allow the new color of line , again called @xmath209 , to represent the bimodule @xmath427 , then we may define the following maps : @xmath428{ipidown.eps } } } \endxy } } & 0 & \begin{array}{cc } { \xy ( 0,-5)*+{1 { \otimes}1 { \otimes}1 { \otimes}1}="1 " ; ( 0,5)*+{1 { \otimes}1}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } & { \xy ( 0,-5)*+{1 { \otimes}{x_i}{\otimes}1 { \otimes}1}="1 " ; ( 0,5)*+{({x_i}+ { x_{i+1 } } ) { \otimes}1 - 1 { \otimes}{x_{i+2}}}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } \end{array } \\ \\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{ipidown.eps } } \endxy } } & 0 & { \xy ( 0,-5)*+{1 { \otimes}1}="1 " ; ( 0,5)*+{1 { \otimes}1 { \otimes}1 { \otimes}1}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } \\ \\ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.4 , angle=180]{pipdown.eps } } } \endxy } } & 0 & \begin{array}{cc } { \xy ( 0,-5)*+{1 { \otimes}1 { \otimes}1 { \otimes}1}="1 " ; ( 0,5)*+{1 { \otimes}1}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } & { \xy ( 0,-5)*+{1 { \otimes}{x_{i+2}}{\otimes}1 { \otimes}1}="1 " ; ( 0,5)*+{1 { \otimes}({x_{i+1}}+ { x_{i+2 } } ) - { x_i}{\otimes}1}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } \end{array } \\ \\
{ \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{pipdown.eps } } \endxy } } & 0 & { \xy ( 0,-5)*+{1 { \otimes}1}="1 " ; ( 0,5)*+{1 { \otimes}1 { \otimes}1 { \otimes}1}="2 " ; { \ar@{|- > } " 1";"2"};\endxy}\end{aligned}\ ] ] @xmath346{6valent.eps } } \endxy } } { \stackrel{\mbox{\scriptsize{def}}}{=}}{\vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{6valentdef.eps } } \endxy}}\ ] ] @xmath429{6valent.eps } } } \endxy } } { \stackrel{\mbox{\scriptsize{def}}}{=}}{\vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.3 , angle=180]{6valentdef.eps } } } \endxy}}\ ] ] we call the first map @xmath430 below . once again , composing the two maps that go through @xmath431 and @xmath209 to get an endomorphism of @xmath209 will yield the identity map of @xmath209 , and the same is true respectively of @xmath432 .
then the equation ( [ eqn - threelines ] ) , which is a decomposition of the identity of @xmath431 , actually follows from this relation in @xmath19 : @xmath335{threelines1.eps } } \endxy}}={\vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{threelines2.eps } } \endxy } } - { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{threelines3v2.eps } } \endxy } } \label{eqn - threelinesbad}\ ] ] there is a more explicit statement to be derived from this relation , completely analogous to the two line case , with projection and inclusion maps @xmath433 ( include the minus sign on the right picture in @xmath412 ) .
similarly we get a decomposition of @xmath432 via the same relation with the colors switched .
the auxilliary module used here is in fact the indecomposable soergel bimodule @xmath134 where @xmath434 .
one can also think of @xmath209 as the formal object in the karoubi envelope that corresponds to the common `` summand '' of @xmath431 and @xmath432 .
the equations above actually represent a morphism in @xmath19 , with the bimodule map being its image under the functor @xmath319 . to state the result without extending the calculus to the karoubi envelope
, one may merely observe that there is a map @xmath435 and vice versa given by precomposing with the appropriate 6-valent vertex ; there is another map @xmath436 given by precomposing with @xmath437 , and a map backwards given by precomposing with @xmath438 . then ( [ eqn - threelines ] ) exactly yields that @xmath439 .
we would like to tie up a loose end and show that these auxilliary maps are actually @xmath1-bimodule maps , which would then imply that the 6-valent vertex is a bimodule map . to show this ,
one should calculate where @xmath430 sends the element @xmath440 , and show that one gets the same image after moving an @xmath170-invariant polynomial between @xmath441 and @xmath442 , or @xmath443 and @xmath444 , and after moving an @xmath445-invariant polynomial between @xmath442 and @xmath443 .
in fact , the algorithm for computing @xmath430 of this element is to slide @xmath442 to the left and @xmath443 to the right , leaving behind either @xmath23 or @xmath180 , and then computing the image .
effectively then , you need only compute this algorithm on @xmath446 , and show that you get the same answer after moving an @xmath445-invariant polynomial between @xmath331 and @xmath179 .
this is a straightforward calculation , although it has so many terms that it shall not be shown here . in conclusion , @xmath315 is an equivalence of categories , and theorem [ mainthm ] is proven .
we have already shown that @xmath447 , without needing to investigate in depth any morphisms except those from @xmath53 to @xmath28 an increasing sequence .
we can deduce some additional facts about general morphisms which are not used elsewhere in this paper .
these facts may come in handy when constructing the analogous category for arbitrary coxeter systems .
proofs of the nontrivial propositions can be found in section [ sec - proofs ] .
the question which naturally arises is whether or not the colors on the boundary determine which colors can appear in the graph . for @xmath448 , ( i.e. for a coxeter subgraph of @xmath449 )
we let @xmath450 be the category defined analogously but where edges can only be labelled by colors in @xmath220 . if colors which do nt appear on the boundary are not needed in the graph after reduction , then the natural inclusion of @xmath450 into @xmath21 is fully faithful , as one would expect .
the case of edge colors is easy . if @xmath33 is the lowest ( resp .
highest ) color in the graph and @xmath33 does not appear in the boundary , then any graph can be reduced to a graph without that color .
this follows immediately from the same proof as corollary [ cor - empty ] .
any graph with only one color @xmath33 on the boundary can be reduced to a graph solely of that color . using the above
, we may eliminate all other colors .
the general case is not as easy to prove graphically , but still holds true .
[ prop - colorelimination ] let @xmath371 be a graph where the color @xmath34 does not appear in the boundary .
then @xmath371 can be reduced to graphs not containing the color @xmath34 at all .
one can prove this proposition merely by counting the dimension of hom spaces .
we can think of any graph with boundary as a morphism in @xmath21 from @xmath53 to @xmath29 .
we have already `` calculated '' the space of such morphisms inductively .
namely , if @xmath29 has no repeated colors , then every morphism is a polynomial with boundary dots , and if @xmath29 has repeated colors , we apply biadjunction or idempotent decompositions to consider the hom space as the sum of the hom spaces of its summands .
since neither biadjunction nor idempotent decompositions adds any new colors to the graph , then we know that any graph must simplify to a morphism constructed purely out of the colors on its boundary !
it would be interesting to have a purely graphical proof of this result , similar to the other proofs which follow .
as described in section [ subsec - sbim ] , one usually constructs soergel bimodules with respect to the @xmath18-dimensional fundamental or geometric representation , instead of the @xmath17-dimensional standard representation .
this amounts to working over the ring @xmath452 / ( x_1+x_2 + \ldots + x_{n+1 } ) $ ] , which is a quotient of @xmath1 by the first elementary symmetric function @xmath451 . since @xmath451 is symmetric , it is in the center of the category @xmath21 ( it slides freely under all lines ) , and one may easily take the quotient category , setting @xmath453 , without changing any of the diagrammatics .
there is a functor from this quotient category to the appropriate category of soergel bimodules in @xmath454 , which is an equivalence of categories , so this diagrammatical quotient category also categorifies the hecke algebra .
one advantage to passing to the quotient @xmath453 is that , after inverting a suitable integer , we may remove the boxes from our list of generators . according to relation ( [ doubledot ] ) , the double dot colored @xmath33 is equal to @xmath455
. thus linear combinations of double dots of colors @xmath456 will give us the @xmath30-span of @xmath457 inside the space of linear polynomials .
this span will not include @xmath173 in @xmath1 , which is why we require at least one box as an additional generator .
however , it is easy to check that , if @xmath17 is invertible in @xmath30 , then @xmath173 is in the @xmath30-span after passing to the quotient @xmath458 . as an example , when @xmath459 , the double dot is equal to @xmath460 , and passing to @xmath461 , we get @xmath462 . thus ,
if @xmath17 is invertible in @xmath30 , one can eliminate boxes from the quotient calculus altogether , replacing them with linear combinations of double dots .
we need now to check that each of our relations holds true for soergel bimodules .
it so happens that there are 28 relations that need to be checked independently , including color variations , and these 28 imply the remainder .
these checks are completely straightforward and require little imagination . with a little imagination ,
however , there are some simple tricks that make checking almost all of these relations entirely trivial . * three polynomial slides ( [ eq - slide1])([eq - slide3 ] ) , * seven frobenius relations : 2 equalities for biadjointness ( [ biadjoint ] ) , 2 equalities twisting a trivalent vertex ( [ twistmerge ] ) , 2 equalities twisting a dot ( [ twistdot1 ] ) , and one version of associativity ( [ assoc ] ) , * three equalities for the dot and needle relations ( equality ( [ doubledot ] ) and first equalities in ( [ dotspacedot ] ) , ( [ needle ] ) ) , * two color variants of the equation ( [ eqn - threelines ] ) for the idempotent decomposition of @xmath431 and @xmath432 , * two equalities for twisting a 6-valent vertex ( either two colors , or one color in both directions ) , see equation ( [ eqn - ipipiprot ] ) , * two color variants for adding a dot to a 6-valent vertex ( equation ( [ eqn - ipipipdot ] ) ) , * two color variants of the 90 degree rotation version of double overlap associativity ( [ eqn - ipipipass ] ) , ( [ eqn - ipipipasswdot ] ) which imply the remainder of the double overlap associativity relations , * six equalities with distant strands : the r2 and r3 moves ( [ eqn - r2move ] ) , ( [ eqn - r3move ] ) , one equality for twisting a 4-valent vertex ( [ eqn - ijijrot ] ) , three equalities for pulling the distant - colored strand across a dot ( [ eqn - ijijdot ] ) , trivalent vertex ( [ eqn - pullfarthrutrivalent ] ) , and 6-valent vertex ( [ eqn - pullfarthru6valent ] ) respectively , * triple overlap associativity ( [ eqn - threecolorass ] ) . for brevity
, we will call any element which is a tensor product @xmath463 a _
1-tensor_. enddot , split , and the 4-valent and 6-valent vertices all send 1-tensors to 1-tensors .
this simple fact is already enough to prove relations ( [ assoc ] ) , ( [ eqn - r2move ] ) , ( [ eqn - ijijdot ] ) , ( [ eqn - pullfarthrutrivalent ] ) , and ( [ eqn - r3move ] ) , since the bottom bimodule in each of these pictures is generated by the 1-tensor .
there are several choices to make when checking the various relations .
once the twisting relations are shown , one is free to prove any twist of the other relations .
also , we may choose which set of generators of the source bimodule to check equality on . finally , whenever a cup or a startdot appears , there are two equivalent ways to write the result , and one might be easier than the other to evaluate .
an arbitrary module @xmath162 of length @xmath24 will have @xmath464 generators as an @xmath1-bimodule , where @xmath339 is the number of different colors that appear .
the reason for this is that between each pair of lines labelled @xmath33 , consecutive if you ignore the other colors , a variable @xmath173 ( or alternatively @xmath180 ) might get stuck , and the number of such pairs is @xmath465 . more explicitly , @xmath162 will be generated by _ any _ @xmath464 linearly independent tensors corresponding to the power set of such pairs , where the tensor corresponding to a subset of pairs has either @xmath173 or @xmath180 somewhere between each pair ( not counting possible variables corresponding to other pairs ) .
it s a mouthful , but an example clarifies it .
the following is an example of a set of generators for @xmath466 : each picture represents a tensor , where by convention a blank area is filled with a 1-tensor .
reading across , the second picture corresponds to the pair of @xmath51 strands , the third to the pair of @xmath42 strands , and the fourth to the pair of @xmath33 strands .
since these three are linearly independent , they take care of all the generators with a linear term . clearly the fourth picture could have also worked for the @xmath42 strands , but if we had chosen it as such , we would have had to choose a new linearly independent vector for the @xmath33 strands
. a clever choice of which generators to use may greatly simplify a calculation by reducing the number of terms in intermediate steps .
there are two main instances when this occurs .
either 6-valent vertex with strands @xmath33 and @xmath42 will send @xmath467 or @xmath468 to a single tensor , while it may send @xmath469 to the sum of two tensors , thus doubling the work we need to do in the remainder of the calculation .
also , @xmath470 entering a 6-valent vertex @xmath431 in the second slot may by moved across the @xmath33-strand to the left for a simple calculation , while @xmath471 leads to a more complicated solution .
in particular , the choice of generators above makes the verification of triple overlap associativity ( [ eqn - threecolorass ] ) rather straightforward .
there is a nice graphical way of doing calculations , which we will demonstrate in perhaps the most difficult case , the highest degree generator .
this graphical method is fairly self - explanatory .
we keep track of the image at every stage in the calculation , which is one term as shown , with all blank spaces being filled by 1-tensors .
an arrow indicates either bringing a symmetric polynomial through a line , or applying a 6-valent vertex to a tensor .
we leave it as an exercise to the reader to verify , using this graphical method of calculation , that both sides of the triple overlap associativity relation , as displayed above , send the other 7 generators to the same elements .
precisely , both sides send the 1-tensor to a 1-tensor , and the other generators to 1-tensors with polynomials in various slots : the 2nd and 4th generators to @xmath359 in the last slot , the third generator to @xmath474 in the last slot , the 5th to @xmath475 in the last slot , the 6th to @xmath476 in the first slot , the 7th to @xmath477 in the first slot , and the 8th as shown above to @xmath476 in the first slot and @xmath474 in the last slot . when @xmath180 enters a merge , it is sent to @xmath300 .
similar pictures can be drawn for the cap , and even simpler pictures can be drawn for the split and enddot , where no arrows are needed since 1-tensors are sent to 1-tensors . if you allow sums of such pictures , one can handle startdot and cup as well . with this method in place , checking ( [ eqn - pullfarthru6valent ] ) and both color variants of double overlap associativity using the following twist is particularly easy .
choose generators so that it is always @xmath180 passed to a 6-valent vertex , and not @xmath173 or @xmath359 .
when there is a cup or a startdot , this creates two terms , each of which must be evaluated separately .
however , the terms can be written as @xmath478 or @xmath479 , so we may choose the one whichever is more convenient .
often the problem of having two terms is very temporary .
for example , consider what happens to the 1-tensor under the map the cup creates two terms , the first with a 1 under the cap and the second with @xmath173 under the cap .
but the first term is annihilated immediately by the cap , and the cap eats the @xmath173 from the other term , to return back a 1-tensor .
so long as a cup or a startdot appears right next to a cap or a merge , one of the two terms is always immediately annihilated .
this is the case for ( [ biadjoint ] ) , ( [ twistmerge ] ) , ( [ twistdot1 ] ) , which all send 1-tensors to 1-tensors as a result . after a quick polynomial slide ,
the same is true for ( [ eqn - ijijrot ] ) , and using the more convenient choice so the @xmath180 ends up underneath the 6-valent vertex , a quick calculation shows it for ( [ eqn - ipipiprot ] ) as well .
we have now given arguments or quick methods for almost all the relations .
the remaining ones are trivially easy to check by hand .
the rest of the one color relations can be checked on a single generator , the 1-tensor : ( [ doubledot ] ) both sides send @xmath23 to @xmath455 ; ( [ dotspacedot ] ) both sides send the 1-tensor to @xmath478 ; the needle relation , where a 1-tensor is killed by a cap .
@xmath343{dotspacedot.eps } } \endxy } } \ \
= \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.5]{multxi.eps } } \endxy } } \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.8]{line.eps } } \endxy } } \ - \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.8]{line.eps } } \endxy}}\ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.5]{multxp.eps } } \endxy}}\ ] ] @xmath343{doubledot.eps } } \endxy } } \ \ = \ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.5]{multxi.eps } } \endxy } } \ -
\ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.5]{multxp.eps } } \endxy}}\ ] ] @xmath332{needle2.eps } } \endxy } } \ \ = \ 0\ ] ] only both color variants of ( [ eqn - threelines ] ) and ( [ eqn - ipipipdot ] ) remain .
these can be checked on the 1-tensor @xmath209 and the generator @xmath480 regardless of whether we are in @xmath481 or @xmath482 .
@xmath483{threelines1.eps } } \endxy}}={\vcenter{\xy ( 0,0)*{\includegraphics[scale=.2]{threelines4.eps } } \endxy } } - { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.2]{threelines3.eps } } \endxy } } & { \xy ( 0,-5)*+{w}="1 " ; ( 0,5)*+{w}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } & { \xy ( 0,-5)*+{x}="1 " ; ( 0,5)*+{x}="2 " ; { \ar@{|- > } " 1";"2"};\endxy}\\ \\ { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.2 , angle=180]{ipipipdot1.eps } } } \endxy } } = { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.2 , angle=180]{ipipipdot2.eps } } } \endxy } } + { \vcenter{\xy ( 0,0)*{\reflectbox{\includegraphics[scale=.2 , angle=180]{ipipipdot3.eps } } } \endxy } } & { \xy ( 0,-5)*+{w}="1 " ; ( 0,5)*+{w}="2 " ; { \ar@{|- > } " 1";"2"};\endxy } & { \xy ( 0,-5)*+{x}="1 " ; ( 0,5)*+{1 { \otimes}1 { \otimes}1 { \otimes}y}="2 " ; { \ar@{|- > } " 1";"2"};\endxy}\\ \end{array}\ ] ] in this section , we provide graphical proofs for a series of propositions which state when the relations of @xmath21 allow us to perform various reduction moves . recall that the _ strict _ @xmath33-colored needle move takes a graph @xmath371 , whose @xmath33-graph in some neighborhood is a needle , and expresses @xmath371 as a linear combination of graphs @xmath394 whose @xmath33-graph has a dot in that neighborhood instead .
the _ ( nonstrict ) _
@xmath33-colored needle move is the same , except that the linear combination may also include graphs @xmath394 which are simpler than @xmath371 , but have arbitrary @xmath33-graphs .
we only require use of the nonstrict versions in this paper , but we prove the strict version when it occurs .
because more than three colors may be required for some proofs , we use extra line styles that designate other arbitrary colors , and are very explicit about what colors they are allowed to be . a thin line will still always represent @xmath33 and a thick line @xmath42 , but we will be lax about other colors .
the only significant part of this lemma is that we can still apply the relations mentioned , even in the presence of arbitrary polynomials in each region .
however , it is an immediate observation that any polynomial inside a region bounded by at least two lines of distant colors may be slid entirely out of the region , via a slide which does not affect the underlying graph . in general
, one must account for the existence of polynomials .
for instance , the relation ( [ eqn - threelines ] ) does not immediately imply that we can apply the graph reduction which sends two 6-valent vertices to a combination of the lines and the `` h '' with dots , since the composition of the two 6-valent vertices might have @xmath173 in an internal region , which can not be slid out .
one can still effectively move a polynomial from one side of a line to the other , using relation ( [ dotspacedot ] ) , but this leaves an extra term where the line is replaced by a pair of dots .
suppose the @xmath33-graph of @xmath371 contains two dots connected by an edge .
then in @xmath371 , the `` edge '' connecting them can only be a series of 4-valent vertices with various @xmath34-strands for @xmath34 distant from @xmath33 .
since dots can be slid across distant - colored strands by ( [ eqn - ijijdot ] ) , we may slide the dots until they are connected directly by an edge , and just use ( [ doubledot ] ) to reduce the graph .
suppose the @xmath33-graph of @xmath371 contains a trivalent vertex connected to a dot .
in @xmath371 , the trivalent vertex is either a trivalent or 6-valent vertex @xmath234 .
again , in @xmath371 the dot may be slid across distant - colored strands until it is connected directly to @xmath234 .
if @xmath234 is trivalent then isotopy allows us to contract the dot , while if @xmath234 is 6-valent , ( [ eqn - ipipipdot ] ) sends @xmath371 to the sum of two simpler graphs ( note : one is the graph after dot contraction , but one is not , so we can not necessarily apply the strict dot contraction move ! ) .
let @xmath371 be a sequence of lines @xmath29 , such that @xmath33 does not appear .
then @xmath371 is equal in @xmath21 to a sum of idempotents , such that each idempotent factors through a sequence of lines @xmath48 where @xmath51 and @xmath42 each appear no more than once , and such that the idempotents do not introduce any new colors not present in @xmath29 .
this proof is akin to the combinatorial argument that the relations defining the hecke algebra or the symmetric algebra are enough to take any complicated word in @xmath486 or @xmath487 to a reduced form .
we use induction on the length of @xmath29 .
if @xmath488 appears in @xmath29 for some @xmath3 , then using ( [ eqn - threelines ] ) , we may factor through @xmath489 instead , modulo morphisms that factor through shorter length sequences .
if any color appears twice consecutively in @xmath29 then we may apply ( [ twolines ] ) to replace the two adjacent lines with an `` h '' , which factors through a sequence of lines of shorter length .
if @xmath490 appears for @xmath34 distant from @xmath3 , then applying the r2 move we can factor through @xmath491 .
none of these procedures added any new colors .
so if we consider the sequence @xmath29 as a word in the symmetric group @xmath492 , then any non - reduced words will reduce to smaller length sequences , and any two reduced words for the same element will factor through each other , modulo smaller length sequences .
we may now use the easily observable fact that any element of the symmetric group can be represented by a reduced word which uses @xmath493 at most once , or @xmath494 at most once ( although not both ! ) .
the obvious corollary is that any element of the symmetric group which has a reduced word presentation that does nt use @xmath170 , can be given a reduced word presentation which uses @xmath495 and @xmath445 at most once ( since all @xmath487 for @xmath496 commute with all @xmath497 for @xmath498 , we can view the @xmath497 as forming a word in @xmath499 , and the @xmath487 as forming a word in @xmath500 ) .
assume @xmath51 does not appear , although the @xmath42 case is analogous .
induct on the number of colors present in the graph : the base case of one color is clear . then we are attempting to apply the @xmath33-colored connecting move to some region of the graph which looks like this : @xmath384 is some graph composed entirely of the colors @xmath501 and @xmath502 .
in fact , in some neighborhood @xmath384 is just a sequence of lines , and by the earlier lemma , we get a sum of graphs where , in a smaller neighborhood , this sequence contains at most a single @xmath42 strand ( and no @xmath51 strands ) . for each term , apply the r2 move to bring the @xmath33-strands past all the other colors so that they are separated by either nothing or a single @xmath42 strand .
we have not yet altered the @xmath33-graph at all
. then equation ( [ twolines ] ) or equation ( [ eqn - threelines ] ) ( respectively ) will allow the strict @xmath33-colored connecting move .
[ cor - between ] suppose we are given a graph @xmath371 and a neighborhood of an @xmath33-colored strand in @xmath371
. then @xmath371 is equivalent in @xmath21 to a linear combination of graphs @xmath394 , which are equal to @xmath371 outside the neighborhood , and has the same @xmath33-graph .
the sequence of lines crossing the @xmath33-strand in @xmath394 contains @xmath52 and @xmath503 each at most once .
consider the neighborhood of an edge colored @xmath33 in a soergel graph @xmath371 .
then in this neighborhood there may be a sequence @xmath29 of lines all distant from @xmath33 which cross over the @xmath33-strand , but in this sequence @xmath33 , @xmath51 , and @xmath42 do not appear : the horizontal line is @xmath33 , the vertical lines are @xmath29 . according to the above lemma , we may look at @xmath29 right above where it hits the @xmath33-strand , and replace it with a sum of idempotents which factor through `` nice '' sequences where @xmath52 and @xmath503 each appear at most once .
consider each term in the sum : @xmath33 is distant from every color appearing in the idempotent , so one can slide the @xmath33-strand past every morphism in the idempotent until it runs through the nice sequence of lines , and the rest of the idempotent is now outside of some smaller neighborhood of the @xmath33-strand in the new graph .
also , all polynomials can be moved outside the neighborhood as well .
@xmath337{linenbhd2.eps } } \endxy } } { \longrightarrow}{\vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{linenbhd3.eps } } \endxy } } { \longrightarrow}{\vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{linenbhd4.eps } } \endxy}}$ ] we apply induction on the number of colors present in the graph to prove these two propositions simultaneously .
both propositions hold when @xmath33 is the only color in the graph , by ( [ eqn - associativity ] ) , ( [ needle ] ) , polynomial slides , and ( [ needlewitheye ] ) , where the latter is needed because an arbitrary polynomial may be within the eye of the needle .
the inductive hypothesis then implies , with the previous propositions , that for any graph with fewer colors , we may apply _ all _ the basic @xmath3-colored moves , when the graph does not contain either @xmath504 or @xmath505 . in this case proposition [ prop - onecolor ]
says that we may reduce the @xmath3-graph to a simple tree or forest . within each region of @xmath371 delineated by the @xmath33-graph
the color @xmath33 is absent , so there are fewer total colors and we may reduce both the @xmath51-graph and the @xmath42-graph inside this region to a simple tree . so suppose the @xmath33-colored graph contains a needle .
the trivalent vertex of the needle is either trivalent or 6-valent in @xmath371 , and the edge of the needle may contain a series of 4-valent vertices .
so a neighborhood of the needle in @xmath371 looks like one of the pictures below .
we may now reduce the @xmath42 and @xmath51 graph of @xmath384 , since @xmath384 does not contain the color @xmath33 . in the first case , the @xmath506 graph has no boundary , so both reduce to the empty graph . in the second case ,
one of the two has a single boundary line , and the other has none , so one graph reduces to the empty graph , and the other to a single dot . within the reduced @xmath384 , this dot
may be slid under other lines until it is a boundary dot , as in the proof of the earlier dot move propositions .
thus we may assume that , after reduction , our neighborhood looks like where now @xmath384 does not contain any of the colors @xmath507 .
note that @xmath384 may have arbitrary polynomials in its various regions .
but then by lemma [ lem - pullthru ] we can pull the line forming the needle through all of @xmath384 , thus completely ignoring @xmath384 from the picture !
there still may be a polynomial in the eye of the needle .
we have effectively reduced to the 2-color case on the right , or the one color case on the left .
we know the one color case works . to check the 2-color case , we use ( [ eqn - ipipipdot ] ) followed by other reduction moves .
@xmath508{proof6.eps } } \endxy}}\ \ { \rightarrow}\ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{proof7.eps } } \endxy } } \ + \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{proof8.eps } } \endxy } } \ \ { \rightarrow}\ \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{proof9.eps } } \endxy } } \ + \ { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{proof10.eps } } \endxy}}$ ] each trivalent vertex of the @xmath33-graph may be either trivalent or 6-valent in @xmath371 .
we are forbidding the case when one is 6-valent with @xmath42 and the other with @xmath51 , so without loss of generality with assume that a 6-valent vertex has @xmath42 as the other color . the neighborhood of this subgraph in @xmath371 looks like one of the following cases .
note that all polynomials may be assumed to be outside the neighborhood .
arbitrary lines distant from @xmath33 may intersect the middle strand .
however , we may use corollary [ cor - between ] to assume that the sequence of lines passing through the middle strand contains at most one instance of @xmath52 and @xmath503 . in the first two cases ,
all such lines may be slid one by one to the right over the trivalent vertex , removing them from the neighborhood . in the third case
, all lines not labelled @xmath52 can be slid to the right or to the left , and since there is at most one line labelled @xmath52 , then at most one line remains .
if there had been multiple lines labelled @xmath52 , then additional lines labelled @xmath509 or higher may have been stuck between them , but we can eliminate this possibility . the case where the 6-valent vertices use @xmath51 instead is obviously analogous . @xmath392{xdiagram.eps } } \endxy } } \quad { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{case2.eps } } \endxy } } \quad { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{proof14.eps } } \endxy } } \quad { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.3]{proof15.eps } } \endxy}}$ ]
the remaining two cases use the same trick : they replace the interior lines on the top and bottom of the graph with the corresponding sum of idempotents , and then resolve each one with double or triple overlap associativity .
the remaining two cases will not allow the strict associativity move , only the non - strict one .
@xmath346{twolines.eps } } \endxy } } = { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{twolinesalt1.eps } } \endxy } } + { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{twolinesalt2.eps } } \endxy } } - { \vcenter{\xy ( 0,0)*{\includegraphics[scale=.4]{twolinesfinal2.eps } } \endxy}}\ ] ] applying this to the thick edges on top of case iii , and symmetrically to the bottom , we get a sum of graphs which either have dots on thick strands connected to 6-valent vertices ( implying that the graphs can be simplified ) or look like the following , with _ no _ polynomial in any interior region : we may apply double overlap associativity to a subgraph of this diagram ( ignoring the top and bottom thick trivalent vertices ) , which will have the final result of applying @xmath33-colored associativity to @xmath371 .
the strict move was not applied because of the extra terms which involved dots attached to 6-valent vertices . to the subgraph of the left picture which ignores the top and bottom 6-valent vertices
, we may apply triple overlap associativity ! ( note that this recent operation has performed associativity on the @xmath33-graph , while actually increasing the number of total 6-valent vertices , making the @xmath42 and @xmath52 graphs more complicated . ) in the right picture , we again slide the dot underneath and combine it with the 6-valent vertex on top to get two diagrams we may apply double overlap associativity to both of these , ( [ eqn - ipipipass ] ) to the first and ( [ eqn - ipipipasswdot ] ) to the second . again , this results in applying the @xmath33-colored associativity move to @xmath371 . because of this , our algorithm towards reducing graphs has always been to reduce the smallest color , then reduce the next color within each region delineated by the previous , and so forth , so that at each stage the color @xmath51 is absent .
we could not start with a non - extremal color , because the above diagram might show up . despite the pessimism of the previous remark towards reducing non - extremal colors , proposition [ prop - colorelimination ]
implies that we can still do a lot .
there should be a purely graphical proof of this proposition , but it is complicated by the failure of @xmath33-colored associativity when both adjacent colors are present .
j. bernstein , i. b. frenkel , and m. khovanov , a categorification of the temperley - lieb algebra and schur quotients of @xmath511 via projective and zuckerman functors , _ selecta math .
_ * 5 * , no.2 ( 1999 ) , 199 - 241 .
i. b. frenkel , m. khovanov , and c. stroppel , a categorification of finite - dimensional irreducible representations of quantum sl(2 ) and their tensor products , _ selecta math .
_ * 12 * ( 2006 ) , 379431 , math.qa/0511467 .
r. rouquier , categorification of @xmath512 and braid groups , in _ trends in representation theory of algebras and related topics ( quertaro , mexico , 2004 ) , contemp . math . _
* 406 * , ams , providence , 2006 , 137167 .
|
the monoidal category of soergel bimodules can be thought of as a categorification of the hecke algebra of a finite weyl group .
we present this category , when the weyl group is the symmetric group , in the language of planar diagrams with local generators and local defining relations .
|
scalar - tensor ( st ) gravitational theories are the most firm candidates for extensions of general relativity ( gr ) .
a great part of their interest comes from the fact that they are induced naturally in the reduction to four dimensions of string and kaluza - klein models@xcite , resulting mostly in the form of a brans - dicke ( bd ) type of st theorybransdicke , often involving also non - minimal coupling to matter , leading to the so called fifth force@xcite .
it is also interesting that st theories are shown to be mathematically equivalent to theories with action depending non - linearly on the ricci scalar , the so called @xmath0 theories@xcite .
finally , st theories are possibly the simplest extension of gr that could accommodate cosmological issues as inflation and universe - expansion acceleration , as well as possible space - time variation of fundamental constants@xcite . on the other hand , observational and experimental
evidence puts strong limits to the observable effects of a possible scalar field .
for example , in the case of a massless scalar the bd parameter @xmath1 is constrained by precise solar - system experiments to be a large number ( @xmath2)bertotti . in this way
, st gravity phenomenology appears to be very similar to that of gr , thus putting strong limits to possible experimental verifications .
there is however a very interesting extension of st gravity put forward by mbelek and lachize - rey@xcite , which could allow electromagnetic ( em ) fields to modify the space - time metric far more strongly than predicted by gr and standard st theories .
the theory was applied in cosmological@xcite and galactic@xcite contexts , and in@xcite it was used to explain the discordancy in the measurements of newton gravitational constant as due to the effect of the earth s magnetic field .
the key new element of that theory is an additional , external scalar field @xmath3 , minimally coupled to gravity . in@xcite it was shown that a general st theory that includes an external field @xmath3 with the mentioned characteristics , and with the magnitude of the coupling derived in@xcite , can explain the unusual forces on asymmetric resonant cavities recently reported@xcite .
we will consider the weak - field limit of a st theory with action given by ( si units are used ) @xmath4 d\omega \notag \\ & & -\frac{\varepsilon _ { 0}c}{4}\int \sqrt{-g}\lambda \left ( \phi \right ) f_{\mu \nu } f^{\mu \nu } d\omega -\frac{1}{c}\int \sqrt{-g}j^{\nu } a_{\nu } d\omega \notag \\ & & + \frac{1}{c}\int \mathcal{l}_{mat}d\omega .
\label{skk}\end{aligned}\ ] ] in ( [ skk ] ) the internal , non - dimensional scalar field is @xmath5 , while the external scalar field is @xmath3 .
these fields have vacuum expectation values ( vev ) @xmath6 and @xmath7 , respectively .
@xmath8 represents newton gravitational constant , @xmath9 is the velocity of light in vacuum , and @xmath10 is the vacuum permittivity .
@xmath11 is the lagrangian density of matter .
the other symbols are also conventional , @xmath12 is the ricci scalar , and @xmath13 the determinant of the metric tensor @xmath14 . the bd parameter @xmath15 is considered a function of @xmath5 , as it usually results in the reduction to four dimensions of multidimensional theories@xcite .
the function @xmath16 in the term of the action of the em field is of the type appearing in bekenstein s theory and other effective theories@xcite .
the em tensor is @xmath17 , given in terms of the em quadri - vector @xmath18 , with sources given by the quadri - current @xmath19 . @xmath20 and @xmath21 are , respectively , the potential and source of the field @xmath3 .
the source @xmath21 contains contributions from the matter , em field and the scalar @xmath5 . the model for @xmath21 proposed in@xcite is @xmath22where @xmath23 is the trace of the energy - momentum tensor of matter , @xmath24 variation of ( [ skk ] ) with respect to @xmath25 results in ( @xmath26 is the usual electromagnetic energy tensor , and @xmath27 the energy tensor associated to the scalar @xmath5)@xmath28 + t_{\mu \nu } ^{\phi } \notag \\ & & + \frac{\phi } { 2}\left ( \nabla _ { \mu } \psi \nabla _ { \nu } \psi -\frac{1}{2}% \nabla ^{\gamma } \psi \nabla _ { \gamma } \psi g_{\mu \nu } \right ) \notag \\ & & + \frac{\phi } { 2}\left ( u+j\psi \right ) g_{\mu \nu } .
\label{glm}\end{aligned}\ ] ] variation with respect to @xmath5 gives @xmath29 , \end{aligned}\ ] ] which can be rewritten , using the contraction of ( [ glm ] ) with @xmath30 to replace @xmath12 , as @xmath31 -\frac{\partial j}{\partial \phi } \psi \phi , \label{phi}\end{aligned}\ ] ] where it was used that @xmath32 .
the non - homogeneous maxwell equations are obtained by varying ( [ skk ] ) with respect to @xmath33 , @xmath34with @xmath35 the vacuum permeability .
finally , the variation with respect to @xmath3 results in @xmath36 having included @xmath8 , it is understood that @xmath5 takes values around its vacuum expectation value ( vev ) @xmath6 .
the scalar @xmath3 is also dimensionless and of vev @xmath7 .
these equations can be approximated in the weak - field limit keeping only the lowest significant order in the perturbations @xmath37 of the metric @xmath38 about the minkowski metric @xmath39 , with signature ( 1,-1,-1,-1 ) , and of the scalar fields about their vev @xmath40 and @xmath41 @xmath42@xmath43@xmath44@xmath45 j^{\mu } -f^{\mu \nu } \partial _ { \nu } \left ( \lambda _ { 0}^{\prime } \phi -\overline{h}/2\right ) \label{max0}\]]@xmath46where @xmath47 , @xmath48 , and @xmath49with @xmath50 . in these equations
only the em sources were included , the effect of the matter terms either being negligible or included in the local gravitational field . using the source @xmath21 given in ( [ source ] ) , with only the em terms we finally obtain the complete set of equations for the em field ( making explicit the electric and magnetic field vectors @xmath51 and @xmath52 , respectively ) [ linearem ] @xmath53where the auxiliary field @xmath54 is defined as @xmath55 , and the electromagnetic sources were redefined as @xmath56 j^{\mu } $ ] .
the constant @xmath57 is @xmath58 which , according to@xcite , has a value of order@xmath59 to study the propagation of electromagnetic waves we consider the case of a vacuum with uniform and static electric and magnetic fields @xmath60 and @xmath61 , so that one can linearize the system ( [ linearem ] ) in the perturbations as [ full ] @xmath62 starting with this system we consider now different simple configurations . for the case without zero order electric field , @xmath63 , and perturbations @xmath64 , @xmath65 , @xmath66 one has from ( full ) @xmath67 for propagation along the magnetic field , @xmath68 , with @xmath69 @xmath70 , one has @xmath71 replacement of the first two equations in the last results in@xmath72 = 0.\ ] ] it is easy to see that the last equation can only be satisfied if @xmath73the usual dispersion relation for em waves in vacuum . for the kind of propagation
considered we thus have the standard plane em wave for @xmath74 and @xmath65 , with only the addition of a longitudinal component of amplitude@xmath75 we consider now propagation perpendicular to the zero order field .
taking @xmath76 , @xmath77 , @xmath78 , @xmath79 , the general system ( [ full ] ) can be reduced to@xmath80 in the case @xmath81 , this system has non - trivial solution only for the dispersion relation@xmath82from which , @xmath83 , with ( @xmath84 ) @xmath85 in the case @xmath86 , one also has @xmath83 , with@xmath87 . \label{gammab0}\ ] ] using the equation for @xmath88 in ( [ full ] ) one can estimate that @xmath89with @xmath90 a characteristic length of the system , so that the @xmath91 terms can be neglected in the expression ( [ gammab0 ] ) , to obtain@xmath92much larger that in the case @xmath81 . for the case without zero order magnetic field , @xmath93 , and perturbations @xmath64 , @xmath65 ,
@xmath66 one has @xmath94 for propagation along the electric field , @xmath95 , with @xmath69 @xmath70 , one then has @xmath96 again , it is easy to see that the last relation is satisfied , for arbitrary longitudinal component @xmath97 , only if @xmath98 , with the standard plane em wave relations for @xmath74 and @xmath65 ; while the longitudinal component @xmath97 must be zero in order to satisfy ( [ dtheta ] ) . in this way , the standard plane em wave is the only solution in this case . for propagation perpendicular to the zero order field , one has now @xmath99 , @xmath100 , @xmath101 , @xmath102 , and the general system ( [ full ] ) can be reduced to@xmath103 if @xmath86 , one has @xmath83 , with the same value ( [ small ] ) as in the case of @xmath63 , while if @xmath81 one has the much larger value@xmath104 from the previous subsections it is seen that the more noticeable effects are obtained for propagation perpendicular to the zero order fields , with the appropriate polarization of the wave in each case
. moreover , comparing ( [ gammalineare ] ) with ( [ gamalinear ] ) it is clear that , from a practical point of view , magnetic fields are preferable . in any case , when the beam traverses a length @xmath105 , the relative variation of the amplitude @xmath106 of any field is given by @xmath107 for the case of a magnetic field of 1 t , with @xmath86 , one can thus estimate from ( [ chi ] ) and ( [ gamalinear])@xmath108 although this effect is relatively large , there is the problem that both , growing and decreasing modes are always present , so that a wave entering the region with the static magnetic field results in a superposition of both modes , and so the variation of amplitude of the growing mode cancels with that of the decreasing mode at first order in @xmath109 , and the effect is only observable at second order , much hindering the experiment .
there is however a further possibility .
it was argued inminotti that , in order for the theory to be consistent with the lack of strong gravitational effects due to the magnetic field of the earth , the non - linear terms in eqs .
( [ phi ] ) and ( [ psi ] ) should come into play . in this way , for the case of a static magnetic field outside its sources one can write @xmath110 , with @xmath111 , so that equations ( [ phi ] ) and ( [ psi ] ) for the static case are@xmath112which have the exact solutions @xmath113 , so that @xmath114 , thus largely reducing the source of the gravitational force .
this solution for the case of the earth s magnetic field is compatible with the proposal in@xcite , in which the solution with @xmath115 was used , if @xmath116 with these considerations , the field @xmath88 is simply given by@xmath117with @xmath118 . in this case
the system ( [ full ] ) can be written as ( with @xmath63)@xmath119where the term with @xmath120 can be neglected as it is small , since from ( [ full ] ) one can estimate that @xmath121with @xmath90 a characteristic length of field variation .
proceeding as before one has@xmath122 for propagation perpendicular to @xmath61 one has@xmath123where
@xmath124 is the angle between @xmath61 and @xmath125 . as a result
the dispersion relation is that of a normal em wave , @xmath126 , and the only anomalous effect is the presence of a small longitudinal component of the electric field . in the case of propagation
along @xmath61 one has@xmath127so that @xmath128 , and @xmath83 , with@xmath129 this effect is similar in magnitude to that in ( [ gamalinear ] ) , but with the advantage that only one sign is possible , so that there are not coexisting growing and decaying modes , and the growth ( or decay ) could be observed at first order .
due to the relevance of first order effects in the propagation of em waves , it seems plausible that the use of simple and readily available nowadays fiber optics would allow the verification of the theoretical results of the previous section .
especially , the last result of subsection 2.3 shows a direct method for measuring the additional amplitude change caused by the propagation of a single mode inside an ordinary polymer fiber .
given the significance of separating between alternative extentions of general relativistic theories for modern cosmology we propose that such an experiment is of great importance due to its simplicity .
specifically , our proposal is to get a sufficiently large fiber appropriately coiled which , with existing materials , can be made to easily reach a km of total distance for the propagating mode . by taking the logarithm of eq ( [ deltaa ] ) as @xmath130 , for the amplitude variation to be in db units , we see that a magnetic field of 2 t would result in an amplitude variation of 1 db in 1 km distance .
it is possible to reduce such a distance by an order of magnitude only through a large magnetic field of about 10 t or more which can be produced in current nmr devices while use of superconducting elements could reach even higher values .
actually , recent reports from the nhmfl at los alamos claim a 100 t machine is already operational @xcite .
at the moment we will only assume the strongest existing rare earth magnets like boron - neodymium for a tabletop experiment where sufficiently high accuracy power meters are available .
the central idea is to detect the difference between measurement on the fiber coil , with and without the b field .
present day power meters have an accuracy noise threshold of about @xmath131 db . fortunately , existing manufacturers may be able to provide bobbins totalling 25 km of fiber or more so that a measurement of 10 - 20 db of additional amplitude variations is in principle possible .
proposed configuration of the optical fiber and helmholtz coils.,scaledwidth=100.0% ] with respect to the fiber coiling process , one has to take into account that any angles introduced to the fiber material introduce additional attenuation to any propagating mode .
technical data for existing fiber materials suggest that there should be a certain curvature with angles small enough not to cause severe damping during normal propagation .
this can be achieved with a flattened coil frame like the one shown in fig.[fig1 ] .
as we are not interested in all the engineering details of an actual experiment we only emphasize the main points where care must be taken using some simplified configurations . in the flattened fiber coil of fig.[fig1 ]
care must be taken so that on the upper path the fiber is parallel to the direction of the external magnetic field .
the return path , though , must be outside the region of influence or else the amplitude variation effect will be cancelled and no difference will be measured .
for this reason we also made the flattened electromagnets shown in such a way that the homogenized flux of the applied b field will only affect the upper part of the fiber s path .
it is also possible to make up an homogeneous magnetic field using neodymium magnets in a special configuration known as a cylindrical `` halbach array '' @xcite .
such arrays have been in use for a long time in magnetic trains , very fast brushless motors and similar electrical engineering applications .
in such a case , the upper or lower part of the flattened fiber coil should be put inside the region of homogeneous b flux of a halbach cylinder .
we have here reported for the first time some new results on the possible gravitational influence on classical em fields in scalar - tensor extentions of general relativity .
we also used the linearized version of the perturbed maxwell equations to analyse the propagation of ordinary modes .
the analysis led us to conclude the possibility of easy , low cost experiments with fiber optics that would allow the verification of the said theories .
we believe that the present state of cosmology with the recurring acute problems of inflation and initial conditions , the cmb anisotropy as well as the dark matter and dark energy , fully justifies the continuation of the present research in more areas where evidence can be accumulated experimentally .
|
the effect of static electromagnetic fields on the propagation of light is analyzed in the context of a particular class of scalar - tensor gravitational theories .
it is found that for appropriate field configurations and light polarization , anomalous amplitude variations of the light as it propagates in either a magnetized or electrified vacuum are strong enough to be detectable in relatively simple laboratory experiments .
|
a _ hubble space telescope ( hst ) _ wfpc2 h@xmath1 image of the region deml106
@xcite in the large magellanic cloud ( lmc ) revealed a pair of bright , compact regions embedded within a large shell nebula @xcite .
deml106 has been previously cataloged by @xcite as n30 , and the two compact regions were identified as an knot and designated as n30b . as shown in figure 1 , while the main body of n30b is unremarkable , it is surrounded by a halo with sharp edges on the northern rim facing the luminous b[e ] supergiant hen s22
(= hd34664 ; * ? ? ?
b[e ] supergiants are known to possess stellar winds @xcite .
the nebular morphology and relative location of hen s22 and n30b thus appear to suggest that the halo of n30b is a bow shock produced by dynamical interactions with the wind of hen s22 .
a two - component stellar wind model has been proposed for b[e ] supergiants : ( 1 ) a cool , slow wind along a dusty equatorial disk which gives rise to narrow , low - excitation emission lines of , e.g. , , [ ] , and [ ] , and ( 2 ) a hot , fast polar wind that is responsible for the broad , high - excitation absorption lines of , e.g. , , , and @xcite .
@xmath5 spectra of hen s22 in the 18003200 range are dominated by p cygni type lines of , from which it is derived that the equatorial disk is viewed nearly edge - on and the terminal wind velocity along the disk is @xmath6 km s@xmath3@xcite ; while @xmath5 spectra in the 11501950 range show blue - shifted absorption of and indicating that the polar wind of hen s22 has an expansion velocity @xmath71000 km s@xmath3 @xcite .
n30b is projected at 20@xmath830@xmath8 ( or 57.5 pc for a distance of 50 kpc , * ? ? ?
* ) from hen s22 ; its 24@xmath8 spatial extent ( figure 1 ) has a good chance to be partially located above the disk and interact with the polar wind of hen s22 . to determine the nature and physical properties of the bow - shock - like halo around n30b , we have obtained high - dispersion , long - slit echelle spectra of n30b , its bow - shock - like halo , its surrounding interstellar medium , and hen s22 .
we find that the bow - shock - like halo shows surprisingly broad h@xmath1 line profiles similar to that of hen s22 , suggesting that it is a reflection nebula .
furthermore , the spectral features vary from the east side to the west side of n30b , indicating that _ n30b indeed sees the wind of hen s22 along a range of viewing angles_. recently archived @xmath0 wfpc2 images of n30b obtained by the hubble heritage project provide additional support to the interpretation of a reflection nebula for the halo of n30b . in this paper , we report these observations , and discuss the physical properties of n30b , the nature of its halo , and the implication of our results on the disk and polar winds of the b[e ] supergiant hen s22 .
our @xmath0 wfpc2 images of deml106 ( pi : chu ) were obtained on 1998 november 14 .
the observations were made through the _ f656n _ ( h@xmath1 ) filter for @xmath9 s. the images were processed using the standard @xmath0 data pipeline and then combined to remove cosmic rays and to produce an image with a total exposure time of 2400 s. we further processed the resultant image with the standard procedures outlined in @xcite to extract h@xmath1 fluxes .
two corrections have been applied to these h@xmath1 fluxes due to the @xmath10300 km s@xmath3 red - shift of the lmc .
first , the filter transmission of the red - shifted h@xmath1 line is @xmath11% of the peak transmission , so the extracted h@xmath1fluxes are multiplied by a correction factor of 1.07 .
second , the [ ] @xmath126548 line is red - shifted into the h@xmath1 passband at @xmath13% of the peak transmission . using the [ ]
@xmath126548/h@xmath1 ratio measured from our echelle spectra ( see 2.2 ) , we find that the [ ] @xmath126548 line contributes to @xmath14% of the h@xmath1 flux . therefore , an additional correction factor of 0.96 is applied to the extracted h@xmath1 fluxes .
the @xmath0 wfpc2 images of deml106 for the hubble heritage project ( pi : noll ) were obtained on 2001 october 18 .
the observations were made with the @xmath15 ( wfpc2 @xmath16 ) , @xmath17 ( wfpc2 @xmath18 ) , @xmath19 ( wfpc2 @xmath20 ) , @xmath21 ( [ ] ) , and @xmath22 ( [ ] ) filters for exposure times of @xmath23 , @xmath24 , @xmath25 , @xmath26 , and @xmath27 s , respectively .
note that the broad - band images may contain contributions from nebular lines , e.g. , the @xmath16 band contains the h@xmath28 , h@xmath29 , and h@xmath30 lines and the @xmath18 band contains the h@xmath31 line .
we have processed these images using standard procedures .
we have also adjusted the coordinates using astrometric solutions calculated from the gsc2.2 positions of 9 stars on the wf2 chip ; our final coordinates should have @xmath10 02 accuracy .
the wfpc2 images of n30b are presented in figure 2 .
high - dispersion spectra of deml106 were obtained with the echelle spectrograph on the 4 m telescope at cerro tololo inter - american observatory ( ctio ) from two observing runs , 2000 january 22 and december 7 .
both runs used the 79 line mm@xmath3 echelle grating in the single - order , long - slit observing configuration , in which a flat mirror replaced a cross disperser and a post - slit h@xmath1 filter was inserted to isolate a single order .
the red long focus camera and the 2000@xmath322000 site2k_6 ccd were used to record data .
the 24 @xmath33 m pixel size corresponds to roughly 0.082 along the dispersion and 026 perpendicular to the dispersion .
limited by the optics of the camera , the useful spatial coverage is @xmath103@xmath34 .
the spectral coverage is wide enough to include both the h@xmath1 line and the flanking [ ] @xmath12@xmath126548 , 6583 lines .
a slitwidth of 250 @xmath33 m ( 164 ) was used and the resultant fwhm of the instrumental profile was @xmath35 km s@xmath3 .
the spectral dispersion was calibrated by a th - ar lamp exposure taken in the beginning of the night , but the absolute wavelength was calibrated against the geocoronal h@xmath1 line present in the nebular observations . the journal of echelle observations is given in table 1 .
the two small regions in n30b are aligned roughly along the east - west direction , thus we call these two regions n30b - e and n30b - w , respectively . in the h@xmath1 image ,
apart from the bow - shock - like halo , the bright central regions of these two regions have simple morphologies expected for strmgren spheres .
n30b - e is about 7@xmath8 ( 1.7 pc ) in diameter and shows a bright spot at its northern edge , while n30b - w is 6@xmath8 ( 1.45 pc ) across and has a rather uniform surface brightness .
the @xmath0 wfpc2 h@xmath1 image , with stellar emission excised , can be used to estimate the h@xmath1 luminosities and required ionizing fluxes of the regions in n30b .
we have made two measurements which differ in their assumption as to whether the halo is a reflection nebula or part of the regions . if the halo around the regions is an envelope reflecting the light from hen s22 , a background component similar to the halo
should be subtracted .
the resultant integrated h@xmath1 fluxes for the 7@xmath8 and 6@xmath8 strmgren spheres of n30b - e and n30b - w are @xmath36 ergs @xmath37 s@xmath3 and @xmath38 ergs @xmath37 s@xmath3 , respectively .
these fluxes have been corrected for the filter transmission and [ ] contamination .
if the halo is ionized , a low background exterior to n30b should be used and the integrated fluxes of the regions including the halo would be about 80% larger ; however , as we show in 3.3 this is not the case , so these large fluxes will not be used in further discussions . adopting the extinction of d106 - 10 in n30b - w , @xmath39 = 0.14 @xcite , and a distance of 50 kpc
, we derive h@xmath1 luminosities of @xmath40 ergs s@xmath3 and @xmath41 ergs s@xmath3 and required ionizing fluxes of @xmath42 photons s@xmath3 and @xmath43 photons s@xmath3 for the strmgren spheres of n30b - e and n30b - w , respectively . using the average surface brightness and adopting an emitting path length similar to the diameter of the region
, we have further derived the rms densities of n30b - e and n30b - w , both @xmath1075 h - atoms @xmath44 . to examine the stellar content and to estimate the available ionizing fluxes in n30b ,
we have carried out photometry for stars brighter than @xmath18 = 20 , using the wfpc2 @xmath16 and @xmath18 images .
such bright stars are expected to be of early types , so it is possible to apply constant zeropoints to transform the wfpc2 @xmath16 and @xmath18 to johnson @xmath16 and @xmath18 .
the color - magnitude diagram of johnson @xmath18 versus @xmath45 is presented in figure 3 . using the colors and magnitudes and assuming that these stars are still on the main sequence
, we can deredden the stars and estimate their spectral types .
this is easily done by projecting the position of a star in the color - magnitude diagram along the dereddening direction back to the main sequence .
the photometric measurements and estimated spectral types are given in table 2 .
most of these stars have also been studied by @xcite using ground - based observations .
the differences between her results and the new wfpc2 results are also given in table 2 for comparison .
excellent agreements in photometry are found for stars in isolation , but the ground - based measurements for faint stars in crowded regions may have errors as large as @xmath46 1 mag . in n30b and its vicinity ,
stars brighter than @xmath18 = 20 are all b stars . ionizing fluxes from early - type stars have been given by @xcite down to b3 and @xcite down to b0.5 .
the ionizing fluxes of b0v and b0.5v stars given by the latter are significantly higher than those of the former .
n30b - e contains only one early b star , star # 13 in table 2 ; the ionizing flux of this b2.5v star is @xmath47 photons s@xmath3 @xcite , two orders of magnitude lower than that required to ionize n30b - e .
it is possible that the b[e ] supergiant hen s22 , of spectral type b0b0.5 @xcite , plays a significant role in the photoionization of n30b - e .
n30b - w contains the bright star # 6 , which is saturated in the wfpc2 images , but has been previously classified spectroscopically as a b1.5v star (= d106 - 10 ; * ? ? ?
the ionizing flux of a b1.5v star given by @xcite is 2@xmath3210@xmath48 photons s@xmath3 , which is also about two orders of magnitude lower than the required .
however , extrapolating from the ionizing fluxes of b0v and b0.5v stars given by @xcite , we find that the ionizing flux of the b1.5v star # 6 ( d106 - 10 ) is sufficient to ionize n30b - w .
the relationship between the regions in n30b and the surrounding giant shell deml106 may be assessed from their kinematic properties .
an example of our echelle observations is illustrated in figures 4a and 4b in two stretches to show the h@xmath1 and [ ] @xmath126548 lines from the e - w slit centered near the star # 6 in n30b - w .
the vertical axis is spatial with east on top , and the horizontal axis is spectral with wavelength increasing to the right . for convenience
, we have converted wavelengths to heliocentric velocities of the h@xmath1 line and marked them along the horizontal axis in figure 4 .
( all radial velocities in this paper are heliocentric . )
the narrowest lines with uniform surface brightness along the slit are telluric h@xmath1 and oh lines , located near 0 , @xmath2425 , and + 660 km s@xmath3 .
the nebular h@xmath1 and [ ] @xmath126548 lines can be found at + 300 and @xmath2380 km s@xmath3 , respectively . the h@xmath1 line in figures 4a and 4b shows a bright , narrow component and a faint , broad component .
the spatial extent of the broad component coincides with that of n30b s halo ; this component will be discussed in 3.3 .
the narrow h@xmath1 component arises from ionized gas .
the two h@xmath1 emission peaks near the center of figure 4a correspond to the regions n30b - e and n30b - w , while the fainter h@xmath1 emission extending throughout the entire slit length corresponds to the giant shell in deml106 .
no line splitting is detected in h@xmath1 or [ ] for either n30b or deml106 . to analyze the kinematics of ionized gas quantitatively , we have extracted velocity profiles of the h@xmath1 and [ ] lines along each slit , made gaussian fits , and plotted the results in figure 5 .
the fitted velocities are plotted along the slit positions , with h@xmath1 in filled symbols and [ ] in open symbols ; the fitted fwhm of the h@xmath1 line is plotted as the error bar " at each position , except where data points are densely packed .
the observed fwhm of the h@xmath1 line is the quadratic sum of the instrumental fwhm ( 13.5 km s@xmath3 ) , the thermal fwhm ( 21.4 km s@xmath3 at a temperature of 10@xmath49 k ) , and the turbulent fwhm .
thus the observed h@xmath1 fwhm of @xmath1026 km s@xmath3 for the regions in n30b implies a turbulent fwhm of @xmath106 km s@xmath3 .
this small internal turbulence is consistent with the expectation for regions ionized by early b - type stars , because they lack strong fast stellar winds .
the giant shell in deml106 has a larger observed fwhm in h@xmath1 , 3540 km s@xmath3 , implying a turbulent fwhm of 2431 km s@xmath3 .
the lack of line splitting suggest that this large turbulent fwhm is dominated by turbulence instead of a systematic shell expansion .
further kinematic information is provided by the velocity variations shown in figure 5 .
the average heliocentric velocity of deml106 is @xmath10 300 km s@xmath3 .
gradual velocity excursions of up to @xmath5010 km s@xmath3 from the average velocity are seen . along
the ns slit centered on hen s22 , a systematic blue - shift of @xmath210 km s@xmath3 is seen in the [ ] @xmath126583 line .
( the nebular h@xmath1 line is too contaminated by the stellar emission to be measured near the star . )
these blue - shifts provide the only evidence that the internal motion of deml106 does contain a component of systematic expansion at a level of @xmath1010 km s@xmath3 .
the kinematics of deml106 show that this giant shell , formed by hen s22 and the other massive stars in the ob association lh38 @xcite , is by no means a simple expanding shell with a large central cavity .
the interior of deml106 may still contain a significant amount of gas in the form of dense cloudlets , and the dynamical interactions between the fast stellar winds and the cloudlets contribute to the large turbulent fwhm observed in deml106 .
the systemic velocities of the regions in n30b are similar to that of deml106 , suggesting that n30b is physically associated with deml106 .
the gas in the vicinity of n30b and especially between its two regions shows blue - shifts of @xmath25 to @xmath210 km s@xmath3 from the systemic velocity ( see figure 5 ) .
low - velocity flows can be produced by an evaporation off the ionization fronts in regions or dynamic interactions between stellar winds and regions . in an evaporation flow ,
the gas projected outside the disk would have motions perpendicular to the line of sight and show no offsets from the systemic velocity .
in a dynamic interaction scenario , if the stellar wind source is on the far side of an region , the ablation of the region by the stellar wind would produce a gas flow toward us .
the ubiquitous blue - shifts observed in the vicinity of n30b indicate that the dynamic interaction between the stellar wind of hen s22 and the regions in n30b is responsible for these low - velocity gas flow .
n30b is probably one of the dense clouds within deml106 ; its visibility is attributed to the early - b stars formed in its interior .
we initially speculated that the bow - shock - like halo around n30b was produced by interactions between the fast polar wind of hen s22 and the regions , and expected to see corresponding supersonic kinematic features within the spatial extent of the halo .
the h@xmath1 + [ ] echellogram of n30b in figure 4 shows that the only kinematic feature that matches the spatial extent of the halo is a prominent , broad h@xmath1 emission component .
the continuous distribution of this broad component over 18@xmath8 , from the eastern edge of n30b - e s halo to the western edge of n30b - w s halo , indicates that the halo envelopes the regions of n30b .
the broad h@xmath1 emission is centered near 300 km s@xmath3 and detected over a velocity range greater than 1000 km s@xmath3 .
such high velocities obviously can not be accelerated by the wind of hen s22 ; furthermore , such strong shocks would have produced more dramatic dynamic effects , such as x - ray emission , which is not seen @xcite .
large velocity widths ( @xmath7 1000 km s@xmath3 ) have been observed in other regions , for example , the carina nebula in our galaxy and the giant region ngc5471 in m101 .
the broad hydrogen balmer line emission in the car ii region in the carina nebula has been attributed to the dust - scattered balmer emission of @xmath51 carinae and is detected at as far as 5 pc from @xmath51 car @xcite .
the broad h@xmath1 and [ ] lines in ngc5471 have been suggested to be powered by supernova explosions and fast stellar winds @xcite .
clearly , n30b can not be associated with supernova explosions , since no x - ray emission has been detected from this region @xcite .
furthermore , n30b does not show broad [ ] lines as are seen in ngc5471 , and therefore it should not have the same physical nature as the broad - line region in ngc5471 .
we will thus focus on whether the broad h@xmath1 line from the halo of n30b may be dust - scattered h@xmath1 emission of the b[e ] supergiant hen s22 . to compare the broad h@xmath1 emission from the halo of n30b to that of hen s22
, we display the echellogram and h@xmath1 line profiles of hen s22 in figure 4c d , below the echellogram of n30b .
while they show qualitatively similar profiles with an absorption component blue - shifted from the centroid of the broad h@xmath1 emission , the widths and depths of the absorption are different .
a similar behavior is observed in the reflection nebula car ii near @xmath51 car .
@xcite extracted h@xmath1 line profiles from six positions in car ii ; these broad profiles ( in their figure 3 ) all show an absorption component on the blue side of the nebular line , but the absorption strength decreases radially from @xmath51 car .
@xcite suggested that the spatial variations in the dust - scattered h@xmath1 line in car ii reflected the temporal variability of the h@xmath1 profile of @xmath51 car @xcite .
however , after monitoring the h@xmath1 profiles of @xmath51 car and car ii from 1985 to 1997 , @xcite show that the the absorption feature in the reflected h@xmath1 line is spatially variable across car ii but temporally invariable over the 12-yr span , and conclude that the reflection nebula is located within a 46@xmath52-wide cone of light from @xmath51 car along the axis of its bipolar homunculus nebula .
we suggest that the broad h@xmath1 line from n30b s halo is also dust - scattered h@xmath1 emission from a nearby h@xmath1-emitting star , hen s22 in this case .
the position - dependence of the broad h@xmath1 line in n30b is also caused by different viewing angles , instead of a temporal variability of hen s22 , as discussed further in 3.5 .
the reflection nebula around n30b is not unique in the vicinity of hen s22 .
our echelle observations also detected broad h@xmath1 line profiles in the nebulosity at 20@xmath8 east of n30b , at a much fainter level .
the bow - shock - like morphology of the reflection nebula around n30b is also not unique among reflection nebulae , as a similar nebula has been observed around the b2v star lss3027 in our galaxy @xcite . in both cases ,
the bow - shock - like structure curves with the apex toward an ob association , lh38 for n30b @xcite and stock 16 for lss3027 @xcite .
the relative nebular and stellar positions are highly suggestive that the stellar radiation and winds from the ob association are responsible for the bow - shock - like morphology of the reflection nebula . for n30b ,
the dynamical interaction is further supported by the small blue - shift , @xmath25 to @xmath210 km s@xmath3 , of the ionized gas in the vicinity of the two regions .
this flow velocity , @xmath25 to @xmath210 km s@xmath3 , is subsonic for 10@xmath49 k ionized gas but supersonic for cold neutral gas .
since the reflective envelope of n30b contains dust and mostly neutral gas , flow velocities of 510 km s@xmath3 would be supersonic ; thus the morphology of n30b s halo may indeed be produced by a bow shock .
the nature of n30b s halo as a reflection nebula is supported by the wfpc2 @xmath16 , @xmath18 , and @xmath20 images in figure 2 .
these three broad - band images show similar morphologies among one another , with almost uniform surface brightness across n30b .
hardly any enhancement is seen over the strmgren spheres where bright h@xmath1 emission is present , indicating that the contribution of nebular line emission from the regions is minimal compared to the scattered stellar continuum and line emission .
therefore , we may use the observed broad - band surface brightness of n30b s reflection nebula and a few simple assumptions to determine the likely position of n30b relative to the b[e ] star hen s22 .
first , we assume a single value for the spherical albedo of every dust grain .
we also assume that the reflection nebula is optically thick , i.e. , every photon from hen s22 incident upon the reflection nebula is either absorbed or scattered .
it has been shown that dust scattering has an angular dependence and follows the phase function @xmath53 where @xmath1 is the deviation of the photon from the forward direction , @xmath28 is the spherical albedo of the dust grains , and @xmath54 is a measurement of the asymmetry of the phase function as given by the expression , @xmath55 @xcite .
we have adopted dust grain properties reported by @xcite for wavelengths longward of 3000 , @xmath56 and @xmath57 .
as shown in figure 6 , the scattering angle @xmath1 is related to the projected distance @xmath58 from hen s22 to the reflection nebula by @xmath59 . using the phase function and the brightness of hen s22
, we can calculate the surface brightness of the reflection nebula expected for a given geometry of the reflection .
the broad - band photometry of hen s22 from 1957 to 1991 show stable brightnesses and colors : @xmath60 and @xmath45 = + 0.28 @xcite .
the reddening of hen s22 , @xmath39 = 0.250.30 , is higher than those of the other stars in deml106 because of its edge - on disk @xcite .
the broad - band photometry of hen s22 made by @xcite in 1994 shows a similar @xmath61 , but a higher @xmath39 = 0.46 .
we have adopted an @xmath39 of 0.25 to deredden the johnson @xmath16 magnitude of hen s22 , and converted it to the wfpc2 @xmath16 ( @xmath15 ) magnitude .
we then calculate the expected surface brightness of the reflection nebula in the wfpc2 @xmath16 band for a range of scattering angles . to compare these predictions to observations ,
a reddening appropriate for the reflection nebula needs to be applied . as
the reflection nebula is on the surface of n30b , its extinction should be no greater than the minimum extinction found among stars in n30b , @xmath62 .
we have applied this amount of extinction to the expected surface brightness of the reflection nebula , and the results are presented in table 3 .
the maximum expected surface brightness comes at a scattering angle of @xmath1040@xmath63 , which corresponds to a distance of @xmath108 pc between hen s22 and the reflection nebula .
we have measured an average surface brightness of 1.6 @xmath32 10@xmath64 ergs s@xmath3 @xmath37 arcsec@xmath65 for n30b s reflection nebula in the wfpc @xmath16 band .
this observed surface brightness is 20% higher than the maximum expected surface brightness . considering that our assumptions and calculation are crude , the observed and expected surface brightness of the reflection nebula are in remarkably good agreement .
as discussed in 3.5 , the reflection nebula is above the equatorial disk and sees a less reddened hen s22 , the expected surface brightness of the reflection nebula is thus higher and would be in even better agreement with the observations .
we conclude that the reflection nebula around n30b is consistent with forward scattering by the dust grains , i.e. , n30b is on the near side of hen s22 .
the most likely position of n30b is @xmath108 pc from hen s22 and the scattering angle @xmath1 is @xmath1040@xmath52 .
the broad h@xmath1 line profiles from the reflection nebula around n30b not only change with position ( figure 7 ) but also differ from that of hen s22 ( figure 4 ) .
adopting the most likely geometry of reflection suggested in 3.4 , @xmath66 8 pc and @xmath67 , the light travel time from hen s22 to us via the reflection nebula is 6 yrs longer than the direct travel time .
thus the spectra of the reflection nebula should be compared to the spectra of hen s22 taken 6 yrs earlier ; unfortunately none are vailable with sufficient resolution .
spectra of hen s22 available in the literature were taken from the late 1960 s to early 1980 s . while these observations yield a consistent systemic velocity of hen s22 , @xmath68 km s@xmath3 @xcite , different blue - shifts ( from the systemic velocity ) of the absorption component in the hydrogen balmer lines
have been reported : @xmath270 km s@xmath3 derived from the h8 to h31 lines observed with a reciprocal dispersion of 20 mm@xmath3 in the years 19681975 @xcite , @xmath287 km s@xmath3 from the h@xmath1 line observed with 38 mm@xmath3 in 1979 , and @xmath240 km s@xmath3 from the h@xmath31 , h@xmath28 , and h@xmath29 lines observed with 20 mm@xmath3 in 1982 @xcite .
since these measurements were made for different balmer lines with different spectral dispersion at different epochs , it is not clear whether the differences in the blue - shifts of the absorption component are caused by atmospheric effects ( low and high balmer lines have different profiles ) or temporal variations in the stellar wind .
our echelle observations of hen s22 were made in 2000 january and december with a reciprocal dispersion of 3.4 mm@xmath3 .
these two spectra show almost identical h@xmath1 profiles , with the absorption component detected at 260 km s@xmath3 .
relative to a systemic velocity of 293 km s@xmath3 , the blue - shift of our h@xmath1 absorption is @xmath233 km s@xmath3 . to compare our h@xmath1 profiles to
that reported by @xcite , we have convolved our spectra with a gaussian with @xmath69 = 32 km s@xmath3 , but we find that the overall h@xmath1 profiles do not match that presented in their figure 5 .
this mismatch might be caused by an incorrect wavelength scale in their figure 5 , where the displacement of the absorption dip from the systemic velocity appears to be @xmath7050 km s@xmath3 , significantly less than what they reported in their text . from the above comparisons we find no conclusive or convincing evidence to indicate that the h@xmath1 line profile of hen s22 varied .
furthermore , the @xmath18 magnitude and the @xmath45 color of hen s22 did not change for more than 0.1 mag from 1957 to 1996 @xcite .
therefore , we can not attribute the spatial variations of the broad h@xmath1 line from the refection nebula to hypothetical temporal variations of the h@xmath1 line of hen s22 .
figure 7 shows smooth spatial variations of the broad h@xmath1 profile , especially the absorption component , across n30b , similar to what has been observed in car ii @xcite .
the explanation of this spatial variation is best illustrated by the homunculus nebula of @xmath51 car , as described and analyzed in detail by @xcite .
the bipolar lobes of the homunculus are dusty and scatter the incident starlight of @xmath51 car ; the reflected spectra at higher latitudes show h@xmath1 absorption extending to larger blue - shifts .
as the homunculus is smaller than 1 ly in size , temporal variations in the spectrum of @xmath51 car are negligible over the differential light travel time ; therefore , the latitude - dependence of the h@xmath1 absorption in the reflected spectra must indicate an anisotropic stellar wind from @xmath51 car .
hen s22 has been shown to possess both a slow disk wind and a fast polar wind @xcite , consequently its spectral properties should also be latitude - dependent .
thus , the variations in the absorption component in the reflected h@xmath1 emission at n30b must be a result of viewing hen s22 from different latitudes , too .
the reflected spectra on the west side of n30b resemble better the observed h@xmath1 line of hen s22 , while those on the east side show the absorption component extending to larger blue - shifts , almost @xmath2100 km s@xmath3 from hen s22 s systemic velocity of 293 km s@xmath3 .
these variations suggest that the east side of n30b views the wind of hen s22 from a higher latitude and the west side of n30b is close to the disk plane .
the absorption component in the h@xmath1 line of a b[e ] star arises from the dense , disk component of the wind close to the star , as illustrated in the cartoon in figure 7 of @xcite .
the variations of the absorption component in the reflected stellar h@xmath1 line emission across n30b thus offer an invaluable opportunity to study the transition from the disk component to the polar component of the stellar wind of a b[e ] star .
future high - s / n , high - resolution spectroscopic observations and stellar atmospheric modeling are needed for a better understanding of the stellar wind of the hybrid - spectrum b[e ] supergiant hen s22 .
@xmath0 wfpc2 h@xmath1 images reveal a bow - shock - like halo around the compact regions in n30b on the side that faces the b[e ] supergiant hen s22 .
long - slit , high - resolution spectroscopic observations of the h@xmath1 line show both narrow and broad emission components in n30b .
the narrow component originates from the ionized gas in the regions ; its velocity variations reveal a gas flow at @xmath25 to @xmath210 km s@xmath3 relative to the regions , indicating dynamic interactions with the stellar wind from hen s22 .
the broad h@xmath1 emission component has the same spatial extent as the halo of n30b , but its width ( @xmath71000 km s@xmath3 ) is too high to have been produced dynamically without detectable x - ray emission .
the spectral shape of the broad h@xmath1 emission is similar to that of hen s22 , both having a blue - shifted absorption component .
we suggest that the halo of n30b is a reflection nebula scattering the stellar emission of hen s22 .
@xmath0 wfpc2 broad - band images show morphologies consistent with this interpretation .
the bow - shock - like morphology is probably indeed formed as a bow shock , as the downstream flow velocity of @xmath25 to @xmath210 km s@xmath3 is supersonic to the neutral gas in the halo of n30b .
assuming typical properties of interstellar dust , we have modeled the surface brightness of the reflection nebula of n30b for a range of scattering angles .
we find that the observed surface brightness of the reflection nebula in the wfpc2 @xmath16 band is within 20% from that expected from dust - scattering at a distance of @xmath108 pc from hen s22 for a scattering angle of @xmath1040@xmath52 .
the absorption component in the broad h@xmath1 emission profile varies across the reflection nebula of n30b as a result of different viewing angles .
the west side of n30b lies closer to the equatorial disk plane of hen s22 , while the east side of n30b is at higher latitudes .
we suggest that the anisotropic wind of the b[e ] supergiant hen s22 can be studied by modeling the broad h@xmath1 line profile and the blue - shifted absorption component , taking into account the 3-dimensional geometry .
this unique opportunity could reveal important insights on the physical conditions and evolution of these enigmatic massive stars .
we thank the anonymous referee for pointing out the existence of monitoring observations of @xmath51 car and its reflection nebula car ii .
we also thank nathan smith and nolan walborn for useful discussions , jayanne english for suggesting deml106 to the hubble heritage project , and monica shaw for assistance in initial data analyses .
this research was supported by the _ hubble space telescope _
grant sti6698.01 - 95a .
bensammar , s. , friedjung , m. , muratorio , g. , & viotti , r. 1983 , , 126 , 427 boumis , p. , meaburn , j. , bryce , m. , & lopez , j. a. 1998 , , 294 , 61 castaeda , h. o. , vilchez , j. m. , & copetti , m. v. f. 1990 , , 365 , 164 chen , c .- h .
r. , chu , y .- h .
, gruendl , r. a. , & points , s. d. 2000 , aj , 119 , 1317 chu , y .- h .
1983 , , 95 , 873 chu , y .- h . , chang , h .- w .
, su , y .-
, & mac low , m .-
1995 , apj , 450 , 157 davies , r. d. , elliott , k. h. , & meaburn , j. 1976 , memras , 81 , 89 feast , m. 1999 , in iau symposium 190 , new views of the magellanic clouds , eds .
h . chu , et al .
, 542 henyey , l. c. & greenstein , j. l. 1941 , , 93 , 70 henize , k. g. 1956 , apjs , 2 , 315 lillie , c. f. & witt , a. n. 1976 , , 208 , 64 lopez , j. a. , & meaburn , j. 1986 , revista mexicana de astronomia y astrofisica , 13 , 27 lucke , p. b. , & hodge , p. w. 1970 , aj , 75 , 171 muratorio , g. 1978 , , 33 , 125 oey , m. s. 1996a , apj , 465 , 231 oey , m. s. 1996b , , 104 , 71 panagia , n. 1973 , , 78 , 929 ruiz , m. t. , melnick , j. , & ortiz , p. 1984
, , 285 , l19 schaerer , d. , & de koter , a. 1997 , , 322 , 598 smith , n. 2002 , phd thesis , university of minnesota zickgraf , f .- j . ,
humphreys , r. m. , lamers , h. j. g. l. m. , smolinski , j. , wolf , b. , stahl , o. , 1996 , a&a , 315 , 510 zickgraf , f .- j . , wolf , b. , stahl , o. , leitherer , c. , & appenzeller , i. 1986 , , 163 , 119 zickgraf , f .- j . ,
wolf , b. , stahl , o. , leitherer , c. , & klare , g. 1985 , , 143 , 421
|
narrow - band @xmath0 wfpc2 images reveal a bow - shock - like halo around the region n30b toward the b[e ] supergiant hen s22 located within the larger deml106 nebula in the large magellanic cloud .
high - dispersion spectra of n30b show a narrow h@xmath1 emission component from the ionized gas ; the velocity variations indicate a gas flow of @xmath25 to @xmath210 km s@xmath3 in the vicinity of the regions , which is resultant from interactions with hen s22 s stellar wind and responsible for the bow - shock morphology .
spectra of n30b s halo show broad h@xmath1 profiles extending over @xmath4 km s@xmath3 , similar to that of hen s22 , indicating that the halo is a reflection nebula of hen s22 .
broad - band morphologies of n30b s halo are also consistent with the reflection nebula interpretation
. we use dust - scattering properties and the observed brightnesses of the reflection nebula and hen s22 to constrain the reflection geometry .
the reflected stellar h@xmath1 emission and absorption vary across the reflection nebula as a result of viewing s22 s anisotropic wind across varying angles .
this reflection nebula , together with the edge - on orientation of hen s22 s disk , provides an invaluable opportunity to study the disk and polar winds of a b[e ] supergiant .
|
currently esa monitors the space debris population in low earth orbit ( leo ) in irregular intervals @xcite .
this is done in the frame of 24 hour radar beam - park experiments .
the data from these measurements are taken in order to constrain models of the population evolution of space debris in orbit @xcite .
the most prominent of the european space debris models is master / proof99 , which allows spacecraft operators to assess the risk of a debris impact for their mission .
as the 24 hour experiments only provide a snapshot of the environment , the temporal behaviour of the space debris population is difficult to determine .
routine monitoring of leo is required in order to better constrain the debris models .
the most suitable technique for monitoring leo is radar detection by high - power , large aperture facilities . in order to obtain inexpensive access to such observations ,
esa started in 2000 a study with the sodankyl geophysical observatory ( sgo ) to use the radar facilities of the european incoherent scatter ( eiscat ) scientific association .
these facilities operate up to 2000 hours per year for ionospheric research purposes . in the study ,
the sgo team was able to demonstrate the exploitation of ionospheric measurements for space debris observations .
a comparison of the retrieved , still preliminary , data with the esa master / proof99 model is presented here .
the space debris measurements were performed in the frame of a demonstration campaign in february 2001 . a radar facility located at @xmath3 and @xmath4 near troms , norway , was used .
the uhf radar at troms operates at a frequency of @xmath5 with a pulse repetition frequency of @xmath6 .
the transmitter can transmit peak powers up to @xmath7 through the main antenna , but it was restricted to just below @xmath8 during the measurement campaign .
the @xmath9 antenna was pointed parallel to the field lines of the local geo - magnetic field , azimuth @xmath10 and elevation @xmath11 . during the test campaign two
different phase codes called tau2 and cp1lt were transmitted . of the tau2 data
@xmath12 hours have been analysed as well as @xmath13 hours of the cp1lt data .
for both experiments the pulse length was about @xmath14 .
the range window of the cp1lt experiment covered ranges from @xmath15 to @xmath16 with a gap between @xmath17 and @xmath18 , and tau covered ranges from @xmath19 to @xmath20 with a gap between @xmath21 and @xmath22 .
the strongest of the high - altitude tau2 events during the test campaign began at 22:19:06 utc , 20 february 2001 .
we identified the target as a large catalogued object , a tsyklon upper stage with the cospar designator 1994 - 11 g .
according to the catalogue the object has a total mass of @xmath23 and is cylindrical in shape , with a diameter of @xmath24 , a height of @xmath25 , and a radar cross section of @xmath26 . the orbital information in the tle catalogue , with tle epoch one day before the experiment epoch , lets us predict that it should have passed the radar beam at the off - axis angular distance of @xmath27 .
figure [ fig_comparison1 ] shows the analysis summary plot of the event .
the analysis was done using @xmath28 coherent integration .
= .6 the top panel of figure [ fig_comparison1 ] shows a quantity proportional to the signal to noise power ratio during the beam transit .
the antenna side - lobe structure is clearly visible .
markers indicate those scans where the ratio was larger than the detection threshold of @xmath29 .
the off - centre angular distance of the pass was fit to the data @xcite .
the best fit was achieved by assuming that the transit occurred @xmath30 off - axis .
it seems difficult to reconcile the predicted offset @xmath27 with the inferred value of @xmath30 .
the middle and bottom panels of figure [ fig_comparison1 ] show the measured slant range and doppler velocity .
the solid dark curves represent quadratic and linear fits to the data points .
the large dots represent the predicted values .
the measured range is about @xmath31 larger than predicted .
the slope of the velocity curve is as predicted , but there is a discrepancy of about @xmath32 in the actual velocity values . the circular - orbit , vertical - beam acceleration is @xmath33 , which is consistent with the value @xmath34 from the velocity fit .
the timing accuracy for the catalogued objects is of the order of @xmath35 , while we believe the measured timing to be accurate to within about @xmath36 .
however , the range and velocity discrepancies can not be removed simply by adjusting the relative timing , because the required correction would be about six seconds for the range data , but only about three seconds for the doppler data . a small misalignment of the antenna can
, however , explain the observed deviation .
this example of the detection of a large catalogued object by the eiscat radar demonstrates the usefulness of routine space debris observations for the validation of large - object catalogues .
while collisions of operational spacecraft with large catalogued objects like the upper stage discussed above can be avoided by manoeuvring the spacecraft , objects smaller than @xmath37 , that normally are not found in the catalogue , pose a serious threat to manned and unmanned spacecraft .
complex spacecraft like the international space station alpha are shielded from impacts of objects smaller than about @xmath38 .
thus , the population in the size range between @xmath39 and @xmath37 is particularly important to observe . in the demonstration campaign performed by sgo
, @xmath40 objects in the size range above @xmath38 were detected within the @xmath41 of analysed data .
the measured distribution in size and altitude distributions discussed below are compared to the prediction by the esa master / proof99 model . due to the pointing of the radar beam along the local geo - magnetic field line
, orbital parameters could not be determined from the measurements .
the predicted distribution of orbital elements is discussed below anyhow .
figure [ fig_sizespec ] shows the distribution of object diameter , as determined by the analysis of the reflected signal amplitude , as well as the prediction by the master / proof99 model .
only objects detected in the range window from @xmath21 to @xmath16 have been considered .
= .5 under the assumption of incoherent integration , objects larger than about @xmath42 are detectable reliably in the range window between @xmath21 and @xmath16 . while the majority of the objects predicted to cross the beam are smaller than @xmath43 , sizes of @xmath44 to @xmath45
have been derived for the objects actually detected in the experiment . from the figure it is evident
that while no object with a derived size larger than @xmath46 has been detected , more than @xmath47 of those large objects should have crossed the beam according to the prediction by master / proof99 . also , the number of @xmath45 objects is three times higher than predicted , while the total number of detected objects is in agreement .
apparently , the sizes derived for the objects larger than @xmath37 have been underestimated . because it is unknown where the objects cross the beam pattern
, the eiscat radars can only give a lower bound for the radar cross section .
consequently , the translation of the measured radar cross section to object diameter systematically underestimates the object size .
the distribution of orbital altitudes of detected objects is shown in figure [ fig_altspec ] . at the altitudes of maximum debris density between @xmath48 and @xmath49 ,
@xmath50 objects were detected .
according to the master / proof99 model , the predicted number of objects detectable in this altitude bin by incoherent integration is below @xmath51 , significantly less than actually detected .
this is also true for all but the altitude bin around @xmath52 .
the sensitivity of the radar appears to be higher than for the standard incoherent detection method .
this can be explained by the fact that the detection algorithm exploits the phase information by coherently integrating the detected signal over @xmath36 .
master / proof99 can only simulate radar detection with incoherent integration .
since the detection algorithm used by the eiscat team is based on a coherent integration technique , the predicted number of objects is lower than the eiscat data .
= .5 assuming circular orbits , the inclination of an object s orbit can in theory be derived from the doppler velocity measured by the radar .
however , the functional dependence of the doppler velocity on the orbit inclination varies with the orientation of the radar beam .
as can be seen in figures [ fig_rr180 ] and [ fig_rr90 ] the mapping from doppler velocity to orbit inclination is ambiguous for an antenna azimuth around @xmath53 ( south ) .
an easterly pointing ( antenna azimuth @xmath54 ) will provide an unambiguous relationship between the measured doppler velocity and orbit inclination .
= .6 = .6 as the eiscat transmitter at troms was pointed parallel to the geo - magnetic field lines , which follow the north - south direction , orbit inclinations can not be derived from the data obtained during the measurement campaign . using the proof application of the master / proof99 model , we can however predict the inclination distribution expected for the measurement campaign . the prediction is shown in figure [ fig_incspec ] .
it can be seen that inclinations from @xmath55 to @xmath56 are covered by beam - park experiments with the transmitter at troms .
this covers debris created in classical sun - synchronous orbits as well as debris created by most launches from plesetsk .
the test campaign performed by sgo resulted in raw data on one large space debris object as well as in the size and altitude distribution of @xmath57 small space debris objects .
the comparison with the esa master / proof99 model shows that the conversion from signal amplitudes to object sizes has to be reviewed .
the more straight forward measurement of the objects altitude provides a distribution that is well in accord with the prediction by the model .
as the model is regularly checked against results from other beam - park experiments , we can conclude that the demonstration measurements with the eiscat transmitter at tromsprovides reliable measurements .
the absolute number of detections is significantly larger than the predicted number when assuming incoherent integration as a detection technique .
thus , the coherent integration of the received signal indeed increases the detection sensitivity . as expected , we find that , when linked to ionospheric measurements , we will not obtain orbital information using the eiscat radars .
this restriction can , however , easily be removed if dedicated observation time is acquired . in this case
an optimum antenna pointing will allow us to determine at least doppler inclinations , which are identical to true orbit inclinations for circular orbits . as the proof tool of the new master / proof-2001 model allows to predict the distribution of doppler inclinations
, this would allow us to constrain the statistical orbit properties of the small - size leo space debris population .
in summary we find that the measurements performed during the demonstration campaign in february 2001 prove the value of space debris data obtained by exploiting ionospheric measurements .
the mere amount of data , that will be available in case simultaneous operations with ionospheric measurements in the order of @xmath58 per year can be performed , provides an important contribution to the understanding of the evolution of leo space debris . in order to realise measurements that operate simultaneously with the ionospheric investigations , a real - time detection method has to be developed , because the volume of data that would be needed to record the raw signal from the radar for @xmath59 per year is prohibitive . with real - time detection only the object parameters as well as the signal of the passage will be recorded . a study to develop real - time measurements
will be started by the end of 2002 .
2 natexlab#1#1 klinkrad , h. , j. bendisch , k. d. bunte , h. krag , h. sdunnus , and p. wegener , the master-99 space debris and meteoroid environment model . ,
* 28(9 ) * , 13551366 , 2001 .
markkanen , j. , m. lehtinen , a. huuskonen , and v. vnnen , measurements of small - size debris with backscatter of radio waves . , sodankyl geophysical observatory , 2002 .
mehrholz , d. , l. leushacke , d. banka , beam - park experiments at fgan , this issue .
|
in the effort to obtain low cost routine space debris observations in low earth orbit , esa plans to utilise the radar facilities of the european incoherent scatter scientific association .
first demonstration measurements were performed from 11 to 23 february 2001 . in total
@xmath0 of radar signals were collected .
here we compare these initial measurements with the predictions of the esa master / proof99 model in order to assess the sensitivity as well as the reliability of the data .
we find that while the determination of object size needs to be reviewed , the altitude distribution provides a good fit to the model prediction .
the absolute number of objects detected in the various altitude bins indicates that the coherent integration method indeed increases the detection sensitivity when compared to incoherent integration . in the data presented here integration times from @xmath1 to @xmath2
were used .
as expected , orbit information can not be obtained from the measurements if they are linked to ionospheric measurements as planned .
in addition routine space debris observations provide also useful information for the validation of large - object catalogues .
( 0,0)(0,0 ) ( -50,180)pedas1-b1.4 - 0003 - 02
|
there is great interest in spherically symmetric self - similar solutions to einstein s equations because of their numerous applications in astrophysics and cosmology @xcite .
indeed there is now considerable evidence for the similarity hypothesis , which postulates that spherically symmetric solutions may naturally evolve to self - similar form in a variety of situations .
the status of this hypothesis has been recently reviewed by carr and coley @xcite . in view of this
, it is clearly important to have as complete a classification of spherically symmetric self - similar solutions as possible .
spherically symmetric self - similar solutions have the feature that all dimensionless quantities can be expressed in terms of @xmath3 , where @xmath4 and @xmath5 are suitably chosen radial and time coordinates .
if the source of the gravitational field is a single perfect fluid , cahill and taub @xcite have shown that the only barotropic equation of state compatible with the similarity assumption has the form @xmath6 for some constant @xmath7 , where @xmath8 and @xmath9 are the energy density and the pressure , respectively . in this case , for a given value of @xmath7 , the solutions are described by two parameters and , providing one restricts attention to shock - free perfect fluids with positive pressure ( @xmath10 ) , they have been completely classified .
this classification has been achieved using a combination of the `` comoving '' approach ( in which the coordinates are adapted to the fluid 4-velocity ) and the `` homothetic '' approach ( in which the coordinates are adapted to the homothetic vector ) .
these approaches have been used by carr and coley @xcite ( henceforth cc ) and goliath et al .
@xcite , respectively , although a full understanding of the solutions requires that one combines them @xcite .
a key step in the cc analysis is the derivation of all possible asymptotic behaviors at large and small distances @xcite . for positive pressure ,
there are at least three kinds of behavior at large spatial distances ( usually corresponding to the limit @xmath11 ) : ( 1 ) asymptotically friedmann ( 1-parameter ) ; ( 2 ) asymptotically kantowski - sachs ( 1-parameter ) , though these are probably unphysical for @xmath12 ) ; and ( 3 ) asymptotically quasi - static ( 2-parameter ) .
there are also two families of solutions which exist only when @xmath13 : ( 4 ) asymptotically minkowski at infinite @xmath14 ( 1-parameter ) ; and ( 5 ) asymptotically minkowski at finite @xmath14 but infinite physical distance ( 2-parameter ) . at small spatial distances , the solutions are of four kinds : they contain either ( a ) a black hole singularity or ( b ) a naked singularity at finite @xmath14 ( but zero physical distance ) or they can be connected to @xmath15 via a sonic point , in which case they are either ( c ) static or ( d ) represent a perturbation of the friedmann solution
. the complete family of @xmath10 solutions can now be found by combining the five kinds of large - distance behavior and four kinds of small - distance behavior .
the way in which one connects the large - distance and small - distance solutions depends crucially on whether or not the solution passes through a sonic point .
if the solutions remain supersonic ( or subsonic ) everywhere , then the small-@xmath14 behavior is uniquely determined by the large-@xmath14 behavior . however , if there is a sonic point , the behavior of the solutions is much more complicated because the equations do not determine their behavior uniquely there , so there can be a discontinuity .
indeed , only a subset of solutions are `` regular '' at a sonic point in the sense that the pressure gradient is finite and they can be extended beyond it . because of this feature , the family of solutions with a regular center and a regular sonic point have a band structure , with the solutions which are analytic at the center and sonic point forming a discrete subset of these @xcite .
this feature is very important for naked singularity formation and critical behavior in the gravitational collapse of a perfect fluid @xcite . in this and the accompanying papers @xcite ( henceforth paper ii )
, we take a first step in extending the classification of cc to the negative pressure case ( @xmath16 ) , using a combination of numerical and analytical studies .
in fact , part of the work is already done , because ref .
@xcite does include the asymptotic analysis for @xmath17 . however , there are some errors in that work ( see appendix a ) and the full family of solutions has not yet been analyzed .
this paper will focus particularly on the case with @xmath1 .
although this equation of state violates the strong energy condition and was little emphasized by cc , it may be very relevant for cosmology both in the early universe ( when inflation occurred @xcite ) and at the current epoch @xcite ( when acceleration may be driven by some form of `` dark energy '' ) . in both situations
the matter model exhibits `` anti - gravity '' but the dominant , null and weak energy conditions still hold . as in the positive pressure case
, it should be stressed that our classification does not cover imperfect or multiple fluids or solutions with shocks .
the accelerated expansion of the universe can also be realized by a scalar field with a flat potential ( i.e. quintessence ) .
the non - existence of self - similar black hole solutions embedded in an exact friedmann universe for this case was already demonstrated in ref .
@xcite and this is clearly complementary to the present work .
there are several key differences between the @xmath1 and @xmath18 cases .
a purely formal difference is that the limiting values of @xmath14 for large and small spatial distances in the flat friedmann solution are reversed : large spatial distances now correspond to @xmath19 and small ones to @xmath20 .
the other differences have more physical significance .
the first is that there are no sound - waves , since the sound - speed @xmath21 is imaginary .
this considerably simplifies the analysis , since there can be no discontinuities and solutions are analytic everywhere .
the second is that there is no exact static solution . although there are still asymptotically quasi - static solutions ( a point which was missed in ref .
@xcite ) . on the other hand ,
the kantowski - sachs and asymptotically quasi - kantowski - sachs solutions now become physical .
finally , the solutions which were asymptotically minkowski at large distances are now replaced with solutions which are asymptotically singular at small distances . despite these differences , many features of the cc classification still apply .
in particular , there is still a 1-parameter family of solutions asymptotic to the flat friedmann model at large distances and these are of great physical interest since they may have cosmological applications . in the positive - pressure case ,
some of these solutions are supersonic everywhere and contain black holes which grow at the same rate as the particle horizon @xcite .
others represent density perturbations in a friedmann background which always maintain the same form relative to the particle horizon @xcite .
however , recently it was pointed out that none of these positive - pressure solutions are `` properly '' asymptotic friedmann because they exhibit a solid angle deficit at infinity @xcite
. they may still be relevant to the real universe ( since observations may not preclude such an angle deficit ) but it would be more accurate to describe them as `` quasi - friedmann '' . in the @xmath1 case , we will show that there are genuine asymptotically friedmann solutions .
we will analyze these solutions numerically and exploit these results in paper ii @xcite to interpret the solutions physically .
the key feature is the existence of asymptotically kantowski - sachs and static solutions , both of which are extendible analytically .
this leads to the possibility of cosmological black hole , wormhole and white hole solutions .
the plan of this paper is as follows . in section
ii , we present the basic field equations for spherically symmetric self - similar spacetimes .
we describe some exact solutions in section iii and solutions which are asymptotic to these in section iv .
we summarize our results and discuss their implications in section v.
we consider a spherically symmetric spacetime with the line element @xmath22 where @xmath23 .
we take the matter field to be a perfect fluid with energy - momentum tensor @xmath24 where @xmath9 and @xmath8 are the pressure and the energy density , respectively , @xmath25 is the @xmath26-velocity of the fluid , and we use units with @xmath27 .
we will adopt comoving coordinates , so that the @xmath26-velocity is @xmath28 the field equations can then be written in the following form : @xmath29 where a comma denotes partial differentiation and @xmath30 is the gravitational constant .
the first two equations correspond to the energy - momentum conservation and the next two specify the misner - sharp mass @xmath31 .
five of the above six equations are independent . throughout this paper , we call the direction of increasing ( decreasing ) @xmath5 the future ( past ) direction .
a spacetime is self - similar if it admits a homothetic killing vector @xmath32 , which is defined by @xmath33 where @xmath34 denotes the lie derivative along @xmath32 .
cahill and taub @xcite first investigated spherically symmetric self - similar solutions in which the homothetic killing vector is neither parallel nor orthogonal to the fluid flow vector .
they showed that by a suitable coordinate transformation
such solutions can be put into a form in which all dimensionless quantities are functions only of the dimensionless variable @xmath35 . in this case , @xmath32 is given by @xmath36 the line element in a spherically symmetric self - similar spacetime can be written as @xmath37 a hypersurface @xmath38 of constant @xmath14 is called a similarity surface and is generated by the homothetic killing vector .
the induced metric on @xmath38 is @xmath39 where @xmath40 is the speed of the fluid relative to the surfaces of constant @xmath14 .
the area radius @xmath41 must be positive but @xmath42 and @xmath4 need not be .
( note that cc adopt a convention in which @xmath4 is always positive . )
a similarity surface is spacelike for @xmath43 , timelike for @xmath44 and null for @xmath45 . in the latter case , @xmath46 is also null and the similarity surface is called a similarity horizon .
it is helpful to rewrite eq .
( [ metric ] ) in the conformally static form : @xmath47,\end{aligned}\ ] ] where the conformal factor depends only on @xmath48 . to obtain the causal structure of the self - similar spacetime
, we can then use the analogy with the static spacetime .
this is because the homothetic killing vector , similarity surfaces and similarity horizons are the counterparts of the killing vector , killing surfaces and killing horizons in the static case ( see e.g. ref .
@xcite ) .
the einstein equations imply that @xmath9 , @xmath8 and @xmath31 must have the form @xmath49 where we assume that the energy density is non - negative ( @xmath50 ) .
then the einstein and energy - momentum conservation equations reduce to ordinary differential equations for the non - dimensional functions with respect to the self - similar variable @xmath14 : @xmath51,\label{basic6z}\end{aligned}\ ] ] where a prime denotes a derivative with respect to @xmath52 .
we assume the equation of state has the form @xmath53 , which is the only barotropic one compatible with the homothetic assumption .
the dominant energy condition requires @xmath54 . in this paper
we exclude the value of 1 ( corresponding to dust ) since this needs special treatment and has been analyzed in ref .
we also exclude @xmath55 , corresponding to a cosmological constant , since this is incompatible with self - similarity .
equations ( [ basic1z ] ) and ( [ basic2z ] ) can then be integrated to give @xmath56 where @xmath57 and @xmath58 are integration constants .
the velocity function @xmath59 can be shown to be @xmath60 equations ( [ basic1z])([basic6z ] ) reduce to ordinary differential equations for @xmath42 , @xmath61 and @xmath62 : @xmath63 . \label{m2}\end{aligned}\ ] ] from eqs .
( [ m1 ] ) and ( [ m2 ] ) , we obtain the following relation between @xmath42 , @xmath61 and @xmath62 : @xmath64 .
\label{eq : constraint}\end{aligned}\ ] ] using this and eq .
( [ m1 ] ) , we obtain another constraint between @xmath65 , @xmath42 and @xmath62 : @xmath66
the flat friedmann solution in self - similar coordinates corresponds to @xmath67 where @xmath68 and @xmath69 are constants .
@xmath59 is determined by eq . ( [ speed ] ) as @xmath70 and eq .
( [ 01 ] ) holds trivially . from eqs .
( [ 00 + 11 ] ) and ( [ eq : constraint ] ) , @xmath68 and @xmath69 must satisfy @xmath71 these equations are consistent provided @xmath7 is not @xmath72 or @xmath73 . since @xmath74 and @xmath75 are determined in terms of @xmath7 , @xmath57 and @xmath58 , where @xmath57 and @xmath58 are just gauge constants , there is no free parameter in this solution for a given value of @xmath7 .
introducing constants @xmath76 and @xmath77 in place of @xmath68 and @xmath69 , we obtain : @xmath78 here the constants @xmath79 and @xmath80 can be chosen arbitrarily and @xmath81 the constants @xmath82 and @xmath83 in eqs .
( [ integral1 ] ) and ( [ integral2 ] ) are given in terms of @xmath76 and @xmath77 as @xmath84 where eqs .
( [ eq : friedmann_s_w])([eq : relation2_friedmann ] ) , ( [ eq : friedmann_s ] ) and ( [ eq : friedmann_w ] ) are used .
one can put the metric into a more familiar form , @xmath85 by introducing new coordinates @xmath86 it should be noted that @xmath87 corresponds to @xmath88 for the decelerating case ( @xmath89 ) but to @xmath90 for the accelerating one ( @xmath1 ) . in the @xmath91 case , the homothetic killing vector is parallel to the fluid flow and these equations no longer apply @xcite .
there exists a finite non - zero @xmath92 where @xmath59 crosses 1 .
figure [ fig : friedmann ] shows the conformal diagram of the flat friedmann solution for @xmath1 , including the similarity surfaces .
we can see that the initial singularity is null .
the similarity horizon @xmath93 corresponds to the cosmological event horizon , while @xmath94 and @xmath88 is a regular center .
the dotted and dot - dashed lines denote the singularity and infinity , respectively .
the conformal diagram of the flat friedmann solution for @xmath1 .
there is a similarity horizon at @xmath93 ( @xmath95 ) , corresponding to the cosmological event horizon , and a regular center at @xmath94 and @xmath88 .
the initial singularity at @xmath96 is null , while null infinity is given by @xmath97 .
the dotted and dot - dashed lines denote the singularity and infinity , respectively .
the thin solid curves and lines denote similarity surfaces , i.e. orbits of @xmath98constant . ]
the kantowski - sachs solution in self - similar coordinates corresponds to @xmath99 then eq . (
[ 01 ] ) holds trivially and eq .
( [ eq : constraint ] ) yields @xmath100 .
\label{eq : aks_mass_consistency}\ ] ] from eq .
( [ speed ] ) , we have @xmath101 and hence @xmath102 as @xmath103 .
( [ 00 + 11 ] ) is trivially satisfied to lowest order but determines @xmath69 to the next lowest order : @xmath104 equations ( [ eq : aks_mass_consistency ] ) and ( [ eq : aks_w0 ] ) then give @xmath105 so we can obtain a physical solution only for @xmath1 and there remains no free parameter in this case .
this solution was first obtained in ref .
@xcite , although that paper introduced different variables to deal with the unphysical solutions for the @xmath18 case .
the kantowski - sachs solution in more standard coordinates can be written as @xcite @xmath106 where @xmath107 is a radial coordinate . as expected , @xmath59 is only real for @xmath1 . in the kantowski - sachs solution
, the area of the 2-sphere with constant @xmath5 and @xmath4 does not depend on @xmath4 but only on @xmath5 , so it expands with time .
the topology of the constant @xmath5 spacelike hypersurface is @xmath108 .
there again exists a non - zero finite @xmath92 where @xmath59 crosses 1 .
@xmath88 or @xmath94 for fixed @xmath109 can be analytically extended to negative @xmath4 or negative @xmath14 beyond @xmath110 .
figure [ fig : ks ] shows the conformal diagram of the ( extended ) kantowski - sachs solution for @xmath1 .
the similarity surfaces are also shown .
we can see that the initial singularity is null .
there are two similarity horizons @xmath111 , corresponding to two cosmological event horizons . the corresponding conformal diagram ( fig .
15 ) in ref .
@xcite is incorrect .
the conformal diagram of the kantowski - sachs solution for @xmath1 .
the solution is analytically extendible beyond @xmath110 and @xmath112 .
there are two similarity horizons @xmath111 , both corresponding to cosmological event horizons .
the initial singularity at @xmath96 is null and future null infinity is given by @xmath97 . ]
the static solution would need to have @xmath113 in which case eq .
( [ speed ] ) gives @xmath114 equations ( [ 00 + 11 ] ) and ( [ eq : constraint ] ) then reduce to @xmath115 and @xmath116 , \label{eq : st_mass_consistency}\ ] ] respectively . for @xmath2 , eq .
( [ eq : st_00 + 11_consistency ] ) can be satisfied and there is a static self - similar solution , but there is no such solution for @xmath117 however , as we will see later , the absence of an exact static solution does not preclude the possibility of an asymptotically static solution .
we now focus on self - similar solutions which are asymptotic to the flat friedmann model at large spatial distances , i.e. in which @xmath119 , @xmath120 , @xmath42 , @xmath62 and @xmath65 approach the form given by eqs .
( [ frwsolution1])-([frwsolution3 ] ) as @xmath121 . in this case , it is convenient to introduce new functions @xmath122 and @xmath123 , defined by @xmath124 which describe the deviations from the flat friedmann solution . equations ( [ 00 + 11 ] ) , ( [ 01 ] ) and ( [ speed ] ) can then be written as @xmath125}{2[v^2-(\gamma-1 ) ] } , \label{fulleq1 } \\
b''&=&-(b'-q)\left(1-q+b'+\frac{\gamma-1}{\gamma}a'\right ) \nonumber \\ & & -\left(q+2b'+\frac{1}{\gamma}a'\right)(1-q+b ' ) , \label{fulleq2 } \\
v&=&\frac{b_0}{a_0}z^{1-q}e^{-2b+(1 - 2/\gamma)a } , \label{const}\label{fulleq3}\end{aligned}\ ] ] respectively . equation ( [ eq : constraint ] ) yields another relation : @xmath126.\end{aligned}\ ] ] the flat friedmann solution is given by @xmath127 for all @xmath14 .
spacelike infinity corresponds to @xmath11 for the decelerating case ( @xmath89 ) and @xmath128 for the accelerating case ( @xmath1 ) . note that @xmath129 at large distances if @xmath122 and @xmath123 are finite .
this means that @xmath94 ( @xmath89 ) and @xmath15 ( @xmath1 ) are horizontal lines in a conformal diagram
. the asymptotic form of self - similar solutions which approach the flat friedmann at large distances can be found by neglecting the @xmath130 term and linearizing the equations with respect to @xmath122 , @xmath131 , @xmath132 and @xmath133 .
( it is noted that @xmath123 does not need to be small . ) from eqs .
( [ fulleq1 ] ) and ( [ fulleq2 ] ) , the linearized equations at spacelike infinity become @xmath134 in both cases .
these equations lead to two independent solutions and the general solution is given by a linear combination of these .
the first solution is @xmath135 where @xmath136 @xmath137 is a constant and the dots denote higher order terms . @xmath122 and @xmath123 converge at large distances in both the decelerating ( @xmath138 ) and accelerating ( @xmath1 ) cases .
this solution was first obtained in ref .
@xcite for the radiation case ( @xmath139 ) and in ref .
@xcite for more general @xmath7 .
however , it is not properly asymptotic to the flat friedmann solution because eq .
( [ defbc ] ) shows that there is a residual @xmath140 term in @xmath42 at infinity ,
so eq . ( [ metric ] ) implies that there is a solid angle deficit @xcite .
we describe solutions with this asymptotic behavior as _ asymptotically quasi - friedmann_. the second solution is @xmath141 where @xmath142 is a constant and @xmath143 means @xmath144 in the relevant limit .
the above solution was also found in ref .
the decelerating solution must be discarded , since this diverges at large distances ( @xmath11 ) . on the other hand ,
the accelerating solution is `` properly '' asymptotic to friedmann since @xmath122 and @xmath123 both converge to zero at large distances ( @xmath128 ) , so there is no solid angle deficit . in the following ,
we generally omit the term `` properly '' and simply describe these solutions as _ asymptotically friedmann_. we conclude that there is a 1-parameter family of asymptotically friedmann solutions at large distances for @xmath1 . note that , in the positive pressure case ( @xmath145 ) , it is well known that there is no physical solution which is _ exactly _ friedmann at sufficiently large but finite distances @xcite . although one might envisage attaching an interior black hole solution to an exterior friedmann solution at a sonic point ( since there can be a discontinuity there ) , no such solution is possible .
this conclusion trivially applies when the pressure is negative ( @xmath16 ) because there are no sound - waves in this case .
we are also interested in solutions which are asymptotically friedmann at _ small _ distances from the origin .
we therefore seek solutions which have @xmath146 for @xmath103 and @xmath1 , where the constants @xmath68 and @xmath69 may be different from the exact flat friedmann case .
equation ( [ speed ] ) implies @xmath147 the exact friedmann relation ( [ eq : relation2_friedmann ] ) still applies but eq .
( [ 00 + 11 ] ) now yields to lowest order @xmath148 hence @xmath149 is either proportional to @xmath150 or falls off even faster . therefore eq .
( [ 01 ] ) implies the following asymptotic behavior : @xmath151 , \\
w&=&w_{0}z^{2}\left[1+w_{1}z^{2(1 - 2/(3\gamma))}+\cdots\right],\end{aligned}\ ] ] where @xmath152 all coefficients of higher order terms are determined from @xmath68 and @xmath69 . since the value for @xmath68 can be different from the exact friedmann case , there is a 1-parameter family of asymptotically quasi - friedmann solutions at small distances .
since @xmath153 as @xmath103 , @xmath94 is a vertical line in a conformal diagram .
these solutions can be interpreted as self - similar models with a regular center as @xmath154 for fixed @xmath155 . on the other hand , in the limit @xmath156 for fixed @xmath157 , they correspond to simultaneous big bang models in the comoving time - slicing . if @xmath68 had the exact friedmann value , we would have friedmann itself , so this is the only solution which is properly asymptotically friedmann .
as we have seen , there exists no static self - similar solution for @xmath1 .
however , we will show that there are still solutions which can be described as _ asymptotically quasi - static_. in this case , we assume @xmath158 as @xmath103 .
only if @xmath68 and @xmath69 are the same as in the static solution do we describe such solutions as _ asymptotically static_. equation ( [ speed ] ) implies @xmath159 this solution can be compatible with eq .
( [ 00 + 11 ] ) if and only if @xmath160 as @xmath103 .
this combination appears in the second term on the right - hand side of eq .
( [ 01 ] ) and this can be regarded as a higher order term .
it then follows that @xmath161 from eq .
( [ eq : hot_combination ] ) , @xmath62 must be of the form @xmath162 where @xmath163 from eq .
( [ eq : constraint ] ) , we obtain the relation @xmath164 .
\label{eq : aqs_1st_order_2}\ ] ] to lowest order eq .
( [ 00 + 11 ] ) becomes @xmath165=\frac{\gamma c_{1}^{2}w_{0}^{1 - 2/\gamma } } { 2s_{0}^{4}}+\frac{2(1-\gamma)}{\gamma } \label{eq : aqs_2nd_order_1}\ ] ] and to the next lowest order eq .
( [ 01 ] ) becomes @xmath166 @xmath167 and @xmath168 are determined from eqs .
( [ eq : aqs_1st_order_1 ] ) and ( [ eq : aqs_1st_order_2 ] ) in terms of @xmath68 and @xmath69 , while @xmath169 and @xmath170 are determined by eqs .
( [ eq : aqs_2nd_order_1 ] ) and ( [ eq : aqs_2nd_order_2 ] ) in terms of @xmath68 , @xmath167 , @xmath69 and @xmath168 .
all higher order terms are determined in terms of @xmath68 and @xmath69 .
hence there is a 2-parameter family of asymptotically quasi - static solutions .
since there is no exact static solution , there are no asymptotically static ones .
also @xmath171 as @xmath103 for @xmath1 , so @xmath94 is a horizontal line in a conformal diagram .
if we consider the analytic continuation beyond @xmath94 into the negative @xmath14 region , the metric should remain analytic in terms of the local inertial cartesian coordinates . for
asymptotically quasi - static solutions , one should consider the analytic continuation beyond @xmath172 because @xmath173 and @xmath174 are finite .
for this purpose , it is convenient to see how the proper time @xmath175 behaves for fixed @xmath176 . for this class of solutions , eq .
( [ integral1 ] ) implies @xmath177 so @xmath178 .
when we continue the solution analytically to the negative @xmath5 region beyond @xmath179 , this means that @xmath180 , where @xmath181 is a positive constant depending on @xmath4 and the upper ( lower ) sign corresponds to the positive ( negative ) @xmath5 . to express @xmath62 and @xmath42 as analytic functions of @xmath175 , we use the following unique continuation : @xmath182 this expression gives a taylor series expansion around @xmath183 in terms of @xmath175 .
because of the presence of the odd powers of @xmath175 , this is not reflection - symmetric about @xmath96 .
on the other hand , in the limit @xmath184 for fixed @xmath155 , one has a vacuum of infinite radius which is not asymptotically flat because @xmath185 approaches a non - zero constant .
we call this _ quasi - static spacelike infinity . _ in paper ii
, we will see that there are a class of solutions describing a friedmann universe emergent from a white hole , where a friedmann spacelike infinity and a quasi - static spacelike infinity are connected .
for the _ asymptotically quasi - kantowski - sachs _ solutions , we assume @xmath187 as @xmath103 . from eq .
( [ speed ] ) , we have @xmath188 and hence @xmath102 as @xmath103 . equation ( [ 00 + 11 ] ) is trivially satisfied to lowest order but to the next order requires either @xmath189 or that the right - hand side falls off even faster than this .
equations ( [ 01 ] ) and ( [ eq : hot_combination_2 ] ) imply @xmath190 equation ( [ eq : constraint ] ) yields @xmath191,\end{aligned}\ ] ] while eq .
( [ 00 + 11 ] ) gives @xmath192 at second order eq .
( [ 01 ] ) becomes @xmath193 hence the coefficients of all higher order terms are determined in terms of @xmath68 and @xmath69 .
this means there is a 2-parameter family of asymptotically quasi - kantowski - sachs solutions .
if the values for @xmath68 and @xmath69 are the same as those for the exact kantowski - sachs solution , all coefficients of higher order terms vanish , so the solution is also exactly kantowski - sachs . in other words , the only solution which is asymptotically kantowski - sachs is kantowski - sachs itself . on the other hand ,
even if the first order terms vanish , i.e. @xmath194 , we can see that @xmath170 may not vanish , so @xmath68 and @xmath69 are different from their kantowski - sachs values .
since @xmath153 as @xmath103 for @xmath1 , @xmath94 is a vertical line in a conformal diagram . for asymptotically quasi - kantowski - sachs solutions , one can consider the analytic continuation beyond @xmath88 because @xmath195 and @xmath196 are non - zero and finite for fixed @xmath155 .
it is useful to see how the proper length @xmath197 changes around @xmath88 for fixed @xmath155 . equation ( [ integral2 ] ) implies @xmath198 so @xmath199 .
this means that , when we continue the solution analytically to the negative @xmath4 region beyond @xmath88 , one has @xmath200 , where @xmath201 is a positive constant depending on @xmath5 and the upper ( lower ) sign corresponds to positive ( negative ) @xmath4 . to obtain analytic expressions for @xmath62 and @xmath42 in terms of @xmath197
, we use the following continuation : @xmath202 the above expression gives a taylor series expansion around @xmath203 .
since @xmath204 in the above expression , @xmath205 generically has an extremum at @xmath206 , while @xmath195 does not .
so this expression is not reflection - symmetric with respect to @xmath207 . on the other hand , in the limit @xmath156 for fixed @xmath157
, we have @xmath208 and @xmath209 , which corresponds to an initial singularity . in paper ii
, we will see that this analytical continuation is crucial for obtaining black hole solutions embedded in a friedmann background as well as a class of wormhole solutions connecting one friedmann spacelike infinity to another quasi - friedmann spacelike infinity . here
we seek solutions in which @xmath59 tends to a finite positive value @xmath210 as @xmath103 .
differentiating eq .
( [ speed ] ) and noting that @xmath211 as @xmath103 , we have @xmath212 substituting this relation into eq .
( [ 01 ] ) , we find to lowest order @xmath213 to lowest order this ordinary differential equation has a general solution of the form @xmath214 where @xmath215 are arbitrary constants and @xmath216 one can show that the square root is real for @xmath1 . also eq . ( [ eq : constraint ] ) is consistent only if the first term dominates the second term in eq .
( [ eq : v*_s_possibility ] ) , so @xmath217 is excluded .
one can easily show that the first term on the right - hand side of eq .
( [ 00 + 11 ] ) converges to zero for @xmath1 and so @xmath218 this is plotted in fig .
[ fg : v_inf ] .
@xmath210 for the asymptotically constant - velocity solution as a function of @xmath7 .
although only the part of the solid curve with @xmath1 is relevant to the present analysis , it is extended to @xmath219 for completeness .
the other root for @xmath210 is also plotted with a dashed line but this may be unphysical . ]
note that the solution with similar asymptotic behavior for @xmath220 @xcite has the other branch of the square root .
for this solution , we have @xmath221 the condition @xmath222 yields a relation between @xmath68 and @xmath69 : @xmath223 there is no other relation between them , so there is a 1-parameter family of solutions with this asymptotic behavior . since @xmath224 as @xmath103 for @xmath1
, @xmath94 is a vertical line in a conformal diagram .
we now discuss the physical significance of this asymptotic solution .
as @xmath7 increases from 0 to 2/3 , @xmath225 decreases from @xmath226 to @xmath227 , while @xmath210 decreases from 1 to 1/3 . because @xmath228 , we have @xmath229 , @xmath230 , @xmath231 and @xmath232 as @xmath233 for fixed @xmath155
. therefore , this solution approaches a vacuum region with infinite physical radius .
however , because the fall - off of the energy density @xmath8 is very slow , the solution is far from asymptotically flat .
in fact , both @xmath31 and @xmath185 diverge to infinity .
this solution can not be analytically extended beyond @xmath94 because @xmath234 diverges to infinity as @xmath233 for fixed @xmath155 , while @xmath8 diverges to infinity as @xmath156 for fixed @xmath157 .
we describe these as _ asymptotically constant - velocity _ solutions .
the infinity reached as @xmath184 for fixed @xmath155 will be called the _ constant - velocity spacelike infinity _ ( cf . the asymptotically minkowski solutions for @xmath13 described in cc ) .
we assume that @xmath236 diverges as @xmath235 , while @xmath59 tends to a finite value @xmath237 .
then eq .
( [ speed ] ) implies @xmath238 and eq .
( [ 01 ] ) yields to lowest order @xmath239 where @xmath240 equation ( [ 00 + 11 ] ) implies @xmath241 , while eq .
( [ m1 ] ) gives @xmath242 where the upper ( lower ) sign corresponds to positive ( negative ) @xmath243 . from eq .
( [ speed ] ) , we have @xmath244 and hence @xmath245 is determined from @xmath68 and @xmath69 . to next lowest order , eq . ( [ 01 ] ) implies @xmath246 , \\ w = w_{0}|z|^{-2\gamma/(6 - 5\gamma ) } \left[1 + w_{1}|z|^{(2 - 3\gamma)/(6 - 5\gamma)}+\cdots \right],\end{aligned}\ ] ] where @xmath247 through eq .
( [ speed ] ) , we obtain @xmath248 where @xmath249 and hence @xmath250 to next lowest order eq .
( [ 00 + 11 ] ) then gives @xmath251 which is consistent with eq .
( [ eq : constraint ] ) .
we describe these solutions as _ asymptotically singular_. for fixed @xmath157 , we have @xmath252 , @xmath253 and @xmath254 as @xmath255 . from eq .
( [ eq : singular_m0 ] ) , we have @xmath256 and so the mass is positive ( negative ) for @xmath257 ( @xmath258 ) .
( note that @xmath42 can be negative but @xmath41 is positive even after the extension to the negative @xmath4 region . ) since the physical properties of these two limits are very different , we distinguish these two cases as _ asymptotically positive - mass singular _ and _ asymptotically negative - mass singular_. for the positive - mass singular case , @xmath259 , @xmath260 and @xmath261 ; for the negative case , @xmath262 , @xmath263 and @xmath264 . although @xmath265 in the limit @xmath255 , the first component of the metric ( [ metric2 ] ) is @xmath266 which tends to @xmath267 .
hence the similarity surface becomes spacelike ( timelike ) for the positive ( negative ) mass case .
we will see in paper ii that the positive - mass singular behavior is associated with black hole and white hole singularities , while the negative - mass one is associated with naked singularities .
it should be noted , however , that the mass of the singularity is not constant but is proportional to the time coordinate @xmath5 .
there is a 2-parameter family of solutions belonging to this class .
in this paper , we have analyzed spherically symmetric self - similar spacetimes .
we have shown that the metric of these spacetimes is conformally static , with the conformal structure being determined by the form of the velocity function @xmath59 .
the homothetic killing vector here plays an analogous role to the killing vector in the static case .
we have expressed the einstein field equations in the comoving approach as ordinary differential equations and analytically investigated the possible asymptotic behavior of perfect fluid solutions with @xmath268 and @xmath1 . in paper ii , we focus on solutions which are asymptotic to the flat friedmann universe at large distances because we are interested in solutions embedded in an accelerating friedmann background @xcite . these solutions can be understood in terms of the `` complete classification '' provided by carr and coley @xcite . however
, this classification mainly applies for perfect fluids with positive pressure and we have extended it to @xmath1 .
we find there are eight possible asymptotic behaviors at small , intermediate and large distances in this class of solutions .
however , it should be stressed that there are other asymptotic behaviors , such as kantowski - sachs asymptotes at @xmath269 , and these are discussed elsewhere @xcite .
tables [ table : exact_solutions ] and [ table : asymptotic_behaviors ] summarize the results .
table [ table : exact_solutions ] lists two exact solutions : friedmann ( f ) and kantowski - sachs ( ks ) .
table [ table : asymptotic_behaviors ] lists eight asymptotic behaviors : asymptotically friedmann ( f ) and asymptotically quasi - friedmann ( qf ) for @xmath118 , asymptotically quasi - friedmann ( qf ) , asymptotically quasi - static ( qs ) , asymptotically quasi - kantowski - sachs ( qks ) and asymptotically constant - velocity ( cv ) for @xmath103 , asymptotically positive - mass singular ( pms ) and asymptotically negative - mass singular ( nms ) for @xmath235 .
.[table : exact_solutions]self - similar exact solutions for @xmath1 .
the form of @xmath42 , @xmath62 and @xmath59 , the number of parameters , the analytic continuation and the causal structure are shown .
there is no arbitrary parameter in these solutions . [
cols="^,^,^,^,^,^",options="header " , ] there are three key differences from the positive pressure case . first , there is no sonic point in the negative - pressure case and hence no additional ambiguity associated with the sonic point .
second , these solutions can be properly asymptotic to the flat friedmann model , in the sense that there is no solid angle deficit .
this contrasts with the positive pressure case , where the solutions are only asymptotically quasi - friedmann .
third , the existence of exact and asymptotically kantowski - sachs solutions , which can be extended beyond the timelike hypersurface @xmath206 along the spacelike direction , leads to physically interesting solutions embedded in a friedmann background .
the analytical results of this paper play a crucial role in paper ii , where we investigate the new solutions numerically and interpret them physically .
finally , it should be emphasized that the matter model considered here suffers from dynamical instability for small - scale perturbation . for a perfect fluid with equation of state @xmath268 ,
the derivative @xmath270 is negative for @xmath271 , which violates the condition for dynamically stable equilibrium .
we can also discuss this instability in terms of linear perturbation analysis .
consider a density perturbation with the dependence @xmath272 to linear order in flat spacetime .
then the dispersion relation @xmath273 is obtained . since the wave - number @xmath274 is real , @xmath275 for @xmath271 , the lower sign corresponding to exponential growth of the perturbation .
since this instability exists for arbitrarily short wavelengths , the system is unstable even in the presence of gravity , except for the case of a cosmological constant ( @xmath55 ) .
this means that the matter fields discussed in this paper can only be valid ( i.e. dynamically stable ) at larger scales .
in fact , there are important physical models which have an effectively negative pressure through coarse - graining but which are still absolutely stable at small scales .
for example , the effective equations of state for domain wall networks and string networks correspond to @xmath276 and @xmath91 , respectively @xcite .
it should also be noted that this small - scale dynamical instability does not appear for a quintessence field ( i.e. a scalar field with a flat potential ) .
for example , suppose the scalar potential is given by @xmath277 ( @xmath278 ) .
this can accelerate the friedmann universe for @xmath279 .
the dispersion relation for the scalar field perturbation in flat spacetime becomes @xmath280 , where @xmath281 denotes the background scalar field and the short wavelength approximation is adopted .
thus , this scalar field is stable against short wavelength perturbations .
the authors would like to thank p. ivanov , h. kodama , h. koyama , m. siino and t. tanaka for useful comments .
hm and th are supported by the grant - in - aid for scientific research fund of the ministry of education , culture , sports , science and technology , japan ( young scientists ( b ) 18740162 ( hm ) and 18740144 ( th ) ) .
hm was also supported by the grant no .
1071125 from fondecyt ( chile ) .
cecs is funded in part by an institutional grant from millennium science initiative , chile , and the generous support to cecs from empresas cmpc is gratefully acknowledged .
bjc thanks the research center for the early universe at the university of tokyo for hospitality received during this work .
carr and coley @xcite have also attempted to classify perfect - fluid spherically symmetric self - similar solutions with @xmath282 , although they have not investigated the solutions numerically ( as we do in paper ii ) .
their analysis essentially extends the cc classification @xcite of solutions with @xmath283 and , in principle , includes the @xmath1 case discussed here .
however , a comparison with our results requires care because cc use a different system of notation from us .
this appendix gives the relationship between the two systems and points out that the analysis in ref .
@xcite is not complete .
cc use the same comoving coordinates ( @xmath284 ) as us and the same similarity variable @xmath3 , although a dot rather than a prime denotes @xmath285 .
they also use the same symbols for the velocity function @xmath59 , pressure @xmath9 , density @xmath8 and areal radial distance @xmath41 .
otherwise the mapping from our variables to theirs is as follows : @xmath286 they also express the field equations as odes in the quantities ( @xmath287 ) but they introduce the variable @xmath288 instead of @xmath62 .
note that the odes are expressed in terms of ( @xmath289 ) in ref .
@xcite .
cc correctly identify friedmann and kantowski - sachs as the only exact power - law self - similar solutions and they look for more general solutions which are asymptotic to these .
they also point out that there is no exact static solutions for @xmath290 but they do not notice that there are still asymptotically quasi - static solutions and this is the main reason why their classification is incomplete .
although cc appreciate the importance of extending some solutions from @xmath291 to @xmath292 , they always take the variable @xmath4 to be positive , whereas we allow it to be negative .
their analysis of the asymptotic solutions at large and small distances is closely related to ours and explicitly includes the asymptotically friedmann and asymptotically kantowski - sachs solutions for @xmath1 .
however , they do not distinguish between the asymptotically friedmann and asymptotically quasi - friedmann solutions .
their description of the asymptotic - kantowski - sachs solutions is also somewhat different : because @xmath4 is taken to be positive , they do not extend this solution from @xmath291 to @xmath292 , so they do not notice these solutions are related to wormholes ( cf .
paper ii ) .
cc derive the asymptotically constant - velocity solutions , as well as deriving and plotting eq .
( [ eq : v_star ] ) , but they describe these as `` asymptotically minkowski '' and do fully understand their physical interpretation .
they also identify the asymptotic solutions which represent a central singularity ( i.e. in which @xmath293 at finite @xmath14 ) but they only include them for @xmath294 and omit them for @xmath1 .
they also make no distinction between the positive and negative mass singularities .
they claim to find another small - distance solution for @xmath294 , which they label `` y '' , but we do not find this and it is probably non - physical .
analogue of tables i and ii appears in ref .
@xcite but no analysis of the global structure of the solutions is included , which is another reason why the wormhole connection is missed .
carr and gundlach @xcite have analyzed the global structure of some of these solutions but there is a mistake in the asymptotically kantowski - sachs solutions shown in fig .
15 , where @xmath94 is wrongly identified with the singularity .
b. j. carr and a. a. coley , class .
* 16 * , r31 ( 1999 ) .
b.j . carr and a. a. coley , gen .
* 37 * , 2165 ( 2005 ) .
m. e. cahill and a. h. taub , commun .
* 21 * , 1 ( 1971 ) .
b. j. carr and a. a. coley , phys .
rev . d*62 * , 044023 ( 2000 ) . m. goliath , u. s. nilsson and c. uggla , class . quant . grav . *
15 * , 167 ; _ ibid _ , 2841 ( 1998 ) .
b. j. carr , a. a. coley , m. goliath , u.s .
nilsson and c. uggla , class .
. grav . * 18 * , 303 ( 2001 ) . b. j. carr and a. a. coley , class .
* 17 * , 4339 ( 2000 ) .
a. ori and t. piran , phys .
59 * , 2137 ( 1987 ) .
a. ori and t. piran , phys .
d. * 42 * , 1068 ( 1990 ) .
t. foglizzo and r. n. henriksen , phys .
d. * 48 * , 4645 ( 1993 ) .
t. harada and h. maeda , phys .
rev . d*63 * , 084022 ( 2001 ) . c. r. evans and j. s. coleman , phys .
rev . lett . * 72 * , 1782 ( 1994 ) .
h. maeda , t. harada and b. j. carr , arxiv:0707.0530 [ gr - qc ] , to appear in physical review d ( paper ii ) .
linde , _ particle physics and inflationary cosmology _
( harwood academic publishers , chur , switzerland , 1990 ) . s. perlmutter _ et al .
_ , astrophys . j. * 517 * , 565 ( 1999 ) ; a.g .
riess _ et al .
_ , astron .
j. * 116 * , 1009 ( 1998 ) ; astron .
j. * 117 * , 707 ( 1999 ) .
t. harada , h. maeda , b.j .
carr , phys .
rev . d*74 * , 024024 ( 2006 ) .
b. j. carr and s. w. hawking , mon . not .
r. astron .
* 168 * , 399 ( 1974 ) .
g. v. bicknell and r. n. henriksen , astrophys .
j. * 219 * , 1043 ( 1978 ) .
b. j. carr and a. yahil , astrophys .
j. * 360 * , 330 ( 1990 ) .
h. maeda , j. koga and k - i .
maeda , phys .
rev . d*66 * , 087501 ( 2002 ) .
r. kantowski and r.k .
sachs , j. math .
phys . * 7 * , 443 ( 1966 ) . b. j. carr and c. gundlach , phys .
rev . d*67 * , 024035 ( 2003 ) .
carr , phys .
rev . d*62 * , 044022 ( 2000 ) .
h. maeda and t. harada , in _ general relativity research trends .
horizons in world physics , volume 249 _ , p.123 , ed .
r. albert ( nova science publishers , new york , 2006 ) .
preprint gr - qc/0405113 .
b. j. carr and a. koutras , astrophys .
j. * 405 * , 34 ( 1993 ) .
a. nusser , mon . not .
soc . * 375 * 1106 ( 2007 ) .
carr , t. harada , and h. maeda , in preparation .
a. vilenkin and e.p.s .
shellard , _ cosmic strings and other topological defects _ , ( cambridge university press , cambridge , 1994 ) .
|
based on the asymptotic analysis of ordinary differential equations , we classify all spherically symmetric self - similar solutions to the einstein equations which are asymptotically friedmann at large distances and contain a perfect fluid with equation of state @xmath0 with @xmath1 .
this corresponds to a `` dark energy '' fluid and the friedmann solution is accelerated in this case due to anti - gravity .
this extends the previous analysis of spherically symmetric self - similar solutions for fluids with positive pressure ( @xmath2 ) .
however , in the latter case there is an additional parameter associated with the weak discontinuity at the sonic point and the solutions are only asymptotically `` quasi - friedmann '' , in the sense that they exhibit an angle deficit at large distances . in the @xmath1 case
, there is no sonic point and there exists a one - parameter family of solutions which are _ genuinely _ asymptotically friedmann at large distances .
we find eight classes of asymptotic behavior : friedmann or quasi - friedmann or quasi - static or constant - velocity at large distances , quasi - friedmann or positive - mass singular or negative - mass singular at small distances , and quasi - kantowski - sachs at intermediate distances .
the self - similar asymptotically quasi - static and quasi - kantowski - sachs solutions are analytically extendible and of great cosmological interest .
we also investigate their conformal diagrams .
the results of the present analysis are utilized in an accompanying paper to obtain and physically interpret numerical solutions .
|
binary systems with a compact object are among the brightest sources in galaxies , in particular in the x - ray energy range . their bolometric luminosity is powered by accretion onto relativistic objects like neutron star ( ns ) or black hole ( bh ) .
luminosity of black hole accretors is produced by accretion disks @xcite , while neutron star accretors have additional structures on their surfaces . in case of non
magnetic neutron stars it is a boundary / spreading layer , in which rapidly rotating matter of the accretion disk decelerate to neutron star rotational velocity ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
if the neutron star is strongly magnetized , then the accretion flow can not continue undisturbed till the neutron star and is channelled to neutron star magnetic poles @xcite . in case of actively accreting neutron
stars the falling matter is stopped near the ns surface due to pressure of outgoing radiation @xcite . in a set of papers
it was shown that the settling of the infalling flow to the ns surface might be unstable with respect to presence of so - called photon bubbles the regions with high radiation pressure act like light fluid with respect to heavy fluid of infalling plasma ( see e.g. * ? ? ?
* ; * ? ? ?
it was predicted that these oscillations might be visible in time series of x - ray luminosity of such sources .
depending on the area of the accretion column footprint estimates of the characteristic timescales varied from hundreds of hz to several khz @xcite .
launch of the orbital observatory rossi xte @xcite have provided a tool to search for these predicted features .
the pca spectrometer of rxte observatory had the largest collecting area of all previous x - ray orbital instruments ( it was switched off on jan.5 , 2012 ) and indeed allowed to discover a lot of new fast phenomena ( see e.g. * ? ? ? * ) . among them
there were claims about detection of photon bubbles oscillation @xcite . in this paper
we are trying to obtain the best possible constrains on presence of a rapidly ( at fourier frequency scales around khz ) variable noise component in lightcurves of accretion powered x - ray pulsars .
as an example we study one of the brightest source of this type v0332 + 53 .
in our search we use data of the pca spectrometer @xcite of the rxte observatory @xcite the most straightforward way to search for quasi - periodic oscillations is to study power density spectra of original lightcurves of x - ray sources , calculated as square of amplitude of their fourier transform @xcite : @xmath1 where , @xmath2 amplitude of the fourier transform of the curve at fourier frequency @xmath3 , @xmath4 number of photons within the transformed lightcurve segment of length @xmath5 . due to limited number of photons from x - ray sources , variability of a signal at the highest fourier frequencies
is always dominated by poisson noise . in the abovementioned normalization
the level of poisson noise to be seen by an ideal instrument is 2.0 @xcite in the case of a real instrument , this counting noise of photons is modified by a number of instrumental effects .
pca is a gas proportional counter , consisting of 5 independent identical units ( proportional counter units - pcus ) each with its own electronics .
all events , which trigger the low level discriminator initiate a sequence of analog - to - digital conversions , during which the detector is busy the detector is deadtime - locked .
all units count x - ray photons , but also count a large flux of background events , created by charged particles .
charged particles typically create events at more than one set of the pcu anodes and thus a large fraction of them can be vetoed using anti - coincidence logic .
this explicitly means that the `` good counts '' is only a fraction of total counts , measured by instrument . in case of bright sources
the fraction of good counts is large but still less than unity .
in addition to simple background events pcus also have so called very large events events , which deposit more than @xmath6 kev of energy into the pcu volume .
these events saturate pcu analogue amplifiers and in order to resume nominal operation of the detector , the whole pcu is intentionally disabled for some pre - set time period .
typical settings , which can be found in scientific data is dsvle=1 and dsvle=2 .
exact values of these vle dead - times during the flight of rxte / pca were measured by different methods to be approximately 50 - 70 @xmath7sec and 140 - 170 @xmath7sec correspondingly @xcite .
effects of modification of pure poisson noise in data of rxte / pca were identified in a number of works ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
it was shown that the main predicted effects of timing logic of pca detector chains on poisson noise of counts are indeed seen in real data , but it is exceedingly hard to calculate exact corrections from the first principles and some recorded housekeeping count rates .
thus , in some works it was adopted to use an analytic model of deadtime - modified poisson noise and fit its parameters to real power spectra @xcite .
this approach uses assumption that at high frequencies ( typically above khz ) there is no intrinsic variability of flux of real sources and all recorded variability is due to modified poisson process .
it was shown that modification of the poisson noise in total pca bandpass due to simple deadtime effect and effect of occasional switch - offs due to very large events can be adequately described by formulas ( see e.g. * ? ? ?
* ) : @xmath8 @xmath9 @xmath10 @xmath11 where @xmath12sec deadtime of the instrument per event , @xmath13 deadtime of very large event , @xmath14 bin time of the lightcurve , @xmath15 average count rate in the lightcurve , @xmath16 number of frequency bins in the fourier transform . in these formulas
all count rates are written for single detector .
example of application of these formulas to power spectrum of the brightest source on x - ray sky sco x-1 is shown in fig.[scox1_broad ] .
for this plot we have used observation obsid 10059 - 01 - 03 - 00 , count rate of the source is @xmath17 kcnts / sec/5pcu , exposure time 1.04 ksec .
we adopted here @xmath18sec @xcite .
it is seen that quality of this analytic fit is very good : @xmath19 for 55 degrees of freedom for power density values at fourier frequencies above 400 hz .
dashed curve shows the model of deadtime modified poisson noise without contribution of the vle component . ] in works of @xcite it was noted that apart from above - mentioned modifications of poisson noise in real data some additional modifications are revealed in power density spectra calculated from data in sub - bands of rxte / pca .
in particular , the most striking feature is an appearance of an additional noise at frequencies up to 1 - 1.5 khz in lightcurves , constructed from data at lowest energy ranges of pca ( e.g. channel range 0 - 13 or 0 - 17 out of 256 channels of pca ) . as an illustration of this effect we present power spectra of the brighest x - ray source on the sky sco x-1 from a set of observations i d 96443 - 03 - 01-\{00, ... ,13 } , 96443 - 03 - 02 - 00 ( time resolution @xmath20 sec @xmath21sec ) , total exposure time @xmath22 ksec .
we have caluclated power spectra for three energy bands : channels 0 - 17 ( @xmath23kev , count rate @xmath2475 kcnts / sec/5pcu ) , 18 - 35 ( @xmath247 - 15 kev , count rate @xmath2428 kcnts / sec/5pcu ) and 0 - 249 ( total energy range of pca 2 - 60kev , count rate @xmath25 kcnts / sec/5pcu ) .
the vle deatime setting during these observations is the same ( dsvle=1 ) as during observation , shown in fig.[scox1_broad ] .
it is seen that while the total band ( channels 0 - 249 ) power density spectrum is indeed similar to that on fig.[scox1_broad ] ( apart from different time binning and somewhat larger source count rate ) , the power density spectra from sub - bands ( channels 0 - 17 , channels 18 - 35 ) are strongly different from that .
kev , @xmath24715 kev and 1560 kev ) . ]
it is likely that the origin of these peculiarities is related to physics of detection of x - ray photons in a gas proportional counter . in work of @xcite
it was proposed that this effect might be caused by temporarily drop of the detector gain over some part of the detector volume immediately after any occurred event .
all events depositing energy into detector volume , create a cloud of electron - ion pairs .
electrons drift to the nearest anode and there create an avalanche .
electrons born in this avalanche are rapidly absorbed by the anode , but much slower ions drift to the respective cathode .
during their drift time they can screen the electric field for subsequent events and thus effectively reduce the detector gain .
events , occurred during time periods , when ions from the previous event are drifting towards the cathode , will have slightly lower pulse heights / ascribed energies . thus , any event , which occurred in the detector volume causes temporarily small drop of the gain in this volume .
it means that events , occurred near the boundary between softer and harder bands will be able to switch from harder band to softer band if they will come during this `` floating boundary '' time period
. temporarily drop of the detector gain squeeze the energy scale of the instrument . therefore
only hard energy bands can loose counts .
the softest energy band does not loose counts , but rather gather counts from neighbouring harder energy band . in order to check presence of this effect in real data
we have calculated cross correlations between light curves of sco x-1 in three non - overlapping energy bands ( the same data as used for fig.[scox1_chans ] ) .
if there would be no `` cross - talks '' between counts in these energy channels apart from ordinary dead - time effect due to detector electronics we should not expect to see any effects on delays @xmath26sec .
in fact we do see very strong correlation of these signals up to delays @xmath27sec ( fig.[cc_scox1 ] ) , where intrinsic variability of the source is absent . top panel of fig.[cc_scox1 ] shows cross - correlation between energy bands determined by pca channels 0 - 17 and 18 - 35 , bottom panel - between 18 - 35 and 36 - 249 .
we see that after any recorded event ( zero lag ) there is a gap at positive delays during @xmath28sec , i.e. the hard band count rate is lower than it should be .
we see these missing events at negative delays they switched to be in the soft band .
the effective timescale for this `` floating boundary '' window is approximately @xmath29sec .
the amplitude of the peak at negative delays ( and amplitude of the gap at positive delays ) should depend on the fraction of counts , which appear as additional to the original counts in the soft energy band for the `` extended ion drift time '' period .
crosscorrelation of light curves in harder energy bands ( fig.[cc_scox1 ] , lower panel ) looks a bit different .
we also see a gap at positive delays switching of fraction of harder band photons into softer band , but at the same time we also see a gap at negative delays .
this happens because during finite time period after any event the count rate in this band ( in our case channels 18 - 35 ) was increased by some fraction of counts from harder band ( channels 36 - 249 ) , but at the same time much larger fraction of counts moved from this band to even softer band ( channels 0 - 17 ) , thus in total this energy band still loose some fraction of its counts . the characteristic time scale of the noted effective change of the energy band boundaries can be estimated from cross - correlations : a gaussian fit gives approximately @xmath30sec ( see fig.[cc_scox1 ] ) .
it should be compared to the time of drift of ions in pcu volume .
xenon ion mobility is approximately 0.8 @xmath31 sec@xmath32 v@xmath32 ( e.g. tables of physical & chemical constants , kaye & laby online , www.kayelaby.npl.co.uk ) . for the applied voltage @xmath33
kv / cm the ions drift velocities is @xmath34 cm / msec .
the distance between anodes and cathodes is approximately 0.6 - 0.7 cm ( the total thicknes of pcu layers is approximately 1.3 cm , @xcite ) , therefore the drift time is approximately @xmath35sec .
this estimate is broadly consistent with the timescale , which we inferred from analysis of cross - correlations .
therefore we do see that the pca scheme of readout of events into separate energy bands have relatively long relaxation time and thus lightcurves of sources , recorded in these energy bands are mutually distorted .
we have simulated the abovementioned effect .
we have simulated pure poisson noise of photons and selected only those , which were separated by more than non - paralizable deadtime @xmath36sec .
for all events we have simulated the energy ( channel ) , according to spectral energy distribution of sco x-1 , recorded in obsids 96443 - 03- ... we have adopted the intrinsic source count rate 20 kcnts / sec / pcu in total energy range ( which gives 100 ksec / sec/5pcu , pca spectrometer has 5 independent detector units ) .
_ we have not simulated the vle deadtime effect _ . in order to imitate the temporarily shift of boundaries between energy bands we have assumed that any incoming event with some predefined probability to cause a change of the energy scale of the instrument ( gain ) .
we assumed that the depth of the gain drop depends on energy of the event .
the energy scale ( gain ) dependence on time since that event ( @xmath37 ) was assumed to be : @xmath38\exp\left[-{1\over{2}}\left({dt\over{t_{\rm drift}}}\right)^2\right]\ ] ] , here the ion drift time @xmath39sec , @xmath37 time between current event and that caused the gain drop .
the final energy channel of the event is @xmath40 .
the resulted event list was split into three energy bands 0 - 17 , 18 - 35 and 36 - 249 .
hz ] power spectra of lightcurves ( total exposure time 20 ksec ) obtained in this way in channels 0 - 17 and 18 - 35 are shown in fig.[power_sim_ch0_17 ] .
crosscorrelations of obtained lightcurves are shown in fig.[crosscor_sim ] .
additional component that one can see at power spectrum collected in channel range 0 - 17 can be approximated by a zero - centered gaussian : @xmath41\ ] ] with characteristic width @xmath42 hz .
example of approximation of the power spectrum of sco x-1 soft band lightcurve ( channels 0 - 17 ) by models of deadtime modified poisson noise , i.e using formulas ( [ eqn2 ] , including effect of vle deadtime ) and effect of the `` floating boundary noise '' ( [ eqn3 ] ) , is shown in fig.[power_ch0_17 ] .
the effect of `` floating boundary '' can be approximated by a zero - centerd gaussian with @xmath43 hz .
solid lines shows the best fit model to the modified poisson noise in the lightcurve .
contribution of noise due to `` floating boundary effect '' is illustrated by a dashed curve , it is approximated by a zero - centered gaussian with characteritic width @xmath44 hz . ]
one of the results , which might be affected by the abovementioned instrumental feature is the detection of a continuum noise around khz in bright x - ray pulsar cen x-3 . in work of @xcite
it was shown that power spectrum of the cen x-3 lightcurve , recorded by pca in the soft energy band ( channels 0 - 17 , which corresponds to energies @xmath457 kev ) demonstrates broad kilohertz continuum ( possibly with two quasi - periodic oscillations ) .
shape of this noise component was estimated by fitting formulas ( [ eqn2 ] ) to data above @xmath242 khz
. formulas ( [ eqn2 ] ) were shown to be quite adequate for the total energy band of the rxte / pca , but as we now see not fully adequate for the pca sub - bands . ) to the data at fourier frequencies above 2 khz .
dashed curve is a contribution of non vle - deadtime components . ] in order to demonstrate the importance of effect of `` floating boundary noise '' for case of cen x-3 we have repeated the analysis of @xcite .
we have selected the same dataset ( obsids 20104 - 01 - 01- ... , taken on feb 28.-mar.3 , 1997 ) , and used only time periods when the source was not eclipsed ( count rate in the channel range 0 - 17 is more than 700 cnts / sec/5pcu ) .
this gives us an exposure time @xmath46 ksec .
the power spectrum in the channel range 0 - 17 ( fig.[cenx3power ] ) was constructed over 16 sec segments ( @xmath47 time bins with @xmath48sec ) and then averaged and binned into logarithmically spaced bins .
the energy spectrum of cen x-3 is strongly different from that of sco x-1 . as the number of counts which are able to switch between energy bands depends on the shape of the spectrum we need to make another simulation of the effect of the ion - drift time on power spectra of the source lightcurves .
the energy spectrum of the source was taken from real data ( see fig.[cenx3ensp ] ) .
power spectrum of the simulated lightcurve in the channel range 0 - 17 is presented in fig.[cenx3sim ] .
simulation was performed with the same setup as for sco x-1 but with lower count rate .
the adopted intrinsic count rate of the source in the total energy band is @xmath49 cnts / sec / pcu .
the resulted power spectrum of a simulated lightcurve ( the channel range 0 - 17 , total simulated exposure time @xmath24518 ksec ) is shown in fig.[cenx3sim ] .
it is seen that the `` floating boundary '' effect creates an increase of variability power exactly at the place where it was seen by @xcite .
therefore we should conclude that the increased noise in signal from cen x-3 at frequencies above 100 hz can be interpreted in terms of the described instrumental effect . as an additional ( but not decisive ) argument in favour of artificial nature of this signal
one can mention that the signal ( broad khz continuum ) was not detected in higher energy band ( @xmath50 kev ) , while the average count rate in this band is similar to that in the soft band and typically the amplitude of all qpo - like features does not decrease with energy .
accreting x - ray pulsar in the binary system v0332 + 53 @xcite is one of the brightest sources of this class in our galaxy .
the rxte observatory have performed a number of observations of this source ( e.g. * ? ? ?
* ) with total exposure more than 500 ksec . in order to obtain the best possible sensitivity to fast oscillations in lightcurve of the source
we have selected only time periods , when the total count rate recorded from the source by pca is larger than 4000 cnts / s/5pcu and analysed only data with time resolution not worse than @xmath51sec . in order to avoid problems with the pca energy sub - bands ( described above )
we have used combined data over all pca energy channels ( 0 - 249 ) . in total
it gives us approximately 113 ksec of observations with vle deadtime setting dsvle=1 ( average count rate @xmath247.2 kcnts / sec / pca ) and approximately 53 ksec with vle deadtime setting dsvle=2 ( average count rate @xmath246.2 kcnts / sec / pca ) .
the resulted power spectra are shown in fig.[power_v0332 ] .
we have fitted the obtained power spectra in frequency range 20 - 4096 hz with a model , consisting of deadtime - modified poisson noise ( formulas ( [ eqn2 ] ) ) and a power law .
we see that the quality of fits is good , @xmath52 values are 58.8 and 79.2 for 60 degrees of freedom for power spectra with vle deadtime settings dsvle=2 and dsvle=1 respectively . during fits
we have fixed the vle deadtime at values @xmath53sec ( dsvle=2 ) and @xmath54sec ( dsvle=1 ) .
we clearly see a power law up to the fourier frequencies 100 - 200 hz ( similar to @xcite ) , but no quasiperiodic oscillations or other components , which are different from instrument - modified poisson noise .
the @xmath55 upper limit on the lorentzian qpo with quality factor 4 ( i.e. @xmath56 ) is approximately 0.4 - 0.75% over the fourier frequency range 200 - 1500 hz .
absence of some additional noise component ( apart from a power law continuing from lower fourier frequencies ) at fourier frequencies below 1 - 2 khz allows us to make some conclusions about properties of the accretion column .
according to calculations of @xcite photon bubble oscillations noise appears below 1 khz timescales only if polars caps have relatively large fractional area , @xmath57 of the total area of the neutron star surface .
if the fractional area of the accretion column is smaller than that the typical frequencies of photon bubble oscillations were predicted to be above khz .
therefore , in the framework of calculations of @xcite we can conclude that the fractional area of the accretion column footprint on the ns surface is smaller than @xmath58 .
it is interesting to mention that indications that the area of the accretion column footprint on ns surfaces can be as small as @xmath59 were demonstrated in work of @xcite from completely different physical effect .
basing on upper limit on the plasma cooling time in the accretion column near magnetic white dwarf it was shown that area of the accretion column footprint is smaller than @xmath60 of the wd surface .
such a small fractional area is a result of small geometrical thickness of accretion flow on top of the magnetospheric surface . extrapolating the matter flow lines ( as lines of force of a dipole magnetic field ) to typical radii of nss authors obtained the abovementioned estimate .
we have studied the power density spectra of bright x - ray sources , measured with rxte / pca with the aim to obtain the best constrains on presence of fast ( @xmath24 khz ) noise components in lightcurves of accretion columns on magnetized neutron stars .
* we have demonstrated that separation of rxte / pca counts into sub - bands creates distortions in their noise properties .
we showed that behavior of sub - band lightcurves can be explained if all events , registered by rxte / pca create a sudden slight drop in detector gain , which redistribute counts , registered by pca between energy sub - bands .
the soft sub - band gain additional counts , harder energy sub - bands loose counts . *
crosscorrelations between lightcurves in different energy sub - bands show that this gain returns to the original state at gaussian time scale @xmath61sec .
one of possible explanations of this effect is a screening of the electric field in the detector volume by slowly moving ions created by previous event .
* we have simulated this effect and obtained reasonable agreement with observed behavior of sub - bands power spectra of lightcurves of sources sco x-1 and cen x-3 .
* we have calculated power density spectra of one of the brightest accretion powered magnetized neutron star ( pulsar ) v0332 + 53 .
avoiding the abovementioned problem with the pca energy sub - bands we have used lightcurves of the source only in the total pca energy bandpass .
we have not detected any quasiperiodic oscillations in the fourier frequency range 200 - 1500 hz with upper limits 0.4 - 0.5% .
* in the framework of photon bubble oscillations in accretion column of magnetized neutron stars , their absence at frequencies below 1 khz might indicate that the fractional area of footprints of the accretion column on the ns surface is small , below @xmath60 .
work is supported by grant of russian science foundation 14 - 12 - 01287 basko m. m. , sunyaev r. a. , 1976 , mnras , 175 , 395 blum w. , rierler w. , rolandi l. , 2008 .
`` particle detection with drift chambers '' , springer , isbn : 978 - 3 - 540 - 76683 - 4 bradt h. v. , rothschild r. e. , swank j. h. , 1993 , a&as , 97 , 355 braun a. , yahel r. z. , 1984 , apj , 278 , 349 davidson k. , ostriker j. p. , 1973
, apj , 179 , 585 jahoda k. , markwardt c. b. , radeva y. , rots a. h. , stark m. j. , swank j. h. , strohmayer t. e. , zhang w. , 2006 , apjs , 163 , 401 klein r. i. , arons j. , jernigan g. , hsu j. j .-
l . , 1996 , apj , 457 , l85 klein r. i. , jernigan j. g. , arons j. , morgan e. h. , zhang w. , 1996 , apj , 469 , l119 lamb f. k. , pethick c. j. , pines d. , 1973 , apj , 184 , 271 leahy d. a. , darbro w. , elsner r. f. , weisskopf m. c. , kahn s. , sutherland p. g. , grindlay j. e. , 1983 , apj , 266 , 160 lyubarskii y. e. , syunyaev r. a. , 1988 , sval , 14 , 390 morgan e. h. , remillard r. a. , greiner j. , 1997 , apj , 482 , 993 pringle j. e. , rees m. j. , 1972 , a&a , 21 , 1 revnivtsev m. , gilfanov m. , churazov e. , 2000 , a&a , 363 , 1013 semena a. n. , et al . , 2014 , mnras , 442 , 1123 shakura n. i. , sunyaev r. a. , 1973 , a&a , 24 , 337 syunyaev r. a. , shakura n. i. , 1986 , sval , 12 , 117 sunyaev r. , revnivtsev m. , 2000 , a&a , 358 , 617 terrell j. , priedhorsky w. c. , 1983 , baas , 15 , 979 tsygankov s. s. , lutovinov a. a. , churazov e. m. , sunyaev r. a. , 2006 , mnras , 371 , 19 van der klis m. , 2006 , in : compact stellar x - ray sources . edited by walter lewin & michiel van der klis .
cambridge astrophysics series , no .
39 wei dennis , mit thesis , 2006 zhang w. , jahoda k. , swank j. h. , morgan e. h. , giles a. b. , 1995 , apj , 449 , 930 zhang w. , morgan e. h. , jahoda k. , swank j. h. , strohmayer t. e. , jernigan g. , klein r. i. , 1996 , apj , 469 , l29
|
we report results of our search for fast oscillations in lightcurve of one of the brightest accretion powered pulsars on the sky v0332 + 53 with the help of data of the pca spectrometer of the rxte observatory .
in course of this search we have carefully explored complications appearing if one uses only sub - bands of the total bandpass of the pca spectrometer .
we show that lightcurves collected in the soft sub - band of the pca spectrometer contains an additional instrumental noise , lightcurves of harder sub - bands lack some fraction of the anticipated poisson noise .
we show that this noise is caused by a cross - talk of energy bands , which lasts up to @xmath0sec .
one hypothesis is that these effects are caused by temporarily drop of the pca detector gain after any occurred event due to slowly moving ions in the detector volume . in order to avoid this effect we searched for fast oscillations in flux of v0332 + 53 only in the total bandpass of the pca spectrometer 2 - 60 kev .
we have not detected any quasi - periodic oscillations in lightcurve of the source with an upper limit at the level of 0.5% in the fourier frequency range 200 - 1500 hz .
[ firstpage ] accretion , accretion discs , x - rays : binaries stars : individual : sco x-1 stars : individual : v0332 + 53
|
an accurate knowledge of the physical properties of young stellar objects ( like their mass , age and luminosity ) is important to constrain theoretical pre main sequence evolutionary models . the determination of these properties , however , depends critically on the availability of accurate distances .
unfortunately , since distances to regions of star formation are often uncertain by more than 20 or 30% , errors on the luminosity and age of young stars are typically about 70% .
significant progress has been possible in recent years thanks to very long baseline interferometry ( vlbi ) observations , particularly with the very long base array ( vlba loinard et al .
2005 , 2007 , 2008 ; torres et al . 2007 , 2009 ;
menten et al . 2007 ; xu et al .
owing to the very accurate astrometry delivered by such instruments , trigonometric parallaxes ( and therefore distances ) can be measured very precisely if multi - epoch observations spread over a few years are obtained .
vlbi instruments are only sensitive to high surface brightness emission , and can only detect objects where non - thermal processes are at work .
such non - thermal sources must , therefore , be identified in the regions of interest before their distance can be measured using multi - epoch vlbi observations .
fortunately , many young stars are magnetically active , and do exhibit detectable levels of non - thermal radio emission .
b4v star s1 in ophiuchus ( loinard et al . 2008 ) is known to exhibit non - thermal radio emission , although theoretical arguments suggest it should be radiative .
we will come back to this point in sect .
this type of emission is typically characterized by strong variability , some level of circular polarization , and a negative spectral index .
also , for magnetically active stars , there is a good correlation between x - ray and non - thermal radio emission ( benz & gdel 2004 ) so young stars with detectable levels of non - thermal radio emission are associated with bright x - ray sources . in this work
, we will focus on the star - forming region associated with the serpens molecular cloud ( strom et al .
1974 ; see eiroa 1992 and eiroa et al .
2008 for two recent reviews ) .
more specifically , we will concentrate on the svs 4 region ( strom et al . 1976 ) , an infrared cluster of at least 11 pre - main sequence sources deeply embedded within the serpens core , and one of the densest young stellar sub - clusters known , with a stellar mass density of @xmath6 pc@xmath7 ( eiroa & casali 1989 ) . in the direction of svs 4 , preibisch ( 1998 ) detected a bright x - ray source ( @xmath8 @xmath5 4 @xmath910@xmath10 erg s@xmath11 ; preibisch 2003a ) now known to be associated with the infrared object ec 95 ( preibisch 2003a , eiroa & casali 1992 ) . using ir
spectroscopy , preibisch ( 1999 ) showed that ec 95 is @xmath5 k2 star , and a comparison with theoretical tracks indicates that it is a very young ( @xmath5 10@xmath12 yr ) , intermediate mass star , presumably a precursor of a @xmath5 4 m@xmath13 herbig aebe star . as a consequence , preibisch ( 1999 ) argued that ec 95 is a
_ proto herbig aebe star_. [ cols="<,^,^,^,^,^,^,^,^,^ " , ] the distance obtained here is based on a trigonometric parallax measurement , and , therefore , does not rely on any assumption regarding the nature of the star , or the properties of the interstellar medium in the serpens and aquila rift clouds .
there is no question that ec 95 is located at 414.9@xmath3 pc . on the other hand , we only measured the distance to that _ one star _ , whereas previous works based on indirect determinations ( such as those of straiys et al . 1996 ) relied on much larger samples
one must , therefore , address the issue of the relevance of the distance measured here for the entire serpens complex . in this respect , it is important to point out that ec 95 is not just any star located in the direction of serpens . as we have seen earlier , it is a very young herbig aebe star .
such intermediate - mass stars are not born in isolation , but as part of small clusters .
since ec 95 is one of at least 11 pre - main sequence stars in the compact svs 4 region , there is no doubt that it is physically associated with that small cluster ( see also eiroa & casali 1989 ) , and that the distance measured here is also the distance to svs 4 .
in addition , svs 4 is universally acknowledged to be part of the serpens core region ( e.g. eiroa & casali 1992 ) , and its members ( including ec 95 ) have always been considered part of the serpens young stellar cluster ( eiroa & casali 1992 ) .
thus , we argue that the distance to ec 95 measured here also provides an accurate estimate of the distance to the serpens core .
we note further that the serpens core is only about 5@xmath14 across on the plane of the sky ( 0.6 pc at a distance of 415 pc ) .
thus , depth effects are not expected to add significantly to the error budget on the determination of the distance to the core . as a consequence ,
we argue that the distance to the serpens core should henceforth be assumed to be 415 @xmath2 5 pc . with an angular diameter of about 3@xmath15
, the serpens cloud is much larger than the serpens core . at a distance of 415 pc
, this would correspond to a physical diameter of about 20 pc .
the depth of the serpens complex is also likely to be of that order , and depending on the location of the core relative to the rest of the complex , the mean distance appropriate for the serpens cloud might be up to 20 pc less or 20 pc more than the distance to the core .
thus , we argue that a conservative estimate of the mean distance to the serpens cloud would be 415 @xmath2 25 pc .
this value is significantly larger than the figure reported by straiys et al .
( 1996 ) for the serpens molecular cloud ( @xmath4 = 259 @xmath2 37 pc ) .
since the total extent of the serpens cloud on the plane of the sky is only about 20 pc , it is extremely unlikely to be 150 pc deep .
instead , we argue that several unrelated regions are likely to coexist along the line of sight .
as mentioned above , serpens is one of the many dust clouds forming the large structure known as the aquila rift .
straiys et al .
( 2003 ) have shown that the front edge of that structure is at about 230 pc , and that its depth is about 80 pc .
the discrepancy between our result and that of of straiys et al .
( 1996 ) could be readily explained if the serpens molecular cloud were not physically associated with the aquila rift , but instead located behind it . in this situation , since the method used by straiys et al .
( 1996 ) is sensitive to the position of the first obscuring material located along the line of sight , it would have naturally picked out the foreground aquila rift clouds , rather that the more distance serpens region .
indeed , in their concluding remarks , straiys et al .
( 2003 ) state that _ both w40 and the serpens molecular cloud are seen projected on the very dark foreground created by the dust lanes of the aquila rift_. such a superposition along the same line of sight of different clouds at different distances is not entirely surprising since that line of sight passes only slightly above the galactic plane , but certainly warrants caution when considering individual regions in that direction .
for instance , it is quite plausible that some of the patchy structures seen in the direction of serpens on optical plates are in fact associated with foreground material in the aquila rift .
in this paper , we have presented multi - epoch vlba observations of ec 95 , a young stellar object located in the svs 4 sub - cluster of the serpens molecular core .
our data demonstrate that ec 95 is in fact a tight binary system with a separation of about 15 mas .
the primary ( ec 95a ) appears to be a 45 @xmath0 intermediate - mass herbig aebe protostar , whereas the secondary ( ec 95b ) is most likely a low - mass t tauri . at radio wavelengths ,
the secondary is on average brighter than the primary .
it is , therefore , also likely to dominate the fairly bright x - ray emission from the system .
the primary , on the other hand , contributes most of the infrared and of the bolometric luminosity of ec 95 .
this might naturally explain why the extinction to ec 95 based on infrared observations was systematically found to be much larger than the extinction based on x - ray data .
interestingly , both members of ec 95 appear to be non - thermal radio emitters .
while a low - mass t tauri star such as ec 95b is expected to generate that type of emission , an intermediate - mass such as ec 95a is not .
we discussed several mechanisms that could explain the presence of non - thermal emission on ec 95a , and argue that the observed properties of ec 95a might be most readily explained if it possessed a corona powered by a thin , rotation - driven convective layer .
the trigonometric parallax of ec 95 appears to be @xmath1 = 2.41 @xmath2 0.02 mas , corresponding to a distance of 414.9@xmath3 pc .
we argue that this implies a distance to the serpens core of 415 @xmath2 5 pc , and to the serpens molecular cloud of 415 @xmath2 25 pc .
this is significantly larger than previous distance estimates based on measurements of the extinction suffered by stars in the direction of serpens ( @xmath4 @xmath5 260 pc ) .
a possible explanation for this discrepancy is that these measurements correspond to the distance to clouds associated with the aquila rift located in the foreground of the serpens cloud .
since our observations resolve ec 95 into a binary , future radio monitoring of that system will allow us to measure the dynamical mass of that system .
to our knowledge , this will be the first time that a dynamical mass is obtained for a young herbig aebe system , and it will enable us to uniquely constrain theoretical pre - main sequence evolutionary for intermediate - mass stars .
the orbital period of the system appears to be 1020 years , so observations in the next two decades will be required to obtain an accurate mass estimate . to serpens cloud is about 20 pc across on the plane of the sky , so it is likely to be also about 20 pc deep .
observations similar to those presented here for a sample of young stars distributed across the cloud would help determine the structure of the region , and would allow us to further constrain the mean distance to this important region of star - formation .
|
using the very long base array , we observed the young stellar object ec 95 in the serpens cloud core at eight epochs from december 2007 to december 2009 .
two sources are detected in our field , and are shown to form a tight binary system .
the primary ( ec 95a ) is a 45 @xmath0 proto - herbig aebe object ( arguably the youngest such object known ) , whereas the secondary ( ec 95b ) is most likely a low - mass t tauri star .
interestingly , both sources are non - thermal emitters .
while t tauri stars are expected to power a corona because they are convective while they go down the hayashi track , intermediate - mass stars approach the main sequence on radiative tracks .
thus , they are not expected to have strong superficial magnetic fields , and should not be magnetically active .
we review several mechanisms that could produce the non - thermal emission of ec 95a , and argue that the observed properties of ec 95a might be most readily interpreted if it possessed a corona powered by a rotation - driven convective layer . using our observations
, we show that the trigonometric parallax of ec 95 is @xmath1 = 2.41 @xmath2 0.02 mas , corresponding to a distance of 414.9@xmath3 pc .
we argue that this implies a distance to the serpens core of 415 @xmath2 5 pc , and a mean distance to the serpens cloud of 415 @xmath2 25 pc .
this value is significantly larger than previous estimates ( @xmath4 @xmath5 260 pc ) based on measurements of the extinction suffered by stars in the direction of serpens .
a possible explanation for this discrepancy is that these previous observations picked out foreground dust clouds associated with the aquila rift system rather than serpens itself .
|
stochastic dynamical systems are used as models for various scientific and engineering problems .
we consider the following class of stochastic dynamical systems @xmath15 where @xmath16 are two real independent wiener processes ( white noises ) and @xmath17 are given deterministic functions .
more complicated stochastic systems also occur in applications ( @xcite , @xcite , @xcite ) , @xcite , @xcite , @xcite ) . for
a planar bounded domain @xmath18 , we can consider the exit problem of random solution trajectories of ( [ eqn1])-([eqn2 ] ) from @xmath18 . to this end , let @xmath19 denote the boundary of @xmath18 and let @xmath20 be a part of the boundary @xmath19 .
the escape probability @xmath21 is the probability that the trajectory of a particle starting at @xmath22 in @xmath18 first hits @xmath19 ( or escapes from @xmath18 ) at some point in @xmath20 , and @xmath21 is known to satisfy ( @xcite , @xcite and references therein ) @xmath23 suppose that initial conditions ( or initial particles ) are uniformly distributed over @xmath18 .
the average escape probability @xmath24 that a trajectory will leave @xmath18 along the subboundary @xmath20 , before leaving the rest of the boundary , is given by ( e.g. , @xcite , @xcite ) @xmath25 where @xmath26 is the area of domain @xmath18 .
the residence time of a particle initially at @xmath22 inside @xmath18 is the time until the particle first hits @xmath19 ( or escapes from @xmath18 ) .
the mean residence time @xmath27 is given by ( e.g. , @xcite , @xcite , @xcite and references therein ) @xmath28
the lagrangian view of fluid motion is particularly important in geophysical flows since only lagrangian data can be obtained in many situations . it is essential to understand fluid particle trajectories in many fluid problems . escape probability ( from a fluid domain ) and mean residence time ( in a fluid domain ) quantify fluid transport between flow regimes of different characteristic motion .
deterministic quantities like escape probability and mean residence time can be computed by solving fokker - planck type partial differential equations .
we now use the these ideas in the investigation of meandering oceanic jets . meandering oceanic jets such as the gulf stream are strong currents dividing different bodies of water . recently , del - castillo - negrete and morrison ( @xcite ) , and pratt et al .
( @xcite , @xcite ) have studied models for oceanic jets .
these models are dynamically consistent to within a linear approximation , i.e. , the potential vorticity is approximately conserved .
del - castillo - negrete and morrison s model consists of the basic flow plus time - periodic linear neutral modes . in this paper
, we consider an oceanic jet consisting of the basic flow as in del - castillo - negrete and morrison ( @xcite ) , plus random - in - time noise .
this model incorporates small - scale oceanic motions such as the molecular diffusion ( @xcite ) , which is an important factor in the gulf stream ( @xcite , @xcite ) .
the irregularity of rafos floats ( @xcite , @xcite , @xcite ) also suggests the inclusion of random effects in gulf stream modeling . this random jet may also be viewed as satisfying , approximately in the spirit of del - castillo - negrete and morrison ( @xcite ) , the randomly wind forced quasigeostrophic model .
several authors have considered the randomly wind forced quasigeostrophic model in order to incorporate the impact of uncertain geophysical forces ( @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) .
they studied statistical issues such as estimating correlation coefficients for the _ linearized _ quasigeostrophic equation with random forcing .
there is also recent work which investigates the impact of the uncertainty of the ocean bottom topography on quasigeostrophic dynamics ( @xcite ) .
the randomly forced quasigeostrophic equation takes the form ( @xcite ) @xmath29 where @xmath30 is a space - time wiener process ( white noise ) .
the stream function would have a random or noise component ( @xcite , @xcite ) .
note that @xmath4 is the meridional derivative of the coriolis parameter @xcite @xcite , i.e. , @xmath0 , where @xmath1 is the rotation rate of the earth , @xmath2 the earth s radius and @xmath3 the latitude .
since @xmath1 and @xmath2 are fixed , @xmath4 is a monotonic decreasing function of the latitude .
the deterministic meandering jet derived in del - castillo - negrete and morrison ( @xcite ) is @xmath31 where @xmath32 this @xmath33 is an approximate solution of the usual quasigeostrophic model @xmath34 with random wind forcing or molecular diffusive forcing in the stochastic quasigeostrophic model ( [ qg ] ) , the stream function would have a random or noise component .
we approximate this noise component by adding a noise term to the above deterministic stream function @xmath33 , that is , in the rest of this paper , we consider the following random stream function as a model for a quasigeostrophic meandering jet , @xmath35 the equations of motion for fluid particles in this jet then have noise terms .
we further approximate them as white noises ( or wiener processes ) @xmath36 or more specifically , @xmath37 dt + \sqrt{{\epsilon } } dw_1 , \label{finaleqn1 } \\ dy & = & -ak\ ; sech^2(y ) \sin(kx ) dt + \sqrt{{\epsilon } } dw_2 , \label{finaleqn2}\end{aligned}\ ] ] where @xmath38 , and @xmath16 are two real independent wiener processes ( in time only ) .
the calculations below are for @xmath39 .
note that @xmath4 is now the only parameter [ finaleqn1 ] , [ finaleqn2 ] , as @xmath40 is given and @xmath41 and @xmath42 depend only on @xmath43 $ ] . when @xmath44 , the deterministic jet consists of the jet core and two rows of recirculating eddies , which are called the northern and southern recirculating regions . outside
the recirculating regions are the exterior retrograde regions ; see figure [ unperturb ] . to
we take @xmath18 to be either an eddy or a piece of jet core ( see figures [ eddy ] , [ jet ] ) .
this piece of jet core has the same horizontal length scale as an eddy , and it is one period of the deterministic jet core ( note that the deterministic velocity field is periodic in @xmath45 ) .
thus we call this piece of jet core a unit jet core . to to from ( [ eqn3 ] ) ,
( [ eqn4 ] ) , ( [ eqn5 ] ) , the escape probability @xmath21 that a fluid particle , initially at @xmath22 , crosses the subboundary @xmath20 of the domain @xmath18 satisfies @xmath46 p_x -ak\ ; sech^2(y ) \sin(kx ) p_y & = & 0 , \\ p|_{\gamma } & = & 1 , \\ p|_{{\partial}d - \gamma } & = & 0.\end{aligned}\ ] ] we take @xmath20 to be either top or bottom boundary of an eddy or a unit jet core ( see figures [ eddy ] , [ jet ] ) .
we numerically solve this elliptic system for various values of @xmath4 between @xmath11 and @xmath13 . in the unit jet core case
, we take periodic boundary condition in horizontal ( meridional ) @xmath45 direction , with period @xmath47 .
a piecewise linear , finite element approximation scheme was used for the numerical solutions of the escape probability @xmath21 , and the mean residence time @xmath27 , described by the elliptic equations ( [ eqn3 ] ) , and ( [ eqn6 ] ) , respectively . using a collection of points lying on the boundary , piecewise cubic splines were constructed to define the boundary of the eddy , and the top and bottom boundaries of the jet core .
computational ( triangular ) grids for the eddy and the jet core were then obtained by deforming regular grids constructed for an ellipse and a rectangular region , respectively .
the computed escape probability crossing the upper or lower boundary of an eddy or a unit jet core , for the case of @xmath48 , are shown in figures [ up ] , [ low ] , [ top ] and [ bottom ] . to to to to suppose that the fluid particles are initially uniformly distributed in @xmath18 ( an eddy or a unit jet core ) .
we can also compute the average escape probability @xmath24 that a particle will leave @xmath18 along the upper or lower subboundary @xmath20 , using the formula ( [ average ] ) ; see figures [ eddyaverage ] and [ jetaverage ] .
for an eddy , the average escape probability for fluid particles ( initially inside the eddy ) escape into the exterior retrograde region is smaller than escape into the jet core for @xmath6 , while for @xmath7 , the opposite holds ( figure [ eddyaverage ] ) . thus @xmath49 is a bifurcation point .
also , the average escape probability for fluid particles escape into the exterior retrograde region increases as @xmath4 increases from @xmath11 to @xmath50 ( or as latitude decreases accordingly ) , and then decreases as @xmath4 increases from @xmath50 to @xmath13 ( or as latitude decreases accordingly ) .
thus @xmath51 is another bifurcation point .
the opposite holds for the average escape probability for fluid particles escape into the jet core . for a unit jet core near the jet troughs
, the average escape probability for fluid particles ( initially inside the jet core ) escape into the northern recirculating region is greater than escape into the southern recirculating region for @xmath8 , while for @xmath9 , the opposite holds .
moreover , for @xmath10 , fluid particles are about equally likely to escape into either recirculating regions ( figure [ jetaverage ] ) .
furthermore , for a unit jet core near the jet crests , the situation is the opposite as for near the jet troughs . to to
the mean residence time @xmath27 of a fluid particle , initially at @xmath22 in either an eddy or a piece of jet core ( see figures [ eddy ] , [ jet ] ) , satisfies @xmath52 u_x -ak\ ; sech^2(y ) \sin(kx ) u_y & = & -1 , \\
u|_{{\partial}d } & = & 0\end{aligned}\ ] ] the mean residence times of fluid particles in an eddy or a unit jet core are shown , for the case of @xmath53 , in figures [ eyemean ] , [ jetmean ] . to to the maximal mean residence time of fluid particles initially in an eddy increases as @xmath4 increases from @xmath11 to @xmath12 ( or as latitude decreases accordingly ) , then decreases as @xmath4 increases from @xmath12 to @xmath13 ( or as latitude decreases accordingly ) ; see figure [ eyemax ]
. however , the maximal mean residence time of fluid particles initially in a unit jet core always increases as @xmath14 increases ( or as latitude decreases accordingly ) ; see figure [ jetmax ] . to to
the present work on fluid particle motion in random flows takes into account of fluid particle _ diffusive _ as well as advective motion .
there has been recent work on fluid particle _ advective _ motion ( molecular diffusion ignored ) in time - periodic , quasi - periodic and aperiodic flows ( periodic @xmath54 quasi - periodic @xmath54 aperiodic @xmath54 random ) ; see , for example , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite and @xcite .
our work on random particle motion does _ not _ require that the random part is small .
however it does require that the flow can be decomposed into steady or unsteady deterministic ( drift ) and random ( diffusion ) parts ; otherwise the fokker - planck formalism does not hold .
* acknowledgement . *
j. duan would like to acknowledge the hospitality of the center for nonlinear studies , los alamos national laboratory , the institute for mathematics and its applications ( i m a ) , minnesota , and the woods hole oceanographic institution , massachusetts .
he is also grateful for the discussions on this work with diego del - castillo - negrete ( scripps institution of oceanography ) , greg holloway ( institute of ocean sciences , canada ) , julian hunt ( cambridge university ) , peter kiessler ( clemson university ) , pat miller ( stevens institute of technology ) , larry pratt ( woods hole oceanographic institution ) , and roger samelson ( oregan state university ) .
d. beigie , a. leonard and s. wiggins , chaotic transport in the homoclini c and heteroclinic tangle regions of quasiperiodically forced two dimensional dynamical systems , _ non linearity _ , * 4 * ( 1991 ) , 775 - 819 .
n. malhotra and s. wiggins , geometric structures , lobe dynamics , and lagrangian transport in flows with aperiodic time - dependence , with applications to rossby wave flow , _ j. nonlinear sci .
_ * 8 * ( 1998 ) , 401 - 456 .
p. miller , c. k. r. t. jones , g. haller and l. pratt , chaotic mixing across oceanic jets , in
_ chaotic , fractal and nonlinear signal processing _ " , r. katz ( eds . ) , aip conference proceedings 375 , 591 - 604 , 1996 .
|
stochastic dynamical systems arise as models for fluid particle motion in geophysical flows with random velocity fields .
escape probability ( from a fluid domain ) and mean residence time ( in a fluid domain ) quantify fluid transport between flow regimes of different characteristic motion .
we consider a quasigeostrophic meandering jet model with random perturbations .
this jet is parameterized by the parameter @xmath0 , where @xmath1 is the rotation rate of the earth , @xmath2 the earth s radius and @xmath3 the latitude .
note that @xmath1 and @xmath2 are fixed , so @xmath4 is a monotonic decreasing function of the latitude .
the unperturbed jet ( for @xmath5 ) consists of a basic flow with attached eddies . with random perturbations ,
there is fluid exchange between regimes of different characteristic motion .
we quantify the exchange by escape probability and mean residence time .
for an eddy , the average escape probability for fluid particles ( initially inside the eddy ) escape into the exterior retrograde region is smaller than escape into the jet core for @xmath6 , while for @xmath7 , the opposite holds . for a unit jet core near the jet troughs
, the average escape probability for fluid particles ( initially inside the jet core ) escape into the northern recirculating region is greater than escape into the southern recirculating region for @xmath8 , while for @xmath9 , the opposite holds .
moreover , for @xmath10 , fluid particles are about equally likely to escape into either recirculating regions .
furthermore , for a unit jet core near the jet crests , the situation is the opposite as for near the jet troughs .
the maximal mean residence time of fluid particles initially in an eddy increases as @xmath4 increases from @xmath11 to @xmath12 ( or as latitude decreases accordingly ) , then decreases as @xmath4 increases from @xmath12 to @xmath13 ( or as latitude decreases accordingly ) .
however , the maximal mean residence time of fluid particles initially in a unit jet core always increases as @xmath14 increases ( or as latitude decreases ) .
* key words : * stochastic dynamics , transport and mixing , geophysical flows , escape probability , mean residence time
|
the method of image charges is a classical technique @xcite for electrostatic problems .
its most elementary application is the problem of a point charge in a spherical cavity inside a conducting medium ( or the reciprocal problem of a point charge outside a conducting sphere ) . in 1845 , william thomson ( lord kelvin ) @xcite noticed that the vanishing - potential boundary condition of conductors can be automatically satisfied on the sphere by putting an image point charge at the kelvin point .
a natural extension is the problem of a dielectric sphere inside a different dielectric background , where a single point image charge no longer works . in 1883
carl neumann @xcite discovered that a point image at the kelvin point together with a line image starting from the kelvin point to infinity solves the boundary condition .
this result has been independently re - derived by several authors , up to 1990s ; see the reviews @xcite for more details .
more recently , the problem of a spherical cavity inside an ionic solution has been studied @xcite .
it is again found that the image charge distribution consists of a point image and a line image .
computational method for the line charge density has been developed which works well in the asymptotic limits @xcite that @xmath0 is either large or small , where @xmath1 is the inverse debye length , while @xmath2 is the radius of spherical cavity .
the current work addresses the general case where @xmath0 is neither large nor small .
the main advantage of image charge methods is to represent the effects of polarization charges in terms of point charges or line of charges , and therefore avoiding the task of numerically solving boundary value problems .
this can substantially reduce the computational cost in monte carlo or molecular dynamics simulation of charged systems .
further reduction of computational cost can be achieved by discretizing line images using gaussian quadratures , which effectively approximates a line charge by a few point charges .
the total electrostatic energy of the system can then be represented as coulomb interaction between many point charges .
most interfaces appearing in nature have irregular shapes , and the corresponding green s functions can not be represented in terms of simple distributions of image charges .
hence one may rightfully argue that image charge methods have rather limited applications .
the most important usage of image charge methods , however , arises in the multiscale reaction - field modeling of coulomb many body systems , where the interfaces are _ artificially introduced_. spherical interfaces are almost always used for their simplicity , and for the availability of analytic results for image charges .
research along this direction is of current interest in both simulations and continuum modeling @xcite . due to the long range nature of coulomb interaction , simulation of charged systems
is highly nontrivial .
a proper treatment of boundary conditions is vital in order to obtain physically meaningful results .
periodic boundary conditions can remove artificial boundary effects in a self - consistent fashion and restore the translation symmetry .
combined with ewald summation method @xcite , the cost of computing the total energy of the system with @xmath3 particles scale as @xmath4 .
this can be further reduced to the order of @xmath5 using a mesh - based algorithm such as the particle mesh ewald or particle - particle particle - mesh ewald lattice summation techniques .
the periodic images are however unphysical and may produce artifacts that obscure the real physics . besides , computational cost of ewald - type summation method is still rather prohibitive for large systems in monte carlo simulations , which limits simulation of charged systems to rather small size .
the development of non - eward methods remains important topics ; see fukuda and nakamura @xcite for a recent review .
an attractive alternative is to use the reaction field type of modeling , which is essentially a multi - scale strategy , schematically illustrated in figure [ schem ] . in this approach ,
an artificial ( spherical ) cavity is introduced .
all ions inside the cavity , together with possible mesoscopic objects such as charged proteins and colloids , are treated explicitly using microscopic model ( such as the primitive model @xcite ) together with monte carlo / molecular dynamics , while ions outside the cavity are treated implicitly using appropriate continuum theory , such as linearized poisson - boltzmann ( pb ) theory .
it is known that linearized pb provides an accurate approximation to dilute electrolytes . for any charge inside the cavity ,
then , we must solve the electrostatic green s function problem , where the potential satisfies poisson equation inside the cavity , and satisfies the ( linearized ) pb equation outside .
the resulted green s function provides the pairwise electrostatic interaction ( the force field ) between mobile ions .
this problem can be efficiently solved using the image charge method developed in this work . in this multi - scale modeling approach
, the microscopic model inside the system and the continuous theory outside the cavity really describe the same system .
therefore the parameters of the poisson - boltzmann theory need to be determined self - consistently .
these include the debye length and the effective dielectric constant . for dilute electrolyte
, the debye length can be theoretically calculated as a function of ion densities , while the dielectric constant can be taken to be that of the solvent .
some unphysical artifacts also arise because of the artificial hard wall repulsion of the cavity surface . using statistical mechanics
, one can study this artifact and use extra short range interaction to compensate it .
alternatively , one can also ignore the details inside a thin shell near the cavity surface of thickness 1 - 2 ion diameters . finally , to achieve the balance between efficiency and precision , the radius of cavity should be chosen to be couple of the debye length .
we also note that there can be different levels of modeling inside the cavity . in the so - called hybrid
implicit / explicit model @xcite , both solvent molecules and ions are treated explicitly inside the cavity .
by contrast , in the so - called hybrid primitive / implicit model , the solvent inside the cavity is modeled implicitly as a dielectric medium , while the ions are treated at the level of primitive model .
the reaction field model augmented by our image charge methods is therefore able to provide an accurate treatment of the electrostatic boundary conditions .
its computation cost of the total energy scales as @xmath6 , where @xmath3 here is the total number of particles including all ions as well as their images . while for small systems ( with total particle number typically smaller than 1000 )
this cost is manageable , for large systems , it becomes unrealistic .
fortunately , the computation cost can be dramatically reduced using fast multipole based methods @xcite .
the computational cost for the combined method generally scales as @xmath7 . in this paper
, we mainly focus on the image charge method for the green s function problem of these multi - scale reaction field models .
the statistical mechanical foundation of these reaction field models will be discussed in a separate publication . in the remaining of this work
, we shall first define the kirkwood series for the green s function ( sec .
ii ) and then use the inverse mellin transform to find the image charge representation ( sec .
we further discretize the line image using method of gauss quadrature and construct an efficient numerical scheme for the computation of the reaction potential . in sec .
iv , we compute the reaction potential using our method and quantify the errors .
we also demonstrate the power of multi - scale modeling by monte carlo simulating a dilute symmetric electrolyte .
the reaction potential at a field point @xmath8 is generated by a point image at the kelvin point @xmath9 and a line image extends from @xmath9 to infinity .
the line charge density is given by theorem [ theom ] . ]
as illustrated in figure [ schem2 ] , let @xmath10 be a spherical cavity with radius @xmath2 , centered at the origin , with dielectric constant @xmath11 .
the volume outside the cavity is filled with electrolyte , which is described by linearized poisson - boltzmann theory and is characterized by a debye length @xmath12 and dielectric constant @xmath13 .
consider a point unit charge fixed at @xmath14 inside the cavity , the average potential ( averaged over the statistical fluctuations of electrolyte outside the cavity ) satisfies the following equations : @xmath15 where @xmath16 and @xmath17 are the average potential inside and outside the cavity respectively .
the boundary condition at infinity is @xmath18 throughout the paper we use light italic letter @xmath19 to represent the magnitude of a vector @xmath20 . on the cavity boundary ,
the green s function satisfies the following standard electrostatic interface conditions : @xmath21 the general boundary value problem associated with eqs . -
actually defines the electrostatic green s function , @xmath22 , which equals @xmath16 inside the cavity and @xmath17 outside .
it is this green s function that shall be directly used in the multi - scale reaction field modeling of electrolyte .
the inverse debye length is defined by @xmath23 where @xmath24 is the bjerrum length of the solvent ( @xmath25 for water at room temperature ) , @xmath26 and @xmath27 are the bulk concentration and the valence of the @xmath28th species of ions . the green s function defined in eqs . - has an azimuthal symmetry , hence depends only on @xmath29 and @xmath30 , where @xmath30 is the angle between the source point @xmath14 and the field point @xmath8 , see fig .
[ schem2 ] . the potential inside the cavity can be written as , @xmath31 which is the superposition of the direct coulomb potential @xmath32 and the reaction potential @xmath33 which is a harmonic function .
both potentials can be expanded in terms of spherical harmonics : [ potentials - all ] @xmath34 where @xmath35 is the smaller ( larger ) one between @xmath36 and @xmath29 , while @xmath37 are constants to be determined , and @xmath38 is legendre polynomial of order @xmath39 . the potential outside the cavity
can also be expanded in terms of legendre polynomials : @xmath40 where @xmath41 is the _ modified spherical hankel function _ ( also called the modified spherical bessel function of the third kind ) @xcite , defined by the following series , @xmath42 the coefficients @xmath43 in eqs .
( [ potentials - all ] ) can be found by solving the standard electrostatic boundary conditions eq .
( [ boundary conditions ] ) .
we shall define the kelvin point @xmath9 via @xmath44 for source charge inside the cavity , we always have @xmath45 using the orthogonality of the spherical harmonics , we obtain the following expansion of the reaction potential , widely known as _ kirkwood series expansion _
@xcite : @xmath46 with the harmonic coefficients given by @xmath47 where @xmath48 and @xmath49 the kirkwood series expansion has been used in the calculation of the reaction potential inside a spherical cavity in simulations @xcite .
it converges slowly when a source charge approaches to the cavity surface , which prevents its wide application in dynamical and monte carlo simulations .
in this section , we develop an image charge representation for the reaction potential using finite mellin transformation .
let @xmath50 be a function defined in the interval @xmath51 $ ] .
its _ finite mellin transform _ , @xmath52 is defined as @xcite @xmath53 = \int_0 ^ 1 f(t ) \ , t^{s-1}dt,\ ] ] where @xmath54 is generically a complex variable .
the original function @xmath50 can be expressed in terms of @xmath52 by the _
inverse mellin transform _
@xmath55 = \int_{\sigma - i\infty}^{\sigma+i\infty } f(s)\ , t^{-s}ds.\ ] ] here the integration is carried out over a vertical line , @xmath56 , in the complex plane .
the dual functions @xmath57 form a _ finite mellin transform pair_. the finite mellin transform is closely related to the one - sided laplace transform .
let @xmath58 , then @xmath59=\mathcal{l}[f(e^{-x});s ] .
$ ] likewise , the inverse transform can also be expressed in terms of the inverse laplace transform .
the finite mellin transform is equivalent to the usual mellin transform for the function with compact support in the finite interval .
the usual mellin transform can be defined in terms of the two - sided laplace transform .
two finite mellin transform pairs shall be useful in our discussion below , [ pair ] @xmath60 where @xmath61 is the dirac delta function .
the mellin transformation of the second pair is defined for @xmath62 .
we shall consider @xmath63 in eq .
( [ kernel ] ) as a function of @xmath39 and analytically continue the integer variable @xmath39 into the complex plane .
this can be done through using the integral representation of bessel functions , for which @xmath64 and the modified bessel function of the second kind ( see , e.g. ref .
@xcite , pp .
917 ) @xmath65 for @xmath66 .
let @xmath50 be the inverse finite mellin transformation of @xmath63 , on @xmath67 , to avoid cluttered notations .
] we have @xmath68 as given by eq .
( [ kernel ] ) , the harmonic coefficient @xmath63 is finite for all @xmath69 , which ensures the existence of the finite mellin transform . substituting eq .
( [ mellin-2 ] ) into the kirkwood series eq .
( [ kirkwood ] ) , and using @xmath70 , the reaction potential can be re - expressed as @xmath71 let us further define a vector @xmath72 .
as @xmath73 decreases from @xmath74 to @xmath75 , the vector @xmath76 runs from the kelvin point @xmath77 to infinity along the radial direction .
we shall see that this is precisely the loci of the image charge line .
let @xmath78 be the magnitude of vector @xmath76 , we have @xmath79 . since we are only interested in the field point inside the cavity , we have @xmath80 , hence we can sum the series in eq .
( [ rf ] ) using the expansion eq .
( [ potentials - all-1 ] ) and obtain @xmath81 the last integral represents the potential generated by one dimensional distribution of image charges along the radial direction , which starts from the kelvin point @xmath9 and extends to infinity .
the linear charge density is @xmath82 .
the geometry is illustrated in fig .
. therefore we arrive at our main result in this work .
[ theom ] let @xmath83 be a finite mellin transform pair with the conjugate variables @xmath73 and @xmath39 , with @xmath63 given in eq . .
the green s function problem has the following image charge representation : @xmath84 where the line charge density is @xmath85 and @xmath86 .
the image charge representation for the reaction potential eq . ( [ image ] ) is equivalent to the kirkwood series representation , eq . .
it is advantageous because the line integral can be efficiently discretized by gauss quadrature .
a few gauss points can provide approximations with accuracy as high as desired @xcite .
physically , this amounts to approximating line image by a few point images .
the following limit of the bessel function can be established ( using eq .
( [ kn - aspmptotics ] ) ) @xmath87 combining with eq .
( [ kernel ] ) we obtain the large @xmath39 limit of the coefficients @xmath63 : @xmath88 the coefficients eq .
( [ kernel ] ) can therefore be decomposed into two parts @xmath89 where @xmath90 vanishes as @xmath39 goes to infinity .
the inverse mellin transform of a constant @xmath91 is a delta function @xmath92 , while the inverse mellin transform of the function @xmath90 is generally a continuous function in the interval @xmath51 $ ] .
the linear charge density therefore can be decomposed into ( with @xmath93 ) : @xmath94 the first term corresponds to a point image at the kelvin point , and the second term corresponds to a continuous line image extending from the kelvin point to infinity .
the simplest limit is @xmath95 , where the exterior of the cavity becomes a conductor . in this limit
, we easily see from eq .
( [ kernel ] ) that @xmath96 , and @xmath97 .
the image charge distribution reduces to a single point image at the kelvin point , a well known result .
when the dielectric constants inside and outside the cavity are the same , i.e. @xmath98 , the point image vanishes , but the line image persists .
this is precisely the case of multi - scale reaction field model for electrolytes .
let us review the simple case where there is no screening ions outside the cavity , i.e. @xmath99 . in 1883 , neumann @xcite found an exact expression for the reaction field , in terms of a point image charge at the kelvin point and a line ] where @xmath100 .
this formula has been re - derived independently by various authors in different fields of applications @xcite ; also see lindell s review @xcite for the summary of history .
the mellin transform method was also used by lindell and collaborators @xcite for finding image charges of the poisson equation in layered media .
the neumann s integral expression can be easily derived from the general result eq .
( [ image ] ) . for @xmath101 ,
the harmonic coefficients eq .
( [ kernel ] ) reduce to , @xmath102 its inverse mellin transform can be exactly calculated using eq .
: @xmath103 substituting this back into eq .
( [ image ] ) we find neumann s result eq .
( [ neumann ] ) .
the application of the neumann s result in molecular dynamics can be found in two recent papers @xcite .
the reciprocal problem of a source charge placed outside of a sphere was also applied in monte carlo simulations of colloidal systems @xcite .
historically , what is widely used in computer simulations of biological systems is only single image charge approximations , see works by friedman @xcite and abagyan and totrov @xcite .
these methods are of the first or second order accuracy in the dielectric ratio @xmath104 .
they fail to be accurate if the ratio @xmath105 is not small . for @xmath106 ,
the inverse mellin transform of @xmath107 can not be exactly calculated .
one way to proceed is to use the following asymptotic approximation @xcite , @xmath108 which gives the correct leading order asymptotics both for @xmath109 and for @xmath110 . substituting it back into eq .
( [ kernel ] ) leads to @xmath111 where @xmath112 .
the inverse mellin transform of this can be easily found .
this approximation works well for the case of small @xmath67 ( small cavity ) and large @xmath67 ( large cavity ) .
let us first look at the large @xmath39 limit of the bessel function @xmath113 . for sufficiently large @xmath39 , the largest term in the sum eq .
( [ knu - def ] ) is given by @xmath114 .
we can therefore rewrite the summation as @xmath115 where we have defined @xmath116 .
we can expand the function being summed into asymptotic series in terms of @xmath117 and extend the upper limit of summation from @xmath39 to @xmath118 .
the resulting summation then can be calculated order by order in @xmath117 .
this can be conveniently done using wolfram mathematica .
for example , up to order of @xmath119 , we have @xmath120 it then follows that @xmath121 generically , therefore , the leading order term of @xmath63 scales as @xmath117 . in the most interesting case @xmath122 , however , this term vanishes and we have @xmath123 now consider the limit where the source charge approaches the cavity boundary , we have @xmath124 .
the reaction potential acting on the source charge then ( see eq .
( [ kirkwood ] ) ) becomes @xmath125 therefore the line image strength must vanish at the kelvin point . in the multi - scale reaction field model , we typically have @xmath126 and @xmath127 of order of unity . all methods discussed above
fail in this case . to obtain accurate approximation , we approximate the harmonic coefficients @xmath63 by a rational function of @xmath39 ( i.e. _ pad approximation _ ) : @xmath128 with constants @xmath129 and @xmath130 for all @xmath28 .
clearly eq .
is a special case of eq .
( [ rational - app ] ) with @xmath131 .
the second term on the right hand side is the order @xmath132 $ ] pad approximant and @xmath133 refer to the degree of polynomials in the numerator and in the denominator respectively .
] of the function @xmath90 , and the rational polynomial preserves the asymptotics , @xmath134 the coefficients @xmath135 are solved by the nonlinear least square method . for given @xmath67 and dielectric ratio @xmath105 ,
the constants in the expansion can be simply determined by a minimization of the total @xmath136 error of the first @xmath137 terms of the harmonic coefficients , @xmath138 ^ 2 .
\label{minim}\ ] ] the newton iteration scheme or other iteration algorithms can be applied to solve this nonlinear optimization problem . to demonstrate the quality of pad approximation
, we list in table [ error1 ] the relative errors for the cases of @xmath139 and 3 with various parameters @xmath67 and @xmath105 : @xmath140 ^ 2\left/\sum_{n=0}^n m_n(u)^2\right.,\ ] ] where @xmath141 is the numerical solution of the nonlinear least square problem . we take @xmath142 since 51 multipoles in the kirkwood series already provide sufficiently high accuracy .
it should be pointed out that the minimization solution depends on the initials and may be not a global minimization .
nonetheless , the results in table [ error1 ] clearly show the remarkable precision of the pad approximation .
for example , the worst case for @xmath143 has a relative error @xmath144 . in comparison , the asymptotic method eq .
( [ xdc ] ) yields much larger error , and completely breaks down for @xmath145 .
.relative errors of pad approximations to the harmonic coefficients for different parameters @xmath67 and @xmath105 . in the minimization
, @xmath142 is taken . the asymptotic solution eq .
( [ xdc ] ) @xcite is also given for comparison . [
cols="^,^,^,^,^",options="header " , ] to illustrate the utility of our image charge method , we apply it in a reaction - field monte carlo simulation of electrolyte .
we artificially introduce a spherical cavity with radius @xmath2 in the electrolyte , and model all charges in the cavity using the primitive model and simulate them using monte carlo method .
all ions outside the cavity , together with the solvent , are treated implicitly using linearized pb theory , characterized by two parameters : the dielectric constant @xmath146 and the inverse debye length @xmath1 . for any ion inside the cavity , the effects of the electrolyte outside the cavity is to introduce a reaction potential , which can be calculated using our image charge method .
it is important to note that the inverse debye length @xmath1 characterizing the medium outside the cavity shall be self - consistently determined by the simulation of the ions inside the cavity .
we run canonical monte carlo simulations of the primitive model using the standard metropolis criterion @xcite for particle displacements .
the ions are modeled by hard spheres with diameter @xmath147 and with a point charge of valence @xmath148 at its center .
the ions are mobile in the solvent medium with dielectric permittivity @xmath149 .
the effective hamiltonian of the system is given by @xmath150 the self energy @xmath151 is given by @xmath152 where @xmath153 is the bjerrum length and @xmath154 is the boltzmann factor .
we take @xmath155 for the water permittivity at room temperature .
the infinite potential is due to the presence of the hard wall at @xmath156 the pairwise interaction energy @xmath157 is given by @xmath158,\ ] ] as is well known , the reaction potential is symmetric under permutation of two variables : @xmath159 .
we first calculate the density distribution by taking a system with @xmath160 and the salt concentration is @xmath161 which corresponds to a debye length @xmath162 .
we calculate both the cases with and without the reaction field . in the former case ,
we use different numbers of image charges varying from 4 to 6 , with the parameters for numerical quadratures the same as those in the upper panel of figure [ accuracy ] . for each setting
, we run @xmath163 mc cycles for each particle to obtain samples for the statistics of particle number in each spherical shell with thickness @xmath164 . in figure [ rdf ]
we show the radial distribution function of anions , i.e. , the normalized density , @xmath165 .
the distribution of cations is similar .
evidently , without accounting for the reaction field , the density is higher near the center of cavity and lower near the cavity boundary .
the difference in the density is up to 3 percent .
when the reaction field is taken into account , the particle density shows variation less than @xmath166 in most regions inside the cavity .
there is substantial deviation of density near the cavity boundary .
this is due to the presence of the artificial hard wall on the boundary , see eq .
( [ self - energy - hamiltonian ] ) . this problem can be cured by adding to the self energy a short ranged correction , or by introducing a buffer zone of a certain thichness ( @xmath167 ) .
the latter approach has been discussed in literature @xcite .
we shall present statistical mechanical discussion of the former approach in a separate publication .
finally , in figure [ rdf2 ] , we plot the results for three different bulk ionic concentrations , where @xmath168 and @xmath169 respectively .
these results again show the necessity of treating the reaction potential , as well as the accuracy and efficiency of our image charge methods .
it is also interesting to note that as the debye length decreases the density becomes flatter , suggesting that the reaction - field monte carlo models becomes more precise . for the case of salt concentration @xmath170 , for example ,
the debye length is approximately half of the cavity radius , while the density variation becomes less than @xmath171 ( excluding the thin region affected by the hard wall artifacts ) .
respectively . the corresponding debye lengths are @xmath172 and @xmath173 respectively .
top panel : without reaction field ; bottom panel : with reaction field . for computation of the reaction potential , @xmath174 and @xmath175 are adopted.,title="fig : " ] respectively .
the corresponding debye lengths are @xmath172 and @xmath173 respectively .
top panel : without reaction field ; bottom panel : with reaction field . for computation of the reaction potential , @xmath174 and @xmath175 are adopted.,title="fig : " ]
in summary , we have developed an image method for charges inside a spherical cavity that is immersed in an ionic solution .
our method is useful for multiscale reaction field models of electrolytes and other more complicated charged systems .
we derive an analytic expression for the reaction potential in terms of one dimensional image charge distribution , and discuss a highly accurate and efficient algorithm for discretizing the image line charge .
we also apply our method to a reaction field monte carlo simulations of @xmath176 electrolytes . and demonstrate the accuracy and efficiency of the new algorithm .
simulation of charged systems is computationally expensive , therefore is always limited to small system size . in a physical system of such size
, the total charge may fluctuate away from zero to a noticeable extent .
these fluctuations can not be taken into account in canonical ensemble simulations . in another word ,
grand canonical ensemble must be used to capture the charge fluctuations of small systems .
ewald - summation method , which is so far the most popular simulation methods for charged systems , are based on periodic boundary conditions , and are difficult to be incorporated with grand canonical ensemble . by strong contrast , reaction - field type of models , besides being more intuitive , can be easily adapted to a grand canonical monte carlo simulation .
this shall be the topic of a separate publication .
the authors acknowledge the financial support from the natural science foundation of china ( grant numbers : 11101276 , 11174196 , and 91130012 ) and chinese ministry of education ( ncet-09 - 0556 ) .
z. x. acknowledges the financial support from the alexander von humboldt foundation for a research stay at the institute for computational physics , university of stuttgart .
the authors thank professors wei cai and chunjing xie for helpful discussion .
|
we revisit the image charge method for the green s function problem of the poisson - boltzmann equation for a dielectric sphere immersed in ionic solutions . using finite mellin transformation ,
we represent the reaction potential due to a source charge inside the sphere in terms of one dimensional distribution of image charges .
the image charges are generically composed of a point image at the kelvin point and a line image extending from the kelvin point to infinity with an oscillatory line charge strength .
we further develop an efficient and accurate algorithm for discretization of the line image using pad approximation and finite fraction expansion .
finally we illustrate the power of our method by applying it in a multiscale reaction - field monte carlo simulation of monovalent electrolytes .
poisson - boltzmann equation , finite mellin transform , green s function , multiscale reaction field model , fast algorithm , pad approximation , finite fraction expansion 35j08 ; 65r10 ; 78a35 ; 82d15
|
we have used the hartree - fock random phase approximation ( hf - rpa ) to study the interacting electron gas in a quantum wire .
the spectra of intersubband spin - flip excitations reveal a considerable red shift with respect to single - particle hf energies . that signals on appearance of collective intersubband spind - density excitations due to the exchange interaction .
the long wavelength dispersions of the intrasubband collective spin - density excitations are linear , but the sound velocities are renormalised due to the exchange interaction and screening
. the in - phase intrasubband charge - density excitation has the long wavelength form @xmath0^{1/2}$ ] .
we found good qualitative agreement of our results with experimental observations .
a semiconductor quantum wire can be fabricated by applying a voltage with a microstructured gate to a 2d electron gas .
the single - particle energy ( spe ) spectrum typically consists of subbands separated by several @xmath1 . at a 1d electron density about @xmath2@xmath3
more than one subband can be occupied . in recent years progress has been made in spectroscopic study of such systems . in angular
resolved raman spectra of gaas quantum wires@xcite , collective spin - density excitations ( sde ) and charge - density excitations ( cde , or plasmons ) were observed .
the measured spectra cover from low to high frequency , and for low - frequency intrasubband excitations , the wave - vector dependence of the spin - wave energy was found to be linear .
the correct interpretation of these experiments allows us not only to understand the interesting physical processes , but also to access important physical parameters .
the exchange interaction is crucial to the collective intrasubband and intersubband sde .
similar to the direct long range coulomb interaction which leads to the depolarisation shift of single - particle excitations and to the appearance of collective plasma modes , exchange interaction gives rise to the red - shift of spe . if the red - shift is sufficiently large , the collective sde with a sizable oscillator strength splits off the continuum of spe . for semiconductor quantum wells
, such split - off appears in the hartree - fock random phase approximation ( hf - rpa)@xcite .
thus , it is important to perform a hf - rpa analysis on collective electron excitations in a realistic gaas quantum wire with full coulomb interaction , and compare the results with measured spectra@xcite .
the selfconsistent and conserving@xcite hf - rpa is suitable for this task , because we will calculate the two - particle spectra but not single - particle properties . as in using any approximation
, the hf - rpa calculation also contains error . however , our hf - rpa results agree very well with experimental measurements . in this letter we will first outline the hf - rpa method to express the spin or charge correlation functions in terms of the corresponding spin- or charge - density induced matrix , which satisfies an eigenvalue matrix equation . in a realistic quantum wire , for high energy intrasubband excitations and for excitation spectra when more than one subband is occupied , the self - consistent equations have to be solved numerically .
however , for the case that only one subband is occupied , the analytical expressions of sde and cde are derived . with additional approximation applied to the hf - rpa analytical solutions
, they reduce to known results .
this may be coincidental , but is outside the central theme of this letter . to define the quantum wire , let us start from an electron gas in a narrow quantum well with interfaces parallel to the @xmath4-@xmath5 plane , and the well width shorter than any other relevant length scales .
we consider the realistic experimental situation that only the lowest subband in the quantum well is occupied , so the motion of electrons in the @xmath6 direction can be ignored . by applying a properly designed gate potential ,
a quantum wire along the @xmath4 axis is fabricated with a longitudinal constant electrostatic potential , but a transverse parabolic one @xmath7/2 along the @xmath5 axis , where @xmath8 is the electronic effective mass .
the quantum wire has a finite length @xmath9 , and we assume periodic boundary conditions . in the absence of electron - electron interaction , the single particle eigenfunctions are simply @xmath10=@xmath11 , with the corresponding eigenenergies @xmath12=@xmath13+@xmath14+@xmath15 , where @xmath16 is the subband index , @xmath17=@xmath18 , and @xmath19 the harmonic oscillator wave function .
the transverse confinement length of the electrons is @xmath20=@xmath21^{1/2}$ ] .
when the electron - electron interaction is turned on , we will use the hf approximation @xmath22 to derive a complete orthonormal basis @xmath23 of quasi - particle hartree - fock states , where @xmath24 is the hartree - fock energy , @xmath25 is the fermi occupation factor and @xmath26 the dielectric constant of the surrounding medium . in terms of this basis set , we will use the corresponding time dependent hf ( hf - rpa ) to calculate the charge - density and the spin - density correlation functions , which describe the self - consistent linear response of the electron gas to an external perturbation . in this letter
, we restrict ourselves to the nonmagnetic ground state . for inelastic light scattering not close to the band gap resonance , the raman intensities in polarised and depolarised scattering geometries are proportional , respectively , to the imaginary parts of the charge - charge correlation function and spin - spin correlation function@xcite .
if we define @xmath27 as the fourier transform of the charge - density operator and @xmath28 the fourier transform of the spin - density operator along the spin quantisation axis , then , these correlation functions are @xmath29 | > $ ] with @xmath30=@xmath31 , where @xmath32 denotes a thermodynamic average . in hf - rpa , they can be expressed as @xmath33 in terms of the @xmath30-density induced matrix @xmath34 .
the charge - density induced matrix satisfies @xmath35 \ , , \label{indcharge}\end{aligned}\ ] ] where @xmath36=@xmath37 is the coulomb matrix element in the basis of the hf single - particle states .
when @xmath30=@xmath38 , there is no direct coulomb interaction ( hartree term ) between the excited spins .
if the exchange term ( fock term ) is neglected , the spin - density excitation spectra are simply given by the quasi - particle hartree - fock energies . in hf - rpa ,
the spin - density excitations are shifted from the quasi - particle hartree - fock energies due to the exchange terms . as a result
, collective spin - density excitation may appear in the spectra .
this shift is a measure of the effective strength of the exchange term in the system .
we will use the above hf - rpa analysis to investigate collective excitations in a multisubband system .
this has to be done numerically , and the first step is to derive the quasiparticle basis set @xmath23 by solving ( [ hfground ] ) self - consistently via iteration
. the functional basis must be sufficient large in order to ensure the required accuracy for the quasiparticle energy spectra and the spatial electron density of the ground state . when using this self - consistent quasiparticle basis set to calculate collective excitations , the required accuracy for the peak position of an excitation energy is the experimental linewidth @xmath390.1 @xmath1 .
the intrinsic materials parameters of a gaas quantum wire required in our calculation are @xmath8=@xmath40 and @xmath26=12.4 , where @xmath41 is the electron mass .
the extrinsic sample parameters are the total electron density @xmath16 , the transverse potential @xmath42 , and the length of the quantum wire @xmath9 .
our calculations are based on a sample with @xmath16=10.4@xmath43@xmath44 @xmath45 , which has been investigated experimentally in ref .
@xcite .
the transverse potential @xmath42 is determined by the experimental conditions . since the experimental sample does not have a parabolic potential as can be seen from the intersubband cde
, we will adjust @xmath46 to fit the subband electron occupations .
when we set @xmath46=7.9 @xmath1 ( @xmath20=120 ) , the derived self - consistent hartree - fock subband spacing between the two lowest subbands is 5.4 @xmath1 at the zone center and 5.6 @xmath1 at the fermi level .
the electron densities in the two occupied subbands are @xmath47=6.4@xmath43@xmath44 @xmath45 for the lowest subband ( labeled by @xmath48=0 ) , and @xmath49=4.0@xmath43@xmath44 @xmath45 for the second lowest subband ( labeled by @xmath48=1 ) .
the physical properties of the quantum wire are insensitive to its length , provided that the wire is sufficiently long .
therefore , we set @xmath9=1.0 @xmath50 m .
the experiment@xcite was done at a temperature 1.7 k , which is the temperature used in our calculation .
let @xmath51 and @xmath52 be , respectively , the fermi wave - vector and the fermi velocity of the @xmath48th subband . in the limit of long wavelength @xmath53@xmath540 , the transverse excitation is forbidden ( or allowed ) if the excitation energy is low ( or high ) , and so the corresponding physical processes are dominated by intrasubband ( or intersubband ) transitions .
when @xmath53 increases to @xmath55-@xmath56 , the low energy intersubband transitions are activated .
however , collective excitations with large energy and/or large @xmath53 are strongly damped . to illustrate the main features of long wavelength excitations ,
let us consider a very simply case by neglecting the electron self - energies and the vertex correction .
in this case we obtain analytical results for both the sde and cde energies .
the sde dispersion is linear in @xmath53 , with sound velocities @xmath57 and @xmath58 for intrasubband excitations in the two lowest subbands
. with two subbands occupied , the cde dispersion has two branches .
the in - phase mode is @xmath59 where @xmath60 is the fourier transform of the coulomb potential @xmath61-@xmath62+@xmath63-@xmath64^{-1/2}$ ] , and the out - of - phase mode is @xmath65 such two - band sde and cde dispersions are exactly the same results as schulz@xcite obtained for a two - subband tomonaga - luttinger model ( tlm)@xcite .
when we turn on the intrasubband and intersubband exchange interaction , as well as the screening , the sde and cde energies will be modified from the above expressions , and have to be derived numerically . in the polarised spectrum , the higher energy region
is dominated by strong intersubband cde . at @xmath53=0 ,
the calculated cde spectrum consists of a single peak which is characteristic to transverse parabolic confinement potentials@xcite . in the region of finite @xmath53
the intersubband hf spe energies show up at 5.5 @xmath1 . in the spin - flip
depolarised raman spectrum , we obtained the red shift of intersubband sde with respect to hf spe .
the sde has a dominant weight of the spectrum and appears at a resonance energy of 2.1 @xmath1 , which is shifted from the hf spe , indicating the importance of vertex correction .
these overall features of our numerical results have been observed experimentally@xcite . in fig .
[ cdesde ] we show the calculated intrasubband sde and cde in the region of energy and frequency for which accurate experimental data are available .
the mode with strong intensity is labeled as sde0 , and the next mode with weaker intensity is labeled as sde1 .
the inset gives their dispersions , which are linear for small @xmath53 , with the corresponding sound velocities @xmath66=0.9@xmath57 and @xmath67=0.7@xmath58 .
these sound velocities are smaller than the respective fermi velocities because of the exchange screening of sde induced by the intersubband virtual transitions , and the renormalisation due to the intra- and intersubband exchange interaction .
the experiments@xcite seem to have detected only @xmath66 , which corresponds to the high energy sde near the fermi wave vector @xmath55 of the lowest subband .
it is probably the low electron density in the second lowest suband that causes the velocity @xmath67 of the sde around @xmath56 to be too small to be observed experimentally .
the spin velocity @xmath66 agrees quantitatively with the experiment@xcite . for sufficiently large @xmath53 , because of the finite length of our quantum wire , the landau damping shows up in fig .
[ cdesde ] in the form of enhanced intensities of satellite single - particle peaks around the sde , instead of a broad band if the quantum wire is infinitely long .
the mode sde1 is not damped , since it does not enter the region of spes .
the dispersions of the cde and sde are also plotted in the inset in fig .
[ cdesde ] . in the long wavelength limit
, the in - phase cde+ can be well fitted with @xmath0^{-1/2}$ ] according to ( [ 2charge+ ] ) , as expected for 1d electron gas@xcite .
the out - of - phase cde- mode , corresponding to ( [ 2charge- ] ) with an expected linear dispersion@xcite , appears in our calculation as a weak band with a sound velocity @xmath68 .
this happens to be the same as the sound velocity of the sde0 , an accidental result for this specific sample .
the cde- mode has a much lower intensity compared to the cde+ mode .
checking against the experiment@xcite , we believe the the lower energy band observed in the low frequency polarised raman spectrum is our calculated cde- mode . the decay of the plasmon ( cde+ ) and the cde- at higher @xmath53 due to the landau damping is stronger than the decay of sde0 and sde1 , as shown in fig .
[ cdesde ] . after the above complete study of a realistic quantum wire , perhaps it is worthwhile to mention some surprising findings for the special case that only the lowest subband occupied at zero temperature , namely , the pure 1d system including spin degrees of freedom . in this case we can drop the band index . for sde , the eigenvalue equation has the form @xmath69 where @xmath70=@xmath71 for the quasi-1d coulomb potential @xmath61-@xmath62+@xmath72^{-1/2}$ ] , which has a logarithmic singularity @xmath73 as @xmath53@xmath540 . since the hartree self - energy is canceled by the potential energy due to the positive background charges , the difference in exchange self - energy at the fermi energy , @xmath74=@xmath75+@xmath76-@xmath75-@xmath76 is @xmath77 $ ] .
we define @xmath78 by @xmath79@xmath80@xmath81-@xmath82 , and at @xmath17=@xmath55 rewrite ( [ sdeone ] ) as @xmath83 \nonumber \\ & & = 2\pi\left [ \hbar\omega -\hbar qv_0 - \delta_x(q)\right ] y(k_0 ) \ , .
\label{add1}\end{aligned}\ ] ] in the limit of long wavelength , by substituting @xmath84@xmath39@xmath85 and @xmath86@xmath39@xmath87 into the above equation , we see that the contributions of coulomb potential in the exchange energy and the vertex corrections , both behave as @xmath88 , exactly cancelling each other .
hence , the @xmath88 behaviour of low - energy single - particle excitation due to the exchange self - energy is removed from the sde and the plasmon energies , and so only the exchange terms at momentum @xmath89 are important . an equation similar to ( [ add1 ] ) for @xmath90 at @xmath91 can also be derived . in the long wavelength limit
, we find the sde energy @xmath92 where @xmath93=@xmath94 .
the sound velocity is reduced with respect to the bare fermi velocity .
we have also solved for the cde and found the plasmon dispersion @xmath95 for a long - range coulomb potential @xmath60@xmath96@xmath97 , the plasmon energy is not affected by the exchange interaction within the hf - rpa .
these results ( [ 1spin ] ) and ( [ 1charge ] ) , derived with full coulomb interaction , differ from those obtained by schulz@xcite for the tlm including a nonsingular backscattering matrix element at @xmath89 , which mixes the right- and the left - traveling modes . nevertheless , in ( [ sdeone ] ) , if we replace the hartree - fock energy by the bare single - particle energy , and omit the singular part around v(q=0 ) in the vertex correction , then our results exactly reduce to schulz s results for the sde and the plasmon energies .
whether this finding , as well as ( [ 2charge+ ] ) and ( [ 2charge- ] ) , is simply a coincidence remains to be clarified . to close this letter , we should mention that according to our analysis , the frequency of the lowest energy sde decreases as @xmath53 increases towards @xmath56 , and then vanishes at a value of @xmath53 close to @xmath98 .
this behaviour suggests a low temperature intrinsic instability of the electron gas against the formation of spin - density waves ( sdw ) , namely , the peierls instability . when this instability emerges from our mean - field analysis
, we must take into account the sdw long - range order in the ground state .
on the other hand , the results of the tlm with backscattering predict the absence of sdw long - range order , but instead the appearance of a slowly decaying wigner crystal at @xmath99@xcite .
we guess that the tlm with backscattering describes the real ground state better than hf - rpa .
hence , we believe that our analysis based on a ground state without sdw long range order is a better approximation than that including sdw in the ground state .
a. b. would like to thank a. sudb and l. j. sham for stimulating discussions .
this research was supported in part by a norfa grant .
|
|
a two - level system ( 2ls ) interacting with a quantum field electromagnetic field ( emf ) in particular has proven to be a very useful model for a wide range of problems from atomic - optical @xcite and condensed matter @xcite processes to quantum computation @xcite . for the latter application stringent limits in maintaining the coherence of the the 2ls ( called qubits )
are required .
this prompted us to revisit the theoretical structure of the 2ls model , paying special attention to its coherence properties .
treatment of spontaneous emission and relaxation are standard textbook topics , whereas decoherence and dissipation , especially in the context of quantum computation , are the focus of more recent investigations @xcite .
because of the familiarity of the model ( see ii.a ) and its theoretical and practical values , we do not need to emphasize the general motivation , but can go right to the point about the aim and results of this paper .
the description of this system generally comprises of two parts : 1 ) spontaneous emission in the 2ls , and 2 ) decoherence due to the interaction of 2ls with the em field , treated as a bath .
the first part allows little room for disagreement , as it can be obtained from elementary calculations .
the second part on decoherence is more subtle .
environment - induced decoherence @xcite has been studied extensively in recent years primarily based on models of quantum brownian motion ( qbm ) @xcite for the interaction of a simple harmonic oscillator ( brownian particle ) with a harmonic oscillator bath ( hob ) at a finite temperature , leading to a reasonably good understanding of its characteristic features .
decoherence of a 2ls in an em field has been studied by a number of authors , notably @xcite , and their dissipative and decoherent behavior are reported to be similar to that of a qbm in a harmonic oscillator bath .
the progression in three stages quiescent , vacuum fluctuation - dominated and thermal fluctuation - dominated , separated by the cutoff frequency and the thermal de broglie frequency ( wavelength ) are indeed characteristic of the qbm results @xcite . our findings , in contrast , are in stark disagreement from that reported in the literature .
we work with the standard 2ls - emf model @xcite and obtain an exact master equation for depicting non - markovian dynamics .
solution of this equation for the reduced density matrix of the 2ls shows that the decoherence rate is close to the relaxation rate .
this is in first appearance rather couter - intuitive , and different from all previous findings . upon careful deliberation
we realize that the ` intuition ' researchers ( including us at the start ) have acquired for dissipation and decoherence are based on the qbm model which influenced the choice of model in the investigation of decoherence for a 2ls .
however , we find that such a commonly invoked intuition for qbm in a hob fails to apply to that of a two - level atom ( 2la ) interacting with an electromagnetic field ( emf ) with the commonly assumed type of resonance coupling in quantum optics .
+ _ decoherence in qbm
_ + physically , when we say that decoherence of the system of a brownian oscillator proceeds in a very short time as it is brought in contact with an environment , a hob at some temperture , we are usually conjuring a model with bilinear @xcite ( or polynomial @xcite ) coupling of the oscillator - bath coordinates , and a ohmic or subohmic spectral function @xcite in the bath .
intuitively , the bath needs to have many degrees of freedom , preferably acting independently of each other so incooperatively that the phase information in the system will be dispersed to the largest extent amongst the many bath degrees of freedom and affords little chance or takes inordinately long time to be revived or reconstituted ( recoherence @xcite ) . the opposite picture ( of very long decoherence time )
is exemplified by two coupled subsystems where no coarse - graining is introduced , or for system - environment couplings which maintain some high level of coherence , or for an environment whose degrees of freedom have long correlation times like in a zero temperature , supraohmic bath .
the case of a ( spin ) particle or ( plasma ) wave interacting with an averaged ( collective ) variable from the environment , such as the mean field , showing landau damping in vlasov dynamics is another example @xcite . just as in the spin echo phenomena ( e.g , chap . 3 @xcite ) , the basic physics in this case is not dissipation in the boltzmann sense , but statistical mixing @xcite .
we will see that this example is of more physical relevance to our problem than the qbm .
+ _ coherence in the 2ls _ + for the 2la - emf system , one clear distinction between an em field as an environment and a system of harmonic oscillators as bath is that the field ( coupled to a detector ) has an intrinsic spectral density function , not to be chosen arbitrarily .
for example , it has been shown @xcite that a conformal scalar field in two dimensions coupled to a monopole detector has an ohmic character while in four dimensions it is supraohmic .
barone and caldeira @xcite showed that the spectral density function for em fields with momentum coupling to an oscillator detector is supraohmic .
these density functions would show very different decoherence behavior from the high temperature ohmic hob used in many discussions of decoherence , the latter case is what the general folklore is based on .
but the most important distinction from qbm is that the 2la couples with the emf in the discrete number basis for the field , unlike the continuous amplitude basis in the qbm .
this fact ( which is true in the rotating wave and dipole approximation ) implies that the 2ls plus emf system is a _ resonant _ one .
hence even though the em field has just as many ( large number of ) modes as the hob , only a very small fraction of them in a narrow range of the resonance frequency are efficiently coupled to the atom .
this is the root cause for the very different qualitative behaviour between the qbm and the 2ls as far as decoherence is concerned .
one extreme case is that of a single mode field described by the jaynes - cummings model , where rabi nutation takes place and the atom - field remains largely a coherent system .
( for a coherent field , the probability for the atom to be found in the excited state at time t regardless of the state of the field ) obeys a poisson distribution .
this distribution in the photon number induces a spread in the rabi frequencies , and causes collapse and revival of the rabi nutation .
these are distinct features of quantum coherence @xcite . ) adding all modes to the field we see spontaneous emission and the decay of the atom .
the probability of an initially excited atom ( remaining in the excited state ) decays exponentially in the wigner - weisskopf form ( characteristic of markovian processes ) with relaxation time constant @xmath1 . for purely radiative decay
the decay time @xmath2 of the inversion is half the decay time @xmath3 of the polarization .
there is no large order of magnitude differences between dissipation and decoherence time ( which in typical qbm high temperature conditions could be as high as 40 @xcite ) .
in fact it is perhaps inappropriate to talk about dissipation for a 2la - emf system because the conditions for a bath to actuate such a process is lacking .
the transition from excited to ground state is closer in nature to relaxation ( in the spin echo sense ) than dissipation . in a cavity where excitation of the atom from the field ( absorption ) balances with emission , it is more appropriate to refer to the resonant state of the atom - field as a coherent system . in these senarios
the distinction between qbm and 2la can not be clearer .
+ _ difference between qbm and 2ls _ + so what led earlier authors to make the qualitative claim that 2ls decoheres easily ?
we think the confusion arises when the picture of qbm dissipation and decoherence is grafted on the 2la - emf system indiscriminantly .
if the field which acts as the environment is a phonon field ( from ion vibrations , see , e.g. , @xcite ) and if the coupling is of the non - resonant type , then there is no disagreement .
decoherence should follow the qbm pattern as reported by many authors .
such sources ( including atomic collisions in a cavity @xcite ) can be important for some setups .
however , when one claims that the em field can decohere a 2ls ( with which it is coupled in a resonant way , as in the standard model ) that is where we disagree .
quantitatively , the model for the 2ls used by most authors for the discussion of decoherence inspired by qbm type of behavior has the atom in a @xmath4 state ( the diagonal pauli matrix ) coupled to the field mode operators @xmath5 .
this type of coupling term ( call it @xmath4 type for convenience ) commutes with the hamiltonian of the system , and admits a diagonalization in the eigenbasis of the hamiltonian .
the field is coupled to the atom as a whole and thus is insensitive to the 2 level transition activity . in particular it does not probe the resonance or coherent properties of the two level atom , which is the most important feature , for quantum computation .
by contrast the standard model for 2la - emf which we studied has a @xmath6 coupling ( call it standard coupling ) to the field modes which highlights the 2 level activity of the atom and the field .
this coupling considered in the standard model is indispensible , i.e. , it _ can not _ be removed from the two - level atom as it _
defines _ it and will be present in any realistic situation .
what then is the origin of the qbm type of contribution to the 2ls ?
if one accepts an environment other than the em field , the question comes down to the characteristics of the experimental apparatus . for well-
prepared ion traps we would expect it to be rather unimportant . if the em field is the only field present , we can still ask if a qbm type of coupling term with the em field would appear , and if yes , how strong would its effect be ? this would be a useful way to accomodate the two different types of coupling terms .
recall that the standard model is derived under the dipole and rotating wave approximation . in the next section
we will show that the @xmath4 type of coupling appears only in the next order expansion after the dipole approximation .
since these are good approximations for a large class of atomic states when the atom is nonrelativistic , the contribution from the qbm type of coupling used in @xcite should be negligible and its ensuing decoherent effect insignificant . in this sense the em field does not in leading order of approximation act like a bath in the qbm way , and coherence in a 2la - emf system
is quite well preserved ( excepting other processes , e.g. @xcite ) .
our puzzle over the result on decoherence in the 2ls reported in the literature was what prompted us to begin this study . without letting any familiar and convenient analogy influence our judgement , and without any preconceived notion , we choose to perform a first - principles calculation of the two - level atom ( 2la)-electromagnetic field ( emf ) system making as few assumptions and covering as wide a range of conditions as possible .
we use the influence functional method @xcite to take into account the full backreaction of the field on the atom , while adopting grassmannian variables for the 2la and the coherent state representation for the emf .
we find exact master equations for the full ( non - markovian ) dynamics in the cases of a free quantum field and a cavity field at zero temperature . in the next section we present the model and the formalism . a detailed derivation of our model is contained in appendix a. in sec .
3 we derive the master equations . in sec
4 we study different mode composition of the field , including that of an atom in a cavity .
we end in sec . 5 with a discussion of our findings and their implications .
this is the first in a series of papers on 2latom and quantum decoherence .
the subsequent papers will treat 2la- emf interaction at finite temperature , for em fields in a coherent and squeezed state , and for multipolar models ( where coupling other than the minimal is assumed ) .
we will also tend to collective qubit systems and moving atoms interacting with an em field .
these results will have corresponding applications in atom optics and quantum computation problems .
our model for atom - field interaction is the standard one ( see appendix a for details ) @xcite . the total hamiltonian for a ( stationary ) atom interacting with a quantum electromagnetic field ( emf ) under the dipole , rotating wave ( rw ) and two - level ( 2l ) approximation is given by = _0 s_z + _ where @xmath7 are the creation and annihilation operators for the kth normal mode with frequency @xmath8 of the electromagnetic field ( thus for the field vacuum @xmath9 = \delta_{\bk,\bk'}$ ] , for all @xmath10 . ) , and @xmath11 is the frequency between the two levels .
here @xmath12 where @xmath13 are the standard 2x2 pauli matrices with @xmath14 , etc .
the coupling constant @xmath15 where @xmath16 and @xmath17 is the dipole matrix element between the eigenfunctions @xmath18 of the electron - field system , @xmath19 is the unit polarization vector ( @xmath20 are the two polarizations ) , and @xmath21 is the spatial mode functions of the vector potential of the electromagnetic field ( in free space , @xmath22 , @xmath23 is the volume of space . ) . under the dipole approximation @xmath24
is evaluated at the position of the atom @xmath25 . since @xmath26 , @xmath27 , we will choose a mode function representation such that @xmath28 is real . to see how this
could possibly be related to the @xmath4 type of coupling with hamiltonian ( used by e.g. , @xcite for the study of decoherence in 2ls ) = _0 s_z + _ we examine the next term after the dipole approximation in ( a.15 ) .
this has a contribution to @xmath29 even when @xmath30 which is equal to @xmath31where @xmath32and @xmath33 is a constant given by @xmath34this generates an additional coupling term @xmath35where @xmath36this gives the lowest order @xmath4 type of coupling in a 2la -emf system .
the ratio of the coupling @xmath37 of the @xmath4 type in eq .
( ii.3)to the dipole coupling @xmath28 in eq .
( ii.1 ) is @xmath38 thus the @xmath4 type of coupling generated from the 2la- emf interaction will be significant only for very high frequencies @xmath39 of the em field , a point intuitively clear from the meaning of the dipole approximation .
+ since feynman and vernon @xcite invented the influence functional method this formalism has been applied to treat the brownian motion of a harmonic oscillator interacting with a harmonic oscillator bath by many authors @xcite .
the two - level system in tunneling has been discussed in detail by leggett et al @xcite , but the derivation of a master equation by this method which can traverse the non markovian regimes has not yet been carried out .
we shall perform such a calculation for a two level system , with the aid of grassmaninn variables convenient for treating fermions , and the coherent state representation in a path integral form .
we construct the coherent state of the combined atom - field system as @xmath40 where @xmath41 , @xmath42 a complex number , denotes the em field coherent states and @xmath43 , @xmath44 a grassmannian or anticommuting number , denotes the electron coherent state . the transition amplitude between the initial state ( i ) at @xmath45 and the final state ( f ) at @xmath46
is expressed formally as @xcite ( here we suppress the index @xmath47 , @xmath48}\ ] ] where the action is @xmath49=\bar{z}z(t)+\bar{\eta } \eta ( t)-\int_{0}^{t}ds(\bar{z}\dot{z}+\bar{\eta } \dot{\eta } + \frac{i}{\hbar } \hspace{0.1cm}h(\bar{\eta } , \eta , \bar{z},z)\ ] ] here @xmath50 is the q - symbol of the hamiltonian @xcite and there is an implied summation over field modes .
@xmath51 in ( ii.8 ) we have substracted a constant term @xmath52 to ( ii.1 ) so that the ground state now has zero energy .
henceforth we set @xmath53 .
the hamiltonian in equation ( 2.8 ) is not a c - number function ; it has terms that are odd .
one might then question the validity of equation ( 2.6 ) for the path integral ; it clearly exists as a formal expression , but its evaluation with a saddle point method , that is based on the hamiltonian of equation ( 2.8 ) might be problematic .
we dispell this doubt for the vacuum case with an operaytor method proof of the master equation in appendix b. it shows that at least for the vacuum case tha saddle point evaluation yields the correct result .
the general cases need separate considerations . for many qubits coupled to the em field vacuum
, we believe that the path integral method yields a simpler treatment than the operator method . the integration is over all paths satisfying @xmath54 we assume initially that the density matrix of the total system+environment is factorizable @xmath55 . only at that time would @xmath42 and @xmath44 be pure complex and grassmannian numbers respectively .
as the system evolves , both @xmath56 and @xmath42 contain grassmann and c - number parts .
the mixing of even and odd parts ( note @xmath57 is odd ) comes about as the initially factorized atom state becomes `` dressed '' . in the open system philosophy ,
as we are interested in the averaged effect of the field on the atom , the atom is considered as the ` system ' while the field as the ` environment ' .
the path integral is performed over the variables @xmath42 , while @xmath58 are treated as external sources . when only one field mode is considered , we have |z_f;t | z_i;0_,| = dz d|z\ { satisfying the boundary condition ( ii.9 ) for @xmath42 .
we use the saddle point method . minimizing the action yields the following equations
@xmath59 with solutions @xmath60 using these for the transition amplitude ( ii.11 ) with the minimum value for the action , we obtain @xmath61 -g ^{2}\int_{0}^{t}ds\int_{0}^{s}ds^{\prime } e^{-i\omega a prefactor in the coherent state path integral is equal to one .
now the influence functional due to this single mode reads @xmath62 & = & \int \frac{d\bar{z}_{i}dz_{i}}{\pi } \frac{d\bar{z^{\prime } } _ { i}dz_{i}^{\prime } } { \pi } \frac{d\bar{z}_{f}dz_{f}}{\pi } e^{-\bar{z}_i z_i - \bar{z}_i ' z_i ' - \bar{z}_f z_f } \nonumber \\ & & \times \langle \bar{z}_{f};t|z_{i};0\rangle _ { \eta , \bar{\eta } } \hspace{0.2cm}\langle \bar{z}_{i}|\rho _ { 0}|z_{i}^{\prime } \rangle \hspace{0.2cm}\langle \bar{z}_{i}^{\prime } ; 0|z_{f};t\rangle _ { \eta , \bar{\eta } } \end{aligned}\ ] ] where the completeness relation for ( unnormalized ) coherent states has been used @xmath63 writing with an obvious identification @xmath64 we can use the identity @xmath65 to obtain @xmath66=e^{\bar{\beta}^{\prime } \beta -(d+d^{\prime } ) } \ ] ] for an initial vacuum state @xmath67 . substituting
, we get the contribution to the influence functional from one mode @xmath68 & = & \exp \left\ { g_k^{2}\int_{0}^{t}ds\int_{0}^{s}ds^{\prime } \left [ \bar{\eta}^{\prime } ( s)\eta ( s^{\prime } ) e^{-i\omega _ k ( s - s^{\prime})}\right.\right .
\nonumber \\ & & \left.\left.+\bar{\eta}^{\prime } ( s^{\prime } ) \eta ( s)e^{i\omega_{k}(s - s^{\prime } ) } -\bar{\eta } ( s)\eta ( s^{\prime } ) e^{-i \omega_{k}(s - s^{\prime } ) } - \bar{\eta } ^{\prime } ( s^{\prime } ) \eta ( s)e^{i\omega_k ( s - s^{\prime } ) } \right ] \right\}\end{aligned}\ ] ] the influence functional for all modes @xmath69 is finally given by @xmath70 & = & \exp \left\ { \int_{0}^{t}ds\int_{0}^{s}ds^{\prime } \left(\mu ( s - s^{\prime } ) [ \bar{\eta } ^{\prime } ( s)-\bar{\eta } ( s)]\eta ( s^{\prime } ) \right.\right . \nonumber \\ & & \left.\left.+\mu ^{*}(s - s^{\prime } ) \bar{\eta}^{\prime } ( s^{\prime } ) [ \eta(s)-\eta ^{\prime}(s)]\right)\right\}\end{aligned}\ ] ] in terms of the kernel @xmath71
having computed the influence functional we have an expression for the reduced density matrix propagator @xmath72\eta ( s^{\prime } ) + \mu ^{*}(s - s^{\prime } ) \bar{\eta } ^{\prime } ( s^{\prime } ) [ \eta ( s)+\eta ^{\prime } ( s)]\right ) \right\}\end{aligned}\ ] ] where the summation over all paths obey the boundary conditions ( ii.10 ) and evaluation and get @xmath73 & = & 0 \\ \dot{\bar{\eta}}-i\omega \bar{\eta } + \int_{0}^{s}ds^{\prime } \mu^{*}(s - s^{\prime } ) \bar{\eta } ^{\prime } ( s^{\prime } ) -\int_{s}^{t}ds^{\prime } \mu ( s - s^{\prime } ) [ \bar{\eta } ^{\prime } ( s^{\prime } ) + \bar{\eta } ( s^{\prime } ) ] & = & 0\end{aligned}\ ] ] it will turn out that only the solution of the first two of these equations will contribute to the path integral .
we will therefore write ( s)=_iu(s)|^(s)=|_f^|u(s ) where @xmath74,@xmath75 are the solutions to equations ( iii.3),(iii.4 ) under the condition u(0)=|u(t)=1 now equation ( iii.3 ) is a linear integrodifferential equation of first order and as such can be solved with the use of the laplace transform and the convolution theorem .
it is easy to show that @xmath76 where @xmath77 is the laplace transform of the kernel ( ii.24 ) and @xmath78 is a real constant larger than the real part of the poles of the integrand .
it turns out that this function @xmath79 contains all necessary information for the computation of the density matrix propagator . substituting our expressions ( iii.8 ) and ( iii.9 ) ) in ( iii.1 )
we can obtain the following expression for the propagator @xmath80 \bar{\eta } ^{\prime}_i \eta _ i \right)\ ] ] since we are using coherent state path integrals we have departed in our evaluation from the standard saddle point approximation used on configuration space path integrals . in these cases ,
the standard procedure is to distinguish the imaginary part of the kernel as corresponding to dissipation and consider only its contribution when performing the saddle point evaluation .
the resulting equations are then the classical dissipative equations of motion .
but in the case of the coherent state path integral , there is no correspondence between extrema of the action and actual classical paths .
hence , there is no sense in splitting the kernel @xmath44 into real and imaginary part , and the saddle point evaluation should be carried out for the whole of the exponential .
it is a standard procedure now to find the master equation @xcite .
we compute the time derivative of the propagator : @xmath81 the next step is to remove from the above equation the dependence on the initial values .
this is done with the use of the following identities @xmath82 note that we are suppressing ( for ease of notation ) symbols denoting left or right grassmann differentiation . in all our expressions we implicitly assume that differentiation with respect to @xmath83 is always right and with respect to @xmath84 always left . for the density matrix at time @xmath85 @xmath86
we obtain the evolution equation @xmath87 this is one of our main results : the master equation for the two - level atom interacting with a an environment of electromagnetic field at its vacuum state . the effect of the field is contained within the function @xmath74 which can be determined by the solution of equation ( iii.3 ) or equivalently by the computation of the contour integral ( iii.9 ) . in the next section we are going to find explicit expressions for @xmath74 for particular choices of the field configuration .
let us return for the moment to equation ( iii.14 ) and write this in an operator language .
it is easy to verify that : @xmath88 where @xmath89 .
if we write @xmath90 the master equation reads @xmath91 + \gamma ( t ) \{s_+s_-,\rho \ } - 2 \gamma ( t ) s_- \rho s_+\ ] ] where @xmath92 the first term corresponds to the unitary hamiltonian evolution , only now the effect of the environment has induced a time dependent shift in the value of the frequency , the second term is time dependent dissipation and the third corresponds to noise . to show how the standard results are regained , and to understand the meaning of the new function in the master equation ,
let us consider the physical process of spontaneous emission . start with a generic initial density matrix =( cl 1-x & y + y^ * & x ) its corresponding q - symbol is @xmath93 if we evolve it with the density matrix propagator ( iii.10 ) we obtain for the state at time @xmath85 @xmath94 corresponding to _ t= ( cl |uu(1-x ) & uy + ) considering the case @xmath95 we get for the probability of spontaneous emission p(10,t)=1-|uu also we should remark that the rate of decoherence in the energy eigenstates is governed by the absolute value of the function @xmath74 ( the off- diagonal terms ) .
but on the other hand @xmath74 itself determines the rate of nergy flow from the atom to the environment .
hence for our particular choice of initial state ( vacuum ) we find that decoherence and relaxation time are essentially identical .
we shall use this equation to study decoherence in an ion trap in a later paper .
our master equation ( iii.17 ) depends solely on the function @xmath96 , which in its turn is determined by the kernel @xmath97 . in this section
we will try to give some analytic expressions for this function in various different cases . to connect with known results @xcite ,
let us start with the case when the field contains only a single mode with frequency @xmath98 .
then @xmath99 will read @xmath100 and @xmath101 the integrand has two poles at the solutions of the equation @xmath102 given by @xmath103^{1/2}}{2}=-i\omega _ { 1,2}\ ] ] hence @xmath104 this result is in agreement with standard ones @xcite now we consider the case of the vacuum electromagnetic field in free space i.e. not constrained by a cavity .
the kernel will read then ( using equation ( ii.24 ) ) @xmath105 where @xmath106 note the factor of @xmath107 in ( iv.6 ) coming from the two photon polarisations and that we in view of ( ii.3 ) we have written @xmath108 .
since the integral ( iv.7 ) is not convergent , we will introduce an exponential cut - off in the higher frequency modes .
the presence of the cut - off is of physical significance since we do not expect high electromagnetic modes to couple with our two - level atom .
hence the kernel @xmath109 will read @xmath110 the laplace transform of @xmath109 is then @xmath111 where @xmath112 denotes the exponential integral function analytically continued to the complex domain . at the limit @xmath113
this is essentially @xmath114 where @xmath115 is the euler - macheronni constant and the logarithm is taking values in the primary branch .
thus @xmath77 reads ( @xmath116 is here @xmath117 obtained by the integration by parts of the laplace transform ) @xmath118 note that the cut - off @xmath119 affects significantly @xmath120 only at large values of @xmath42 , which essentially correspond to the very short time limit , i.e the time where the two - level atom starts getting acquainted with the photon reservoir . at larger times ( @xmath121 )
we do not expect the cut - off to contribute significantly in the evolution .
this is a rather typical behaviour in quantum brownian motion models , provided that the ultraviolet cut - off of the environment is much larger than the natural frequencies of the system . to evaluate the integral we first have to find the poles of the denominator .
we can do that in a perturbation expansion .
first let us absorb the divergent @xmath116 factor in a frequency renormalisation .
@xmath122 so that we need find the zeros of the function @xmath123 .
looking for a solution in the vicinity of @xmath124 we find @xmath125 we have a pole with a negative real part and we can verify numerically ( also physically expected ) that there is no pole with greater real part .
this means when evaluating the inverse laplace transform we can ignore the contribution of the branch - cut at @xmath126 ( being on the right of the pole ) and hence after some time , where all possible other poles with absolutely larger value of their real part will have stopped contributing the solution will be @xmath127 this implies a markovian time evolution and the identification of decoherence - relaxation time with @xmath128 .
this is a general feature of the presence of a continuum of modes as can be seen from the case of an atom transparent to all modes but the ones in a strip , say @xmath129 $ ] containing the resonance frequency .
it is easy to verify that in this case we have again @xmath130 hence defining again @xmath131 we can find the pole at @xmath132 note that the real part of the pole comes from the presence of a minus sign in a logarithm of some real valued object .
hence in the case where the atom s frequency is outside the strip of interacting modes there will be no dissipation .
this feature separates us from the qbm case , characterising the atom - field system as primarily a resonant one .
let us now consider the case of the atom lying within a cavity consisting of two parallel plates at distance @xmath133 .
the field satisfies dirichlet boundary conditions on the surface of the plates , hence the modes in the normal direction to the plates are multiples of @xmath134 .
the kernel then reads @xmath135 hence we can compute its laplace transform @xmath136 where @xmath137 is defined by @xmath138 and appears in equation ( iv.20 ) through analytical continuation in the complex plane .
this integral can actually be computed at the limit of vanishing @xmath139 ( @xmath113 ) - see reference @xcite , equation 3.427.4 .
@xmath140 + o(\epsilon ) \end{aligned}\ ] ] note that @xmath77 is a sum of the term of case 2 ( @xmath141 ) and a finite one ( no dependence on @xmath119 ) .
the logarithm of the @xmath142-function gives a countable number of branch cuts at @xmath143 , @xmath144 positive integer , the resonance modes of the cavity .
again the important pole has a negative real part .
we can again compute the pole perturbatively .
it lies at @xmath145 clearly the logarithm of the gamma functions is the term out of which the real part of the pole appears .
since the real part of the pole is negative the branch cut is excluded from the integration contour and hence @xmath146 here @xmath142 gives a dissipation constant . .
in figure 1 we give a plot of the real part of the pole ( @xmath147 ) as a function of the frequency @xmath148 .
note that it has sharp maxima on the resonance points , implying persistence of coherence . already from the approximation ( iv.23 )
we observe that the difference in the renormalized frequency from the case @xmath149 ( @xmath150 ) is finite .
unfortunately perturbation expansion is not reliable when @xmath151 is close to the resonance frequencies ( this corresponds to negative integers arguments in the @xmath1-function where it diverges ) and for this regime we have not been able to get any analytic results . in figure 2
we have plotted the dependence of @xmath152 - \omega[\infty ] $ ] as the frequency changes .
this effect of the frequency shift for an atom within the cavity is well known , as well as its relation to the casimir effect @xcite .
let us now integrate what we have found and look at the overall picture .
the physics of a 2la - emf system at zero temperature is characterized by a number of time constants : + 1 ) the inverse natural frequency @xmath153 + 2 ) the inverse coupling constant @xmath154 + 3 ) the relaxation time constant @xmath155 + 4 ) the cavity size @xmath156 ( divided by c ) first consider a zero temperature field in free space , thus ignoring factors 4 ) .
start with only one mode in the field in resonance with the atom , then the system undergoes rabi nutation with frequency @xmath157 , where @xmath144 is the photon number in the field . the collapse time ( assuming a large mean photon number @xmath158 ) is @xmath159 , and revival time is @xmath160 .
atom excitation becomes significant in a time much greater than @xmath161 but shorter than @xmath162 .
( this is the condition for a first order perturbation theory to give reasonable results . ) for a large number of modes , spontaneous emission occurs at the relaxation time scale @xmath163 which we found to be the same as the decoherence time the time for the off diagonal elements of the reduced density matrix to decay ( sec .
4 ) . when the mean number of photons in the field is large ( @xmath164 ) , they become comparable to the collapse time .
this is a measure of the coherence in the atom - field system , and is controlled mainly by their coupling and the photon number in the field .
we see that with the resonance condition , the nature of decoherence in 2ls is very different from the qbm situation , where phase information in the brownian particle is efficiently dispersed in the many modes in the bath coupled almost equally to the system . as we remarked in the introduction ,
the identification of the phase information and energy flow from the 2ls to its environment is similar to the spin echo phenomena ( landau ` damping ' ) which is based on statistical mixing rather than dissipation .
the mathematical distinction lies between considering the system coupled to the discrete number basis ( our model ) and the continuous amplitude basis ( qbm ) of the environment .
the latter case essentially produces noise that drives the system in a way insensitive to its own intrinsic dynamics . while in the former case , the coupling respects the internal dynamical structure of the 2ls and allows it to keep its coherence . to see how the distribution of modes in a field changes the picture , the cavity field calculation in sec .
4 is useful . as shown in fig .
1 , the relaxation constant develops peaks and minima .
the resonance effect is enhanced by a cavity size commensurate with the natural frequency of the 2la and dissipation weakens .
narrow band resonance fluorescence as well as inhibition of spontaneous decay by frequent measurements the quantum zeno effect
are interesting phenomena which our equations can provide finer details .
non - markovian processes involve memory effects ( nonlocal in time ) . for the qbm problem , except for the case of high temperature ohmic bath which gives markovian dynamics , other types of spectral density ( supraohmic ) or at low temperatures , the dynamics of the system
is non - markovian @xcite . when the reaction time of the bath is comparable to or faster than the natural time scale of the system ( @xmath165 ) , one also expects to see non - markovian behavior .
by contrast , the 2la is quite different : at zero temperature there is only one timescale @xmath166 that determines both decoherence and relaxation .
there is no memory effect and hence the process is markovian .
we expect that in finite temperature the dynamics of the 2la will be nonmarkovian @xcite .
this is because there are more ways for the atom and the field to get entangled , and the memory effects of their interaction would presumably persist . in conclusion we find that the 2ls interacting with an em field is far more coherent than what is commonly believed , the misconception probably arising from the mistaken identification of this system with the brownian model of an oscillator interacting with a harmonic oscillator bath . + + + * acknowledgement * we wish to thank phillip johnson and adrian dragulescu for a close reading of our paper and making useful suggestions , and sanjiv shresta for checking the formulae and solutions to some nonlocal differential equations numerically and providing us with the figures .
sanjiv shresta also first noticed a sign disrepancy in a sample calculation , which alerted us to possible ambiguities in the grassmanian variable approach for finite temperature and coherent state fields .
we also thank dr juan pablo paz for pointing out a mistake in equation ( iii.17 ) of an earlier version of this paper .
this work is supported in part by nsf grant phy98 - 00967 .
in this appendix we give a rather detailed derivation of the hamiltonian for a nonrelativistic atom interacting with a second - quantized electromagnetic field under the dipole , rotating wave and two level approximations .
this is to facilitate the comparison of our model ( ii.1 ) with @xmath6 coupling with that used by others ( ii.3 ) with @xmath4 coupling .
( see introduction ) . to make this also useful for later papers in this series
, we have included atomic motion .
note that the convention here is closer to @xcite than that used in the text which is closer to @xcite .
the conversion is explained in footnote 2 .
the dynamics of a moving atom ( mass @xmath167 , momentum @xmath168 ) whose electrons ( charge @xmath169 , mass @xmath170 ) interact with an electromagnetic field ( vector potential @xmath171 , coulomb potential @xmath23 ) is decribed by the ( classical ) hamiltonian h = + ( - e ) ^2 + e v ( ) + h_b where @xmath172 is the hamiltonian for the electromagnetic field . expanding out
one can write this as h = h_a + h_e + h_b + h_c where h_a = describes nonrelativistic atom motion , h_e = + e v ( ) decribes the dynamics of the ( one ) electron , while h_c1 = - and h_c2 = ^2 describe the coupling between the electron and the field .
the second term makes no contribution to one - photon processes and will be ignored .
we will refer to @xmath173 as the unperturbed hamiltonian , and @xmath174 the interaction hamiltonian . in a second - quantized form
, the hamiltonian for the radiation field is given by h_b = _ _^ where @xmath175 , @xmath176 are the creation and annihilation operators for the kth normal mode of a free massless vector field .
thus for the field vacuum @xmath177 , @xmath178 , @xmath179 = \delta_{\bk,\bk'}$ ] , for all @xmath10 . we can perform a harmonic decomposition of the vector potential of the electromagnetic field ( , t ) = _ ( ) ^where , assuming the field is contained in a box of size l , the spatial mode functions @xmath180 is given by _ ( ) = l^-3/2 _ f _ ( ) here @xmath181 is the unit polarization vector and @xmath20 are the two ( transverse ) polarizations . in free space , @xmath182 .
we assume that electron motion is much faster than the motion of the atom , thus it sees a stationary central coulomb potential around the center of mass of the atom .
denoting the ( time - independent , nonrelativistic ) electronic wave function eigenstate belonging to the eigenvalue @xmath183 by @xmath18 , i.e , @xmath184 , we can write the hamiltonian for the electron in the second - quantized form as h_e = _
i e_i a_i^ a_i where @xmath185 labels the bound states of the eletron ( we assume vanishing probability for the atom to ionize ) and @xmath186 and @xmath187 are the creation and annihilation operators . as fermions they obey the anticommutation relations @xmath188 .
to perform perturbation theory , the electronic wave function of the interacting system is expanded in terms of the eigenfunctions @xmath18 of the unperturbed hamiltonian , with basis formed by the direct product of the electron and the field states thus the electron field operator @xmath189 can be expanded as ( ) = _ i a_i _ i ( ) . with this , the interaction hamiltonians is given by h_i = -^(*x * ) ( * p * ) ( * x*)d^3x or in terms of @xmath190 operators h_i = _ i , j , a_i^ a_j ( g_ijb _ + |g_ijb_^ ) where g_ij = - _ i(*x * ) _ ( * x * ) _ j(*x*)d^3x . + * dipole approximation * + now consider conditions when the spatial variation of the vector potential @xmath171 of the electromagnetic field is small compared to the electronic wave function @xmath191 , one can expand @xmath192 in @xmath193 around the position of the atom @xmath194 : e^i = e^i [ 1 + - ( ) ^2 + ... ] the dipole approximation amounts to keeping just the leading term .
doing so , we can take the field mode function @xmath195 out of the integration above and evaluate it at the atomic position . to evaluate @xmath196
we make use of = = [ x_i , h_e ] yielding , _
j d^3x = i _ij
d_ij where @xmath197 , and @xmath17 is the dipole matrix element , @xmath26 . define d_ij -_ij _ note that @xmath27 . with this under the dipole approximation ,
g_ij = d_ij f _ ( ) + * rotating wave approximation * + in the interaction picture , recalling that the time evolution of the ladder operators are given by ^(t ) = a_i^e^i _it ( t ) = a_j e^-i _j t ( t ) = b _
e^-i _t the interaction hamiltonian @xmath198 in the interaction picture becomes h_i = _ i , j , g_ij a_i^ a_j b _
e^i(_ij - _ ) + _ ij |g_ij a_i^ a_j b_^ e^i(_ij + _ ) where @xmath199 .
we see there are two types of oscillatory terms present : @xmath200 .
processes most effective in the absorption or emission of a photon by the atom correspond to those with near resonance frequency @xmath201 .
assuming @xmath202 ( @xmath203 ) , the first type with @xmath204 has a rapidly oscillating phase and its contribution is small compared with the second type with @xmath205 whose stationary phase at near resonance gives a large contribution .
physically the first type corresponds to either the excitation of the atom _ along with _ the emission of a photon or the relaxation of the atom _ along with _ the absorption of a photon , which is less probable than the second type corresponding to the excitation of an atom upon the absorpsion of a photon or the relaxation of an atom with the emission of a photon .
we shall therefore ignore the first type of terms , which amounts to working under the rotating wave approximation .
this is the second major approximation in this standard model .
+ * two - level atom * + let us now consider the idealized case when the atom has only two electronic states , @xmath206 corresponding to @xmath207 , with energies equal to @xmath208 .
( the two states can interchangeably be labeled as @xmath209 , or @xmath210 or @xmath211 . ) thus @xmath212 .
thus _ a = ( a_2^a_2 - a_1^a_1 ) _ z _0 s_z where we have introduced a pauli matrix ( 2x2 ) representation @xmath14 for the interaction hamiltonian above , under the rwa , in the @xmath213 contribution to the summation , the first line containing @xmath214 is kept , while the second line is dropped .
the reverse is true for the @xmath215 term .
the interaction hamiltonian ( in the interaction picture ) now becomes h_i = _ introducing the pauli matrix representation for the fermion operators @xmath216 and @xmath217 , and defining @xmath218 ( recall @xmath219 ) we can write the interaction hamiltonian ( in the heisenberg picture ) in a simple form : _ i = _ g _ ( s_+ b_f _ ( ) + s_- b_^|f _ ( ) ) ) therefore the total hamiltonian for our model of a moving atom interacting with a quantum electromagnetic field under the dipole , rotating wave and two - level approximation is given by = + _0 s_z + _
we use the resolvent decomposition of the propagator @xmath220 then by writing @xmath221 we can expand the resolvent and get @xmath222 when we act the expanded resolvent in any vector @xmath223 , ( @xmath224 denotes an eigenstate of the hamiltonian of the two level system ) we see that + only expanded terms that contain alternating sequences of @xmath225 and @xmath226 survive .
this makes the summation much easier . also if we note
that @xmath227 hence we can compute the matrix elements by resumming the expansion @xmath228 we can then verify that the only nonzero elements @xmath230 are the following and their conjugates @xmath231 then it is easy to check that this reproduces the propagation as given by equation ( 3.21 ) that was obtained through the influence functional method .
indeed @xmath232 is exactly the same as the one defined by equation ( 3.9 ) .
i. l. chuang et al , science * 270 * , 1633 ( 1995 ) ; d. p. di vencenzo , science * 270 * , 255 ( 1995 ) .
p. w. shor , phys . rev . *
a52 * , r2493 ( 1995 ) .
j. i. cirac and p. zoller , phys . rev
. lett . * 74 * , 4091 ( 1995 ) .
a. ekert and r. jozsa , rev .
phys . * 68 * , 733 ( 1996 ) .
see , e.g. , w. h. zurek , physics today * 44 * , 36 ( 1991 ) ; prog .
. phys . * 89 * , 281 ( 1993 ) .
e. joos and h. d. zeh , z. phys .
* b59 * , 223 ( 1985 ) ; d. giulini _ et al _ , _ decoherence and the appearance of a classical world in quantum theory _ ( springer verlag , berlin , 1996 ) .
|
we perform a first- principles derivation of the general master equation to study the non - markovian dynamics of a two - level atom ( 2la ) interacting with an electromagnetic field ( emf ) .
we use the influence functional method which can incorporate the full backreaction of the field on the atom , while adopting grassmannian variables for the 2la and the coherent state representation for the emf .
we find exact master equations for the cases of a free quantum field and a cavity field in the vacuum . in response to the search for mechanisms to preserve maximal coherence in quantum computations in ion trap prototypes , we apply these equations to analyse the decoherence of a 2la in an emf , and fine that decoherence time is close to relaxation time .
this is at variance to the claims by authors who studied the same system but used a different coupling model .
we explain the source of difference and argue that , contrary to common belief , the emf when resonantly coupled to an atom does not decohere it as efficiently as a bath does on a quantum brownian particle . the master - equations for non - markovian dynamics derived here
is expected to be useful for exploring new regimes of 2la - emf interaction , which is becoming physically important experimentally .
c i u plus 1000pt minus 1000pt # 1 # 1= to # 1= to # 1nucl .
phys .
# 1 # 1phys .
lett .
# 1 # 1phys .
rev .
lett .
# 1 # 1phys . rev . *
d * # 1 # 1phys . rev . * a * # 1 # 1prog .
theor .
phys .
# 1 #1astron . and astrophys .
# 1 # 1@xmath0 # 1 # 1)([#1 ] ) # 1(#1 ) = 8.5 in = 6.5 in = -.5 in = 0.in = 0.in addtoresetequationsection
|
a network is an abstract entity consisting of a certain number of nodes connected by links or edges .
the number of nodes that can be reached from a reference node @xmath4 in one step is called its degree denoted by @xmath5 . if equal number of nodes can be reached in one step from all the nodes , the network is said to be regular or homogeneous .
a regular lattice where nodes are associated with fixed locations in space and each node connected to equal number of nearest neighbours , is an example of a regular network .
however , in the general context of complex networks , it is defined in an abstract space with a set of nodes @xmath6 and a set of links denoted by @xmath7 . as the spectrum of @xmath8 values of the nodes increases ,
the network becomes more and more irregular and complex . over the last two decades
, the study of such complex networks has developed into a major field of inter - disciplinary research spanning across mathematics , physics , biology and social sciences @xcite .
many real world structures @xcite and interactions @xcite can be modeled using the underlying principles of complex networks and analysed using the associated network measures @xcite . in such contexts ,
the corresponding complex network can be weighted @xcite or unweighted and directed @xcite or undirected depending on the system or interaction it represents . in this paper , we restrict ourselves to unweighted and undirected networks and the possible extensions for weighted and directed networks are discussed in the end .
the topology or structure of a complex network is determined by the manner in which the nodes are connected in the network .
for example , in the case of the classical random graphs ( rg ) of erds and r ' enyi ( e - r ) @xcite , two nodes are connected with a constant and random probability @xmath2 . in contrast , many real world networks are found to have a tree structure with the network being a combination of small number of _ hubs _ on to which large number of individual nodes are connected @xcite .
an important measure that distinguishes between different topologies of complex networks is the degree distribution @xmath9 that determines how many nodes in the network have a given degree @xmath8 . for the rgs
, @xmath9 is a poisson distribution around the average degree @xmath10 @xcite while many real world networks follow a fat - tailed power law distribution given by @xmath11 , with the value of @xmath3 typically between 1 and 3 @xcite .
such networks are called _ scale free _ ( sf ) @xcite due to the inherent scale invariance of the distribution . though topology is an important aspect of a complex network , that alone is not sufficient to characterize and compare the interactions that are so vast and diverse .
a number of other statistical measures have been developed for this purpose , each of them being useful in different contexts .
two such commonly used quantifiers are the clustering coefficient ( cc ) and the characteristic path length ( cpl ) .
there are also characteristic properties of local structure used to compare the complexity of networks in particular cases , such as , the hierarchy or community structure @xcite in social networks and motifs @xcite and super family profiles @xcite in genetic and neuronal networks . however , a single index that can quantify the diversity of connections between nodes in networks even with different topologies , is the heterogeneity measure @xcite .
it is also indicative , in many cases , of how stable and robust @xcite a network is with respect to perturbations from various external parameters .
an important example is the technological network of north american power grid @xcite .
recent studies have also revealed the significance of the heterogeneity measure in various other contexts , such as , epidemic spreading @xcite , traffic dynamics in networks @xcite and network synchronization @xcite .
the network heterogeneity has been defined in various ways in the literature which we will discuss in detail in the next section where , we will also present the motivations and need for a new measure . while all the existing measures are based on the degree correlations @xmath5 and @xmath12 of nodes @xmath4 and @xmath13 in the network , the measure proposed in this paper uses only the degree distribution @xmath9 to compute the heterogeneity of the network .
however , we show that this new measure varies directly with the @xmath8 spectrum , or the spectrum of @xmath8 values in the network , and hence gives a true representation of the diversity of node degrees present in the network . in other words , it serves as a single index to quantify the node diversity in the network . in this work
, we also include a class of networks not considered so far in the context of heterogeneity measure in any of the previous works .
these are complex networks constructed from the time series of chaotic dynamical systems , called recurrence networks @xcite .
they have a wide range of practical applications @xcite and the measures from these networks are used to characterize strange attractors in state space , typical of chaotic dynamical systems , as discussed in v. the diversity of node degrees in
the rns was actually one of the motivations for us to search for a heterogeneity measure that could be used to compare the structural complexities of different chaotic attractors through the construction of rns .
our paper is organized as follows : in the next section , we discuss briefly all the previous measures of heterogeneity and give reasons why we have to look for a new measure .
the measure that we propose is based on the idea of what we consider as a completely heterogeneous network of @xmath1 nodes , that is illustrated in iii .
the proposed measure of heterogeneity is presented in
iv while v and vi are devoted to computation of this new measure for various synthetic as well as real world networks .
our conclusions are summarised in
if we carefully analyze the heterogeneity measures proposed in the literature , it becomes clear that two different aspects of a complex network can be quantified through a heterogeneity measure .
they are the diversity in node degrees and the diversity in the structure of the network .
for example , the initial attempts to measure the heterogeneity try to capture the diversity in the node degrees of the network and were mainly motivated by the random graph theory .
the first person to propose a measure of heterogeneity was snijders @xcite in the context of social networks and it was modified by bell @xcite as the variance of node degrees : @xmath14 where @xmath10 represents the average degree in the network .
though this is still one of the popular measures of heterogeneity , its applicability is mainly limited to rgs where one can effectively define an average @xmath8 .
another measure was proposed by albertson @xcite as : @xmath15 which is a sum of the local differences in the node degrees in the network .
this index also is not completely adequate in quantifying correctly the heterogeneity of networks with different topologies .
apart from the above two measures defined in the context of social networks , another measure @xcite has recently been proposed to quantify the degree heterogeneity .
it uses a measure of inequality of a distribution , called the gini coefficient @xcite , which is widely used in economics to describe the inequality of wealth . here
a heterogeneity curve is generated using the ratio of cumulative percentage of the total degree of nodes to the cumulative percentage of the number of nodes .
the heterogeneity index is then measured as the degree inequality in a network .
though the authors compute heterogeneity of several standard exponential and power law networks , the measure turns out to be very complicated and works mainly for networks of large size with @xmath16 . in short ,
none of these measures , though useful in particular contexts , truly reflects heterogeneity as represented by the diversity of node degrees in a network .
a comparative study of the above heterogeneity measures has been done by badham @xcite .
the second aspect of heterogeneity discussed in the literature is the topological or structural heterogeneity possible in a complex network which is especially important in real world networks .
an example for this is the measure proposed by estrada @xcite recently , given by @xmath17 which can also be normalised to get a measure @xmath18 within the unit interval @xmath0 $ ] as : @xmath19 a comparison of the completely heterogeneous networks ( see text ) with @xmath20 , @xmath21 , @xmath22 and @xmath23 . in each case , all the possible @xmath8 - values from @xmath24 to @xmath25 are present in the network as shown .
one degree ( one @xmath8 value ) has to be shared by two nodes since the @xmath26 node will have the degree of any one of other nodes .
it is empirically shown that this degree of @xmath26 node , denoted by @xmath27 , is automatically fixed ( if the network has all possible degrees from @xmath24 to @xmath25 ) and is @xmath28 if @xmath1 is even and @xmath29 if @xmath1 is odd .
for example , for @xmath20 and @xmath21 , @xmath30 and for @xmath31 and @xmath23 , @xmath32 and so on . ] change in the degree distribution for a typical complex network as it is transformed from complete homogeneity to complete heterogeneity , for @xmath33 . ]
if we analyse this measure closely , we find that it is basically different with respect to the earlier measures .
the reason is that the measure proposed by estrada is based on the randic index @xcite given by @xmath34 now , the randic index was originally proposed @xcite as a topological index under the name _ branching index _ to measure the branching of carbon atom skeletons of saturated hydrocarbons . this index is so designed to get extremum value for the `` star '' structure which is the most heterogeneous branching structure and is bounded by values given by @xmath35 with the limiting values for the star structure and a regular lattice . to be specific
, estrada defines heterogeneity through an irregularity index for each pair of nodes @xmath36 , where @xmath37 if @xmath38 and @xmath39 for @xmath40 and @xmath41 .
it is obvious that a measure based on this definition will be maximum for a `` star network '' of @xmath1 nodes compared to all other networks since there are @xmath25 connections with @xmath36 having maximum value .
the above discussion makes it clear that estrada s measure elegantly captures the structural aspect of heterogeneity associated with a complex network .
this is also evident in the results given by the author .
out of all possible branching structures , the heterogeneity is maximum for the star structure . while the star network has @xmath42 , the values for networks with other topologies are much less with a typical sf network having @xmath43 .
this measure is important in the context of real world networks with different topology and structure and can be used to classify such networks as shown by estrada .
our focus here is the heterogeneity associated with the diversity in node degrees ( analogous to the earlier attempts of heterogeneity ) to propose a measure applicable to networks of all topologies .
an important difference is that we use the frequencies of the node degrees , rather than @xmath5 directly , to define this measure .
we show that , as the spectrum of @xmath8 values in the network increases , the value of the measure also increases correspondingly .
we call the measure proposed here _ degree heterogeneity _ in order to distinguish it from the measure in @xcite . also , the two measures capture complimentary features of heterogeneity in a complex network
. a network having high heterogeneity in one measure may not be so in the other measure and vice versa .
for example , the star network is nearly homogeneous in our definition of heterogeneity , as shown below .
it is also possible to correlate the robustness or stability of a network with the measure proposed here , with the sf networks having comparatively high value of heterogeneity . on the other hand ,
the star network is most vulnerable since disruption of just one node can destroy the entire network . to define the new measure
, we require a network with a limiting value of heterogeneity to play a role similar to that of star network in the earlier measure .
this network is presented in the next section .
finally , the heterogeneity measure that we define below can be shown to have direct correspondence with the entropy measure of a complex network @xcite , characterized by the standard shannon s measure of information @xmath44 .
in particular , this measure can be so adjusted to get the value zero for completely homogeneous networks and the value @xmath45 for the completely heterogeneous case as defined by us in this work .
though there are attempts to represent heterogeneity through entropy @xcite , we prefer to view the two measures as two sperate aspects of a complex network .
entropy is usually associated with the rate at which a process or system ( evolving ) generates information . in the case of a network ,
each node can be considered as an information hub and links as channels for dissipating information . for the completely homogeneous network ,
no new information is generated while it tends to be maximum when the diversity in the node degrees is maximum . in this sense ,
entropy and heterogeneity are closely related and both have values normalized in the interval @xmath0 $ ] . however , while entropy is a dynamic measure , heterogeneity is basically a static measure characterizing the structure and diversity of connections between the nodes and not directly concerned with the information transfer .
that is why traditionally the two measures have been treated seperately , though the values of both for the extreme cases can be made identical .
moreover , the measure that we propose below has the following advantages : \i ) only the degree distribution of the network is required to compute the heterogeneity in contrast to all the previous measures proposed so far .
\ii ) the specific condition that we apply for the completely heterogeneous network provides analytical values for heterogeneity in terms of network size .
\iii ) based on the proposed measure , we are able to give a structural characterization index for a chaotic attractor through the construction of a complex network called recurrence network .
here we present what we consider as the logical limit of a completely heterogeneous network of @xmath1 nodes .
the reader may find that this is an ideal case .
nevertheless , it helps to put the concept of heterogeneity of a complex network in a proper perspective . consider an unweighted and undirected complex network of @xmath1 nodes , with all the nodes connected to the network having a degree of at least one . if all the nodes have the same degree @xmath8 , the network is completely homogeneous with the degree distribution @xmath9 being a @xmath46 peaked at @xmath8 .
let us now consider the other extreme where no two nodes have the same degree .
the maximum possible degree for a node is @xmath25 .
let the nodes be arranged in the ascending order of their degree .
it is obvious that the @xmath26 node will have to take a degree equal to that of any one of the other nodes having degree from 1 to @xmath25 . to find out what degree is possible for the @xmath26 node under the given condition
, we start with taking small number of nodes as shown in fig .
[ f.1 ] , where we show @xmath47 different cases of @xmath1 ranging from @xmath47 to @xmath23 . in each case
, the @xmath26 node is represented as a _ pentagon shape _ with its degree denoted as @xmath27 .
it is clear that if all the node degrees are to be different , there is only one possible value of @xmath27 for the @xmath26 node , which is @xmath48 if @xmath1 is even and @xmath49 if @xmath1 is odd . a snapshot of different types of complex networks in the increasing order of their heterogeneity(@xmath50 ) , taking @xmath51 in all cases . from top to bottom , the nature of the network varies from completely homogeneous , star , rg , sf and finally to completely heterogeneous network .
the degree distribution and @xmath8 spectrum are also shown for each case to indicate that @xmath50 is a measure of the diversity in the node degrees . ]
we now give a simple argument that this result is true in general for any @xmath1 .
the degree of node 1 is @xmath24 which means that it is connected only to the node with degree @xmath25 .
that is , node 1 is not connected to @xmath26 node .
node 2 is connected only to two nodes with degree @xmath25 and @xmath52 and hence it is also not connected to node @xmath1 . by induction , one can easily show that the @xmath53 node is connected only to nodes with degree @xmath25 , @xmath52, ....... @xmath54 .
suppose @xmath1 is even .
when @xmath55 , this node is connected to nodes with degree from @xmath25 to @xmath48 . to avoid self loop , this node should be connected to node @xmath1 .
thus all nodes with higher degree from @xmath48 to @xmath25 are connected to node @xmath1 whose degree becomes @xmath48 . by a similar argument
, one can show that the degree of @xmath26 node is @xmath49 if @xmath1 is odd .
let us now consider the degree distribution @xmath9 of this completely heterogeneous network .
all the nodes have different @xmath8 values and only two nodes share the same @xmath8 value , @xmath27 .
one can easily show that : @xmath56 our definition of heterogeneity is derived in such a way that this network has maximum heterogeneity , which is done in the next section .
it is very well accepted that a network of @xmath1 nodes with all nodes having equal degree @xmath8 is a completely homogeneous network with @xmath9 being a @xmath46 centered at @xmath8 .
the value of @xmath8 can be anything in the range @xmath57 and all these networks have heterogeneity measure zero , for any @xmath1 . in principle , the heterogeneity of a network should measure the diversity in the node degrees with respect to a completely homogeneous network of same number of nodes .
all the measures defined so far in the literature directly use the @xmath8 values present in the network for computing the heterogeneity measure . here
we argue that a much better candidate to define such a measure is @xmath9 rather than @xmath8 . since @xmath9 is a probability distribution , as the spectrum of @xmath8 values increase , the value of @xmath9 gets shared between more and more nodes with the condition @xmath58 . in other words ,
this variation in @xmath9 reflects the diversity of node degrees and hence the heterogeneity of the network . a typical variation of @xmath9 as the network changes from complete homogeneity to complete heterogeneity
is shown in fig .
note that for rgs , this variation in @xmath9 is with respect to @xmath59 , with @xmath10 being the average degree , while for sf networks , it is with respect to @xmath60 . to get the heterogeneity measure
, we first define a heterogeneity index @xmath61 for a network of @xmath1 nodes as the variance of @xmath9 with respect to the peak value corresponding to the completely homogeneous case : @xmath62 the condition implies that the summation is only over @xmath8 values for which @xmath63 . for a completely homogeneous network ,
@xmath9 is non zero only for one value of @xmath8 , say @xmath27 , and @xmath64 , making @xmath65 , for all @xmath1 .
we now consider the other extreme of completely heterogeneous case . from the results in the previous section for the completely heterogeneous case
, we have @xmath66 putting the values of @xmath9 and simplifying , we get @xmath67 this is the maximum possible heterogeneity measure for a network of @xmath1 nodes . for large @xmath1 , as a first approximation , we have @xmath68 for finite @xmath1 , its value is @xmath69 and as @xmath16 , @xmath70 . to define the heterogeneity measure @xmath71 for a network , we normalize the heterogeneity index of the network with respect to the completely heterogeneous network of same number of nodes to get the value in the unit interval @xmath0 $ ] : @xmath72 if @xmath1 is sufficiently large , say @xmath73 as is the case for most practical networks , @xmath74 and @xmath75 .
degree distribution of e - r networks ( rgs ) for four different @xmath2 values with @xmath1 fixed at @xmath76 .
the @xmath8 spectrum is shown below the degree distribution .
the value of @xmath50 and @xmath10 are also indicated in each case .
note that @xmath50 varies directly with the degree diversity or the spectrum of @xmath8 values in the network . ]
variation of @xmath50 with @xmath2 for rgs for a fixed value of @xmath1 , as shown .
we expect the profile for a higher @xmath1 value to be within that of a lower @xmath1 as @xmath50 decreases with @xmath1 for any fixed @xmath2 . ]
we note the following features regarding @xmath50 : \i ) it is defined here for unweighted and undirected complex networks and represents a unique measure applicable to any network independent of the topology or degree distribution and increases with the diversity in the node degrees .
\ii ) however , certain topologies have inherent limitations in diversity . for example
, @xmath50 for a star network is very close to zero and hence the star network is nearly homogeneous in our definition .
this is because , in the star topology , the degree of only one node is different from the rest of the nodes .
\iii ) since we use the counts of the node degrees rather than directly @xmath5 to find @xmath50 , we can not express the measure in terms of the elements of the laplacian matrix , as some authors have done .
\iv ) for two networks of the same size @xmath1 independent of the topology , the measure we propose has a direct correspondence with the degree diversity in the network . to show this explicitly
, we present the _
k spectrum _ , the spectrum of @xmath8 values in the network in the form of a discrete line spectrum . in fig .
[ f.3 ] , we compare some standard networks in the increasing order of their @xmath50 , taking @xmath51 . for each network , we show the degree distribution
( as histogram ) , the @xmath8 spectrum and the value of @xmath50 . note that , of different topologies , the sf network is the most heterogeneous . here
the star network has a reasonably high value of @xmath50 since @xmath1 is only @xmath77 .
we also show the completely heterogeneous network with @xmath78 , for comparison .
\v ) the heterogeneity index @xmath61 is defined as a measure normalized with respect to the size of the network @xmath1 . for large @xmath1 , since @xmath79 , the measure @xmath50 can also be used to compare the heterogeneities of two networks even if @xmath1 is different .
this is especially important for real world networks where @xmath1 varies from one network to another , as discussed in vi .
however , a network with larger @xmath1 generally tends to have lower @xmath50 since , to keep the same heterogeneity , the range of non zero @xmath8 values should also increase correspondingly . in other words ,
a network with @xmath80 nodes attains complete heterogeneity if the @xmath8 values range from @xmath24 to @xmath81 whereas , to attain complete heterogeneity for a network of @xmath82 nodes , the @xmath8 values should range from @xmath24 to @xmath83 .
the above result also implies that for any network that is evolving or growing , for example the sf network where the nodes are added with preferential attachment @xcite , the value of @xmath50 generally keeps on decreasing with increasing @xmath1 . in the next section
, we numerically study the variation of @xmath50 with different network parameters for various synthetic networks .
in this section , we analyze @xmath84 different classes of complex networks , namely , the rgs of erdos - renyi , the sf networks and the networks derived from the time series of chaotic dynamical systems , called recurrence networks ( rns ) whose details are discussed in v.c . for rgs , the degree distribution is poissonian centered around an average degree @xmath85 where @xmath2 is the probability that two nodes in the network is connected . in fig .
[ f.4 ] , we show the degree distribution and the @xmath8 spectrum for rgs of @xmath47 different @xmath2 values with @xmath1 fixed at @xmath76 .
the values of @xmath50 for all these networks are also shown .
the main result here is that the value of @xmath50 increases correspondingly with the range of @xmath8 values for a fixed @xmath1 .
we next consider how @xmath50 depends on @xmath2 and @xmath1 , the two basic parameters of the rg .
the effect of changing @xmath2 for a fixed @xmath1 as well as changing @xmath1 for a fixed @xmath2 are to shift the average @xmath8 value of the nodes in the rg .
since the degree distribution is approximately gaussian for large @xmath1 , the spectrum of @xmath8 values depends directly on the variance of the gaussian profile . as @xmath2 increases from zero for a fixed @xmath1 , the spectrum of @xmath8 values and hence @xmath50 increase correspondingly .
due to the obvious symmetry of the network with respect to the transformation @xmath86 , as @xmath2 increases beyond @xmath87 , @xmath50 starts decreasing .
thus the maximum value of @xmath50 is obtained for @xmath88 for any fixed @xmath1 . on the other hand , by increasing @xmath1 for any fixed @xmath2 , one expects the gaussian profile of the degree distribution to become sharper , thus decreasing @xmath50 .
these results are compiled in fig .
[ f.5 ] for three values of @xmath1 .
note that the minimum @xmath2 value that can be used for @xmath89 is @xmath90 and this decreases as @xmath1 increases . in the figure , we show the results starting from @xmath91 .
higher values of @xmath1 would involve very large computer memory requirements for large @xmath2 .
however , we have checked the variation of @xmath50 with @xmath1 for smaller @xmath2 values , say @xmath92 and @xmath93 , for @xmath1 up to @xmath94 and have found that the decrease is approximately exponential
. degree distributions ( inset ) and the distributions in log scale along with the @xmath8 spectrum for synthetic sf networks with four different values of @xmath3 and @xmath1 fixed at @xmath76 . in all cases ,
the values of @xmath50 and the minimum degree @xmath95 of the network are also shown .
as the spectrum of @xmath8 values increases , @xmath50 increases correspondingly for a fixed @xmath1 .
note that there appears to be a second peak with a gap in the distribution for @xmath96 .
this is size dependent effect due to the presence of many @xmath8 values having @xmath9 very close to zero .
it is also evident from the @xmath8 spectrum shown below each distribution .
for example , in the case @xmath97 where the @xmath8 spectrum is almost continuous without a visible gap , the scaling becomes more evident . ]
variation of @xmath50 with @xmath1 for synthetic sf networks with two different values of @xmath95 .
the same variation is shown in the inset in log scale indicating a clear power law in both cases . ] for sf networks , the degree distribution obeys a power law @xmath11 . to construct the sf network synthetically
, we use the basic scheme proposed by barabasi et al . @xcite . in this scheme , we start with a small number of initial nodes denoted as @xmath98 . as a new node
is connected , a fixed number of edges , say @xmath99 , is added to the network .
this number represents the minimum number of node degree , @xmath95 , in the network .
the new edges emerging at node creation are distributed according to the preferential attachment mechanism .
the two parameters , @xmath98 and @xmath95 , determines the value of @xmath3 as the network evolves .
we have constructed sf networks of different @xmath3 by changing both @xmath98 and @xmath95 . in fig .
[ f.6 ] , we show the degree distribution and the corresponding @xmath8 spectrum for sf networks with four different @xmath3 and @xmath95 , with @xmath1 fixed at @xmath76 .
we find that the @xmath8 spectrum and hence the value of @xmath50 depend directly on @xmath95 as can be seen from the figure .
in other words , for a sf network of fixed @xmath1 , @xmath50 increases as the value of @xmath95 increases .
the variation is approximately linear for @xmath95 in the range @xmath24 to @xmath100 .
more interesting is the variation of @xmath50 with @xmath1 for a fixed @xmath95 . in fig .
[ f.7 ] , we show the variation of @xmath50 as @xmath1 increases from @xmath82 to @xmath101 for two different sf networks with @xmath102 and @xmath100 .
this variation is also shown in the inset in a log scale in the same figure indicating that @xmath50 varies as @xmath103 , where the value of @xmath104 is found to be @xmath105 for @xmath102 and @xmath106 for @xmath107 for the given range of @xmath1 values .
however , we do not claim that this variation is , in general , a power law since we have only tested a limited range of @xmath1 values .
this needs to be explicitly tested with other alternatives with a wider range of @xmath1 values .
recently , a new class of complex networks has been proposed for the characterization of the structural properties of chaotic attractors , called the recurrence networks ( rns)@xcite .
they are constructed from the time series of any one variable of a chaotic attractor . from the single scalar time series
, the underlying attractor is first constructed in an embedding space of dimension @xmath108 using the time delay embedding @xcite method .
any value of @xmath108 equal to or greater than the dimension of the attractor can be used for the construction of the attractor .
the topological and the structural properties of this attractor can be studied by mapping the information inherent in the attractor to a complex network and analyzing the network using various network measures . to construct the network , an important property of the trajectory of any dynamical system is made use of , namely , the recurrence @xcite . by this property
, the trajectory tends to revisit any infinitesimal region of the state space of a dynamical system covered by the attractor over a certain interval of time . to convert the attractor to a complex network
, one considers all the points on the embedded attractor as nodes and two nodes @xmath4 and @xmath13 are considered to be connected if the distance @xmath109 between the corresponding points on the attractor in the embedded space is less than or equal to a recurrence threshold @xmath110 .
selection of this parameter is crucial in getting the optimum network that represents the characteristic properties of the attractor .
the resulting complex network is the rn which , by construction , is an unweighted and undirected network .
the adjacency matrix @xmath111 of the rn is a binary symmetric matrix with elements @xmath112 ( if nodes @xmath4 and @xmath13 are connected ) and @xmath113 ( otherwise ) .
more details regarding the construction of the rn can be found elsewhere @xcite .
here we follow the general framework recently proposed by us @xcite to construct the rn from time series .
top panel shows the variation of @xmath50 with @xmath1 for rns constructed from lorenz ( filled triangle ) and henon ( filled circle ) attractor time series .
bottom panel shows the variation of @xmath50 with @xmath108 for fixed @xmath1 for lorenz ( filled triangle ) and duffing ( filled circle ) attractors . in both graphs ,
the error bar comes from the standard deviation of values for @xmath50 computed from time series with ten different initial conditions . ] for generating the time series , we use the equations and the parameter values given in @xcite for all chaotic systems . for continuous systems ,
we have used the sampling rate @xmath114 for generating the time series .
the time delay used for embedding is the first minimum of the autocorrelation .
we first study how the value of @xmath50 varies with the number of nodes @xmath1 for rns . in fig .
[ f.8 ] ( top panel ) , we show the results for the lorenz attractor and the henon attractor .
it is evident that for large value of @xmath1 , @xmath50 converges to a finite value . we have checked and verified that this is true for other low dimensional chaotic attractors as well .
it is found that once the basic structure of the attractor is formed , the value of @xmath50 remains independent for further increase in @xmath1 .
in other words , the range of @xmath8 values increases with @xmath1 to keep the value of @xmath50 approximately constant .
this result also follows from the statistical invariance of the degree distribution of the rn as has already been shown @xcite .
next , we consider the variation of @xmath50 with embedding dimension @xmath108 .
this is also shown in fig .
[ f.8 ] ( bottom panel ) for two standard chaotic attractors .
it is clear that the value of @xmath50 converges for @xmath115 in both cases .
this is because , the fractal dimension of both these attractors are @xmath116 .
we have already shown @xcite that the degree distribution of the rn from any chaotic attractor converges beyond the actual dimension of the system .
thus , @xmath50 turns out to be a unique measure for any chaotic attractor independent of both @xmath108 and @xmath1 . from the construction of rns ,
the range of connection between two nodes is limited by the recurrence threshold @xmath110 .
hence the degree of a node in the rn and the probability density around the corresponding point over the attractor are directly related .
for example , for the rn from a random time series , every node has degree close to the average value @xmath10 since the probability density over the attractor is approximately the same .
one can show that the degree distribution of the rn from a random time series is gaussian for large @xmath1 .
thus , the @xmath8 spectrum of the rn is indicative of the range in the probability density variations over the attractor , which in turn , is characteristic of the structural complexity of the attractor .
we have already shown that the measure @xmath50 proposed here is indicative of the diversity in the @xmath8 spectrum . moreover
, it is found to have a specific value for a given attractor independent of @xmath108 and @xmath1 .
it is well known that the statistical measures derived from the rns characterize the structural properties of the corresponding chaotic attractor . in particular , since every point on the attractor is converted to a node in the rn and the local variation in the node degree is a manifestation of the variation in the local probability density over the attractor , the measure @xmath50 can serve as a single index to quantify the structural complexity of a chaotic attractor through rn construction . in table
i , we compare the values of @xmath50 for rns constructed from several standard chaotic attractors . in all cases ,
the saturated value of @xmath50 converged upto @xmath117 is shown . in each case , ten different rns are constructed changing the initial conditions and the average is shown with standard deviation as the error bar .
the results indicate that that among the continuous systems compared , the lorenz attractor is structurally the most complex while in the case of 2d discrete systems , lozi attractor is found to be the most diverse in terms of the probability density variations .
finally , it will also be interesting to see how the value of @xmath50 is affected by adding noise to the chaotic time series . to test this ,
we generate data adding different percentages of noise to lorenz data .
when the value of @xmath50 is computed , it is found that the value reduces systematically with the increase in the noise percentage and approaches the value of noise for a noise level @xmath118 .
this result clearly indicates that the measure will be useful for analysing the real world data .
the value of @xmath50 for random time series with @xmath119 and @xmath120 is found to be @xmath121 . to get a comparison with the values of conventional complex networks , we compute @xmath50 for a rg with @xmath122 that gives the same @xmath10 as that of the rn from random time series and a typical sf network with @xmath123 with @xmath119 in both cases .
the average of ten different simulations is taken .
we find that @xmath124 for the rg which is exactly same as the rn from random time series and @xmath125 for the sf network .
.comparison of @xmath50 for several standard chaotic attractors . [ cols="<,^,^,^,^,^,^,^",options="header " , ] since we have to restrict to the case of unweighted and undirected networks , we could use only a small subset from the very large variety of real world networks that are mostly weighted or directed . to extend the measure to directed networks , one has to consider the in - degree and out - degree distributions and find the heterogeneity separately . in order to generalise the measure to weighted networks ,
the distribution of the weight or strength of the nodes in the network @xcite , rather than the simple degree distribution is to be considered and define the measure accordingly . for example , for unweighted and undirected networks , all the links are equivalent and hence the degree of @xmath126 node @xmath5 is just the sum of the links connected to node @xmath4 . on the other hand , for weighted networks ,
each link is associated with a weight factor @xmath127 and hence the degree @xmath5 should be generalised to the sum of the weights of all the links attached to node @xmath4 : @xmath128 thus the degree distribution needs to be generalised to the _ strength distribution _ @xmath129 , which is the probability that a given node has a strength equal to @xmath130 @xcite .
the equation for heterogeneity for weighted networks can be modified accordingly .
however , it should be noted that the weight factors are assigned based on different criteria depending on the specific system or interaction the network tries to model .
modifying the measure by incorporating the specific aspects of interaction , the measure itself becomes network specific .
the measure that we propose here is independent of such details and is representative of only the diversity of node degrees in a network , determined completely by the simple degree distribution .
complex networks and the network based quantifiers have become useful tools for the analysis of many real world phenomena .
physical , biological and social interactions are increasingly being modeled and characterized through the language of complex network .
an important measure for the characterization of any complex network is its heterogeneity measured in terms of the diversity of connection reflected through its node degrees . here
we introduce a measure to quantify this diversity which is applicable to networks of different topologies .
this measure is minimum ( equal to zero ) for a completely homogeneous network with all @xmath131 . to get the upper bound for the measure
, we consider the logical limit of heterogeneity possible in a network of @xmath1 nodes where nodes of degree varying from 1 to @xmath25 are present , whose heterogeneity is normalised as 1 .
while considering this network of limiting heterogeneity , we also prove that the degree that is repeated ( or shared by two nodes ) is @xmath28 if @xmath1 is even and @xmath132 if @xmath1 is odd .
also , the measure that we propose here uniquely quantifies the diversity in the node degrees in the network which is characteristic of the type and range of interactions the network represents .
the diversity also depends on the topology of the resulting network .
for example , for rgs , the diversity is limited since most degrees are centered around the average value @xmath10 while the sf networks are comparatively more diverse due to the presence of hubs .
the proposed measure can quantify this diversity in the node degrees irrespective of the topology of the network . by applying the proposed measure
, we compute the heterogeneity of various unweighted and undirected networks , synthetic as well as real world .
we study numerically how the heterogeneity varies for rg with respect to the two parameters @xmath2 and @xmath1 , while for sf networks the variation of heterogeneity with respect to @xmath3 as well as @xmath1 are analysed . to illustrate the practical relevance of the measure ,
we analyse the rns constructed from the time series of chaotic dynamical systems and highlight its utility as a quantifier to compare the structural complexities of different chaotic attractors .
as has already shown by us @xcite , the nonlinear character and chaotic dynamics underlying the time series can be distinguished from the rns through the usual characteristic measures of complex networks like cc or cpl .
however the subtle differences between the degree distributions of rns from different chaotic systems are not clearly evident from their cc or cpl .
we find these can be quantified uniquely using the proposed measure of degree heterogeneity .
* data accessibility * : we use the open source software gephi _
www.gephi.org/ _ for the construction of all networks .
all the codes used for computing the heterogeneity and other network measures are available at _ https://sites.google.com/site/kphk11/home . _ * author contributions * : the initial idea for the paper started from the discussions of rj and kph .
the codes for computations of network measures were developed in association with rm .
the interpretation of the results and the expert guidance for the manuscript were finalised after a series of discussions with ga .
all the authors gave the final approval of the manuscript .
* acknowledgements * : the authors thank one of the anonymous referees for a thorough review of the manuscript and especially for pointing out the correspondence of our heterogeneity measure with the entropy measure of complex networks .
rj and kph acknowledge the computing facilities in iucaa , pune .
|
we propose a novel measure of degree heterogeneity , for unweighted and undirected complex networks , which requires only the degree distribution of the network for its computation .
we show that the proposed measure can be applied to all types of network topology with ease and increases with the diversity of node degrees in the network .
the measure is applied to compute the heterogeneity of synthetic ( both random and scale free ) and real world networks with its value normalized in the interval @xmath0 $ ] . to define the measure ,
we introduce a limiting network whose heterogeneity can be expressed analytically with the value tending to 1 as the size of the network @xmath1 tends to infinity .
we numerically study the variation of heterogeneity for random graphs ( as a function of @xmath2 and @xmath1 ) and for scale free networks with @xmath3 and @xmath1 as variables . finally , as a specific application
, we show that the proposed measure can be used to compare the heterogeneity of recurrence networks constructed from the time series of several low dimensional chaotic attractors , thereby providing a single index to compare the structural complexity of chaotic attractors . * keywords : complex networks , heterogeneity measure , recurrence network analysis *
|
the work of d.h.e .
gross @xcite has recently pointed out the attention on the fact that a microcanonical description of systems which may display phase transitions is in general more adequate than the traditional canonical one .
this is because the canonical description may `` smear out '' important information contained in the microcanonical description which is richer @xcite .
for example negative specific heats , which have recently been observed experimentally in mesoscopic systems @xcite , can be accounted for in the microcanonical ensemble but not in the canonical one @xcite .
indeed it is well known that canonical ensemble and microcanonical ensemble are not in general equivalent , even when the thermodynamic limit is considered @xcite .
the statistical mechanical analysis of physical systems based on canonical ensemble is quite well established and universally agreed upon .
roughly speaking one has to compute the partition function @xmath0 and derive the thermodynamics of the system from the free energy @xmath1 .
things are not quite so broadly agreed upon in the case of the microcanonical ensemble .
in fact since the pioneering works of boltzmann and gibbs two possibilities were given for the microcanonical analysis of physical systems , which correspond to two different definitions of entropy ( see for example the textbook of gibbs @xcite or the more recent textbook of huang @xcite ) . following @xcite and
@xcite we shall refer to these two entropies as `` surface entropy '' and `` volume entropy '' .
the surface entropy is defined as : is equal to @xmath2 . ]
@xmath3 where @xmath4 with @xmath5 denoting dirac delta function .
sometimes this is also referred to as boltzmann entropy or boltzmann - planck entropy .
the volume entropy is defined as : @xmath6 where @xmath7 with @xmath8 denoting heaviside step function .
the quantities @xmath9 and @xmath10 are related through the differential equation @xcite : @xmath11 where the prime symbol denotes derivation with respect to @xmath12 .
these entropies are named surface entropy and volume entropy because they are calculated as the logarithm of the area of the hyper - surface of constant energy in phase space and the volume of phase space that it encloses respectively . with reference to the literature about microcanonical phase transitions
the surface entropy is certainly the most popular .
for example barr _ et .
@xcite method based on large deviation techniques uses the surface entropy .
the surface entropy is used also in rugh s microcanonical formalism adopted in ref .
the strongest advocate of surface entropy is perhaps gross @xcite .
nonetheless pioneers of microcanonical phase transitions , such as thirring @xcite and lynden - bell @xcite , used the volume entropy .
the two entropies coincide in the thermodynamic limit but when the number of degrees of freedom of the system under study is small relevant differences may appear , therefore it is necessary to choose properly .
some authors @xcite have already pointed out that the surface entropy is not adequate when dealing with small systems because it does not account properly for finite - size effects . on the other hand
there is a number of theoretical reasons to prefer the volume entropy when the number of degrees of freedom becomes small .
here we shall review these reasons and we will illustrate the employment of the volume entropy in the study of phase transition with a small lennard - jones chain which displays a region of negative heat capacity .
in this section we will summarize some old and recent results concerning the volume entropy .
these results indicate that , no matter the number of degrees of freedom of the system under study , the volume entropy always provides a good mechanical analogue of thermodynamic entropy .
for example helmholtz @xcite proved that the logarithm of the area enclosed in phase space by the trajectory of a 1-dimensional system ( i.e. , the 1d volume entropy @xmath13 ) provides a mechanical analogue of physical entropy in the sense that if one considers the quantities @xmath14 and @xmath15 where @xmath16 is the kinetic energy , @xmath17 is an external parameter on which the hamiltonian depends and @xmath18 denotes time average , then , @xmath19 this result , known as helmholtz theorem @xcite , says that the volume entropy is a good mechanical analogue of thermodynamic entropy in the sense that it reproduces exactly the fundamental law of thermodynamics ( i.e. the _ heat theorem _ ( [ eq : heattheo ] ) ) .
the helmholtz theorem has been recently generalized to multi - dimensional ergodic ( i.e. metrically indecomposable ) systems ( see ref
. @xcite or ref .
@xcite for a different but equivalent approach ) .
the resulting generalized helmholtz theorem essentially states that the volume entropy reproduces the heat theorem _ no matter the number of degrees of freedom @xmath20_. the same can not be said about surface entropy which has been proved to reproduce it only up to corrections of the order @xmath21 @xcite . as a matter of fact
gibbs presents the volume entropy in his celebrated _ principles of statistical mechanics _ @xcite as the entropy that naturally satisfies the fundamental principle of thermodynamics ( that is the heat theorem ) @xcite .
hertz @xcite pointed out that the volume entropy is an adiabatic invariant already in 1910 , and based his approach to statistical mechanics on it . among the textbooks that adopt the same approach , those of mnster @xcite , becker @xcite , and the more recent book of berdichevsky @xcite are worth mentioning .
adiabatic invariance is another good property of volume entropy because it reproduces quite well clausius requirement that `` for every quasi static process in a thermally isolated system which begins and ends in an equilibrium state , the entropy of the final state is equal to that of the initial state '' @xcite .
of course the surface entropy is not an adiabatic invariant , although it becomes approximately such as the number of degrees of freedom increases @xcite .
very recently it has been also proved that non - adiabatic transformations occurring in isolated systems which are initially in a state of thermal equilibrium always result in an increase of the expectation value of the volume entropy @xcite .
this result too does hold _ no matter the number of degrees of freedom @xmath20 _ and can not be proved in general for surface entropy .
thus the volume entropy explains quite satisfactorily also clausius law of entropy increase `` for every non quasi static process in a thermally isolated system which begins and ends in an equilibrium state , the entropy of the final state is greater than that of the initial state '' @xcite .
recently more and more authors are becoming aware of the theoretical value of volume entropy .
for example , on the basis of a laplace transform technique for the microcanonical ensemble , pearson _ et .
@xcite reached the conclusion that the volume entropy `` is the most correct definition for the entropy , even though it is unimportant for any explicit numerical calculation '' , meaning that in the thermodynamic limit the difference with surface entropy becomes negligible . on the other hand for small system
, such intrinsic correctness of the volume entropy becomes very important .
adib @xcite argues that the finite size corrections to surface entropy found in ref .
@xcite would be unneeded if the volume entropy were used instead .
it is worth mentioning that the volume entropy has another property that is particularly important for small systems which have negative heat capacity , namely it is a naturally _ nonextensive _ entropy . according to lynden - bell @xcite , in fact
, systems with negative heat capacity are necessarily _ nonextensive_. the property of nonextensivity of volume entropy follows directly from the composition rule of enclosed volumes , @xmath22 @xmath23 , of two systems with total energy @xmath24 @xcite : @xmath25 which is not a simple multiplication but a form of convolution which accounts for all possible partitions of energies between the two systems .
it has to be stressed that , despite of what is often stated in literature , the surface entropy is nonextensive too , as the composition rule for surface integrals @xmath26 is the convolution , not the multiplication @xcite : @xmath27 in sum , the volume entropy accounts for certain basic principles of thermodynamics , like the heat theorem and clausius formulation of the second law equally well for large and small systems , whereas the surface entropy accounts for them only in the case of large system .
for this reason it is the most appropriate _ mechanical analogue of thermodynamic entropy _ when dealing with small systems .
according to the helmholtz theorem @xcite the mechanical analogue of thermodynamic entropy of a one dimensional system is @xmath28 where @xmath29 denote the turning points of the trajectory .
if the potential is such that there is only one trajectory per energy level ( ergodicity ) , then @xmath13 satisfies eq .
( [ eq : heattheo ] ) @xcite .
nonetheless , if the system has more than one trajectory per energy level for a certain range of energies and the system is on _ one _ of them , still we can use the above formula and say that the heat theorem is satisfied as long as the energies considered lye within that energy range . in this case @xmath30 and @xmath31 would be calculated as time averages over the _ actual _ trajectory and @xmath13 would be given by the area enclosed by _ that _ trajectory only .
let us illustrate this with a practical example .
let us consider a 1d chain composed of three particles which interact via a lennard - jones potential .
let us fix the position of two of them and let us place the third one in between , so that the first two particles act as walls of a 1d box .
let us now study the behavior of the particle inside the box .
let the interaction potential be : @xmath32 and let us place the `` walls '' at @xmath33 .
then the particle in the box is subject to the following potential : @xmath34 for values of @xmath17 larger than a certain critical value @xmath35 , this system has a critical energy @xmath36 such that for energy below @xmath37 ergodicity is broken and there are two trajectories per energy level . above @xmath38 the dynamics is ergodic and there is only one trajectory per energy level . in a lennard - jones box of size @xmath39 .
the separatrix corresponds to the critical energy @xmath40 . below the critical energy
there are two distinct trajectories ( the dynamics is not ergodic).,width=8 ] figure [ fig : phasespace ] shows a contour plot of various energy levels in phase space for a particle of mass @xmath41 in the lennard - jones box of size @xmath39 . for energy @xmath42 we have a separatrix . below @xmath38 the curve of constant energy splits into two disconnected curves , whereas for values of @xmath12 larger than @xmath38 we have only one curve . below the critical energy the volume integral @xmath10 is given by the area enclosed by
_ one _ of the two possible trajectories .
as the energy crosses the critical value the integral @xmath10 jumps discontinuously . in formulae we have : @xmath43 \nonumber\\ \times \int \left[2m(e-\varphi(x , v))\right]^{1/2}_+ dx\end{aligned}\ ] ] the symbol @xmath44^{1/2}_+$ ] denotes a function that is equal to @xmath45 for @xmath46 and is null otherwise .
the volume entropy , which is calculated by taking the logarithm of the expression above , then displays a jump at the critical energy as well . in a lennard - jones box of size @xmath39
the discontinuity of entropy at the critical energy signals a discontinuous phase transition.,width=8 ] figure [ fig : entropy ] shows a plot of @xmath47 as a function of @xmath12 , for the values @xmath41 and @xmath39 .
the critical energy is @xmath40 . in a lennard - jones box of size @xmath39 .
the curve displays a region of negative specific heat . at the critical energy the temperature goes to zero.,width=8 ]
figure [ fig : temperature ] shows the temperature plotted against the energy calculated , according to eq .
( [ eq : heattheo ] ) , as : @xmath48 there is a region of negative slope in the graph which correspond to a _ negative heat capacity_.
the example provided in the previous section is perhaps too simple to be of interest to any specific physical problem .
nonetheless it illustrates qualitatively the mechanism of microcanonical phase transition as captured by the volume entropy .
such phase transitions are associated with the crossing of separatrix trajectories , for which the dynamics of the system has no finite time scale .
the figures show neatly that at the separatrix energy the entropy has a discontinuous jump , the temperature goes to zero , and for energies below the critical energy we have a region of negative heat capacity .
these are not specific features of the system studied potential studied with volume entropy @xcite and has been predicted for the pendulum @xcite . ] .
whenever a separatrix is crossed there is a sudden _ open - up _ of a larger portion of phase space for the trajectory to _ enclose _ which leads to a discontinuity in the entropy .
further , at the separatrix , the period of motion , which for a well known theorem of classical mechanics is given by @xmath49 @xcite , becomes infinite .
therefore the temperature , _
i.e. _ , @xmath50 goes to zero . since the temperature @xmath51 is definite positive , below the critical energy there _
necessarily _ is a region of negative slope , that is negative heat capacity .
the appearance of a negative heat capacity is associated with a convex intruder in the entropy ( see fig . [
fig : entropy - inset ] ) which signals the approach to the separatrix from below .
it is important to notice that using the surface entropy would lead to a _ drastically different _ result . in this case
the temperature would be calculated as @xmath52 , which might not tend to zero at the critical energy !
note also that @xmath53 is not proportional to the average kinetic energy and can be negative .
therefore , in agreement with ref . @xcite we believe that surface entropy is not suited for low dimensional systems with broken ergodicity . the volume entropy could be used to address microcanonical phase transitions in small dimensional systems with either long or short range interactions , like the @xmath54 model , chains of particles interacting via lennard - jones potential @xcite or the hamiltonian mean field model @xcite .
all these models are expected to undergo a breaking of ergodicity @xcite , thus there are separatrix trajectories and possible phase transitions that the volume entropy can detect .
the advantage of using the volume entropy is that it provides a good mechanical analogue of thermodynamic entropy even for small system , thus accounting properly for the finite - size effects . as the development of technology is allowing experimentalists to probe the thermodynamic behavior of smaller and smallers systems
, this is becoming an increasingly important task .
the main limitation of the present approach is that it is restricted to classical statistical mechanics , thus it does not account for quantum - mechanical phenomena .
|
when studying the thermodynamic properties of mesoscopic systems the most appropriate microcanonical entropy is the volume entropy , i.e. the logarithm of the volume of phase space enclosed by the hypersurface of constant energy . for systems with broken ergodicity ,
the volume entropy has discontinuous jumps at values of energy that correspond to separatrix trajectories .
simultaneously there is a convex intruder in the entropy function and a region of negative specific heat below such critical energies .
we illustrate this with a simple model composed of a chain of 3 particles which interact via a lennard - jones potential .
|
back in the dawn of prehistory , when i was a student , murray gell - mann was an inspiration to me . his work dominated the particle physics landscape that i entered then : strangeness , @xmath0 theory , the eightfold way , quarks , current algebra , the renormalization group , and so much more , and laid the basis for the developments that have occurred since .
these fundamental contributions are now so embedded in the fabric of particle physics that perhaps we sometimes forget to remember them and celebrate their originator with all the respect he deserves .
it is therefore a pleasure for me to participate in this meeting honouring murray and his achievements , and a privilege to be given the opportunity to speak here about the next chapter in particle physics that is now unfolding .
there is a standard list of open questions beyond the standard model of particle physics .
what is the origin of particle masses , and are they due to a single elementary higgs boson , or to something else ?
why are there so many types of matter particles , and is the answer to this question related to the origin of matter in the universe ? what is the nature of the cold dark matter that makes up some 80% of the matter in the universe ? how to unify the fundamental forces ? how to construct a quantum theory of gravity ?
the lhc may be able to address all these questions .
one of its main motivations has been to solve the mass problem , and its experiments should tell us definitively whether or not there exists a higgs boson resembling that in the standard model .
there is a large class of models in which cold dark matter is composed of particles that were in thermal equilibrium in the early universe , in which case they should weigh @xmath1 tev , and be produced at the lhc .
one example of such a theory is supersymmetry , which would also assist in the unification of the fundamental forces .
measuring the masses of supersymmetric particles , if they exist , would be a great way of testing predictions based on such theories .
supersymmetry and extra dimensions are key aspects of string theory , the only promising candidate for a consistent quantum theory of gravity , which could be tested in very novel ways if the lhc produces microscopic black holes .
what are the prospects that the lhc might cast light on these enticing scenarios ?
in the standard model , particles acquire their masses from couplings to a universal scalar field whose associated quantum , called the higgs boson , has become the ` holy grail ' of particle physics .
direct searches for the higgs boson by experiments at the lep accelerator established the lower bound @xmath2 gev @xcite .
precision electroweak data are sensitive to @xmath3 through quantum corrections , and yield the preferred range @xcite @xmath4 the 95% confidence - level upper limit on the mass of the higgs boson is 157 gev if only the precision electroweak data are used , or 186 gev if the lep direct lower limit is included .
recently , the tevatron experiments cdf and d0 have excluded a range of heavier masses @xcite : @xmath5 a combined fit to all the data , shown in fig . [
fig : gfitter ] , yields the asymmetric estimate @xcite @xmath6 at the 68% confidence level . if the higgs mass is large , so is the higgs quartic self - coupling @xmath7 , and renormalization - group effects cause it to blow up at some relatively low scale @xmath8 , as seen in fig .
[ fig : eeghr1 ] , heralding the appearance of new non - perturbative physics . on the other hand ,
if @xmath3 is small , negative renormalization by the large @xmath9-quark yukawa coupling drives @xmath10 , leading to an instability in the electroweak vacuum , unless new physics such as supersymmetry intervenes @xcite .
only a narrow range of @xmath11 gev is compatible with the survival of the standard model at all scales up to the planck mass
. this would be the ` maximal conceivable disaster ' scenario for the lhc : a single standard model higgs boson and nothing else !
the precision electroweak data favour small values of @xmath3 , and the combination with the tevatron exclusion ( [ tevatron ] ) excludes the blow - up scenario at the 99% confidence level @xcite .
the unstable - vacuum scenario is preferred , but the ` disaster ' scenario is not even disfavoured at the 1-@xmath12 level .
the standard list of the primary physics objectives of the lhc includes the search for this higgs boson ( or whatever replaces it ) , the nature of dark matter , the primordial plasma that filled the universe when it was less than a microsecond old , and matter - antimatter asymmetry .
it is worth remembering that the cross sections for producing interesting heavy new particles at the lhc are typically @xmath13/tev)@xmath14 and possibly with additional factors @xmath15 , much less than the total cross section @xmath13/@xmath16 .
typical cross sections for producing the higgs boson or supersymmetric particles are @xmath17 of the total cross section : looking for them will resemble searching for a needle in 100,000 haystacks !
for this reason , the higgs boson will not appear as soon as the lhc produces high - energy collisions .
the left panel of fig .
[ fig : pofpa ] shows the amount of luminosity required at 14 tev in the centre of mass either to exclude the higgs boson at the 95% confidence level or to claim a 5-@xmath12 discovery , as a function of its mass .
we see that a couple of hundred inverse picobarns could get the lhc into the exclusion business , while several inverse femtobarns would be needed to guarantee detection at any mass .
there is an intermediate range of masses @xmath18 gev where detection via @xmath19 decays is relatively easy , but in the preferred low - mass range ( [ all ] ) these decay modes are less important and higgs detection becomes more difficult .
the stakes in the higgs search are high , since it is key to many puzzles in cosmology as well as particle physics . how is electroweak symmetry broken ? is there such a thing as an elementary scalar field ? what is the fate of the standard model at high scales ?
did mass appear when the universe was a picosecond old through an electroweak phase transition ?
did cp - violating higgs interactions help create the matter in the universe ? did another elementary scalar field , the inflaton , cause ( near-)exponential expansion of the early universe , and hence make the universe so big and old ?
the typical scale of vacuum ( dark ) energy in the higgs potential is some 60 orders of magnitude larger than the measured value : why is there so little dark energy ? discovering the higgs boson , or proving that it does not exist , may not answer all these questions , but it may be our best experimental probe of them . the most exciting higgs scenario for the lhc may be that it is proven _ not _ to exist .
that would really shake us smug theorists out of our torpor !
the best alternative known to me would be to break electroweak symmetry by boundary conditions in extra dimensions @xcite .
finding evidence for extra dimensions might be easier to explain to a lay audience than an ill - named god particle .
there are many motivations for supersymmetry : its beauty , it would render the hierarchy of mass scales more natural , it predicts a light higgs boson , it stabilizes the electroweak vacuum , it facilitates grand unification , and it is apparently needed for the consistency of string theory . here
i focus on the fact that supersymmetry could provide the dark matter required by astrophysics and cosmology @xcite . in many supersymmetric models ,
there is a multiplicatively - conserved quantum number called @xmath20-parity , that may be represented as @xmath21 , where @xmath22 is spin , @xmath23 is lepton number , and @xmath24 is baryon number .
it is easy to verify that known particles have @xmath25 , whereas their putative supersymmetric partners , differing in spin by 1/2 , would have @xmath26 .
the conservation of @xmath20 parity would imply that sparticles are produced in pairs , that heavier sparticles decay into lighter ones , and that the lightest supersymmetric particle ( lsp ) is stable , because it has no legal decay mode .
hence , it should still be around after being produced early in the big bang , and could provide the needed dark matter .
presumably , the lsp is some neutral , weakly - interacting particle , otherwise it would have bound to ordinary matter and been detected by now . in this case , the favoured signature for supersymmetry at colliders is missing transverse energy carried away by the invisible dark matter particles .
there are important constraints on supersymmetry due to the absence of sparticles at lep and the tevatron , the lep lower limit on @xmath3 and the consistency of @xmath27-quark decays with the standard model . some hint of new physics at the tev scale may be provided by the measurement @xcite of the anomalous magnetic moment of the muon , @xmath28 , that could be explained by supersymmetry , although there are still uncertainties in the standard model calculation of @xmath28 @xcite .
the measured density of dark matter , @xmath29 , provides a very tight constraint on some combination of supersymmetric model parameters , if the lsp provides the dark matter .
the interplay of these constraints is shown in fig .
[ fig : tb10 ] for one particularly simple supersymmetric model with universal supersymmetry - breaking parameters @xmath30 and @xmath31 assumed at the gut scale , the cmssm @xcite .
we have recently made a global supersymmetric fit using a frequentist approach to analyze the precision electroweak data , the lep higgs mass limit , the cold dark matter density , @xmath27-decay data and ( optionally ) @xmath28 .
we combined the likelihood functions from these different observables to construct a global likelihood that can be used to infer preferred regions of the supersymmetric parameter space @xcite .
we see in fig .
[ fig : willis ] that the preferred regions of the @xmath32 planes in both the cmssm and a model with common non - universal supersymmetry - breaking contributions to the higgs masses ( nuhm1 ) correspond to relatively low masses where the relic lsp density is brought into the wmap range by coannihilations with light sleptons , particularly the lighter stau .
the ` focus - point ' region at large @xmath31 is disfavoured , principally but not exclusively by @xmath28 .
if one drops this constraint , considerably larger ranges of @xmath31 and @xmath30 would be allowed , though small values are slightly preferred by other data @xcite .
[ fig : discovery ] compares the preferred regions of these @xmath32 planes with the 5-@xmath12 discovery reach of the lhc with given amounts of integrated luminosity at certain centre - of - mass energies .
the 5-@xmath12 discovery reach of the lhc with 1/fb of luminosity at 7 tev in the centre of mass would probably include all the 68% confidence - level regions in fig .
[ fig : discovery ] .
if supersymmetric particles are this light , they would contribute significantly to the electroweak radiative corrections that test the standard model at the quantum level . using low - energy precision measurements as input , one may use radiative corrections to predict high - energy observables such as @xmath33 , @xmath34 and @xmath3 . within the standard model ,
the predictions of @xmath33 and @xmath34 have been very successful , and one would hope that their success would not be undermined in supersymmetric extensions of the standard model . on the one hand , supersymmetry predicts a restricted range for @xmath3 , tending to reduce an uncertainty in the calculations of the radiative corrections . on the other hand , light sparticles
could themselves contribute significantly to the radiative corrections . as seen in fig .
[ fig : mtmw ] , the net effect of supersymmetry in the context of the global fits to the cmssm and nuhm1 introduced above is to _ reduce _ the predicted ranges of @xmath33 and @xmath34 @xcite .
the predictions are in both cases in good agreement with the measured values : indeed , the supersymmetric predictions for @xmath34 are even slightly better than those in the standard model .
another fashionable scenario for physics beyond the standard model is the appearance of large extra dimensions .
their possible existence was suggested decades ago , and given extra momentum by string theory , which appears to require their appearance at some distance scale .
however , our current understanding of string theory does not offer any firm guidance as to their possible sizes .
it has been suggested that , if they are large they might solve or at least alleviate the mass hierarchy problem .
perhaps i am showing my age , but i find large extra dimensions less attractive than supersymmetry .
nevertheless , the final arbiters will be experiments , particularly at the lhc .
they may find kaluza - klein excitations of standard model particles , or they may discover missing energy leaking into extra dimensions , or they may discover that gravity becomes strong at the tev scale and lhc collisions produce microscopic black holes .
this last speculation has sparked concerns that have no scientific basis .
the same theories that predict microscopic black holes also predict that they decay via hawking radiation , with interesting grey - body factors that would be fascinating tests of string theory or any other quantum theory of gravity .
even if they were stable , the continued existence of the earth and other celestial objects tells us to fear nothing from the lhc @xcite .
do i hear the sound of other axes being ground ?
following the first start of the lhc on september 10th , 2008 and the electrical fault nine days later that laid the lhc low , there was jubilation on november 20th , 2009 when the lhc was restarted .
this was redoubled 3 days later when the first collisions were observed at 900 gev in the centre of mass .
soon afterwards , the beams were successfully ramped up to 1.18 tev and the highest - energy human - made collisions were seen . remarkably quickly
, the lhc experiments were able to reconstruct the decays of known particles and remeasure their masses .
two - photon decays of the @xmath35 and @xmath36 were quickly seen , but seeing the higgs will take a while , as discussed above : no higgs yet ! both atlas and cms have seen multi - jet events , and have shown distributions of the missing transverse energy .
so far , these agree only too well with monte carlo simulations : no supersymmetry yet !
multi - jet events occur at rates compatible with qcd : no hawking - decaying black holes yet !
while i was writing up this talk , the lhc produced its first collisions at 3.5 tev per beam on march 30th , 2010 , and fig .
[ fig : first ] shows some of the first events observed by the atlas and cms experiments .
the default lhc operating scenario is to collide at 3.5 tev per beam until the end of 2011 , aiming to accumulate 1/fb of integrated luminosity .
depending on the lhc running experience during this period , the lhc beam energy may be increased slightly during this period . as illustrated in fig .
[ fig:7tev ] , this should enable the lhc to at least equal the tevatron sensitivity to an intermediate - mass higgs boson weighing between 150 and 180 gev .
moreover , it would give the lhc a reach for new physics such as supersymmetry that would extend beyond the tevatron .
indeed , by discovering or excluding the gluino , this first physics run of the lhc may already be able to tell us whether supersymmetric are light enough to be produced with a 500 gev linear @xmath37 collider @xcite .
a long shutdown is then planned for consolidation of the lhc and its injectors , including the lhc magnet interconnects and training of the dipole magnets , making it possible to run the lhc at or close to its design energy of 7 tev per beam .
further in the future , there will be at least major upgrade of the lhc , incorporating linac4 and new interaction - region insertions .
the scope of possible further upgrades ( an spl ?
a higher extraction energy for the ps booster ?
replacement of the ps ?
new collision insertions ?
crab cavities ? ) are still under discussion , and will be decided only in the light of operational experience with the lhc .
back in 1982 , while she was prime minister of the uk , mrs . thatcher visited cern and was introduced to british physicists .
when she was told i was a theoretical physicist , she asked me what do you do ? " i explained that my job was to think of things for the experiments to look for , and hope they find something different .
would nt it be better if they found what you predicted ?
" asked mrs .
t. , who always liked _ her _ ideas to be vindicated .
my response was that , in that case , we would not be learning anything really new .
likewise , i sincerely hope that the lhc will be remembered in the history of physics for something _ not _ described in this talk .
it is a pleasure to thank harald fritzsch for his kind invitation to this enjoyable event , and k. k. phua and his team for their generous hospitality . above all , it is a pleasure to thank murray for his inspiration over the years .
r. barate _
[ aleph , delphi , l3 and opal collaborations , lep working group for higgs boson searches ] , phys . lett .
b * 565 * ( 2003 ) 61 [ arxiv : hep - ex/0306033 ] .
aleph , cdf , d0 , delphi , l3 , opal and sld collaborations , lep electroweak working group , tevatron electroweak working group and sld electroweak and heavy flavour groups , arxiv:0911.2604 [ hep - ex ] .
t. aaltonen _ et al . _
[ cdf and d0 collaborations ] , phys .
* 104 * ( 2010 ) 061802 [ arxiv:1001.4162 [ hep - ex ] ] .
h. flacher , m. goebel , j. haller , a. hocker , k. moenig and j. stelzer , eur .
j. c * 60 * ( 2009 ) 543 [ arxiv:0811.0009 [ hep - ph ] ] .
j. r. ellis and d. ross , phys .
b * 506 * ( 2001 ) 331 [ arxiv : hep - ph/0012067 ] .
j. ellis , j. r. espinosa , g. f. giudice , a. hoecker and a. riotto , phys .
b * 679 * ( 2009 ) 369 [ arxiv:0906.0954 [ hep - ph ] ] .
j .- j . blaising , a. de roeck , j. ellis , f. gianotti , p. janot , g. rolandi and d. schlatter , + http://council-strategygroup.web.cern.ch/council-strategygroup/bb2/ + contributions / blaising2.pdf .
m. davier , a. hoecker , b. malaescu , c. z. yuan and z. zhang , arxiv:0908.4300 [ hep - ph ] .
j. ellis and k. a. olive , chapter 8 of _ ` particle dark matter : observations , models and searches ' _ edited by gianfranco bertone , pp .
142 - 163 , cambridge university press hardback isbn 9780521763684 , arxiv:1001.3651 [ astro-ph.co ] . o. buchmueller _ et al . _ ,
jhep * 0809 * ( 2008 ) 117 [ arxiv:0808.4128 [ hep - ph ] ] . o. buchmueller _ et al .
_ , eur .
j. c * 64 * ( 2009 ) 391 [ arxiv:0907.5568 [ hep - ph ] ] .
o. buchmueller _ et al .
_ , phys .
d * 81 * ( 2010 ) 035009 [ arxiv:0912.1036 [ hep - ph ] ] .
s. b. giddings and m. l. mangano , phys . rev .
d * 78 * ( 2008 ) 035009 [ arxiv:0806.3381 [ hep - ph ] ] ; j. r. ellis , g. giudice , m. l. mangano , i. tkachev and u. wiedemann , j. phys .
g * 35 * ( 2008 ) 115004 [ arxiv:0806.3414 [ hep - ph ] ] .
atlas collaboration , + http://atlas.web.cern.ch/atlas/public/evtdisplay/events.html .
|
high - energy collisions at the lhc are now starting .
the new physics agenda of the lhc is reviewed , with emphasis on the hunt for the higgs boson ( or whatever replaces it ) and supersymmetry .
in particular , the prospects for discovering new physics in the 2010 - 2011 run are discussed .
cern - ph - th/2010 - 074
|
the fusion hindrance that appears in heavy ion reactions has been known for many years . now
the mechanism of fusion hindrance is gradually understood as to be due to an extra internal barrier between the touching configuration and the compound shape after overcoming the coulomb barrier @xcite .
however what causes the internal barrier is a difficult problem .
one of the hypotheses is that the internal barrier could be thought as the conditional saddle point in the liquid - drop potential as well as could be attributed to an effective barrier due to the dissipation of the incident kinetic energy @xcite .
since there are two barriers for the fusion , its theoretical description is divided into two consecutive steps : one is from infinity to the contact configuration passing over the coulomb barrier , and another one is from the contact point to the compound state overcoming the internal barrier .
this so - called two - step model explains the extra - push energy and furthermore gives an energy dependent fusion cross sections @xcite .
the measured residue cross section can be calculated by plugging a statistical evaporation model on the two - step model .
to evaluate the amplitude of the hindrance to the fusion , the size of the inner barrier is not enough .
the dissipation mechanism also plays an important role and one has to solve dynamically the diffusion over this barrier .
this was done analytically in one dimensional case assuming a parabolic barrier @xcite . here
, we will focus our attention on the appearance of the hindrance to the fusion that is due to the saddle point height measured from the energy of the touching configuration formed by the projectile and target nuclei of the incident channel .
in other words , there is no hindrance for cases without internal barrier .
therefore , the border between the normal and the hindered fusion is given by the condition that the contact configuration be on the top of the saddle point .
for mass - symmetric reactions , the fusion path between the touching configuration and the compound shape can be easily determined since the system evolves only along the radial degree of freedom . in order to determine
if the fusion hindrance occurred for a given reaction , we simply compare the position of the barrier and the touching configuration : the fusion exists while the barrier is located inner side of the touching point and vice versa .
@xcite systematically analyzed different symmetric reactions and drew a clear border between the two kinds of reactions ( with and without fusion hindrance ) .
how is the situation for mass - asymmetric cases ?
generally this problem becomes difficult because the system crosses a ridgeline , not a saddle .
it is then difficult to determine an effective reaction coordinate .
however there is a way to avoid the difficulty : let the reactions evolve starting from the contact point on the potential surface without any random force , and then check if the compound state is reached .
if the composite system evolves toward the compound direction , it means that there is no barrier between contact point and compound state , and thus the fusion hindrance does not exist .
otherwise , the fusion hindrance occurs . with this method , the current paper studies the occurrence of the fusion hindrance for asymmetric reactions .
@xcite show that the neck of the amalgamated system also plays an important role in the hindrance to the fusion .
thus , one of the related questions is about the evolution of neck degree of freedom : is there also a fusion hindrance in the neck evolution , like for radial evolution ?
we will answer to these questions in the present paper .
there are several ways to parameterize the shape of the amalgamated system . in this paper
we use the so - called two - center parameterization @xcite based on three important parameters which are : the distance between two centers @xmath3 , the mass - asymmetry parameter @xmath0 , and the neck parameter @xmath4 .
the dimensionless parameter @xmath3 is defined as follows : @xmath5 where @xmath6 denotes the distance between two centers of the harmonic potentials , and @xmath7 the radius of the spherical compound nucleus .
the mass - asymmetry parameter is defined as usual , @xmath8 where @xmath9 and @xmath10 are the mass numbers of the colliding nuclei .
the neck parameter @xmath4 is defined by the ratio of the smoothed height at the connection point of the two harmonic potentials and that of spiked potential . in this description
, nuclear shape is defined by equipotential surfaces with a constant volume .
for example , @xmath11 means no correction , i.e. , complete di - nucleus shape , while @xmath12 means no spike , i.e. , flatly connected potential , which describes highly deformed mono - nucleus .
thus , the neck describes shape evolution of the compound system from di - nucleus to mono - nucleus .
the initial value of @xmath3 , @xmath0 and @xmath4 , for a given reaction at the touching point is @xmath13 respectively . here
@xmath14 ( @xmath15 ) stands for the mass number of the projectile ( target ) . with this parameterization ,
the liquid drop model ( ldm ) potential is adopted to calculate the energy of the touching di - nucleus .
these three parameters are connected by the potential , the inertia and friction tensors . for the sake of comprehension
, we will study the neck and the radial degrees of freedom separately .
in this section we focus our attention on the potential landscape seen by the neck parameter .
let us recall that in the two - step model , the fusion hindrance is described as the additional barrier hampering the evolution from the contact point to the compound nucleus . for @xmath16kr+@xmath17pb at various relative distances.,width=275,height=261 ] in order to show the direction of evolution of the neck degree of freedom , the ldm potential is plotted as a function of @xmath4 in fig .
[ fig-1 ] , where the different lines correspond to different distances between the projectile and target .
the figure shows two different directions of evolution : ( i ) at small @xmath3 ( @xmath18 or @xmath19 ) the potential decreases with reducing neck parameter ( thicker neck ) which means that at contact the neck becomes thick by the driving effect of the potential .
( ii ) at large @xmath3 ( @xmath20 or @xmath21 ) , the potential increases with reducing neck parameter around @xmath22 and a bump exists at a critical value of @xmath4 .
this indicates that an extra energy should be provided to reach the thick neck , i.e. to form a mono - nucleus . in other words ,
the system is hindered to form a compound nucleus .
in fact , the frontier between the two cases can be determined by the slope of the potential at the point @xmath23 : @xmath24 here @xmath25 denotes for the ldm potential . positive @xmath26 and negative @xmath26 correspond to the first case and the second case , respectively .
pd + @xmath27pd .
( a ) ldm potential as a function of @xmath4 ; ( b ) components of the ldm potential , surface potential ( above zero ) and coulomb potential ( below zero ) , as a function of @xmath4 ; ( c ) di - nuclear shapes at @xmath23 .
the dashed , solid and dash - dotted lines are for @xmath28 , 1.67 , 1.8 , respectively .
the slope @xmath26 equals to @xmath29 at @xmath30.,width=610,height=212 ] the reasons why the potential as a function of @xmath4 shows different features are displayed in fig .
[ fig - shape ] .
the potentials in this figure are referred to the potential at the spherical compound state . at small relative distances @xmath3 ( the red dashed line ) ,
reducing @xmath4 from @xmath23 means decreasing the surface and thus the corresponding surface potential , while the coulomb potential becomes larger ( fig .
[ fig - shape](b ) ) . since the surface potential reduces faster than the increasing of the coulomb potential , the total potential becomes smaller with reducing @xmath4 , as shown by the red dashed line in fig .
[ fig - shape](a ) .
when the distance between the projectile and target increases ( black solid line in fig .
[ fig - shape ] ) the decreasing speed of the surface area becomes smaller . at a critical distance @xmath31 ( for instance @xmath32 for @xmath27pd + @xmath27pd ) , the decreasing of the surface potential equals the increasing of the coulomb potential with reducing @xmath4 around @xmath4 = 1 , so that the slope at @xmath23 becomes zero , as shown by the solid line in fig .
[ fig - shape](a)(b ) . while at larger @xmath3 , shown by the blue dash - dotted line , the surface area is stable with reducing @xmath4 around @xmath4 = 1 and thus the total potential is increasing while neck starts to evolve from @xmath11 .
the @xmath33 and @xmath34 are the proton number and neutron number of the compound nucleus , respectively
. see text for more details.,width=371,height=282 ] in fact we only need to know the situation at contact ( @xmath35 ) . by simply checking the value of @xmath26 at @xmath35 , we know whether the evolution of the neck degree of freedom is hindered or not . with this method ,
different combinations of the @xmath14 and @xmath15 for different neutron number @xmath36 and proton number @xmath37 of a compound nucleus are systematically studied and the results are shown in fig .
[ fig-2 ] .
the borderlines , corresponding to different @xmath0 , are plotted connecting the reactions for which @xmath26 equals 0 at @xmath11 and @xmath35 .
the line of @xmath38-stability is also plotted . inside the borderlines , the slope
@xmath26 is greater than 0 and the reactions are not hindered , while outside the lines it is the opposite . comparing the area encircled with different lines
, it also shows that the number of reactions without neck fusion hindrance for small @xmath0 is smaller than for large @xmath0 .
this is due to the fact that more symmetric reactions correspond to larger coulomb repulsion force which leads the coulomb potential to increase more quickly with reducing @xmath4 .
since the encircled area is very large even for @xmath0 = 0 , the neck evolution is not hindered for most of the realistic reactions .
we have just concluded that the neck evolution is not hindered for most of the realistic reactions .
in fact , once at contact a strong potential slope drives the neck degree of freedom toward a mono - nucleus shape . on the other hand ,
the radial and mass - asymmetry degrees of freedom face a potential barrier for systems with hindered fusion . as a consequence ,
the neck degree of freedom evolves faster than the other two .
it has reached equilibrium before the diffusion over the potential barrier has started @xcite .
as most of the di - nucleus evolve along @xmath3 and @xmath0 direction with an equilibrated neck parameter , we study the fusion hindrance in the rest two degrees of freedom at the equilibrated value of @xmath4 . for different symmetric or asymmetric reactions to form @xmath39po.,width=271,height=274 ]
the neck parameter at equilibrium can be determined through the average of @xmath4 via @xmath40 where @xmath41 , and @xmath42 is the temperature of the system . as shown in fig .
[ fig - aver - neck ] , @xmath43 differs slightly for different reactions . in most cases
, @xmath44 is close to 0.1 .
therefore we take @xmath45 in the next calculations .
this value is also used in the calculation of fusion cross section for @xmath46ca induced reactions in the two - step model , and with which the residue cross sections are calculated combining the statistical evaporation model @xcite . by introducing a shell energy reduction factor , the calculated residue cross sections show good agreement with experimental data from ca + u to ca + bk .
based on the ldm , when the projectile and target evolve from the contact point to the compound state , the parameters \{@xmath0 , @xmath3 } change from \{@xmath47 , @xmath48 } to \{@xmath0 = 0 , @xmath3 = 0 } in the same time .
however because the ldm potential as a function of @xmath0 at fixed @xmath3 and @xmath4 is similar to a parabolic potential centered at @xmath49 , the evolving system is not hindered in the mass - asymmetry direction .
therefore we only need to consider the fusion hindrance in the radial direction .
since the friction in the @xmath0 and @xmath3 direction is close to each other , the radial direction can not be de - coupled from the mass - asymmetry direction .
thus in the analysis of the radial fusion hindrance we will consider @xmath0 and @xmath3 direction in the same footing .
it is easier to understand the radial fusion hindrance in symmetric reactions by simply comparing the contact point and the saddle point , shown in fig .
[ fig - rfh - sym ] . in this figure
@xmath4 has been fixed to 0.1 . for reactions with light nuclei as @xmath50zr
+ @xmath50zr and @xmath51mo + @xmath51mo , the saddle points are located at the right hand side of the contact line ( @xmath52= 1.587 ) , which means that projectile and target at contact will evolve towards each other automatically by the driving force of the ldm potential and finally form the compound state . while for the reaction @xmath27pd + @xmath27pd , it is opposite : when the two touching @xmath27pd reach to the neck equilibrium ( point `` c '' in fig .
[ fig - rfh - sym ] ) , they need to overcome an extra - barrier to reach the compound state . in other words , the reaction is hindered , or the reaction has radial fusion hindrance .
this is another fusion hindrance at the contact state besides the neck fusion hindrance . ) , respectively .
the vertical dashed line corresponds to the contact point @xmath53 for symmetric reactions.,width=270,height=279 ] however this method can not be applied to asymmetric reactions because the saddle point changes into a ridgeline . but
similar to the symmetric case , as we can imagine , if the contact point is located inside ( left hand side of ) the ridge line the reaction will reach the compound state without radial fusion hindrance , and v.v .
unfortunately , it is not easy to get the ridgeline from a potential surface . to find the fusion path by diffusion
, one has to solve a langevin equation as shown in eq .
( 6 ) of ref .
@xcite . here
, we employ another method : switching off the random force , setting the initial momentum in @xmath3 and @xmath0 direction as zero and then studying the evolution track of the reaction by solving the newtonian equation .
the friction tensor is calculated with the wall - and - window formula @xcite .
if the system reaches the compound state it means that the contact point is located inside the ridgeline , and thus the system is not hindered .
contrarily the contact point is located outside the ridgeline and extra - push energy is necessary to form a compound nucleus .
in fact this method can also be used for the symmetric case .
-stability are also plotted .
the abscissa and ordinate stands for the neutron number @xmath34 and proton number @xmath33 of the compound nucleus , respectively . ,
width=369,height=270 ] with this method we study the radial fusion hindrance in symmetric and asymmetric reactions
[ fig - rfh ] shows the location of the borderline between hindered and non - hindered reactions for various asymmetries @xmath0 .
for the reactions located on the borderlines , the contact points are just located on the ridgeline , while for the reactions located inside the borderlines and outside the borderlines , the contact points are located inside and outside of the ridgeline , respectively .
therefore all of the reactions located outside of the ridgelines need extra - push energy to reach the compound state , or in other words , these reactions have radial fusion hindrance .
[ fig - rfh ] also shows that the number of reactions without fusion hindrance at large @xmath0 is larger than that at small @xmath0 as it is well known experimentally .
this can be easily understood : more symmetric reactions ( with small @xmath0 ) have larger repulsive coulomb potential ( @xmath54 ) and the ridgeline is closer to or even shifted into the left side of the contact points . in the same time , with reducing @xmath0 , the contact point @xmath52 becomes larger , as indicated in eq .
( [ eq-2.5 ] ) .
this further reduces the number of reactions without fusion hindrance at more symmetric reactions .
it is commonly known from experiment that the reactions @xmath50zr
+ @xmath50zr , @xmath51mo + @xmath51mo are not hindered in the fusion reaction while the reaction @xmath27pd + @xmath27pd is @xcite .
the three reactions are also plotted in fig .
[ fig - rfh ] : zr+zr and mo+mo are located inside the line of @xmath55 while pd+pd is located outside , showing the agreement with theoretical analysis . it is interesting to see that the lower two dots ( @xmath56hg : zr+zr and @xmath39po : mo+mo ) is located inside all of the lines , showing that all of the reactions to form @xmath56hg and @xmath39po are not hindered . while the reactions to form @xmath57u ( the top dot marked by `` @xmath27pd+@xmath27pd '' ) is different : because compound nucleus is located outside of the line with @xmath58 0.7 and inside of the line with @xmath59 0.7 , then the reactions with @xmath58 0.7 are hindered while the reactions with @xmath59 0.7 are not . fig .
[ fig - rfh ] also shows that the number of reactions without fusion hindrance is limited for each asymmetric parameter @xmath0 , and each line for the corresponding @xmath0 has a @xmath60 .
for example , @xmath61 for @xmath62 and @xmath63 for @xmath64 .
according to the extra - push theory@xcite , @xmath65 for @xmath55 and @xmath66 for @xmath64 if @xmath67 .
the comparison shows that the realistic calculations give smaller @xmath60 for larger @xmath0 , while for the reactions with smaller @xmath0 , the two models give similar @xmath60 . comparing the figs .
[ fig-2 ] and [ fig - rfh ] , it is also found that the encircled area in fig .
[ fig - rfh ] is much smaller than the area in fig .
[ fig-2 ] , and the former one is completely included in the latter one .
this indicates that the reactions without radial fusion hindrance are apparently without neck fusion hindrance , while the reactions with neck fusion hindrance have obviously radial fusion hindrance .
.,width=277,height=200 ] the symmetric case can be compared with the parameterized ldm proposed by swiatecki et al.@xcite . in ref .
@xcite , the proton number @xmath33 and the mass number @xmath68 of the compound nucleus , for which the contact point of the symmetric reaction is located on the saddle point of the parameterized ldm , satisfies @xmath69 plotting together the above eq .
( [ eq-4 ] ) and the borderline at @xmath55 of fig .
[ fig - rfh ] , we get fig .
[ fig - compare ] .
it is clear that the lower limit of @xmath34 for fixed @xmath33 is almost the same between two models , but the upper limit of @xmath34 is quite different : the parameterized ldm gives much larger @xmath70 than the two - center ldm model .
this is probably caused by the too few parameters used to describe the so complicated liquid drop energy in ref .
fusion reactions between two colliding heavy ions may be hindered or not .
this is explained by the appearance of an inner barrier after the contact point that hampers the formation of a compound nucleus .
the liquid drop potential landscape described with the two - center parameterization for the deformation allows us to study the frontier between hindered and non - hindered reactions as shown in fig .
[ fig-2 ] and [ fig - rfh ] .
the neck and radial degrees of freedom may both have to overcome an inner barrier to reach the compound shape .
for each of them , we can draw a borderline that classifies reactions with respect to the hindrance .
it appears that for available nuclei , the neck degree of freedom is not hindered .
a strong potential slope always drives it toward a mono - nucleus shape .
it is not the case for the radial degree of freedom .
we obtained a borderline that is far more restrictive , i.e. the hindrance to the fusion appears for compound nuclei having a charge larger than @xmath71 when they are formed by a symmetric reaction .
the frontier is located at larger values of @xmath33 for asymmetric reactions , as it is observed experimentally .
the exact location of the border between hindered and non - hindered reactions might depend on the parameterization
. it would be therefore very useful to have some experimental results to assess the models . finally , it is important to note that in this paper , we have just studied the appearance of the hindrance to the fusion .
the amplitude of the hindrance , i.e. the extra - energy that is necessary to the system to fuse , also depends on dynamical parameters as the friction coefficient @xcite . the work is supported in part by the national natural science foundation of china ( nos . 10905021 , 10979023 , 10979024 ) , the key project of the ministry of education of china ( no . 209053 ) , the zhejiang provincial natural science foundation of china ( no .
y6090210 ) , and the qian - jiang talents project of zhejiang province ( no .
2010r10102 ) .
the authors also acknowledge supports and hospitality by rcnp osaka university , ganil , and huzhou teachers college , which enable us to continue the collaboration .
this work is also partly supported by the c3s2 computing center in huzhou teachers college .
|
the fusion hindrance in heavy ion collisions is studied in the framework of the two - center liquid drop model .
it appears that the neck and the radial degrees of freedom might both be hampered by an inner potential barrier on their path between the contact configuration to the compound nucleus .
heavy ion reactions with and without the two kinds of fusion hindrance are classified through systematic calculations .
it is found that the number of reactions without radial fusion hindrance is much smaller than that without neck fusion hindrance , and for both kinds of fusion hindrance the number of reactions without fusion hindrance at small mass - asymmetry parameter @xmath0 is smaller than that at large @xmath0 . in the formation of a given compound nucleus ,
if a reaction with @xmath1 is not hindered , then other reactions with @xmath2 are also not hindered as it is well known experimentally .
|
submillimeter - selected galaxies ( smgs ) generally refer to the population detected in the pioneering lensed and blank deep field surveys using the submillimeter common user bolometer array ( scuba ; holland et al .
1999 ) instrument on the james clerk maxwell telescope ( jcmt ) ( e.g. , smail , ivison & blain 1997 ; hughes et al .
1998 ; barger et al . 1998 ; eales et al . 1999 ) .
they are regarded as the high redshift ( z@xmath323 ; chapman et al .
2005 ; chapman et al .
2003 ; see also simpson et al . 2004 )
counterparts to the population of luminous and ultraluminous infrared galaxies ( lirgs and ulirgs ) detected by the infrared astronomical satellite , _ iras _ , typically out to [email protected] ( soifer et al .
1984 ; joseph & wright 1985 ; and soifer , neugebauer & houck 1987 ) , since they share similar properties ( fundamentally , that the bulk of the bolometric luminosity is emitted in the restframe far - ir , powered by a combination of star formation and agn activity . ) . as a necessary consequence of the selection at longer submillimeter wavelengths ,
smgs tend to have higher bolometric luminosities , and those with comparable luminosities to _ iras_-selected ulirgs tend to have cooler dust temperatures ( blain et al .
2004 ) . between the two are the infrared space observatory ( _ iso_)-selected ulirgs ( aussel et al .
1999 ; rowan - robinson et al . 1999 ; elbaz et al .
2002a ) , which occupy an intermediate redshift space ( z@xmath31 ) , and have dust masses in - between the _ iras _ and scuba - selected ulirgs ( chapman et al . 2002
; blain et al . 2004
; this is also a region of parameter space that will be shared by _ spitzer_-selected galaxies , see , e.g. , yan et al . 2004 ) .
sources selected at 350@xmath1 m are expected to be predominantly lirgs and ulirgs at 1@xmath63 ( the k - correction changes from negative to positive with increasing redshift as the observing window moves towards and past the wien peak ; see , e.g. , guiderdoni et al .
1998 ; khan , 2005 , in preparation ) .
galaxies at z@xmath31 are likely to be the dominant source of the integrated far - ir background ( puget et al .
1996 ; fixsen et al . 1998 ; elbaz et al .
2002b ) ; also , observations suggest that after a relatively steep rise from z=0 to @xmath31 , the cosmic star formation rate ( csfr ) appears to flatten off between z=1 and @xmath34 , with evidence for a slow decline at higher redshift ( lilly et al .
1996 ; madau et al . 1996 ; connolly et al .
1997 ; steidel et al . 1999 ; gabasch et al .
hence 350@xmath1m - selected sources are an important probe of the epoch of peak star formation rate in the universe . in this letter
we report on the detection of a single source above 3@xmath0 at 350@xmath1 m .
it is , to the best of our knowledge , the first short - wavelength submillimeter - selected galaxy ( 200500@xmath1 m ) and is denoted as ssg 1 in this letter ( but by standard convention , smm j143206.65 + 341613.4 ) .
ssg 1 is the first object discovered _ purely _ by its 350@xmath1 m emission , through blank deep observations of the botes field in a program designed to search efficiently for new sources at 350@xmath1 m ( the sharc ii unbiased deep survey , suds ; khan et al . , 2005 , in preparation ) .
verification of the detection is confirmed by coincident sources subsequently identified in the _ spitzer _ mips 24@xmath1 m filter , all four _ spitzer _ irac filters , and j , k@xmath2 , i and r band , enabling us to place constraints on the photometric redshift , luminosity and nature of this source .
this letter presents these fluxes , plus important flux upper limits in other bands , to characterize this source , and with this understanding we consider the implications for other sources that may be discovered at 350@xmath1 m .
eight hours of data were obtained using the second generation submillimeter high angular resolution camera ( sharc ii ) at the caltech submillimeter observatory ( cso ) on mauna kea , hawaii , in january and march 2004 .
sharc ii is a 350@xmath1m - optimized camera ( dowell et al .
2003 ) built around a @xmath7 element close - packed bolometer array ( moseley et al .
it achieves a point - source sensitivity of @xmath8 in good weather .
the 384 pixels of the sharc ii array image a region of around @xmath9 on the sky .
its filled absorber array provides instantaneous imaging of the entire field of view , sampled at roughly 2.5 pixels per nominal beam area .
the beam profile was measured on known compact sources , and was verified to be within three per cent of the diffraction - limited beamwidth of @xmath10 .
all observations were taken using the dish surface optimisation system ( leong et al .
2003 ) , which corrects for the primary mirror deformation as a function of zenith angle , to improve the telescope efficiency and the pointing . for these data the in - band zenith atmospheric opacity ( @xmath11 ) ranged from 1.0 to 1.4 , corresponding to a zenith transmission of around 30 per cent .
our observations were centred on the botes deep field ( de vries et al .
2002 ) , at position ra@xmath12 , dec@xmath13 ( j2000 ) .
the data were reduced using the standard cso reduction software , crush ( kovcs , 2005 , in preparation ) . this software implements a self - consistent least - squares algorithm to solve for the celestial emission , taking into account instrumental and atmospheric contributions to the signal .
the skymap is calibrated with the flux and point spread function based on observations of callisto taken throughout the observing period at similar elevations .
[ -0.15 cm ] observed wavelength & flux density & instrument survey + [ 0.1 cm ] & & + [ -0.15 cm ] x - ray ( 0.57kev)&@xmath14c.g.s & _ chandra_xbotes + 4220 & @xmath15 0.1@xmath1jy & kpno ndwfs + 6590 & 1.0 @xmath4 0.1@xmath1jy & kpno ndwfs + 8081 & 3.1 @xmath4 0.2@xmath1jy & kpno
ndwfs + 1.24@xmath1 m & 10.5 @xmath4 1.1@xmath1jy & kpno flamex + 2.16@xmath1
m & 42.0 @xmath4 1.4@xmath1jy & kpno flamex + 3.6@xmath1
m & 80.0 @xmath4 2.9@xmath1jy & _ spitzer _ irac shallow + 4.5@xmath1 m & 65.8 @xmath4 3.6@xmath1jy & _ spitzer _ irac shallow + 5.8@xmath1 m & 55.3 @xmath4 17.3@xmath1jy & _ spitzer _ irac shallow + 8.0@xmath1 m & 81.7 @xmath4 15.1@xmath1jy & _ spitzer _ irac shallow + 24@xmath1 m & 0.61 @xmath4 0.04mjy & _ spitzer _ mips irs / mips gto + 70@xmath1 m & @xmath15 40mjy & _ spitzer _ mips irs / mips gto + 160@xmath1 m & @xmath15 100mjy & _ spitzer _ mips irs / mips gto + 350@xmath1 m & 23.2 @xmath4 7.9mjy & sharc
ii suds + 21 cm & @xmath15 63@xmath1jy & westerbork
deep 1.4ghz + an oversampled @xmath16 fit is used to determine the position of the source and the flux per beam .
the best - fitting position is at 14:32:[email protected] + 34:16:13.4@xmath17 ( j2000 ) ( quoting 3@xmath0 position uncertainties ) , with a flux per beam of 23.2 @xmath4 7.9mjy ( the 1@xmath0 uncertainty comprises the background noise , primary calibrator and gain calibration errors )
. the signal - to - noise in the detection is 3.6 .
the additional photometry for ssg 1 ( table 1 ) is from the _ chandra _ xbotes survey ( murray et al . 2005 ) , noao - ndwfs ( jannuzi & dey 1999 ) , flamex ( elston et al . , 2005 , in preparation ) , _
spitzer_-irac shallow survey ( eisenhardt et al . 2004 ) , _ spitzer_-mips ( irs and mips gto teams ) and the westerbork deep 1.4ghz survey ( de vries et al .
the upper limits are quoted to 3@xmath0 .
the corresponding multiwavelength positions and probability of coincident detection are given in table 2 .
[ -0.15 cm ] & band & p(accidental overlap ) + & & & + [ 0.1 cm ] & & & + [ -0.15 cm ] 14:32:06.61 & + 34:16:12.0 & r & 9.5% + 14:32:06.55 & + 34:16:11.6 & k@xmath2 & 2 .. 9% + 14:32:06.58 & + 34:16:12.0 & 3.6@xmath1 m & 9.4% + 14:32:06.60 & + 34:16:11.5 & 24@xmath1 m & 0.2% + 14:32:06.65 & + 34:16:13.4 & 350@xmath1 m & n / a + within the 99% 350@xmath1 m confidence region there are two candidate sources resolved in the optical and nir bands ( figure 1 ) .
we assume that the optical - nir counterpart of ssg 1 is the source coincident with the mips 24@xmath1 m detection .
the source to the east of this ( at irac position ra@xmath18 , dec@xmath19 /j2000 ) is not detected by mips , and is designated as ssg 1e for the purposes of this letter .
using the fluxes presented in table 1 , we can use several methods to constrain the redshift , reddening , infrared luminosity and dust temperature of ssg 1 .
it can be seen in figure 2 that the optical - nir sed of ssg 1 is characterized by a prominent bump associated with the continuum emission of the stellar populations .
this bump peaks at 1.6@xmath1 m in the rest - frame , providing a very good constraint on the redshift , and also indicates ssg 1 is more likely to be starburst rather than agn - dominated ( which are usually associated with a featureless power - law spectrum ; see , e.g. , egami et al .
2004 ) .
[ -0.15 cm ] source & photometric redshift & model & a@xmath20 + & & & + [ -0.15 cm ] ssg 1 & 1.0@xmath21 & impz & [email protected] + & 0.99@xmath22 & stardust2 & [email protected] + ssg 1e & [email protected] & impz & [email protected] + impz ( babbedge et al . 2004 ) uses only the optical - nir photometry when fitting templates with bayesian statistics .
the best - fitting redshifts are z=1.0@xmath23 for ssg 1 and [email protected] for ssg 1e respectively ( 1@xmath0 ) , and are listed in table 3 .
it is worth noting that if , as impz suggests , both galaxies are at the same distance , and interacting , it would make the ssg pair a widely separated interacting ulirg system ( a local example of which is iras 091111007 ; khan et al .
2005 ) .
stardust2 ( chanial et al . , 2005 , in preparation ) uses a mid - ir to radio spectral library constrained by the ir - radio correlation and a variety of local , _ iras _ , _ iso _ and scuba color - color correlations in addition to a stellar synthesis model for the fuv to nir window ( devriendt et al .
this provides a simultaneous constraint on the thermal dust emission as well as the redshift .
the effective dust temperature is found following the methodology given in chapman et al .
( 2002 ; 2005 ) .
the model assumes @xmath24 , @xmath25 , and @xmath26 , and that a single component is responsible for both the optical and ir emission . because of the degeneracy in fitting @xmath27 and the total - ir luminosity , @xmath28 ( 81100@xmath1 m ) , it is the 350@xmath1 m flux in conjunction with the radio upper limit that constrains the infrared luminosity . the radio upper limit ( table 1 )
is used to find @xmath16 as follows : @xmath29 with the radio @xmath30 term being @xmath31 ^ 2/\sigma^2 , f_\nu({\rm 1.4\,ghz } ) > 3\sigma\\ & = & 0 , \quad f_\nu({\rm 1.4\,ghz } ) \leq 3\sigma\end{aligned}\ ] ] where @xmath32 is the modeled spectral energy distribution and 3@xmath0 is the radio upper limit listed in table 1 .
the best - fitting model for ssg 1 from stardust2 returns a redshift of 0.99@xmath22 ( all errors are quoted to 1@xmath0 ; note : stardust2 uses the longer wavelengths , in addition to the optical - nir photometry , to fit the extinction given in table 3 ) .
this model also gives a dust temperature of [email protected] , and a log(@xmath28 ) of 12.02@xmath33l@xmath34 , with 88% of the total bolometric luminosity radiated at wavelengths longward of 5@xmath1 m .
the predicted 850@xmath1 m flux is 2.7@xmath35mjy .
other redshift estimates were independently obtained following an approach that mostly relies on the fit of the 1.6@xmath1 m feature ( see , e.g. , le floch et al .
2004 ) . in agreement with the results derived from impz and stardust2 , these fits of
the stellar bump led to a redshift z@xmath31 using the photometric arp220 sed , and [email protected] using various templates from the library of devriendt et al .
( 1999 ) .
we also use the radio - submillimeter correlation to estimate a minimum redshift for the source , as the radio flux is unlikely to be significantly agn - enhanced .
formally , these correlations should be used as a statistical redshift indicator , so with that caveat we present the minimum redshift of ssg 1 based on the 1.4ghz upper limit ( table 1 ) and an 850@xmath1 m flux derived from the best - fitting stardust2 sed ( figure 2 ) . using the relations of dunne , clements & eales ( 2000 ) and carilli & yun ( 2000a ; 2000b ) we get @xmath36 of 1.2 and 1.5 respectively .
the photometric redshift is more secure than the radio - submillimeter correlation since the latter is prone to systematics caused by uncertain dust temperatures ( see clements et al .
the ( r@xmath37 color of 5.7 for ssg 1 classifies it as an extremely red object ( ero ) , and both impz and stardust2 find this object to be highly reddened ( table 3 ) .
at least a third of the smg population can be classified as eros ( smail et al .
2002 ; webb et al . 2004 ; frayer et al . 2004 ) , which comprise two classes of galaxies : elliptical and dusty star - forming luminous infrared galaxies ( but a significant submillimeter flux is usually indicative of the latter class ) .
eros are thought to comprise a significant fraction of the cosmic star formation density at redshifts of one and higher ( cimatti et al . 2002 ) , the epoch by which the majority of the universe s star formation has taken place ( dickinson et al . 2003 ; rudnick et al . 2003 ) .
determining the effective dust temperature of the source allows us to estimate the total fraction of the bolometric luminosity that is reprocessed by the dust .
our effective dust temperature from stardust2 is [email protected] , which , with our total infrared luminosity of @xmath38 , allows us to directly compare ssg 1 with _
iras _ and scuba - selected ulirgs in the l vs t diagram of blain et al .
we find that ssg 1 is warmer than an equivalent luminosity smg from the chapman et al .
( 2003 ; 2005 ) high redshift sample , and cooler than the _ iras _ moderate redshift counterparts of stanford et al .
( 2000 ) . instead
, it is similar to the intermediate redshift _ iso _
sample of garrett ( 2002 ) .
the notion of ssg 1 being an _ iso _ ulirg analog is further reinforced by the best - fitting 60 - 100@xmath1 m colors from stardust2 , of which the closest template is fn1 - 40 , an _ iso _
cold ( @xmath39=25.7k ) ulirg at z=0.45 ( chapman et al .
the 850@xmath1 m flux ( @xmath33mjy ) predicts ssg 1 to be fainter than the majority of scuba - selected smgs ( @xmath40 , due to survey and instrument limitations ) .
we conclude that ssg 1 is more like _ iso_-selected ulirgs rather than the scuba - selected smgs .
we report the detection of the first short - wavelength submillimeter - selected galaxy ( ssg 1 ) and present optical , nir and ir photometry which we use to constrain the redshift , luminosity and dust temperature of this source .
the photometric redshift estimators are all in agreement , producing a best - fitting redshift of @xmath31 .
the dust temperature of [email protected] and luminosity of @xmath38 make ssg 1 an analog of intermediate redshift _
iso_-selected ulirgs rather than high redshift scuba - selected smgs .
if there exists a population of galaxies with properties similar to ssg 1 , in a redshift space between _ iso _ and scuba , it may be argued that observations at 350@xmath1 m , in conjunction with the current _ iso _ , _ spitzer _ and scuba samples , could describe the ir - luminous population from [email protected] ( although the scuba - selected galaxies will be more luminous ) .
not only would this population bridge the gap in redshift space , but also the gap in `` dust temperature space '' ( and consequently `` dust mass space '' ) .
sharc ii thus complements scuba in revealing warmer smgs that might be missed otherwise .
furthermore , finding the lower luminosity smgs is necessary for an accurate understanding of the submillimeter luminosity function . from our understanding of the first 350@xmath1m - selected object , ssg 1
, we infer that observations at 350@xmath1 m , far from being superfluous to scuba , will be of great use in determining the nature and evolution of the luminous infrared galaxy population .
the caltech submillimeter observatory is supported by nsf contract ast-0229008 .
this work is based in part on observations made with the _ spitzer _ space telescope , which is operated by the jet propulsion laboratory , california institute of technology under nasa contract 1407 .
this work made use of images and data products provided by the noao deep wide - field survey ( jannuzi & dey 1999 ) , which is supported by the national optical astronomy observatory ( noao ) .
noao is operated by aura , inc .
, under a cooperative agreement with the national science foundation .
the flamex survey acknowledges support from noao and nsf ( ast-9875448 , ast-0407085 , ast-0436681 ) , and technical assistance from s.n .
raines .
we would like to thank the anonymous referee for their insightful comments which have significantly improved this letter .
we also thank tom phillips and the cso for observing time and assistance during our runs , and darren dowell , colin borys and attila kovcs for instrument and data reduction support .
we express our gratitude to our fellow gsfc co - is on suds : bob silverberg and dave chuss .
we also thank suds co - i rick arendt for continual support on the data analysis .
thanks jon gardner , rob ivison and steve eales for very helpful discussions .
jannuzi , b. t. , dey , a. , 1999 , in `` photometric redshifts and the detection of high redshift galaxies '' , asp conference series , vol .
191 , edited by r. weymann , l. storrie - lombardi , m. sawicki , and r. brunner .
isbn : 158381 - 017-x , p. 111
joseph r.d . ,
wright g.s . , 1985 , mnras , 214 , 87 puget , j .- l . ,
abergel , a. , bernard , j .-
, boulanger , f. , burton , w. b. , desert , f .- x . ,
hartmann , d. , 1996 , a&a , 308 , l5 rowan - robinson , m. , et al .
, 1999 , the universe as seen by _
iso_. eds .
p. cox & m. f. kessler .
esa - sp 427 . , p. 1011
|
we report the detection of a 3.6@xmath0 350@xmath1m - selected source in the botes deep field .
the source , the first short - wavelength submillimeter - selected galaxy ( ssg 1 ) , was discovered as part of a blank field extragalactic survey using the 350@xmath1m - optimised submillimeter high angular resolution camera ( sharc ii ) at the caltech submillimeter observatory . with multiwavelength photometry from noao - ndwfs ( r and i band ) , flamex ( j and k@xmath2 ) , _ spitzer _ ( irac and mips ) and the westerbork 1.4ghz deep survey ( radio upper limit ) , we are able to constrain the photometric redshift using different methods , all of which suggest a redshift @xmath31 . in the absence of long - wavelength submillimeter data we use sed templates to infer that this source is an ultraluminous infrared galaxy ( ulirg ) with a dust temperature of 30@xmath45k , occupying a region of luminosity - temperature space shared by modarate redshift _ iso_-selected ulirgs ( rather than high redshift scuba - selected smgs ) .
sharc ii can thus select smgs with moderately `` warm '' dust that might be missed in submillimeter surveys at longer wavelengths .
|
star formation theory for decades developed under the assumption that magnetic flux is frozen in highly conducting interstellar gas , unless a process of ambipolar diffusion carries neutrals across the magnetic field lines @xcite .
the flux freezing is , however , violated in turbulent fluids due to fast reconnection @xcite ( henceforth lv99 ) .
this prediction is now not only supported by successful numerical tests @xcite and observations @xcite , but also by formal mathematical derivations based on modern understanding of the lagrangian properties of mhd turbulence ( @xcite ) .
the diffusion of magnetic fields mediated by reconnection in turbulent fluids was predicted to be important for star formation by lazarian ( 2005 ) .
the process should be widely spread as turbulence is ubiquitous in the interstellar media ( armstrong et al .
1994,@xcite , chepurnov & lazarian 2010 ) . the corresponding process was termed _
reconnection diffusion _ in analogy with the ambipolar diffusion , the accepted process in the standard picture for magnetic flux removal from molecular cloud cores . according to the latter mechanism ,
the violation of magnetic flux freezing is possible through the drift of neutrals passing through nearly perfectly frozen - in ions with the magnetic field .
the reconnection diffusion presents a potent alternative to the ambipolar diffusion process ( see @xcite for reviews ) .
the numerical demonstration of the reconnection diffusion was first performed by @xcite for the case of diffuse media and collapsing molecular clouds ( see also @xcite ) .
recently @xcite ( hereafter sgl12 ) have performed simulations of the formation of a keplerian disc in a turbulent , strongly magnetized low - mass cloud core ( with a sink particle in the center with m @xmath1 solar - mass and an initial uniform magnetic field in the z direction with intensity 35 @xmath0 g ) and found that the reconnection diffusion was efficient in this set up , decreasing the magnetic flux during the disk build up .
sgl12 proposed that this mechanism can provide the solution of the so - called `` magnetic braking catastrophe '' discussed in the literature @xcite .
the latter conclusion was criticized in @xcite ( hereafter s+12 ) who performed amr simulations of the collapse of a 100 solar mass turbulent cloud core permeated by a magnetic field ( with 1.3 mg in the center and declining radially outwards with r@xmath2 ) .
they introduced sink particles in the cloud above a density threshold and detected the formation of several protostars around which keplerian discs with typical sizes of up to 100 au built up .
then , they examined a few mechanisms that could be potentially responsible for lowering the magnetic braking efficiency and thus , allowing for the formation of the keplerian discs and concluded that none was necessary in their models , nor even reconnection diffusion .
they argued that the build up of the keplerian disk was a consequence of the shear flow generated by the turbulent motions in the surroundings of the disk ( which carry large amounts of angular momentum ) .
the lack of coherent rotation in the turbulent velocity field would not allow the development of toroidal b components and toroidal alfvn waves that could remove outward magnetic flux , in spite of the small values of the mass - to - flux ratio that they considered in their models ( around @xmath3 ) . according to s+12 ,
any effects like misaligned magnetic fields and angular momentum vectors , reconnection diffusion or any other non - ideal mhd effects seem not to be necessary .
they conclude that `` turbulence alone provides a natural and at the same time very simple mechanism to solve the magnetic braking catastrophe '' .
s+12 conclusion above was based on the calculation of the mean mass - to - flux ratio within a sphere around the disk with a radius much larger than the disk ( i.e. , r = 500 au ) .
this ratio @xmath0 was computed taking the volume - weighted , mean magnetic field evaluated in this sphere , in combination with the sphere mass m , normalized by the critical value .
they found that at these scales @xmath0 varies around a mean of 2 - 3 and is comparable with the initial value in the core ( which is also the overall initial value in the massive cloud ) and also to the mhd simulations without turbulence .
however , they have also found that in some cases @xmath0 increases at smaller radius and eventually reaches values above 10 at radii @xmath4 100 au ( i.e. , nearly 5 times larger than the initial value ) .
therefore , s+12 detected flux transport within the keplerian disk , at least in some of the disks formed .
they did not consider that this could be due to transport arising from reconnection diffusion , because at these scales the velocity structures are already well ordered in their models .
thus s+12 concluded that numerical diffusion was the possible source of flux loss . in this letter we put this conclusion to scrutiny and
argue that the increase seen in s+12 is @xmath5 and due to reconnection diffusion in agreement with both theoretical expectation and the results of sgl12 .
now , let us go back to the sgl12 results . focusing on the formation of a keplerian disk in a turbulent cloud core with a single sink , sgl12 clearly found flux loss during the process of the disk build up , as indicated from the analysis of their figures 1 to 3 . in order to benchmark their results ,
sgl12 also performed simulations of non - turbulent hydrodynamic as well as ideal mhd , and highly resistive mhd models .
they found that the ideal mhd model is unable to produce a rotationally supported disk due to the magnetic flux excess that accumulates in the central regions , while the mhd model with artificially enhanced resistivity produces a nearly - keplerian disk with dimension , radial and rotational velocities , and mass similar to the pure hydrodynamical model , and the turbulent model also produces a nearly - keplerian disk , but less massive and smaller ( r @xmath6 100 au ) , in agreement with the observations . for illustration
, figure a.1 of the appendix , reproduces the diagrams as obtained in sgl12 , i.e. , radial profiles for the rotational velocity , the radial velocity , the mass and the magnetic field for the set of models described above , but considering two different resolutions for the turbulent model ( a 256@xmath7 resolution , as in sgl12 , and a 512@xmath7 resolution model ) . as we see , both resolutions produce similar results ( see more details in the appendix ) .
in addition , figure 1 depicts three - dimensional diagrams of three snapshots of the turbulent model computed by sgl12 . considering the three mhd disk formation models investigated by sgl12 ( i.e. , an ideal collapsing cloud core with no turbulence , a highly resistive core with no turbulence , and an ideal turbulent core ) , figure 2 shows the time evolution of the gas mass , the average magnetic flux and the mean mass - to - magnetic flux ratio , @xmath0 , which was calculated employing the same equation ( 1 ) of s+12 , for these three models .
these quantities were computed within a sphere surrounding the central region for three different radii : a large one ( r=1000 au ) , which encompasses the large scale envelope where the disc is build up ( similarly as in s+12 ) , an intermediate ( r=500 au ) , and a small one ( r=100 au ) which corresponds to the region where the disc is later formed .
figure 2 shows that the turbulent model of sgl12 starts with an average @xmath8 and finishes with @xmath9 within @xmath10 au ( see figure 2 , bottom right panel ) .
therefore , as in s+12 , this result suggests no significant variations in @xmath0 . besides
, these values reveal no significant changes with respect to the ideal mhd model either
. however , the values of both , the turbulent and the ideal mhd model at this scale , are also comparable to those of the resistive model where we clearly know that @xmath11 @xmath12 large magnetic flux loss . how to interpret these results then ? when averaging over the whole sphere of radius @xmath10 au around the disk / system , the real value of @xmath0 at the small disk scales ( r @xmath13 100 au ) is hindered by the computed overall values in the envelope .
therefore , it is not enough to compute this average value to conclude that there is no flux loss in the process of the disk build up .
as we decrease the radius of the sphere at which the average @xmath0 is computed , we clearly see that the magnetic flux of the turbulent model becomes comparable to that of the resistive model ( see middle panels of figure 2 ) , specially at the scale of the keplerian disk build up ( @xmath14 100 au ) and , in consequence , there is an increase of @xmath0 with time in the turbulent model with respect to the ideal mhd model .
the final value of @xmath15 within the disk region , therefore , nearly 5 times larger than the initial value in the whole cloud .
thus , similarly to s+12 , there is no significant variations of @xmath0 with time when considering its average over the whole sphere that contains both the turbulent envelope and the disk / sink .
but , in the final state the resulting value of @xmath0 within the disk is larger than the initial value in the cloud .
a similar trend is also found for the non - turbulent @xmath16 mhd model , where the imposed explicit artificial resistivity leads to magnetic flux loss which in turn allows the build up of the keplerian disk .
these results clearly indicate that a nearly constant value of the average value of @xmath0 with time over the whole disk@xmath17envelope system is not a powerful diagnostic to conclude that there is no significant magnetic flux transport in protostellar disk formation ( as suggested by s+12 ) .
when examining the ideal mhd model , there is one important point to remark .
@xmath0 , which should be expected to be constant with time in an isolated system , is also slightly growing within @xmath18 au in this model .
this is due to the adopted open boundaries in the system and to the volume averaging of the magnetic flux . to understand this behaviour
, we must inspect the time evolution of both the mass and the average magnetic flux in the system which are shown in the top and middle diagrams of figure 2 , respectively . actually , in all models the total mass ( envelope / disk plus accreted gas into the sink ) increases with time due to a continuous mass aggregation to the system entering through the open boundaries . also , there is a growing of the magnetic flux with time which is at least in part due to a continuous injection of magnetic field lines into the system through the open boundaries .
both effects , i.e. , the increase in mass and magnetic flux could compensate each other and produce a nearly constant @xmath0 with time . however , there is another effect to be noticed . the magnetic flux in the middle diagrams of figure 2 was not computed within a @xmath19 ( accreting ) volume with the gas , but at a fixed sphere radius .
if we had followed a fixed amount of accreting gas with time then , we would have obtained a constant number of magnetic field lines and therefore , a constant magnetic flux within this comovimg volume in the ideal mhd case .
this is in practice very difficult to compute from the simulations because of the complex geometry of the turbulent magnetic field lines .
however , the key point here is not to obtain the exact value of the magnetic flux for the ideal mhd or the other models in a comoving volume , but to realize that at the scale of the disk build up ( @xmath20100 au ) , the magnetic flux of the turbulent mhd model , which is initially comparable to that of the ideal mhd model , decreased to a value similar to that of the resistive model at the time that the disk has formed ( @xmath2025,000 to 30,000 yr ) , as indicated by the middle left panel of figure 2 .
this is a clear indication of the removal of magnetic flux from the disk build up region to its surrounds in the turbulent model . to help to better clarify the analyses above ,
we have also plotted in figure 3 @xmath0 as a function of the mass for the three different regions considered in figure 2 .
each @xmath21 in figure 3 has been normalized by its initial value : @xmath22 / \left [ 0.13/\sqrt{g } \right],\ ] ] where @xmath23 is the initial value of the magnetic field and @xmath24 is the initial radius of the sphere containing the mass @xmath25 ) .
these diagrams provide a way to evaluate @xmath0 in comoving parcels with the gas .
inside 100 au , figure 3 shows that in the turbulent model @xmath21 is larger than in the ideal mhd model and comparable to the resistive model at the largest masses .
this indicates a smaller amount of magnetic flux in the turbulent and resistive models inside the disk region . as we go to the larger radii , @xmath0 becomes more and more comparable in the three models , in consistency with the results of figure 2 . therefore , based on the results above , we conclude that the flux loss ( and the increase of @xmath26 within the disk ) in sgl12 turbulent models is real and is due to the action of reconnection diffusion , as discussed in detail in sgl12 ( see also santos - lima et al .
2010 , lazarian 2011 , de gouveia dal pino et al .
2012 , lazarian et al .
2012 , leo et al .
2012 ) .
we must note that the flux transport by reconnection diffusion is faster where turbulence is stronger and faster .
this is a fundamental prediction from lv99 fast reconnection theory which was numerically tested in high resolution simulations of cloud collapse in santos - lima et al .
( 2010 ; see also the appendix ) . in the sgl12 turbulent simulations , while the disk is built up by the accretion of the turbulent gas in the envelope that surrounds the sink , reconnection diffusion is fast and causes magnetic flux loss at the same time that it allows the turbulent shear to build up a keplerian profile in this _ collapsing _ material .
this means that the material that formed the keplerian disk out of the accretion of the turbulent envelope _ has already lost magnetic flux _ when it reaches its final state and that is why the final value of @xmath0 is much larger within the disk radius ( @xmath13 100 au ) .
in other words , the mass - to - flux ratio increase that is detected in the final keplerian disk is due to removal of magnetic flux from the highly turbulent envelope material while this material was accreting and building up the disk , i.e. , _ before _ the final state .
after the keplerian disk is formed ( in r @xmath13 100 au ) , the operation of reconnection diffusion inside this region decreases because turbulent structures are smaller and slower there .
fortunately , in terms of magnetic breaking , high values of reconnection diffusion are no longer needed because the magnetic field flux excess has been already removed during the accreting phase and disk build up . as a result , the argument given by s+12 that reconnection diffusion could not explain the increase of @xmath0 in their tests within the disk scales because the fluid is no longer turbulent there , is not correct .
we have also plotted the magnetic field @xmath27 versus the density @xmath28 in different times as in s+12 and found variations which are significant as we decrease the radius where the average of b and @xmath28 are computed , in consistency with the results of figures 2 and 3 and the discussion above .
figure 4 shows these plots for 30,000 yr within spheres of radii equal 100 , 500 , and 1000 au . at the 1000 au scale , both the ideal and the turbulent mhd models are comparable and follow approximately the @xmath29 trend , as in s+12 .
however , as we go to the smaller scales and specially to the 100 au scale , the two models clearly loose this correlation , similar to the resistive model in all scales .
this effect in the resistive model is clearly a natural consequence of the diffusion of the magnetic field from the inner denser regions to the less dense envelope regions .
the turbulent model tends to follow the same trend : we note that for a given density , the magnetic field is smaller in the resistive model than in the turbulent model which in turn , has a smaller magnetic field than in the ideal model ( see left panel in figure 4 ) , in consistence with the previous results . in the case of the mhd model , the nearly constant magnetic field with density at the 100 au scale is due to the effect of the geometry . at this scale ,
the built up disk dominates , but the averaging is performed over the whole sphere that encompasses the region .
this includes also the very light material above and below the disk which has magnetic field intensities as large as those of the high density material in the disk .
( the same effect explains also the larger magnetic field intensities in the low density tail at the 500 and 1000 au scale diagrams @xmath30 middle and right panels , specially for the ideal mhd model . )
the geometry at 100 au obviously also affects the turbulent model in the same way , however we have found from the simulations that in this case the amount of low density gas carrying high intensity magnetic field below and above the disk is smaller than in the ideal mhd case .
this is because in this case the loss of the @xmath31 correlation at the 100 au scale is also affected by the diffusion of the magnetic field as in the resistive model , in consistence with the analyses of figures 2 and 3 .
although the initial conditions in sgl12 are different from those in s+12 , the build up of the keplerian disks by the accretion of the turbulent envelope around the sinks is quite similar to the setup in sgl12 , so that we can perform at least qualitative comparisons between the results . in particular , both models consider initially supercritical cores , i.e. , initial @xmath0 larger than unity .
we remember that in figures 2 and 3 only the values of @xmath0 corresponding to the accreted gas mass were plotted in order to allow an easier track of the mass and magnetic flux evolution of the disk and envelope material .
nonetheless , the total @xmath0 in the sgl12 models are larger than unity when including the sink , as in s+12 models ( see sgl12 ) .
another important parameter in this analysis of magnetic flux transport is the initial ratio between the thermal and magnetic pressure of the gas , @xmath32 , which is also similar in both models ( @xmath33 in the center of the cloud in the s+12 models , while @xmath34 in the whole core of the sgl12 models ) . as a matter of fact
, sgl12 chose an initial value of @xmath32 smaller than unity in the core in order to show that reconnection diffusion could be able to remove the magnetic flux excess even from an initially magnetically dominated gas and thus solve the magnetic braking problem . in the case of the s+12 models , it is possible that due to their setup , sink regions far from the center of the cloud do not have this constraint on @xmath32 . in this case , such regions might not require , in principle , removal of magnetic flux to allow the formation of a rotating disk and thus this initial condition would naturally avoid the magnetic braking problem . however ,
when ingredients such as rotation and turbulence are introduced , the accreting history in the different sinks within this cloud may change completely depending on the relative strength of the turbulence and magnetic field . in this sense
, the formation of a keplerian disk is very sensitive to the @xmath35 conditions of the region around the sink ( rather than the global initial conditions of the entire cloud ) and therefore , the track of the detailed evolution of the magnetic flux around each sink region at the scale of the disk build up would be required in order to evaluate the real evolution of @xmath0 around each sink .
such an analysis is missing in s+12 work .
based on the discussion in the previous paragraphs , it is natural to assume that the increase in @xmath0 detected in some of the s+12 models within the disk radius is @xmath5 , as in sgl12 model .
this increase could be simply an evidence that flux loss was very efficient during the disk build up in the turbulent envelope around this sink .
the fact that they find a final @xmath26 in the disk which is much larger than the initial value in the cloud suggests that even if numerical resistivity is operating in the inner regions , a substantial magnetic flux excess was removed by turbulent reconnection diffusion when the disk was still forming from the accreting envelope .
regarding the potential role of the numerical resistivity , we can make some quantitative estimates .
the relevant scales for the reconnection diffusion to be operative are the turbulent scales , from the injection to the dissipation scale , i.e. , within the inertial range scales of the turbulence which are larger than the numerical viscous scale . in the simulations of sgl12 with a resolution of @xmath36
this scale is approximately of 8 cells . to evaluate the relative role of the numerical dissipation on the evolution of the magnetic flux at scales near the dissipation range
, we can compare the advection and the diffusion terms of the magnetic field induction equation at a given scale .
considering the magnetic flux variations of the sgl12 turbulent model within a @xmath37 au scale at @xmath38 kyr , we find that the ratio between these two terms , which gives the magnetic reynolds number , is @xmath39 , where @xmath40 au and we have considered the radial infall velocity as a characteristic velocity of the system in this region , @xmath41 km / s ( see figure a.1 ) .
the approximate numerical resistivity for the employed resolution in sgl12 is @xmath42 .
therefore , although present , the numerical dissipation was not the dominant ingredient driving the change of the magnetic flux inside @xmath37 au in the sgl12 turbulent model . examining the case of the s+12 models with increase in @xmath0 , considering that their magnetic reynolds number must be even larger at 100 au ( i.e. , the numerical viscosity is even smaller ) due to their higher resolution , then in their case there should be no significant magnetic flux removal either at 100 au due to numerical resistivity because of the same arguments above ( see also the appendix ) .
since the * s+12 * authors do not provide the details of the magnetic field , turbulence , and density intensity within their keplerian disks , it is hard to argue whether there was some significant flux loss or not in the other cases that they investigated where no increase was detected in the averaged @xmath0 over a large volume .
it is also possible that some of these disks developed in regions where the local magnetic fields were not strong enough to cause magnetic braking and prevent the growth of the keplerian disk . in these cases ,
even if flux loss by reconnection diffusion is occurring it would be undetectable .
in the previous section , we have demonstrated that an analysis which is based solely on the computation of the average value of the mass - to - magnetic flux ratio ( @xmath0 ) over the whole envelope that surrounds a newly formed keplerian disk , is not adequate to conclude that there is no significant magnetic flux loss in simulations of disk formation .
this averaging masks significant real increases of @xmath0 in the inner regions where the disk is build up out of the turbulent envelope material that is accreting .
actually , we have demonstrated that this is what happens on the build up of the keplerian disk both , in the turbulent and in the resistive models of sgl12 , where magnetic flux loss has been detected .
while the average @xmath0 computed over the large scale envelope / disk does remain nearly constant with time , the value of @xmath0 inside the formed keplerian disk is 5 times larger than the initial value in the cloud core .
similarly , some of the disks formed in the turbulent s+12 models also revealed a value of @xmath0 5 times larger within the disk than the initial cloud value , while the average @xmath0 over the whole envelope surrounding the disk was nearly constant with time .
in sgl12 models it has been found that the reduction of the magnetic braking efficiency during the building of the disk was due to the action of reconnection diffusion .
the latter process is based on the lv99 theory , as remarked , which in turn , has been tested with high resolution numerical simulations @xcite as well as , by lagrangian analysis of turbulent mhd fluids @xcite .
in addition , reconnection diffusion has been successfully tested numerically for the first time in @xcite with very high resolution simulations which , taken together , allows us to claim that the process must be occurring in all magnetized turbulent environments , including regions of accretion disk formation ( see also leo et al .
2012 , de gouveia dal pino et al .
therefore , we suggest that the increase in @xmath0 found in s+12 s disk is real and , as in sgl12 , is caused by reconnection diffusion rather than numerical effects .
while the shear flow generated by the turbulent motions in the surroundings of the disk ( which carry large amounts of angular momentum ) allows the build up of the rotationally supported disk , it also removes the magnetic flux excess due to fast turbulent reconnection , which otherwise may prevent the formation of the disk . during the build up of these disks out of turbulent envelopes around the sinks embedded in a massive cloud core ( as in s+12 models )
, some envelopes may have strong local magnetic support ( low local @xmath0 ) which may prevent the keplerian disk formation unless magnetic flux is removed , and some not .
in the former case , we argue that reconnection diffusion is reducing the effect of the magnetic braking . in the latter case , even in the presence of magnetic flux loss induced by reconnection diffusion in the turbulent flow ,
its effect is marginal and therefore difficult to detect , because the magnetic field is dynamically unimportant .
naturally , both sgl12 and s+12 models suffer from numerical effects at small scales .
however , we have provided quantitative arguments that evidence that numerical resistivity should have a minor effect in both cases .
besides , it is important to emphasize that reconnection diffusion is not just an empirically explored phenomenon , as remarked above . in particular , it has been shown that the lv99 theory does not depend on the microphysics of local reconnection events @xcite .
therefore , reconnection diffusion should be represented correctly by numerical simulations at the scales where the turbulence is not damped ( see also the appendix and lazarian 2011 , santos - lima et al .
2010 , de gouveia dal pino et al .
2012 , leo et al .
since this condition is satisfied at the scales investigated in the numerical simulations here discussed , reconnection diffusion is expected to be represented correctly in these simulations .
r.s.l acknowledges support from the brazilian agency fapesp ( 2007/04551 - 0 ) and e.m.g.d.p from fapesp ( 2006/50654 - 3 ) and cnpq ( 300083/94 - 7 ) .
al acknowledges the support the nsf grant ast 1212096 , the vilas associate award as well as the nsf center for magnetic self - organization in laboratory and astrophysical plasmas .
the calculations presented here were performed in the supercomputer alfa - crucis of the astrophysical informatics laboratory lai of iag , astronomy department , university of so paulo ( funded by fapesp : 2009/54006 - 4 ) .
99 allen a. , li z .- y . , shu f. h. , 2003 , apj , 599 , 363 armstrong j. w. , rickett b. j. , spangler s. r. , 1995 , apj , 443 , 209 chepurnov a. , lazarian a. , 2010 , apj , 710 , 853 ciaravella a. , raymond j. c. , 2008 , apj , 686 , 1372 de gouveia dal pino e. m. , santos - lima r. , lazarian a. , leo m. r. m. , falceta - gonalves d. , kowal g. , 2011 , iaus , 274
, 333 de gouveia dal pino e. m. , leo m. r. m. , santos - lima r. , guerrero g. , kowal g. , lazarian a. , 2012 , phys , 86 , 018401 elmegreen b. g. , scalo j. , 2004 , ara&a , 42 , 211 eyink g. l. , 2011 , phrve , 83 , 056405 eyink g. l. , lazarian a. , vishniac e. t. , 2011 , apj , 743 , 51 fatuzzo m. , adams f. c. , 2002 , apj , 570 , 210 galli d. , lizano s. , shu f. h. , allen a. , 2006 , apj , 647 , 374 hennebelle p. , fromang s. , 2008 , a&a , 477 , 9 kowal g. , lazarian a. , vishniac e. t. , otmianowska - mazur k. , 2012 , nonlinear processes in geophysics , 19 , 297 kowal g. , lazarian a. , vishniac e. t. , otmianowska - mazur k. , 2009 , apj , 700 , 63 krasnopolsky r. , li z .- y . , shang h. , 2011 , apj , 733 , 54 lazarian a. , 2005 , aipc , 784 , 42 lazarian a. , 2011 , arxiv , arxiv:1111.0694 lazarian a. , esquivel a. , crutcher r. , 2012 , apj , 757 , 154 lazarian , a. , santos - lima , r. , & de gouveia dal pino , e. 2010 , numerical modeling of space plasma flows , astronum-2009 , 429 , 113 lazarian a. , vishniac e. t. , 1999 , apj , 517 , 700 leo , m. r. m. , santos - lima , r. , de gouveia dal pino , e. m. , lazarian , a. 2012 , in prep .
li p. s. , mckee c. f. , klein r. i. , fisher r. t. , 2008 , apj , 684 , 380 lizano s. , shu f. h. , 1989 , apj , 342 , 834 mckee c. f. , ostriker e. c. , 2007 , ara&a , 45 , 565 mellon r. r. , li z .- y . , 2008 , apj , 681 , 1356 mestel l. , spitzer l. , jr . , 1956 , mnras , 116 , 503 mouschovias t. c. , 1976 , apj , 207 , 141 nakano t. , tademaru e. , 1972 , apj , 173 , 87 price d. j. , bate m. r. , 2007 , mnras , 377 , 77 santos - lima r. , de gouveia dal pino e. m. , lazarian a. , 2012 , apj , 747 , 21 santos - lima r. , lazarian a. , de gouveia dal pino e. m. , cho j. , 2010 , apj , 714 , 442 seifried d. , banerjee r. , pudritz r. e. , klessen r. s. , 2012 , mnras , 423 , l40 shu f. h. , 1983 , apj , 273 , 202 sych r. , nakariakov v. m. , karlicky m. , anfinogentov s. , 2009 , a&a , 505 , 791 zweibel e. g. , 2002 , apj , 567 , 962
santos - lima et al .
( 2010 ) have performed a rigorous numerical test of the role of turbulent magnetic reconnection diffusion on the transport of magnetic flux in diffuse cylindrical clouds , considering periodic boundaries and different numerical resolutions between @xmath43 and @xmath44 .
they found very similar results for all resolutions which revealed the importance of the diffusion mechanism above to remove magnetic flux from the inner denser to the outer less dense regions of the clouds .
in particular , they demonstrated that the higher the value of the ratio between thermal and magnetic pressure , @xmath32 ( smaller b ) the more efficient the flux removal due to turbulent reconnection diffusion is .
recently , similar studies with self - gravitating spherical magnetized clouds have confirmed this trend as well ( leo et al .
2012 , de gouveia dal pino et al .
in particular , they have demonstrated that turbulent reconnection diffusion of the magnetic flux is very effective and may allow the transformation of initially subcritical into supercritical cores .
the turbulent reconnection diffusion theory is based on the fact that in the presence of turbulence , magnetic reconnection becomes fast and independent of the ohmic resistivity ( at the scales where the magnetic reynolds number @xmath45 is @xmath46 ) .
this is because there is an increase of the number of magnetic reconnection events boosted by the turbulence , in according with lazarian & vishniac ( 1999 ) reconnection theory ( which has been already tested numerically in kowal et al .
2009 , 2012 ) .
however , a common distrust in turbulent numerical simulations is that turbulence could be enhancing the numerical resistivity itself .
this is unjustified , as proved by the careful numerical analysis performed in kowal et al .
( 2009 , 2012 ) and in santos - lima et al .
if this were true , the turbulent reconnection rate would reduce whenever the resolution of the numerical experiment were increased , which is not the case . besides , the turbulent reconnection diffusion coefficient , which is of the order of the richardson hydrodynamical diffusion coefficient ( @xmath47 for super - alfvnic turbulence ; see e.g. , lazarian 2011 ) , is much larger than the numerical ( or the ohmic ) diffusion coefficient at scales larger than the dissipation scales of the turbulence , so that the effects of turbulent reconnection diffusion on the magnetic flux transport are dominant over numerical diffusion at the relevant scales of the system .
nonetheless , in order to provide more quantitative tests about the reliability of the sgl12 models , we also have run a similar turbulent model to that in sgl12 , but with a resolution twice as large , which is named _
turbulent-512_. we also included in this model an explicit small ohmic resistivity ( @xmath48 ) in order to speed up the numerical computation ( this is however , comparable to the numerical viscosity for this resolution and much smaller than the turbulent diffusion coefficient , so that it does not influence the physical results of the problem ) .
the computational domain in this model is cubic with @xmath49 au of side , and the sink accretion radius is half of the value for the models in sgl12 ( @xmath50 au ) . in figure a.1 , we present the radial profiles for the mean values of the radial and azimuthal velocities , the disk ( + envelope ) mass , and the mean vertical magnetic field at @xmath51 kyr for this model , which are compared with those of the sgl12 models . the overall behaviour of these models is described in detail in sgl12 . here , we only emphasize the main features .
we clearly see that the results of the turbulent models for both resolutions are similar .
a disk with a keplerian velocity profile is also build up in the high resolution model within the same radius ( @xmath52 au ) as in the smaller resolution turbulent model of sgl12 .
the mass of the disk inside this radius is also identical in both models and slightly smaller for larger radii ( in the envelope ) in the higher resolution model .
the vertical magnetic field is also slightly smaller for radii larger than @xmath53 au , but similar to the lower resolution model in the inner regions ( except for some fluctuations due to the different sizes of the sinks , but which are not relevant for the present analysis .
the infall velocity is generally smaller in the higher resolution model .
nonetheless , in this model , a slightly thinner disk develops and if this velocity is averaged only over the higher density gas concentrated at smaller heights around the disk , then the infall velocity profile becomes very similar with that of the smaller resolution model .
this also has to do with the fact that the turbulence in the model with lower resolution decays slightly faster , so that at the period of time considered in figure a.1 , the model with lower resolution has reached already a more relaxed , non turbulent state .
the similarity between the results of both turbulent models indicates that the lower resolution model of sgl12 is reliable and therefore , can be employed in the analysis presented in this work .
|
recent numerical analysis of keplerian disk formation in turbulent , magnetized cloud cores by @xcite demonstrated that reconnection diffusion is an efficient process to remove the magnetic flux excess during the build up of a rotationally supported disk .
this process is induced by fast reconnection of the magnetic fields in a turbulent flow . in a similar numerical study
, @xcite concluded that reconnection diffusion or any other non - ideal mhd effects would not be necessary and turbulence shear alone would provide a natural way to build up a rotating disk without requiring magnetic flux loss .
their conclusion was based on the fact that the mean mass - to - flux ratio ( @xmath0 ) evaluated over a spherical region with a radius much larger than the disk is nearly constant in their models . in this letter
we compare the two sets of simulations and show that this averaging over large scales can mask significant real increases of @xmath0 in the inner regions where the disk is built up .
we demonstrate that turbulence - induced reconnection diffusion of the magnetic field happens in the initial stages of the disk formation in the turbulent envelope material that is accreting .
our analysis is suggestive that reconnection diffusion is present in both sets of simulations and provides a simple solution for the `` magnetic braking catastrophe '' which is discussed in the literature in relation to the formation of protostellar accretion disks .
[ firstpage ] diffusion ism : magnetic fields mhd turbulence star formation accretion disks
|
the search for higgs bosons plays one of the most important roles at high - energy collider experiments at the tevatron and the lhc .
the higgs boson is the remnant of electroweak symmetry breaking in the scalar higgs sector of the standard model ( sm ) and its supersymmetric extensions .
the minimal supersymmetric extension of the sm ( mssm ) requires the introduction of two higgs doublets in order to preserve supersymmetry @xcite .
this leads to the existence of five elementary higgs particles , two cp - even ( @xmath6 ) , one cp - odd ( @xmath7 ) and two charged ( @xmath8 ) states . at lowest order ( lo ) all couplings and masses of the mssm higgs sector
are described by two independent input parameters , which are usually chosen as @xmath9 , the ratio of the two vacuum expectation values @xmath10 , and the pseudoscalar higgs - boson mass @xmath11 . at lo ,
the light scalar higgs mass @xmath12 has to be smaller than the @xmath13-boson mass @xmath14 . including the one - loop and dominant two - loop corrections the upper bound
is increased to @xmath15 gev @xcite .
recent first three - loop results confirm this upper bound within less than 1 gev @xcite .
the couplings of the various neutral higgs bosons to fermions and gauge bosons depend on the mixing angles @xmath16 and @xmath17 .
normalized to the sm higgs couplings , they are listed in table [ tb : hcoup ] .
.[tb : hcoup ] _
higgs couplings in the mssm to fermions and gauge bosons [ @xmath18 relative to sm couplings .
the subscripts @xmath19 denote up- and down - type fermions . _ [ cols="<,^,^,^,^",options="header " , ] the soft supersymmetry - breaking parameters @xmath20 and @xmath21 will be introduced as @xmath22 parameters at an input scale @xmath23 which will in general be of the order of the susy scale . in order to resum large corrections for large values of @xmath24
we define the sbottom masses and couplings to the higgs bosons in terms of the effective bottom mass definition which has been used in ref .
our effective bottom mass runs with five active flavours consistently , i.e. all heavier particles are decoupled , while the @xmath25 definition requires running with the contributions of all supersymmetric particles .
the scale of the strong coupling @xmath26 in @xmath27 within the effective bottom mass is identified with the average mass of the sbottom and gluino states in order to account for the nnlo corrections to a large extent @xcite and is thus different from the scale @xmath28 . ]
@xcite @xmath29 with @xmath30 , where @xmath31 denotes the running bottom mass in the @xmath22 scheme at the scale @xmath28 and @xmath32 are the sbottom and gluino pole masses respectively .
the effective top mass is identified with the running @xmath22 top mass
@xmath33 the use of these effective masses corresponds to the consistent inclusion of the resummed and rg - improved yukawa couplings of the higgs decays @xmath34 everywhere in the squark sectors , too .
the stop / sbottom mass matrix at lo is then given by ( @xmath35 ) @xmath36 \\
\hat m_q(q_0 ) [ \bar{a}_q(q_0)-\mu r_q ] & \tilde{\overline{m}}_{\tilde q_r}^2(q_0 ) + \hat m_q^2(q_0 ) \end{array } \right ] \ , , \ ] ] where @xmath37 denotes the running trilinear @xmath22 coupling at the scale @xmath23 and @xmath38 .
analogous to eqs .
( [ eq : dterms ] ) the @xmath39-terms @xmath40 have been absorbed in the parameters @xmath41 , @xmath42 the stop / sbottom mass matrix is modified by higher - order corrections in the diagonal and off - diagonal entries .
while the corrections to the off - diagonal entries will be treated by the renormalization of the mixing angles , we compensate the radiative corrections to the diagonal matrix elements by shifts in the soft mass parameters @xmath43 , @xmath44 in order to arrive at simple expressions at nlo for the stop / sbottom masses . using the squark pole masses
@xmath45 the tree - level definition @xmath46 of the mixing angle is given by @xmath47 } { m_{\tilde q_1}^2 - m_{\tilde q_2}^2 } \quad , \quad \cos 2\tilde\theta_q = \frac{\tilde m_{\tilde q_l}^2(q_0 ) - \tilde m_{\tilde q_r}^2(q_0 ) } { m_{\tilde q_1}^2 - m_{\tilde q_2}^2 } \ ,
, \label{eq : sqmix}\ ] ] where the shifted squark mass parameters @xmath48 have been used with the radiatively corrected squark pole masses . this definition will be used as the tree - level - like mixing angle with loop - corrected squark masses @xmath45 at nlo , too , but will only play the role of an auxiliary parameter in our analysis as will be explained later . at
nlo the masses of the stop / sbottom eigenstates are given by @xmath49 ^ 2 + 4\hat m_q^2(q_0 ) [ \bar{a}_q(q_0 ) - \mu r_q]^2 } \right ] + \delta m_{\tilde q_{1/2}}^2 \nonumber \\
\delta m_{\tilde q_{1/2}}^2 & = & \sigma_{11/22}(m_{\tilde q_{1/2}}^2 ) + \delta \hat m_{\tilde q_{1/2}}^2 \label{eq : sqmass}\end{aligned}\ ] ] the diagonal parts @xmath50 of the stop / sbottom self - energies can be calculated from the diagrams in fig .
[ fg : sqself ] , ( 130,40)(130,-10 ) ( 0,0)(100,0)5 ( 50,0)(10,20)(0)0.5 ( 130,0)(230,0)5 ( 180,0)(20,0,180)45 ( 260,0)(360,0)5 ( 290,0)(330,0 ) ( 310,0)(20,0,180 ) ( 310,0)(20,0,180)45 ( 390,0)(490,0)5 ( 440,20)(20,0,360)5 ( 440,0)2 ( 88,-3)@xmath51 ( 193,-3)@xmath52 ( 297,-3)@xmath52 ( 10,6)@xmath53 ( 70,6)@xmath54 ( 113,6)@xmath53 ( 143,-13)@xmath55 ( 143,25)@xmath56 ( 175,6)@xmath54 ( 218,6)@xmath53 ( 248,-12)@xmath57 ( 248,25)@xmath58 ( 280,6)@xmath54 ( 323,6)@xmath53 ( 351,35)@xmath55 ( 383,6)@xmath54 + @xmath59 + 4 m_{\tilde q_{1/2}}^2 b_0(m_{\tilde q_{1/2}}^2;0,m_{\tilde q_{1/2 } } ) \label{eq : sigii } \\ & + & \left . \!\!\!\!\ ! 2\!\left [ m_{{\tilde{g}}}^2 + \hat m^2_q(q_0 ) - m_{\tilde q_{1/2}}^2 \!\ ! \mp 2m_{{\tilde{g}}}\hat m_q(q_0 ) \sin 2\tilde\theta_q \right ] \ ! b_0[m_{\tilde q_{1/2}}^2;m_{{\tilde{g}}},\hat m_q(q_0 ) ] \right\ } \ , .
\nonumber\end{aligned}\ ] ] the one - loop integrals are defined as @xcite @xmath60[(k+p)^2-m_2 ^ 2 ] } \nonumber \\
b_1(p^2;m_1,m_2 ) & = & \frac{1}{2p^2}\left\ { a_0(m_1)-a_0(m_2 ) - ( p^2+m_1 ^ 2-m_2 ^ 2 ) b_0(p^2;m_1,m_2 ) \right\ } \ , .\end{aligned}\ ] ] the parameter @xmath61 denotes the t hooft mass of dimensional regularization .
the counter terms @xmath62 are given by @xmath63 \right\ } \nonumber \\ & = & - c_f\frac{\alpha_s}{\pi } \gamma(1+\epsilon ) ( 4\pi)^\epsilon \left\ { \frac{1}{\epsilon } + \log\frac{\bar \mu^2}{q_0 ^ 2 } \right\ } \left\{m^2_{{\tilde{g}}}\mp m_{{\tilde{g}}}\hat m_q(q_0 ) \sin
2\tilde \theta_q ) \right\ } \nonumber \\ & & + \frac{\delta
\hat m_q}{\hat m_q(q_0 ) } \left\ { 2\hat m^2_q(q_0 ) \mp \frac{1}{2}(m_{\tilde q_2}^2 - m_{\tilde q_1}^2 ) \sin^2 2\tilde \theta_q \right\ } \
, , \label{eq : dmsq}\end{aligned}\ ] ] where the tree - level - like mixing angle @xmath46 of eq .
( [ eq : sqmix ] ) has been used and the parameters @xmath64 and @xmath65 have been renormalized in the @xmath22 scheme , @xmath66 the counter term of the effective quark mass @xmath67 for @xmath68 is given by expression used in ref . @xcite . ]
@xmath69 + b_1[\hat m_q^2(q);m_{{\tilde{g}}},m_{\tilde q_2 } ] \right .
\nonumber \\ & & \left .
\hspace*{0.7 cm } + 2 m_{{\tilde{g}}}(\bar a_q-\mu r_q ) \frac{b_0[\hat m_q^2(q);m_{{\tilde{g}}},m_{\tilde q_1 } ] - b_0[\hat m_q^2(q);m_{{\tilde{g}}},m_{\tilde q_2}]}{m^2_{\tilde q_1}-m^2_{\tilde q_2}}\right\ } \ , , \label{eq : dmq}\end{aligned}\ ] ] where @xmath70 is a finite counter term required to restore the supersymmetric relations between the higgs boson couplings to top / bottom quarks and stops / sbottoms within dimensional regularization @xcite .
the term @xmath71 denotes the correction @xmath27 of eq .
( [ eq : deltab ] ) for the bottom mass , while it vanishes for the top mass case , i.e. @xmath72 .
the term @xmath27 in the counter term of the bottom mass cancels the leading term for large values of @xmath24 in the last line of eq .
( [ eq : dmq ] ) so that this counter term is free of large corrections . using the nlo corrected squark pole masses of eq .
( [ eq : sqmass ] ) and the tree - level - like mixing angle @xmath73 of eq .
( [ eq : sqmix ] ) , the shifted squared soft susy - breaking squark mass parameters @xmath74 can be obtained from the relations .
an alternative and equivalent option would be to compensate the shift @xmath75 by a shifted trilinear coupling @xmath76 .
moreover , it should be noted that the shifted parameters @xmath77 and @xmath78 are not equal to each other any more in accordance with the different radiative corrections to the stop and sbottom masses .
] , @xmath79 the tree - level definition of the mixing angle @xmath73 in eq .
( [ eq : sqmix ] ) corresponds to the following counter term at nlo , @xmath80 @xmath81 however , in order to avoid artificial singularities in physical observables for stop / sbottom masses @xmath82 close to each other , in our calculation the mixing angle of the squark fields has been renormalized via the anti - hermitian counter term with @xmath83 or @xmath84 @xcite which leads to a result free of the artificial singularities for stop / sbottom masses @xmath82 close to each other , too .
another option free of these singularities is provided by taking the residue of eq .
( [ eq : sqmixct ] ) for equal masses for the mixing angle counter term @xcite . ] @xcite , @xmath85 where @xmath86 denotes the off - diagonal part of the stop / sbottom self - energy ( see fig .
[ fg : sqself ] ) describing transitions from the first to the second mass eigenstate or _ vice versa _ , @xmath87 \right .
\nonumber \\ & & \left .
\hspace*{2 cm } + \frac{\sin 2\tilde\theta_q}{4 } \left [ a_0(m_{\tilde q_2 } ) - a_0(m_{\tilde q_1 } ) \right ] \right\ } \cos 2\tilde\theta_q \ , .
\label{eq : sig12}\end{aligned}\ ] ] this implies a finite shift @xmath75 to the mixing angle @xmath73 of eq .
( [ eq : sqmix ] ) , @xmath88 which modifies the relations of eq .
( [ eq : mlr ] ) by replacing @xmath89 .
the scale and scheme dependence of the input parameters determining the squark pole masses is explicitly compensated by the corrections @xmath90 to the tree - level relations of eq .
( [ eq : sqmass ] ) , while the scale dependence of the mixing angle is compensated by the finite shift @xmath91 in eq .
( [ eq : sqmix1 ] ) .
it should be noted that the corrected mixing angle fulfills the relation @xmath92 at nlo consistently which will be necessary for the cancellation of the ultraviolet divergences of the nlo corrections to higgs decays into squarks .
the scale of the strong coupling constants @xmath26 in eqs .
( [ eq : sigii ] , [ eq : dmsq ] , [ eq : daq ] , [ eq : dmq ] , [ eq : sig12 ] ) has been identified with the input scale @xmath23 . in order to obtain fully consistent input parameters
( [ eq : deltab ] , [ eq : sqmix ] , [ eq : sqmass ] , [ eq : mlr ] ) would have to be solved iteratively until the squark pole masses do not change anymore . performing this iteration explicitly ,
however , shows that for certain mssm scenarios convergence can not be reached .
this happens in particular for scenarios where the parameters @xmath93 and @xmath94 are close to each other so that the mixing angle @xmath73 is driven towards @xmath95 and @xmath96 towards @xmath97 .
this leads to a situation in which the squark pole masses do not change anymore , but the mixing angles @xmath98 do not correspond to the relations of eqs .
( [ eq : sqmix ] , [ eq : sqmix1 ] ) so that the iterated parameters are inconsistent . this situation can be avoided by stopping the iteration earlier before running into this inconsistent limit .
since the iteration adds contributions beyond nlo to the input parameters and the squark masses , the iteration effects can be considered as arbitrary in a nlo analysis .
we have chosen a procedure where no iteration is performed as described above . with the shifted parameters @xmath99 of eq .
( [ eq : mlr ] ) the squark pole masses are calculated by eq .
( [ eq : sqmass ] ) with vanishing @xmath100 terms and the mixing angles according to eqs .
( [ eq : sqmix ] , [ eq : sqmix1 ] ) , since the nlo corrections are absorbed in the shifted values for @xmath101 . however , for a consistent treatment of the @xmath71 terms of eq .
( [ eq : deltab ] ) iterations for the effective bottom mass @xmath102 are performed for the sbottom sector .
this ensures that the effective bottom mass of the corrected mass matrix corresponds to the nlo sbottom pole masses consistently . the nlo neutral higgs couplings to squarks read in the current eigenstate basis as @xmath103 \label{eq : hsbsbcoup}\end{aligned}\ ] ] with the couplings @xmath104 listed in table [ tb : hsfcoup ] .
note that we use a common renormalization scale @xmath105 for the effective quark mass @xmath106 and the trilinear coupling @xmath65 for simplicity .
the corresponding couplings to the stop / sbottom mass eigenstates @xmath107 are obtained by the appropriate rotations according to eq .
( [ eq : couprot ] ) by the on - shell mixing angle @xmath108 of eq .
( [ eq : sqmix1 ] ) .
the choice of the on - shell mixing angle @xmath108 for these rotations implies that this is also the relevant mixing angle involved in the physical processes .
the on - shell mixing angle is uniquely related to the soft susy - breaking input parameters according to eqs .
( [ eq : sqmix],[eq : sqmix1 ] ) and thus in particular to the parameter @xmath76 . in our numerical analysis
we include the running of the effective quark mass @xmath106 and the trilinear coupling @xmath65 up to the nll level in ( susy)qcd .
note that due to the counter terms for the effective quark masses of eq .
( [ eq : dmq ] ) the heavy particles are decoupled from its running .
we have included 5 ( 6 ) light flavours in the nll running of the effective bottom ( top ) masses .
for the running of the trilinear couplings @xmath65 we have kept the contributions of all coloured particles including the top quark , the squarks and the gluino which is consistent with the counter term @xmath109 of eq .
( [ eq : daq ] ) . the running effective quark mass for @xmath110 light flavours is thus given by @xmath111}{c[\alpha_{s}(m_q)/\pi ] } \nonumber \\
c(x ) & = & \left(\frac{23}{6}\,x\right)^{\frac{12}{23 } } \ , \left[1+\frac{3731}{3174}\,x\right ] \qquad \mbox{for $ q = b \quad ( n_f=5)$ } \nonumber \\
c(x ) & = & \left(\frac{7}{2}\,x\right)^{\frac{4}{7 } } \ , \left[1+\frac{137}{98}\,x\right ] \qquad
\quad \ ; \mbox{for $ q = t \quad ( n_f=6)$ } \ , , \label{eq : msbarev}\end{aligned}\ ] ] where @xmath112 denotes the corresponding quark pole mass .
the nll evolution of the trilinear couplings @xmath65 with @xmath113 quark and squark flavours is determined as the solution of the corresponding two - loop renormalization group equations ( rges ) with respect to the strong coupling constant @xcite and can be expressed as scheme appropriately @xcite . in the @xmath25 scheme
the coefficient @xmath114 is replaced by @xmath115 , the coefficient @xmath116 by @xmath117 and the @xmath22 gluino mass @xmath118 by the corresponding @xmath25 gluino mass in eq .
( [ eq : aqnll ] ) . ]
@xmath119 \left[1+\frac{1}{6}\,\frac{\alpha_{s , susy}(q_0)}{\pi } \right ] \right .
\nonumber \\ & & \left .
\hspace*{3.5 cm } - \frac{16}{27}\,\frac{\alpha_{s , susy}(q_0)}{\pi } \left[\frac{\alpha_{s , susy}^2(\mu_r)}{\alpha_{s , susy}^2(q_0)}-1\right ] \right\ } \label{eq : aqnll}\end{aligned}\ ] ] with the running strong coupling @xmath120 including all quarks , squarks and gluinos , @xmath121 where @xmath122 is determined by the matching condition @xmath123 \right\}\ ] ] between the strong coupling @xmath120 including all quarks , squarks and gluinos and the low - energy qcd couplings @xmath26 with 5 active light flavours .
the @xmath22 mass @xmath118 of the gluino can be obtained from the gluino pole mass @xmath124 , @xmath125 \right\ } \
, , \end{aligned}\ ] ] where @xmath126 .
the sum is performed over all 3 ( s)quark generations .
for the first two generations we neglected the quark masses in the squark mass matrices so that the corresponding squark pole masses are given by @xmath127 ^ 2}{2\tilde{\overline{m}}_{\tilde q_{l / r}}^2(q_0 ) } \log\left| \frac{m_{{{\tilde{g}}}}^2-\tilde{\overline{m}}_{\tilde q_{l / r}}^2(q_0)}{m_{{{\tilde{g}}}}^2 } \right| \right\}\end{aligned}\ ] ]
( 130,90)(0,0 ) ( 0,50)(50,50)5 ( 50,50)(100,100)5 ( 50,50)(100,0)5 ( -15,48)@xmath128 ( 105,98)@xmath55 ( 105,-2)@xmath129 + at leading order the mssm higgs boson decays into squark - antisquark pairs are described by the diagram shown in fig . [ fg : lodia ] .
the corresponding partial decay width reads as @xmath130 ^ 2 \sqrt{\lambda_{ij}}\ ] ] with the two - body phase - space function @xmath131 while higgs boson decays into squarks of the first two generations are generally suppressed , higgs decays into sbottoms and stops develop sizable branching ratios , whenever they are kinematically possible @xcite .
in particular scenarios their branching ratios can reach a level of 8090% @xcite .
( 130,90)(0,0 ) ( 0,50)(50,50)5 ( 50,50)(100,100)5 ( 50,50)(100,0)5 ( 50,50)(30,-45,45)35 ( -15,48)@xmath128 ( 105,98)@xmath55 ( 105,-2)@xmath129 ( 90,48)@xmath56 ( 130,120)(-20,0 ) ( 0,50)(50,50)5 ( 75,75)(100,100)5 ( 75,25)(100,0)5 ( 50,50)(75,75 ) ( 50,50)(30,-45,45 ) ( 50,50)(30,-45,45)35 ( 75,25)(50,50 ) ( -15,48)@xmath128 ( 105,98)@xmath55 ( 105,-2)@xmath129 ( 55,65)@xmath57 ( 55,30)@xmath57 ( 90,48)@xmath58 ( -35,48)@xmath52 ( 115,48)@xmath52 ( 130,120)(-40,0 ) ( 0,50)(20,50)5 ( 50,50)(100,100)5 ( 50,50)(100,0)5 ( 35,50)(15,0,360)5 ( -15,48)@xmath128 ( 105,98)@xmath55 ( 105,-2)@xmath129 ( 30,70)@xmath55 + ( 130,120)(0,0 ) ( 0,50)(50,50)5 ( 50,50)(100,100)5 ( 50,50)(100,0)5 ( 75,75)(8,15)(45)0.5 ( -15,48)@xmath128 ( 105,98)@xmath55 ( 105,-2)@xmath129 ( -40,48)@xmath52 ( 115,48)@xmath52 ( 130,120)(-20,0 ) ( 0,50)(50,50)5 ( 50,50)(100,100)5 ( 50,50)(100,0)5 ( 75,25)(8,15)(-45)0.5 ( -15,48)@xmath128 ( 105,98)@xmath55 ( 105,-2)@xmath129 + ( 130,100)(0,0 ) ( 0,50)(50,50)5 ( 50,50)(100,100)5 ( 50,50)(100,0)5 ( 65,65)(100,65)33 ( -15,48)@xmath128 ( 105,98)@xmath55 ( 105,63)@xmath56 ( 105,-2)@xmath129 ( 130,90)(-20,0 ) ( 0,50)(50,50)5 ( 50,50)(100,100)5 ( 50,50)(100,0)5 ( 65,35)(100,35)33 ( -15,48)@xmath128 ( 105,98)@xmath55 ( 105,33)@xmath56 ( 105,-2)@xmath129 + the susy qcd corrections at nlo have been calculated in refs . @xcite . however , the analytical results will be repeated in this section in our notation and for the investigation of calculational details once consistent input parameters at nlo are included in the theoretical analysis . the susy qcd corrections consist of one - loop virtual corrections , mediated by the feynman diagrams depicted in fig .
[ fg : nlodiav ] , supplemented by the renormalization of the mass , coupling and wave function parameters involved , and real corrections due to gluon bremsstrahlung , see fig .
[ fg : nlodiar ] .
we have performed the loop and phase - space integration within dimensional regularization in @xmath132 dimensions so that ultraviolet and infrared singularities appear as poles in @xmath133 .
the infrared singularities cancel after adding the real and virtual corrections .
the ultraviolet poles disappear after adding the corresponding counter terms which will be discussed in detail in the following , since they have to correspond to the schemes in which the squark masses , mixing angles and couplings are defined .
the wave - function counter terms @xmath134 are determined by normalizing the residues of the diagonalized squark propagators to unity .
the explicit calculation of the diagrams of fig .
[ fg : sqself ] leads to the expressions @xmath135
b'_0(m_{\tilde q_i}^2;m_{{\tilde{g}}},m_q ) \right\ } \ , , \end{aligned}\ ] ] where the derivative of the two - point function is defined as @xmath136 throughout our calculation in this section the quark mass @xmath112 is defined as @xmath137 of eq .
( [ eq : deltab ] ) for decays into sbottom pairs and as @xmath138 of eq .
( [ eq : tmass ] ) for decays into stops .
the mixing angle is renormalized by the anti - hermitian counter term of eq .
( [ eq : sqmixct ] ) .
the trilinear couplings @xmath139 ( @xmath140 ) are defined in the @xmath22 scheme at the scale @xmath105 which is identical to the running couplings in the @xmath25 definition at nlo .
thus , the counter term of @xmath139 can be cast into the form @xmath141 the quark mass counter term @xmath142 is given by @xmath143 with the top and bottom mass counter terms @xmath144 of eq .
( [ eq : dmq ] ) with @xmath145 . the calculation of the diagrams of fig .
[ fg : nlodiav ] for the virtual corrections and those of fig .
[ fg : nlodiar ] for the real corrections leads to the final result after renormalization , @xmath146 + c^\phi_{ct } + c^\phi_{real } \nonumber \\ c^\phi_1 & = & b_0(m_{\tilde q_i}^2;0,m_{\tilde q_i } ) + b_0(m_{\tilde q_j}^2;0,m_{\tilde q_j } ) - b_0(m_\phi^2 ; m_{\tilde q_i } , m_{\tilde q_j } ) \nonumber \\ & & - 2(m_\phi^2-m_{\tilde q_i}^2-m_{\tilde q_j}^2 ) c_0(m_{\tilde q_i}^2,m_\phi^2,m_{\tilde q_j}^2;0,m_{\tilde q_i},m_{\tilde q_j } ) \nonumber \\ c^{h , h}_2 & = & \frac{g_q^{h , h}}{g_{\tilde q_i\tilde q_j}^{h , h } } \left\ { \delta_{ij } m_q^2 \left [ b_0(m_{\tilde q_i}^2;m_{{\tilde{g}}},m_q ) + b_0(m_{\tilde q_j}^2;m_{{\tilde{g}}},m_q ) + 2b_0(m_{h , h}^2;m_q , m_q ) \right .
\nonumber \\ & & \left .
+ ( 2m_{{\tilde{g}}}^2 + 2m_q^2-m_{\tilde q_i}^2-m_{\tilde q_j}^2 ) c_0(m_{\tilde q_i}^2,m_{h , h}^2,m_{\tilde q_j}^2;m_{{\tilde{g } } } , m_q , m_q ) \right ] \right . \nonumber \\ & & + { \cal r}_{ij } m_{{\tilde{g}}}m_q \left [ ( m_{h , h}^2 - 4m_q^2 ) c_0(m_{\tilde q_i}^2,m_{h , h}^2,m_{\tilde q_j}^2;m_{{\tilde{g } } } , m_q , m_q ) \right . \nonumber \\ & & \left .
-b_0(m_{\tilde q_i}^2;m_{{\tilde{g}}},m_q ) - b_0(m_{\tilde q_j}^2;m_{{\tilde{g}}},m_q ) \right ] \right\ } \nonumber \\ c^a_2 & = & \frac{g_q^a}{g_{\tilde q_i\tilde q_j}^a } \left\ { m_q^2 \sin 2\theta_q \left [ b_0(m_{\tilde q_i}^2;m_{{\tilde{g}}},m_q ) - b_0(m_{\tilde q_j}^2;m_{{\tilde{g}}},m_q ) \right . \right .
\nonumber \\ & & \left .
+ ( m_{\tilde q_i}^2-m_{\tilde q_j}^2 ) c_0(m_{\tilde q_i}^2,m_a^2,m_{\tilde q_j}^2;m_{{\tilde{g } } } , m_q , m_q ) \right ] \right . \nonumber \\ & & \left .
-(-1)^i m_{{\tilde{g}}}m_q \left [ m_a^2 c_0(m_{\tilde q_i}^2,m_a^2,m_{\tilde q_j}^2;m_{{\tilde{g } } } , m_q , m_q ) \right .
\nonumber \\ & & \left .
-b_0(m_{\tilde q_i}^2;m_{{\tilde{g}}},m_q ) - b_0(m_{\tilde q_j}^2;m_{{\tilde{g}}},m_q ) \right ] \right\ } \nonumber \\ c^{h , h}_3 & = & -\frac{{\cal s}_{ik } g^{h , h}_{\tilde q_k\tilde q_l } { \cal s}_{lj}}{g_{\tilde q_i\tilde q_j}^{h , h } } b_0(m_{h , h}^2;m_{\tilde q_k } , m_{\tilde q_l } ) \nonumber \\ c^a_3 & = & b_0(m_a^2;m_{\tilde q_i } , m_{\tilde q_j } ) \nonumber \\ \frac{\alpha_s}{\pi } c^{h , h}_{ct } & = & \delta z_{ii}+\delta z_{jj } + \frac{2}{g_{\tilde q_i\tilde q_j}^{h , h } } \left [ \frac{\partial g^{h , h}_{\tilde q_i\tilde q_j}}{\partial m_q } \delta m_q + \frac{\partial g^{h , h}_{\tilde q_i\tilde q_j}}{\partial a_q } \delta \bar
a_q + { \cal t}_{ij } \delta\theta_q \right ] \nonumber \\ \frac{\alpha_s}{\pi } c^a_{ct } & = & \delta z_{ii}+\delta z_{jj } + \frac{2}{g_{\tilde q_i\tilde q_j}^a } \left [ \frac{\partial g^a_{\tilde q_i\tilde q_j}}{\partial m_q } \delta m_q + \frac{\partial g^a_{\tilde q_i\tilde q_j}}{\partial a_q } \delta \bar
a_q \right ] \nonumber \\ c^\phi_{real } & = & c_f \gamma(1+\epsilon ) \left ( \frac{4\pi\bar\mu^2}{m_\phi^2}\right)^\epsilon \left\ { \frac{1-\rho_i-\rho_j}{2\beta } \left [ \frac{\log x_0}{\epsilon } + \log x_0 \log\frac{\rho_i\rho_j x_0}{\beta^4 } \right .
. \nonumber \\ & & \left .
\left . - \frac{1}{2 } \log^2 x_1 - \frac{1}{2 } \log^2 x_2 + 4
li_2\left(\frac{1-x_0}{-x_0}\right ) - 2 li_2(1-x_1 ) - 2 li_2(1-x_2 ) \right ] \right . \nonumber \\ & & \left .
+ \frac{1}{\epsilon } + \log\frac{\rho_i\rho_j}{\beta^4 } + 4 - \frac{1+\rho_i+\rho_j}{2\beta } \log x_0 + \frac{\rho_i\log x_1 + \rho_j\log x_2}{\beta}\right\ } \ , , \label{eq : cphi}\end{aligned}\ ] ] where in @xmath147 we used the abbreviations @xmath148 with the two - body phase - space function @xmath149 of eq .
( [ eq : lambda ] ) and @xmath150 with @xmath151 .
the two mixing matrices @xmath152 and @xmath153 used in the coefficients @xmath154 and @xmath155 are given by @xmath156 the three - point function @xmath157 is defined as @xcite @xmath158[(k+p_1)^2-m_2 ^ 2][(k+p_{12})^2-m_3 ^ 2]}\ ] ] with @xmath159 .
the finite remainders of the mixing angle renormalization at the external squark legs can be cast into the form of the matrix @xmath160 multiplied by the finite shift @xmath161 which includes the corresponding anti - hermitian counter term @xmath162 of the mixing angle as given in eq .
( [ eq : sqmixct ] ) . in eq .
( [ eq : deltath ] ) the singularity for @xmath163 is cancelled by the anti - hermitian counter term .
finally the derivatives of the squark couplings @xmath164 to the higgs bosons are given by @xmath165 the results for the coefficients @xmath166 are ultraviolet and infrared finite after all individual contributions are added up since our renormalized mixing angles fulfill the relation @xmath167 consistently . the scale of @xmath26 in eq.([eq : cphi ] ) is identified with the renormalization scale everywhere apart from the @xmath27 term within the effective bottom mass of eq .
( [ eq : deltab ] ) .
it should be noted that our final results do not contain any leading @xmath24-enhanced corrections of the @xmath27-type anymore due to our scheme choices of the effective masses and couplings at nlo . due to this property
the problems of a naive renormalization program have been solved consistently leading to moderate radiative corrections to the higgs decay widths into squark - antisquark pairs as will be shown explicitly in the next section . by using power counting arguments along the lines of ref .
@xcite it can be shown that our approach resums all leading terms of @xmath168 so that the residual ho corrections are free of these leading terms .
in the current eigenstate basis there are three sources of terms proportional to @xmath24 : _
( i ) _ the higgs couplings to squarks and quarks as in table [ tb : hcoup ] and eq .
( [ eq : hsbsbcoup ] ) which drop out in the relative corrections ; _ ( ii ) _ off - diagonal mass insertions in the sbottom propagators which are power - suppressed by @xmath169 due to the kln theorem and the @xmath169 structure of these insertions ; ( iii ) potential @xmath24-enhancements of the counter terms .
the @xmath22 renormalization of @xmath0 as in eq .
( [ eq : daqp ] ) does not involve any @xmath24 enhanced corrections as can also be inferred from the beta functions of the rges for the trilinear couplings up to three - loop order @xcite .
the @xmath24 enhancement of the bottom - mass counter term entering the corresponding bottom yukawa couplings involved in the higgs couplings to bottom quarks and the sbottom mass matrix is explicitly absorbed in the effective bottom mass of eq .
( [ eq : deltab ] ) which contains the corresponding threshold correction @xmath27 in resummed form @xcite .
therefore the counter terms @xmath170 and @xmath171 are free of @xmath24 enhanced corrections at all orders up to terms which are suppressed by @xmath169 .
close to the threshold @xmath172 the nlo susy
qcd corrections develop coulomb singularities , @xmath173 which agrees with the usual sommerfeld rescattering correction factor @xcite .
these singularities for @xmath174 can be regularized by taking into account the finite decay widths of the squarks .
moreover , these coulomb factors can be resummed systematically .
both effects are expected to be relevant close to threshold , but are beyond the scope of this paper . on the other hand far above the threshold @xmath175 the nlo corrections approach the asymptotic limit @xmath176 \right .
\nonumber \\ & & \left .
\hspace*{0.5 cm } + 2\frac{m_s}{g_{\tilde q_i\tilde q_j}^\phi } \frac{\partial g_{\tilde q_i\tilde q_j}^\phi}{\partial a_q } \left [ -\log\frac{\mu_r^2}{m_s^2 } + \frac{1}{4 } \log^2\frac{m_\phi^2}{m_s^2 } + \frac{\zeta_2}{2 } - 2 \right ] + \delta_\phi \right\ } \nonumber \\ \delta_{h , h } & = & \left [ \frac{m_q}{g_{\tilde q_i\tilde q_j}^\phi } \frac{\partial g_{\tilde q_i\tilde q_j}^\phi}{\partial m_q } - 1 \right ] \left [ \log\frac{m_\phi^2}{m_s^2 } - 2 \right ] + \frac{g^\phi_q m_q^2}{g_{\tilde q_i\tilde q_j}^{h , h } } \delta_{ij } \log\frac{m_\phi^2}{m_s^2 } \nonumber \\ & & + ~{\cal t}_{ij}~\frac{m_q}{m_s } \frac{\cos 2\theta_q}{g_{\tilde q_i\tilde q_j}^\phi } \left [ 1-\frac{1}{2}\log\frac{m_s^2}{m_q^2 } \right ] \nonumber \\ \delta_a & = & 0 \ , , \label{eq : largemh}\end{aligned}\ ] ] where we have identified all supersymmetric masses , @xmath177 , and neglected the quark mass against the other masses . the term @xmath178 is given by @xmath179 thus confirming the absence of large corrections of @xmath180 to the on - shell quark and squark mass as well as the mixing angle counter terms as in eq .
( [ eq : sqmixct1 ] ) and renormalizes the quark mass on - shell , large corrections of @xmath181 emerge for the partial higgs decay widths into squarks . ] . the asymptotic result far above the threshold develops a double logarithmic contribution in the second square bracket of eq .
( [ eq : largemh ] ) which originates from the second diagram of fig .
[ fg : nlodiav ] in the sudakov limit .
it can not be absorbed in an appropriate choice of the renormalization scale @xmath105 but will cancel against the corresponding double logarithm of gluino radiation in association with quark - squark pairs in the large higgs mass limit which has not been taken into account in this work . in the following
we will identify @xmath105 with the corresponding higgs mass @xmath182 as the central scale choice .
the numerical analysis of the neutral higgs boson decays into stop and sbottom pairs is performed for two mssm scenarios , one close to the one of ref .
@xcite where we lifted the gluino and squark masses of the first two generations beyond the mass bounds from the lhc @xcite , and a second scenario with large susy - breaking masses and large mixing in the stop sector as representative cases : @xmath183 where @xmath184 and @xmath185 denote the squark and slepton mass parameters of the first two generations .
the results of this work have been implemented in the program hdecay @xcite which calculates the mssm higgs decay widths and branching ratios including the relevant higher - order corrections @xcite .
we use the rg - improved two - loop expressions for the higgs masses and couplings of ref .
@xcite which yield predictions for the higgs boson masses that agree with the diagrammatic calculations of ref .
@xcite within 34% in general . thus the leading one- and two - loop corrections
have been included in the higgs masses and the effective mixing angle @xmath16 .
consistency of our scheme with the scheme and scale choices of ref .
@xcite requires the evolution of our sbottom parameters @xmath186 and @xmath187 to the scale @xmath188 and the stop parameters @xmath189 and @xmath190 to the scale @xmath191 since the scales @xmath23 and @xmath192 are of similar order of magnitude we neglect resummation effects so that the relations are given by and @xmath193 are not equal , because they are evaluated for different scales @xmath194 and @xmath195 .
we have modified the calculation of ref .
@xcite to account for these differences consistently . ] @xmath196 \right\ } \log\frac{q_b^2}{q_0 ^ 2 } \nonumber \\
\overline{m}^2_{\tilde b_r}(q_b ) & = & \overline{m}^2_{\tilde b_l}(q_0 ) + \left\ { - c_f \frac{\alpha_s(q_0)}{\pi } m_3 ^ 2(q_0 ) + \frac{\alpha_b}{4\pi } x_b \right\ } \log\frac{q_b^2}{q_0 ^ 2 } \nonumber \\
\bar a_t(q_t ) & = & \bar a_t(q_0 ) + \left\ { c_f \frac{\alpha_s(q_0)}{\pi } m_3(q_0 ) + \frac{\alpha_t}{4\pi } \bar a_t(q_0 ) + \frac{3}{2 } \frac{\alpha_b}{\pi } \bar a_b(q_0 ) \right\ } \log\frac{q_t^2}{q_0 ^ 2 } \nonumber \\
\overline{m}^2_{\tilde t_l}(q_t ) & = & \overline{m}^2_{\tilde t_l}(q_0 ) + \left\ { - c_f \frac{\alpha_s(q_0)}{\pi } m_3 ^
2(q_0 ) + \frac{1}{4 } \left [ \frac{\alpha_t}{\pi } x_t + \frac{\alpha_b}{\pi } x_b \right ] \right\ } \log\frac{q_t^2}{q_0 ^ 2 } \nonumber \\ \overline{m}^2_{\tilde t_r}(q_t ) & = & \overline{m}^2_{\tilde t_l}(q_0 ) + \left\ { - c_f \frac{\alpha_s(q_0)}{\pi } m_3 ^ 2(q_0 ) + \frac{\alpha_t}{4\pi } x_t \right\ } \log\frac{q_t^2}{q_0 ^ 2 } \label{eq : transform}\end{aligned}\ ] ] with the abbreviations @xmath197 and the soft susy - breaking higgs mass parameters @xmath198 the top pole mass has been taken as @xmath199 gev , while the bottom quark pole mass has been chosen to be @xmath200 gev , which corresponds to a @xmath22 mass @xmath201 gev .
the strong coupling constant has been normalized to @xmath202 which corresponds to the qcd scale @xmath203 mev for 5 active flavours .
the related susy qcd scale @xmath122 of eq .
( [ eq : als_susy ] ) amounts to @xmath204 kev in scenario a and @xmath205 kev in scenario b. the @xmath22 masses involved in the scales @xmath206 can be derived as @xmath207 gev and @xmath208 gev while the @xmath22 gluino mass at the input scale @xmath23 amounts to @xmath209 tev in scenario a and @xmath210 tev in scenario b. the squark masses in the two scenarios amount in particular to @xmath211 where @xmath212 and @xmath213 denote the up- and down - type squark masses of the first two generations . in fig .
[ fg : vienna_t ] we display the results for the partial decay widths of the heavy neutral higgs particles into stop pairs and the relative corrections in scenario a. since the stops are moderately heavy all decay modes are kinematically allowed for higgs masses above about 560 gev .
the partial decay widths range at the few gev level and turn out to be similar for all higgs bosons above the corresponding thresholds except the scalar higgs decay widths into non - diagonal stop pairs which are smaller . apart from the coulomb singularities at threshold the susy
qcd corrections are of moderate size in our scheme .
( 100,500)(0,0 ) ( 40.0,120.0 ) ( 40.0,-135.0 ) similar results have been obtained for the neutral higgs boson decay widths into sbottom pairs as shown in fig .
[ fg : vienna_b ] .
the moderate size of the full susy
qcd corrections confirms the proper absorption of the @xmath27 terms as in eq .
( [ eq : deltab ] ) . if the trilinear coupling @xmath0 and the bottom quark mass @xmath214 would be renormalized in the on - shell scheme the susy
qcd corrections would increase the lo results by more than an order of magnitude so that the result in the on - shell scheme becomes totally unreliable @xcite .
the partial widths of the non - diagonal higgs decays @xmath215 are small , since the mixing factor @xmath216 is small and thus the contributions of the left - right couplings @xmath217 are suppressed , while the diagonal couplings @xmath218 and @xmath219 cancel each other to a large extent in the evaluation of the coupling @xmath220 of eq .
( [ eq : couprot ] ) . using the scheme of ref .
@xcite for the mixing angle , i.e. defining the mixing angle counter terms as @xmath221 with @xmath222 or @xmath223 , the results agree with ours within less than 1% of eq .
( [ eq : sqmix ] ) develop sizeable differences to the results in our scheme for squark masses @xmath82 close to each other , where the partial decay widths @xmath224 turn out to be negative thus signalizing a basic problem with the tree - level - like mixing angle . ] .
( 100,500)(0,0 ) ( 40.0,120.0 ) ( 40.0,-135.0 ) ( 100,500)(0,0 ) ( 40.0,120.0 ) ( 40.0,-135.0 ) in order to obtain an estimate of the residual theoretical uncertainties we show the renormalization scale dependence of the partial decay widths into stop and sbottom pairs in fig .
[ fg : scale_v ] .
the scale dependences of the partial decay widths into stop and sbottom pairs are significantly reduced at nlo .
varying the renormalization scale by a factor of 2 around the central scale @xmath225 the theoretical uncertainties can be estimated as 510% for decays into stop and sbottom pairs at nlo .
[ fg : br_v ] displays the corresponding branching ratios of the dominant decay modes of the heavy scalar and the pseudoscalar higgs boson in scenario a. it is clearly visible that the decays into stop and sbottom pairs belong to the dominant decay modes for masses above about 300400 gev , i.e. where they are kinematically allowed , reaching branching ratios of up to about 50% in total .
moreover , in scenario a the decay modes into @xmath226 s and neutralinos play a significant role , too .
( 100,500)(0,0 ) ( 40.0,120.0 ) ( 40.0,-135.0 ) in fig .
[ fg : scen_b ] we display the final results for the scenario b. since the masses @xmath227 are close to or larger than 500 gev , the decay modes @xmath228 are kinematically forbidden for @xmath229 tev .
the partial decay widths of decays into stop and sbottom pairs range at the few gev level as in scenario a once they are kinematically allowed .
it can clearly be inferred from fig .
[ fg : scen_b ] that the susy
qcd corrections are of moderate size apart from the threshold regions where the coulomb singularities enhance the size of the corrections .
it should be noted that the results with the sbottom mixing angle renormalized via the tree - level relation of eq .
( [ eq : sqmixct1 ] ) would lead to unphysical negative partial decay widths for heavy scalar higgs decays @xmath230 for higgs masses larger than 1 tev , where these decay channels open up , so that this scheme is strongly disfavoured .
these negative contributions can be traced back to the uncancelled artificial singularity for @xmath231 in the finite remainders @xmath232 of eq .
( [ eq : deltath ] ) , if @xmath233 is replaced by the counter term @xmath234 of eq .
( [ eq : sqmixct1 ] ) @xcite .
( 100,500)(0,0 ) ( 40.0,120.0 ) ( 40.0,-135.0 ) in fig .
[ fg : br_b ] we show the branching ratios of the heavy scalar and the pseudoscalar higgs boson in scenario b as functions of the corresponding higgs masses . as in scenario a the decay modes into stop and sbottom pairs play a significant role , once they are kinematically allowed .
this starts to be the case for the heavy scalar higgs boson @xmath235 already for masses above about 400 gev , where the decay into light stop mass eigenstates opens up .
the pseudoscalar higgs boson can only decay into mixed pairs of a light and heavy stop and sbottom eigenstates so that these decay modes open up for masses above 900 gev .
the decays into charginos and neutralinos contribute significantly for larger higgs masses .
( 100,500)(0,0 ) ( 40.0,120.0 ) ( 40.0,-135.0 ) for smaller values of @xmath24 the decay modes @xmath236 and @xmath237 play a significant role , while the decays into sbottom pairs are usually suppressed .
however , the neutral higgs boson decays into stop pairs still play a significant role and can even be the dominant heavy higgs boson decays in a large higgs mass range .
qcd corrections to these decay modes are of similar size as in the scenarios with large values of @xmath24 analyzed in this work .
in this work we have analyzed the nlo susy
qcd corrections to heavy mssm higgs decays into stop and sbottom pairs and elaborated a consistent scheme and setup for the theoretical nlo results and the corresponding input parameters , i.e. the squark masses and their couplings to the higgs bosons .
a reliable determination of these corrections requires a proper treatment of contributions which are enhanced by @xmath24 .
the derivation of the input squark masses and couplings with nlo accuracy requires a modification of the tree - level relations among the stop and sbottom masses by higher - order corrections , if they should correspond to the physical pole masses .
since the size of these modifications depends on the mssm parameters and pole masses themselves an iterative procedure should be applied .
these iterations , however , do not converge for all scenarios so that we restricted the evaluation to a single iteration step .
the final setup of the input parameters matches nlo accuracy consistently .
a preferred choice of the scheme for the stop and sbottom mixing angles is the on - shell definition via the anti - hermitian counter term which removes spurious singularities for nearly mass degenerate cases which drive the off - diagonal decays @xmath238 to negative and thus unphysical values .
the final nlo corrections are of moderate size apart from the expected coulomb singularities at the decay thresholds .
the residual theoretical uncertainties at nlo have been estimated to be 510% . + * note added in proof . * during completion of this work ref .
@xcite appeared which discusses several options for the renormalization of the stop and sbottom sector with nlo accuracy .
their treatment of the stop and sbottom sectors is different from our approach in particular for the effective bottom mass and the squark mixing angles .
+ * acknowledgments . *
+ we are grateful to a. djouadi , m. mhlleitner , h. rzehak and p.m. zerwas for carefully reading our manuscript and useful comments .
p. fayet , nucl . phys .
* b90 * ( 1975 ) 104 , phys .
* b64 * ( 1976 ) 159 and phys .
* b69 * ( 1977 ) 489 . s. dimopoulos and h. georgi , nucl .
b193 * ( 1981 ) 150 ; n. sakai , z. phys .
* c11 * ( 1981 ) 153 ; k. inoue , a. kakuto , h. komatsu and s. takeshita , prog .
* 67 * ( 1982 ) 1889 , prog .
* 68 * ( 1982 ) 927 [ erratum - ibid .
* 70 * ( 1983 ) 330 ] and prog .
* 71 * ( 1984 ) 413 .
see e.g. g. degrassi , s. heinemeyer , w. hollik , p. slavich and g. weiglein , eur .
j. * c28 * ( 2003 ) 133 .
s. p. martin , phys .
* d75 * ( 2007 ) 055005 ; r. v. harlander , p. kant , l. mihaila and m. steinhauser , phys .
* 100 * ( 2008 ) 191602 , erratum _ ibid . _
* 101 * ( 2008 ) 039901 ; p.
kant , r. v. harlander , l. mihaila and m. steinhauser , arxiv:1005.5709 [ hep - ph ] .
s. schael _ et al . _
[ aleph , delphi , l3 and opal collaborations ] , eur . phys .
j. * c47 * ( 2006 ) 547 .
m. s. carena _ et al .
_ [ higgs working group collaboration ] , arxiv : hep - ph/0010338 .
atlas collaboration , technical design report , cern
lhcc 9914 ( may 1999 ) ; g. l. bayatian _ et al . _
[ cms collaboration ] , j. phys .
* g34 * ( 2007 ) 995 .
a. djouadi , j. kalinowski , p. ohmann and p. m. zerwas , z. phys .
* c74 * ( 1997 ) 93 .
a. bartl , h. eberl , k. hidaka , t. kon , w. majerotto and y. yamada , phys .
* b402 * ( 1997 ) 303 ; a. arhrib , a. djouadi , w. hollik and c. jnger , phys .
* d57 * ( 1998 ) 5860 . c. weber , k. kovarik , h. eberl and w. majerotto , nucl .
* b776 * ( 2007 ) 138 .
h. eberl , k. hidaka , s. kraml , w. majerotto and y. yamada , phys .
* d62 * ( 2000 ) 055006 . s.p .
martin and m.t .
vaughn , phys .
* b318 * ( 1993 ) 331 .
l. j. hall , r. rattazzi and u. sarid , phys .
* d50 * ( 1994 ) 7048 ; r. hempfling , phys .
* d49 * ( 1994 ) 6168 ; m. carena , m. olechowski , s. pokorski and c. e. wagner , nucl .
* b426 * ( 1994 ) 269 ; m. carena , d. garcia , u. nierste and c. e. wagner , nucl .
* b577 * ( 2000 ) 88 . j. guasch , p. hfliger and m. spira , phys . rev . * d68 * ( 2003 ) 115001 .
d. noth and m. spira , phys .
* 101 * ( 2008 ) 181801 and arxiv:1001.1935 [ hep - ph ] ; l. mihaila and c. reisser , jhep * 1008 * ( 2010 ) 021 .
a. crivellin , l. hofer , j. rosiek , jhep * 1107 * ( 2011 ) 017 . g. t hooft and m.j.g .
veltman , nucl .
* b153 * ( 1979 ) 365 ; g. passarino and m.j.g .
veltman , nucl .
* b160 * ( 1979 ) 151 .
w. beenakker , r. hpker , t. plehn and p.m. zerwas , z. phys .
* c75 * ( 1997 ) 349 - 356 .
n. baro , f. boudjema , phys .
d80 * ( 2009 ) 076010 .
s. p. martin , m. t. vaughn , phys .
* d50 * ( 1994 ) 2282 ; y. yamada , phys .
rev . * d50 * ( 1994 ) 3537 ; i. jack and d. r. t. jones , phys .
* b333 * ( 1994 ) 372 ; i. jack , d. r. t. jones , a. f. kord , phys .
* b579 * ( 2004 ) 180 .
m. spira , fortsch .
* 46 * ( 1998 ) 203 ; a. djouadi , phys .
* 459 * ( 2008 ) 1 .
a. sommerfeld , _ atombau und spektrallinien _ ,
vieweg , braunschweig 1939 .
v. khachatryan _ et al . _
[ cms collaboration ] , arxiv:1101.1628 [ hep - ex ] ; the atlas collaboration , arxiv:1102.2357 [ hep - ex ] .
a. djouadi , j. kalinowski and m. spira , comput .
commun .
* 108 * ( 1998 ) 56 ; a. djouadi , j. kalinowski , m. mhlleitner and m. spira , in j. m. butterworth _ et al .
_ , arxiv:1003.1643 [ hep - ph ] ; a. djouadi , m. m. mhlleitner and m. spira , acta phys .
* b38 * ( 2007 ) 635 - 644 .
carena , m. quiros and c.e.m .
wagner , nucl .
* b461 * ( 1996 ) 407 .
manuel walser , diploma thesis , zrich , 2003 .
|
we analyze neutral higgs boson decays into squark pairs in the minimal supersymmetric extension of the standard model and improve previous analyses .
in particular the treatment of potentially large higher - order corrections originating from the soft susy breaking parameters @xmath0 , the trilinear higgs coupling to sbottoms , and @xmath1 , the higgsino mass parameter , is investigated .
the remaining theoretical uncertainties including the susy - qcd corrections are analyzed quantitatively .
-1.0 cm psi
pr1101 neutral higgs boson decays to squark pairs reanalyzed e. accomando@xmath2 , g. chachamis@xmath3 , f. fugel@xmath3 , m. spira@xmath3 and m. walser@xmath4 _ @xmath2 university of southampton , theory group , southampton so17 1bj , united kingdom + @xmath3 paul scherrer institut , ch-5232 villigen psi , switzerland + @xmath5 institute for theoretical physics , eth zrich , ch8093 zrich , switzerland _
|
the telescope array project is a collaboration with 120 scientists from four nations ( japan , usa , korea , and russia ) , with the observatory located in utah , usa @xcite .
the detectors involved in the telescope array experiment consist of surface detectors ( sds ) arranged in an array and fluorescence detector ( fd ) telescopes . fig .
[ ta_map ] shows the detector map ( squares : sd positions , triangles : fd station positions ) .
the sd array consists of 507 sds , arranged over an area of approximately 700 km@xmath0 with 1.2 km spacing between the sds @xcite .
the sds measure the arrival timing and particle densities of extensive air showers ( eass ) using two - layered plastic scintillators of 3 m@xmath1 .
three fd stations are placed around the sd array .
the fds measure the air fluorescence light emitted from eass ; the fd data are analyzed and the longitudinal development of eass is reconstructed to estimate primary energies , arrival directions , and particle species with the aim of studying the nature of ultra high energy cosmic rays . the fd station near the northwest corner of the sd array houses fourteen fds that were relocated from the high resolution fly s eye ( hires ) experiment ( e.g. @xcite ) .
for these fds and for the data obtained using these fds , we can use the same calibration and analysis methods used in the hires experiment @xcite .
the specifications , configurations , and calibrations of the relocated fds can be obtained from previous reports on hires ( e.g. @xcite ) . on the other hand ,
the southeast ( brm ) and southwest ( lr ) fd sites have twelve fds each ; these fds were newly designed for the ta experiment . a picture of the fd station is shown in fig .
[ station_image ] .
an fd consists of a primary mirror and a photomultiplier tube ( pmt ) camera .
we employed a spherical mirror to obtain a wide field of view ( fov ) with a sufficient focusing power .
in addition , the size of the collecting area of the mirror was determined from the maximum distance for detectable eass at the highest energies .
the mirror aperture and curvature radius were determined to be 3300 mm ( area of 6.8 m@xmath1 ) and 6067 mm , respectively .
this setting realizes the detection of eass with a sufficiently high accuracy from a distance of 30 km and with the primary energy of 10@xmath2 ev .
a primary mirror is segmented into eighteen small hexagonal mirrors ; the distance between the parallel sides of the hexagonal mirror was 660 mm . a pmt camera consisting of 16@xmath316 pmts and having an effective area of 860 mm @xmath3 992 mm is set at a distance of 3000 mm from the mirror .
the fov of each pmt is approximately 1@xmath4 , and that of the camera is 15@xmath4 in elevation and 18@xmath4 in azimuth .
[ telescope_image ] shows a schematic view of the fd telescope frame equipped with a pair of telescopes , one each in the upper and lower parts of the frame .
the heights of the upper and lower mirror centers and the upper and lower camera centers from the floor are 5.5 , 1.5 , 6.0 , and 2.8 m , respectively .
the fov centers of the upper and lower telescopes are at elevations of 10.5@xmath5 , and 25.5@xmath5 , respectively . as shown in fig .
[ telescope_image ] , no segment mirror is installed at the center of the primary mirror . at this position ,
a jig was mounted for alignment of the optical system of the telescope when it was constructed .
similarly , during normal operations , a uv light flasher is mounted here as a standard light source for the calibrations and adjustments of pmt gains @xcite .
the projected view of the fd station onto the station floor is shown in fig .
[ telescope_position ] .
the station has six telescope frames ; accordingly , the fov of a station is from 3@xmath4 to 33@xmath4 in elevation and 108@xmath4 in azimuth . to evaluate the accuracies of the measured shower parameters such as arrival direction , primary energy , and longitudinal development , studies of resolutions and systematic uncertainties of fd telescope optics are important .
thus , we ensured that we accurately manufactured , constructed , and installed the fd telescopes with utmost care and after confirmation of their optical characteristics . in this paper
, we describe the elements that constitute an fd telescope and the installation process of the telescopes .
moreover , the reflectance of segment mirrors , which are continuously monitored after the installations , is also presented .
the contents of this paper are as follows . in sec .
[ camera_production ] and sec .
[ mirror_production ] , we report the designs and productions of the pmt cameras and those of the segment mirrors , respectively .
[ mirror_camera_installation ] explains the installation of the cameras and mirrors .
mirror reflectance and its variations are described in sec .
[ mirror_ref_monitor ] .
in this section , we describe the configuration of the newly designed pmt cameras and the results of user acceptance tests .
each camera has 256 pmts .
7000 pmts including spares have been produced .
the pmt used in the ta experiment , hamamatsu r9508 based on r6234 - 01 , is specially manufactured for the experiment .
the photoelectric surface on the top of a tube is hexagonal in shape with a distance of 60 mm between the parallel side ; its effective area is equivalent to that of circle of 57 mm diameter @xcite . on the bottom of the tube ,
a printed circuit board of a bleeder and a preamplifier is installed . in order to reduce the contamination of the night sky background photons into fluorescence signals
, we use a 4 mm thick optical filter ( schott ag , bg3 ) .
the filter is mounted on the surface of each pmt using a self - fusing tape ( hitachi chemical co. , ltd .
, highbon tape no.2 ) and a polyimide film tape ( 3 m , 5434 ) .
the transmittance of bg3 is higher than 80% in the wavelength range of 305@xmath6395 nm , which is within the range of all the major air fluorescence lines @xcite .
the pmts were inspected by the supplier before shipping and were selected based on their cathode and anode sensitivities , dark current , and other basic parameters . in addition , the quantum efficiency ( including collection efficiency ) , and cathode uniformity were measured for sampled pmts , once for 1000 pmts .
we also imposed sufficiently high gains and low noise levels on pmts at the experimental sites .
the minimum required gain was 6@xmath7 with the maximum applied high voltage of @xmath61200 v. after installation of camera , we adjusted pmt gains to absolutely calibrated pmts @xcite and measured the uniformity of the pmt response at its photoelectric surface on the sites @xcite . during fd observations ,
each pmt gain was monitored every one hour using uv flashers .
we exchanged two of the 6144 pmts because the gains of these two pmts had decreased in the first half year .
since then , no pmt has been exchanged .
a pmt camera has been placed at the primary focus of each fd telescope .
[ camera.fig ] shows the schematic view of the pmt camera .
the camera has a uv transparent acrylic window ( kuraray , paraglas - uv00 ) attached in front of pmts for protection from dust .
the transmittance of the window is more than 90% in the wavelength range of 300@xmath6400 nm @xcite .
pmts were fixed on a 16 mm thick plate of extra super duralumin and were arranged on a triangular grid with a clearance gap of approximately 1 mm among them .
the gap geometries were measured based on the uniformity of the camera surface response @xcite .
signals from the pmts and the dc power to the pmts are transferred through shielded lan cables of category 5 ( length : 50@xmath6100 cm ) between the pmts and the patch panels inside a camera .
the patch panels connected to the lan cables are shown in fig .
[ camera_rear_photo.fig ] .
we use a shielded 40 core twisted pair cable ( bando densen co. , ltd .
, bios - a-2820p ) each for sixteen pmts to transfer the pmt signals from the patch panels in a camera to a data acquisition ( daq ) system in the electronics room with temperature control , situated 20 m from the camera location .
the cable length for the upper and lower cameras is 25.5 m and 22.3 m , respectively .
detailed descriptions of the electronics and daq system are presented elsewhere @xcite .
high voltages ( hvs ) are applied to pmts via coaxial cables ( bando densen co. , ltd . ,
1.5d-2v ) using an order - made power supply ( takasago co. , ltd . ) .
the power supply has 256 outputs , and hv value on each output can be controlled and monitored separately through ethernet , with the accuracy of @xmath8 .
this uncertainty corresponds to approximately @xmath9 of the pmt gains under our setting .
the typical applied hv to pmts is @xmath10 v , and the rated maximum voltage is @xmath11 v. as a safety measure to protect human beings , pmts , and the power supply , the power supply is automatically shut down when unusually high values are detected .
a dc power supply ( kenwood , pw18 - 3ad ) is connected to two patch panels to distribute dc voltage of @xmath12 v to the 256 preamplifiers , each connected to one pmt . in this manner
, we prepared one dc power supply for each camera .
in this section , we describe the specifications of the segment mirrors employed in the new fd telescopes of the ta experiment , their pre - shipment and acceptance inspections , and the inspection results .
the 3.3 m aperture spherical mirror of the fd telescope is segmented into eighteen small mirrors , as shown in fig .
[ telescope_image ] .
the mirror is made of borosilicate glass ( schott ag , tempax ) , and each segment is hexagonal in shape with a thickness of 10.5 mm and a distance of 660 mm between the parallel sides . on the surface of the mirror was deposited a 200 nm thick reflection coating of aluminum through vacuum deposition , and on this coating was deposited a 50 nm thick protective coating of @xmath13 through anodization process .
the wavelength that induces maximum reflectance depends on the thickness of the anodized coating .
this thickness was optimized to obtain the maximum reflectance at wavelength of 350 nm .
the optical parameters and fabrication accuracies of segment mirrors depend on the experimental requirement . in order to optimize the optical parameters , we studied the reconstruction accuracies of eass using ray - tracing , eas simulation , and reconstruction programs . from our studies
, we determined the specifications of the optical parameters and the fabrication accuracies ; the curvature radius of the segment mirror was 6067 mm @xmath14100 mm , the spot size at the focal point was smaller than 40 mm ( it is comparable size of the pmt surface ) , and the reflectance in the wavelength range of 300 nm@xmath6400 nm was more than 85% .
segment mirrors were manufactured ( sanko seikohjyo co. , ltd . ) from january to november , 2004 , for the brm station and from march to december , 2006 , for the lr station .
in all , 500 segment mirrors , including spares , were produced .
the procedure for mirror production is as follows .
first , the segment mirror was shaped by heating a planar glass on a ceramic model plate in a temperature - controlled electric oven .
after this spherical mirror fabrication , the mirrors were selected through curvature radius measurement .
next , the mirror surface was coated with 200 nm thick aluminum through the vacuum deposition . to protect the surface , 50 nm thick @xmath13 crystal layer was coated on it ; these @xmath13 crystals were produced from a solution containing ammonium hydroxide , tartaric acid , and ethylene glycol .
after the surface fabrication , the mirrors were selected based on reflectance measurement .
finally , a 150 mm diameter support disk made of the same material as the mirror and a stainless steel flange were bonded to the back of the mirror using glue ( 3 m , dymax 840 ) .
the flange was used to support the segment mirror from the telescope frame . for the acceptance inspections , we measured the curvature radius and spot size at the curvature center . from our ray - tracing studies
, we estimated an acceptable spot diameter of 20 mm , in which 90% of the reflected photons fall , at the center of the curvature .
because of this phenomenon , parallel incident light makes a spot of 40 mm diameter on the camera , in which 68% of reflected photons fall , in the case of a normal fd observation .
we developed a test system to measure the curvature radius of mirrors as the distance corresponding to the minimum spot size .
[ tamed_setup ] shows the schematic view of the system , and fig .
[ tamed_setup_pic ] shows the photograph of the same .
this system consisted of a linear precision motion stage ( range : @xmath14250 mm , accuracy : 0.04 mm , chuo precision industrial co.,ltd . , mm stage als-250-c2p with controller qt - cd1 ) ; a diffused light source , i.e. , a green laser ( kochi toyonaka giken co.,ltd .
, glm - d2 ) with a diffuser plate ; an image scanner ( range : 16 bit , resolution : 2400@xmath64800 dpi , canon , canoscan lide80 ) ; and a laser distance meter ( accuracy : @xmath141 mm , murakami giken , dist pro@xmath15a ) .
the diffused light source was set 100 mm away from the optical axis of the mirror , and the center of the scanner s sensitive area was fixed at the axisymmetrical point to the optical axis .
both the light source and the scanner were mounted on the motion stage and were collectively moved between 6067 @xmath14100 mm from the center of the mirror in 5 mm steps .
next , we determined the curvature radius by searching for the point where the spot size was minimized . from our ray - tracing calculations
, we confirmed that the spot size of the light reflected on the scanner was minimized at the same distance as the curvature radius in spite of the off - axis alignments of the light source and the scanner .
a typical observed spot image is shown in fig .
[ scaner_photo ] . in this analysis
, the image size was defined as the diameter of the circle in which 90% of the detected photons fall ; this circle was centered at the weighted center of the image .
circles in fig .
[ spotsize_distance ] show an example relation between the spot size and the distance from a mirror . from the data points ( circles ) in fig .
[ spotsize_distance ] , we obtain the curvature radius to be 6082 mm . typically , as shown in fig
. [ scaner_photo ] , the image shape was non ideal circle , because the surface of the mirrors was slightly elliptical .
accordingly , we required an additional criterion for elliptical mirrors .
we also measured spot sizes of elliptical mirrors as follows .
first , we fitted a shot image at the beginning of a distance of 5967 mm with a line ( called x axis ) , and another line ( called y axis ) was obtained perpendicular to the x axis .
these axes were fixed during the subsequent calculations .
the shot image was projected on these axes , and the projected spot sizes on these axes were calculated at each distance step that the stage away from the mirror .
finally , we obtained the curvature radii on these axes from the minimum of these plots .
we also required the curvature radii on the two axes to be within 6067@xmath14100 mm .
the projected spot sizes on the x and y axes for the sampled mirror are shown as square and triangle points , respectively , in fig .
[ spotsize_distance ] . from this figure ,
the curvature radii on the x and y axes are 6072 mm and 6087 mm , respectively , and these radii also satisfy the above requirement .
[ mirror_curvature_radius ] shows the distribution of the curvature radius of the accepted mirrors , and fig .
[ spotsize_at_curvature_radius ] shows the distribution of the normal spot size in diameter at the curvature radius .
the curvature radius of the accepted mirrors is 6057 mm ( 1@xmath16 : @xmath620/@xmath1730 mm ) , and their spot size is 12 mm ( 1@xmath16 : @xmath62/@xmath173 mm ) in diameter .
henceforth , 1 @xmath16 indicates 68% confidence level . from the acceptance test
, we found that 5% mirrors were unacceptable . to complement the required number of mirrors , other acceptable mirrors were produced additionally .
we employed two different acceptance tests for the segment mirror reflectance .
one is an accurate measurement in the laboratory for a wide wavelength range for sampled mirrors , which was performed by the manufacturer .
the other is a simple measurement with a portable spectrophotometer for a narrow wavelength range for all the segment mirrors .
as it is difficult to measure the reflectance of the curved and large segment mirrors accurately , the manufacturer produced small flat pieces of mirrors simultaneously with sampled segment mirrors and precisely measured the reflectance of the flat pieces as part of the delivery inspection process .
the reflectance of the small flat mirrors was measured using a spectrometer ( jasco inc .
, ubest v-550 ) , whose range , resolution , and accuracy are 190@xmath6900 nm , 2 nm , and 1% , respectively .
[ mirror_ref_sanko ] shows the typical reflectance of sampled small flat mirrors .
this reflectance is more than 90% in the wavelength range of 300@xmath6400 nm , containing all the major air fluorescence lines .
we use a portable spectrophotometer ( konica minolta , inc .
, cm-2500d ) for acceptance inspections of all the manufactured segment mirrors and for routine monitoring of reflectance .
the range and its resolution are 360@xmath6740 nm and 10 nm , respectively .
the nominal uncertainties of measurement are less than 1% . to confirm the accuracy and stability of the spectrophotometer , we measured the reflectance of a reference mirror eighteen times in the period from june 2008 to november 2010 . the reference mirror ( ocean optics inc .
, stan - ssh - nist ) is quartz - coated ; its reflectance was measured by the national institute of standards and technology ( nist ) , usa . fig . [ nist_mirror_ana ]
shows a comparison between our measurements and those done by nist . from this figure ,
the systematic differences are within @xmath141% and the deviations are within @xmath141% in the concerned wavelength range .
[ nist_hist.fig ] shows the variation in the reflectance of the reference mirror at the wavelength of 360 nm , measured using a portable spectrophotometer .
the stability was 0.6% ( 1@xmath16 ) during the period from june 2008 to november 2010 .
the typical reflectance of a primary mirror composed of eighteen segments is shown in fig .
[ mirror_ref_wavelength ] .
this figure shows that the reflectance in the wavelength range of 360@xmath6400 nm is higher than 90% .
the distribution of the reflectance at 360 nm for all the segment mirrors is shown in fig .
[ mirror_ref_wavelength_360 ] .
this distribution shows that the differences in the reflectance depend on 108 sampling points of each primary mirror and that the variance is less than 1% , which is within measurement accuracies .
[ mirror_ref_wavelength_360_each ] shows the reflectance of the primary mirrors in brm and lr stations , which is the average of eighteen segment mirrors , with 1@xmath16 error bars .
all the telescopes fulfilled the requirement that the averaged reflectance was greater than 90% in the wavelength range of 360 nm .
mirrors and cameras were shipped from japan to our observatory in utah , usa , and these items were installed at brm from february 2005 to july 2006 and at lr in march 2007 . in this section , we describe the installations of the fd telescopes and their alignment . as shown in fig .
[ telescope_position ] , six telescope frames were installed in each station ( brm and lr ) with each frame having two telescopes .
hence , in all , twelve telescopes were installed in each station .
a station building has three rolling doors to protect the telescopes from sunlight , rain , wind , and other natural conditions .
the telescopes are directly exposed to air during fd observation , because no other protection windows were installed .
when the wind speed was higher than 15 m / s , doors were closed to prevent themselves from being stuck .
we required less than 0.1@xmath4 fov deflections of the telescope with wind speeds lower than 15 m / s .
wind - induced deflections of the telescopes were studied by the manufacturer ( mitsui engineering & shipbuilding co. , ltd . ) using simulation programs and by applying wind - tunnel tests on a structure model of a station including telescopes . from their studies , winds from the side of the telescope building induce the maximum deflections in the telescope frame .
because winds pass through the building from one door to another , their effective speed is not reduced by the building .
however , the current design of the frame fulfills our requirement with regard to frame deflections , even in the worst condition .
based on the studies of the manufactures , the design of telescope frame structure was fixed . to install the telescope frames
precisely , we conducted a survey and drew datum lines and points for telescope geometry .
using a gyro - compass , the absolute azimuthal direction was measured with reference to a mountain s peak .
uncertainties of the gyro - compass measurement was 1@xmath18 , which was confirmed from the star observation using the gyro - compass .
three datum lines , shown as solid lines in fig
. [ telescope_position ] , were drawn on the floor with relative accuracies of 1@xmath18 between these lines .
azimuthal lines of each telescope frame , shown as dashed lines in fig .
[ telescope_position ] , were drawn from the datum lines with accuracies of 0.1@xmath5 using a measuring tape . from the azimuthal lines , reference points for the telescope frames
were marked with accuracies of 1 mm .
we installed the telescope frames , which can be separated into frames for cameras and mirrors , as shown in fig .
[ telescope_image ] , in accordance with the following procedures .
the telescope frames were placed and fixed on the reference points , and before installing real cameras , a prototype standard camera was attached to the camera frames .
this prototype standard camera had the same weight and size of a real camera and was equipped with reference lines on the surface for adjusting positions .
a steel cylinder with a diameter of 500 mm is mounted at the center of each primary mirror , as shown in fig .
[ telescope_image ] .
the central cylinder was used to adjust the alignment of segment mirrors and the camera of the telescope .
telescope directions were adjusted by the following procedures .
first , we installed a laser range meter on the central cylinder .
next , we adjusted the elevation angle to the specified value , monitored using a digital tilt meter , ( digital protractor , pro 360 ) , with an accuracy of @xmath140.2@xmath4 .
we adjusted the azimuthal angle of the cylinder for the central axis laser to point the string of a plumb 3000 mm away from the cylinder set exactly above the reference line on the floor .
then , the position of the camera frame was precisely adjusted with reference to the central axis laser .
the upper limit for the installed error on the distance between the camera and the cylinder is @xmath1430 mm .
however , this misalignment was resolved when we installed mounts of segment mirrors whose heights were adjusted to remove the distance error . on the other hand , to reduce the construction cost , we required relatively low accuracies of the direction of the camera , which are within 1@xmath4 .
this affects the negligible uncertainties of the effective area of the camera , the order of which is @xmath19 .
after the adjustment of the camera position , the prototype standard camera was unmounted from the camera frame . finally , the positions of the segment mirror mount were adjusted .
[ mirror_setup ] shows the schematic view of a jig called `` banana1 '' for adjusting the positions of the segment mirror mount .
banana1 was mounted on the central cylinder of mirrors and can be rotated around the central axis of the cylinder .
the equipment had two arms , one of which had three protruding legs of different lengths ; the other arm had a counter weight . by rotating banana1 , we adjusted the length of every segment mirror mount for one of the three legs on banana1 to be aligned with the reference point on the mount surface . at the same time
, the direction of each segment mirror was roughly adjusted using banana1 .
the cameras and mirrors were set on the telescope frame by the following procedures .
we set the segment mirrors on their mounts and precisely aligned each direction by screwing in and out two adjusting bolts on a mount . to adjust the direction of segment mirrors
, we use an equipment called `` banana3 '' , shown in fig .
[ mirror_setup2 ] .
the equipment was mounted on the camera frame , and a white board was perpendicularly set on the optical axis of the primary mirror .
the distance between segment mirrors and the board was 6067 mm , which is the same distance as the curvature radius of the mirrors .
green leds were connected to the white board , 5 , 10 , and 15 cm away from the center of curvature , as shown in fig .
[ banana3_face ] .
when the direction of the segment mirror was adjusted , the led lights were reflected to the symmetrical point about the center of curvature .
the direction was adjusted with an accuracy of less than @xmath20 , which was estimated from the accuracy of center position determination of reflected light of 5 mm ( 1/4 diameter of the reflected lights ) and from the distance between mirror and the white board , 6067 mm . for parallel lights ,
this setting accuracy provided us the spots of 40 mm diameter , in which 68% of reflected photons fall , at the camera surface .
after these adjustments , the segment mirror mounts were locked up .
pmt cameras were mounted on the camera frames with an accuracy of 10 mm on the camera plane .
this accuracy corresponds to @xmath210.2@xmath4 uncertainty of the telescope s pointing direction .
pmt cameras were connected by signal , hv , and dc cables . finally , the wiring was confirmed through test operations . in all , the directions of the telescope are adjusted with a few 0.1@xmath5 accuracies .
these directions were confirmed with a star light analysis , which uses a similar method as in reference @xcite .
our fds record air fluorescence signals with background photons ( bgps ) , as we employ a dc coupling for the pmt signal readout .
thus , we obtain the variations in bgp with a normal trigger rate of 2@xmath63 hz . when a star traverses the fov of a pmt , the number of bgps increases and decreases in a few minutes .
the time variations of bgps are simulated from star positions and telescope directions . using these estimated time variations of bgps
, we studied the telescope directions and spot sizes of the primary mirror .
the correction values of the fov direction of fd obtained from this study are used in our simulation and the eas reconstruction programs .
our results of the star light analysis will be reported in a future paper .
the reflectance of the segment mirrors is monitored every months .
mirror reflectance decreases with time , because our fds are exposed to the air directly and because dust is attached to the mirrors during observations .
in addition , the reflectance of lower mirrors decreases more rapidly than that of upper mirrors .
[ mirror_ref_time_dep ] shows the time variations of the averaged mirror reflectance at 360 nm of a typical lower telescope lr04 .
the largest decrement of 4 % occurred between june 2007 and march 2008 .
when we analyze the fd data from that period , we take this 4 % as the systematic uncertainty of the mirror reflectance . every summer ,
the mirrors are washed using purified water and detergent for sensitive equipments ( alconox , inc .
, liqui - nox ) .
purified water is obtained from tap water filtered through charcoal and deionization filters .
after washing , the mirrors are rinsed using purified water and dried naturally .
[ mirror_ref_time_dep ] shows that the reflectance of the mirror after washing is recovered to the same level as in the first installation .
[ ref_diff ] shows the differences in mirror reflectance before and after washing at each wavelength ; these differences have no clear wavelength dependencies .
based on this result , we mainly monitor the reflectance at 360 nm .
in japan , we have produced pmt cameras and mirrors for use in the air fluorescence detector of the telescope array experiment .
the cameras and mirrors that passed our acceptance inspections were shipped to our experimental site in utah , usa . the telescope frames , mirrors , and cameras were installed at the fd stations .
the directions of the telescope are adjusted with an accuracy of a few 0.1@xmath5 , and the directions are confirmed with a star light analysis .
these parameters are used in our detector simulation and the eas reconstruction programs .
monitoring of the performances of the cameras and mirrors begin according to the following schedules : the gains of pmt camera , hour by hour during fd operation ; mirror reflectance , every few months .
our mirrors are washed every summer , and their reflectance after washing is recovered to the same level as in the first installation .
the telescope array experiment is supported by the japan society for the promotion of science through grants - in - aid for scientific research on specially promoted research ( 21000002 ) `` extreme phenomena in the universe explored by highest energy cosmic rays '' , and the inter - university research program of the institute for cosmic ray research ; by the u.s .
national science foundation awards phy-0307098 , phy-0601915 , phy-0703893 , phy-0758342 , and phy-0848320 ( utah ) and phy-0649681 ( rutgers ) ; by the national research foundation of korea ( 2006 - 0050031 , 2007 - 0056005 , 2007 - 0093860 , 2010 - 0011378 , 2010 - 0028071 , r32 - 10130 ) ; by the russian academy of sciences , rfbr grants 10 - 02 - 01406a and 11 - 02 - 01528a ( inr ) , iisn project no .
4.4509.10 and belgian science policy under iuap vi/11 ( ulb ) .
the foundations of dr .
ezekiel r. and edna wattis dumke , willard l. eccles and the george s. and dolores dore eccles all helped with generous donations .
the state of utah supported the project through its economic development board , and the university of utah through the office of the vice president for research .
the experimental site became available through the cooperation of the utah school and institutional trust lands administration ( sitla ) , u.s .
bureau of land management and the u.s . air force .
we also wish to thank the people and the officials of millard county , utah , for their steadfast and warm support .
we gratefully acknowledge the contributions from the technical staffs of our home institutions and the university of utah center for high performance computing ( chpc ) .
we thank hamamatsu photonics k.k .
, sanko seikohjyo co. , ltd . , quality steel fabricating and welding , inc , t&d maintenance , nitto kogyo corporation , and mitsui engineering & shipbuilding co. , ltd . for their kind support .
100 h. kawai et al . , nucl .
suppl , 175 - 176 , ( 2008 ) 221 h. kawai et al .
, j. phys .
a , 78 ( 2009 ) 108 t. abu - zayyad et al . , nim a , 450 ( 2000 ) 253 t. nonaka et al .
, proc . of 30th icrc , 5 ( 2007 ) 1005 , and now preparing a detailed paper j. n. matthews et al .
, proc . of 30th icrc , 5 ( 2007 ) 1157 h. tokuno et al .
, nim a , 601 ( 2009 ) 364 h. tokuno et al .
, proc . of 30th icrc , 5 ( 2007 ) 1013 , and now preparing a detailed paper y. tameda et al .
, nim a , 609 ( 2009 ) 227 a. taketa et al .
, proc . of 29th icrc , 8 ( 2005 ) 209 , and now preparing a detailed paper p. a. sadowski et al .
, astropart .
phys , 18 ( 2002 ) 237 , circles : 2008 , squares : 2009 , triangles : 2010 ) .
these error bars show the non - uniformity of reflectance just before mirror washing . for visibility ,
the circles and triangles are slightly moved on the horizontal axis . ]
|
since 2007 , the telescope array ( ta ) experiment , based in utah , usa , has been observing ultra high energy cosmic rays to understand their origins .
the experiment involves a surface detector ( sd ) array and three fluorescence detector ( fd ) stations .
fd stations , installed surrounding the sd array , measure the air fluorescence light emitted from extensive air showers ( eass ) for precise determination of their energies and species .
the detectors employed at one of the three fd stations were relocated from the high resolution fly s eye experiment . at the other two stations ,
newly designed detectors were constructed for the ta experiment .
an fd consists of a primary mirror and a camera equipped with photomultiplier tubes . to obtain the eas parameters with high accuracies , understanding the fd optical characteristics
is important . in this paper , we report the characteristics and installation of new fds and the performances of the fd components .
the results of the monitored mirror reflectance during the observation time are also described in this report .
ultra high energy cosmic rays , extensive air showers , air fluorescence light detectors
|
bose - einstein condensation has been observed in magnetically trapped dilute vapors of@xmath6 the alkali elements @xmath7rb @xcite , @xmath8na @xcite , @xmath0li @xcite , and @xmath9h @xcite . at the nanodegree temperatures of these experiments ,
the systems would have frozen solid long ago were they in free space . in the confining trap , however , zero - point motion keeps the atoms apart , and the systems remain gaseous .
the case of @xmath0li is special , however , in that the interatomic interaction is predominantly attractive , as indicated by a negative scattering length .
thus , the condensate in @xmath0li should be less stable than the other cases .
the purpose of this paper is to study the nature of the instability , its onset , and manifestations .
the object of study is the condensate wave function @xmath10 , which gives the probability amplitude for annihilating one particle in the condensate at @xmath11 at time @xmath12 .
we use the time - dependent gross - pitaevskii ( gp ) equation , or nonlinear schrdinger equation ( nlse ) , which corresponds to a mean - field approximation : @xmath13 \psi \nonumber \\
u_{0 } & = & \frac{4\pi \hbar ^{2}|a|}{m } \label{nlse}\end{aligned}\ ] ] where @xmath14 is a negative scattering length , and the external potential is taken to be harmonic : @xmath15 this defines a characteristic length @xmath16 , the width of the unperturbed ground - state wave function : @xmath17 the number of condensate particles enters through the normalization @xmath18 which is a constant of the motion .
another constant of the motion is the hamiltonian @xmath19 \label{hamiltonian}\ ] ] the parameters used in @xmath0li experiments correspond to @xcite @xmath20 where @xmath21 is taken to be approximately equal to the geometric mean of the three circular frequencies of the experimental trap .
equation ( [ nlse ] ) does not take into account the coupling between the condensate and the `` thermal cloud '' of uncondensed atoms , nor does it describe the loss of atoms from the trap due to collisions .
these effects will be considered later . in free space
, the nlse with attractive interactions has an instability known in plasma physics as `` self - focusing '' @xcite , whereby the initial wave function develops a singularity in finite time , corresponding to a local collapse to a state of infinite density . in an external trap
, however , the ground state is apparently stable , as long as @xmath1 is not too large , as indicated by variational calculations using a gaussian trial wave function @xcite .
the width of the gaussian narrows with increasing @xmath1 , and collapses to zero at some critical value @xmath2 this conclusion is borne out by other studies @xcite . in particular , kim and
zubarev obtain an exact upper bound for @xmath22 , which for a gaussian wave function gives @xmath23 ueda and leggett @xcite obtain the upper bound in a variational calculation .
for the @xmath0li parameters ( [ parameters ] ) , this formula gives @xmath24 . in the numerical calculations
described later , we obtain @xmath25 or @xmath26 for the @xmath0li parameters .
this number becomes slightly larger when gain and loss effects are taken into account .
actually , even ignoring collisional loss , the condensate is only metastable for @xmath27 , for it can decay via quantum tunneling .
this effect is not described by the nlse , although an approximate decay amplitude can be obtained by continuation of the nlse to imaginary time .
kagan et al .
@xcite estimated the amplitude by calculating an overlap integral between initial and final states represented by gaussian wave functions , and obtained @xmath28 , where @xmath29 and @xmath30 are respectively the widths of the initial and final wave functions . through numerical calculations ,
shuryak @xcite found that the decay rate is proportional to @xmath31 $ ] .
stoof @xcite wrote down a wkb formula for the decay rate , but did not explicitly evaluate it . using variational wave functions , ueda and
leggett @xcite obtained a rate proportional to @xmath32.$ ] these estimates indicate that the tunneling probability is negligible unless @xmath33 .
however , the wkb approximation breaks down in this neighborhood , and reliable calculations become difficult . in experimental terms
, it is also difficult to observe tunneling in the present system , because the effect is masked by collisional loss , especially near @xmath34 @xcite . for these reasons
, we shall not calculate the tunneling amplitude in this paper ; but we will give a qualitative discussion of the phenomenon , in as far as it pertains to the instability of the system . ueda and huang
@xcite formulated an approach to the stability problem , including tunneling , in terms of the feynman path integral for a transition amplitude . in a saddle - point approximation to the path integral , one obtains the nlse as the saddle - point condition , whose continuation into imaginary time yields the tunneling amplitude . starting with a gaussian initial wave function
, they assume that it remains gaussian , in order to analyze the problem analytically . in this `` gaussian approximation ''
, it was found that , as soon as @xmath35 @xmath22 , the system begins to collapse locally , in a region at the center of the trap .
the size of this region grows as @xmath1 increases . by putting the nlse in hydrodynamic form
, one can see that the pressure becomes negative inside this `` black hole '' , and , consequently , there would be infinite density fluctuations . in the present work ,
we remove the gaussian approximation by performing numerical calculations .
we verify that the picture presented by ueda and huang is correct . for @xmath35 @xmath22 , a `` black - hole ''
does appear at the center of the trap , within which the density has sharp local maxima , whose heights grow as the grid size for the numerical computation is decreased .
this is consistent with the expectation that the density becomes divergent in the continuum limit .
finally , we take into account gain and loss mechanisms by adding phenomenological terms to the nlse .
we are able to study in some detail the growth - collapse cycles of the condensate , as anticipated in earlier work @xcite .
to give an overview of the mechanisms for instability , we think of the condensate wave function at each spatial point as a `` coordinate , '' which moves in an effective potential , consisting of the last two terms in the hamiltonian ( [ hamiltonian ] ) , plus fluctuation corrections arising from the term @xmath36 .
the system is classically stable when this motion is bounded at all spatial points .
we immediately see that the attractive interaction term @xmath37 can lead to instability , for it decreases without bound when the density increases .
the question is whether there is an energy barrier guarding against the free fall .
a precise definition for the effective potential can be modeled after that for a spatially uniform system in quantum field theory@xcite . for a qualitative description ,
however , it is much simpler to use the `` gaussian approximation''@xmath6introduced by ueda and huang@xcite , where we assume that the wave function is gaussian , given by its initial form @xmath38 where @xmath39 we introduce a width parameter @xmath40 by putting @xmath41 thus , @xmath42 corresponds to the ground - state eigenfunction of the harmonic oscillator . with this
, we have @xmath43 where @xmath44\ ] ] and in this approximation
the effective potential is @xmath45 |\psi |^{2}-\frac{u_{0}}{2}% for different @xmath46 .
there is an energy barrier with height @xmath47 ^{2}\ ] ] which is nonzero at @xmath48 , because @xmath49 . at large @xmath50
the effective potential tends to @xmath51 . in reality , of course , the nlse ceases to be valid somewhere along the drop , for other physical effects , such as solidification , come in .
a typical initial wave function @xmath52 is shown on the left panel of fig .
[ aekhfig1 ] . at a given @xmath46
, we can measure the wave function on the left panel , and transfer it to the horizontal axis on the right panel .
if it lands to the left of the barrier maximum for the particular @xmath53 then the system at that point is classically stable , but can decay via quantum tunneling as indicated by the classically forbidden path @xmath54 .
if it lands to the right , the system at that @xmath46 will rapidly collapse to a state of high density . since @xmath52 has a maximum at @xmath48 with value proportional to @xmath55 , the system may only decay via tunneling at any @xmath46 , if @xmath6it is classically stable at @xmath48 .
otherwise , the system in a region about @xmath48 will rapidly collapse .
the condition for stability against collapse is therefore @xmath56 which corresponds to @xmath27 , with @xmath57 the time for the onset of local collapse is expected to be greater than the time needed to establish a quasi - stationary initial wave function .
thus , @xmath40 should correspond to the best gaussian approximation to that wave function .
this expectation is verified by numerical calculations described below , which also show that the gaussian approximation is very good , up to the onset of collapse .
of course , @xmath58 ceases to have meaning once the collapse begins , for @xmath3 .
the numerical calculations for @xmath59 , which is just below @xmath22 , give @xmath60 as mentioned earlier , we will not calculate the tunneling amplitude in this paper , but concentrate on numerical solutions of the nlse in real time .
we seek spherically symmetric solutions to the nlse ( [ nlse ] ) , for the parameters ( [ parameters ] ) for the @xmath0li experiments . for convenience we put @xmath61 with the boundary condition @xmath62 , and the normalization condition @xmath63 from now on , we shall measure distance in units of @xmath16 , and time in units of @xmath64 .
to do this without introducing new symbols for dimensionless distance and time , we formally put @xmath65 .
the nlse ( [ nlse ] ) then takes the form @xmath66 where @xmath67 the initial condition with gaussian form is @xmath68 where @xmath69 we calculate @xmath70 numerically as a complex function , using an implicit difference equation algorithm due to goldberg et al.@xcite .
to find out how various initial states relax to a stationary state , we first solve the equation with an initial wave function uniform within a certain radius , and zero outside .
for @xmath71 , which is somewhat below the critical number , the time development is shown in fig .
[ aekhfig2](a ) .
we see that the modulus @xmath72 appears to be gaussian - like , except for small local fluctuations , after only about one - hundredth of an oscillator cycle . in fig .
[ aekhfig2](b ) , we start with a gaussian with @xmath42 , with the same @xmath1 .
this corresponds to the stationary wave function in the absence of interatomic interaction . in a few oscillator cycles , the width narrows to a stationary value .
if we initially choose @xmath40 near the stationary value , the location of the peak of @xmath72 will slowly oscillate about some mean position .
[ aekhfig2](c ) shows the case when the ratio of the standard deviation to the mean of the peak location is minimized .
this corresponds to @xmath73 , for this choice of @xmath1 .
the wave function in this case is sensibly stationary , and the gaussian approximation is good . in fig .
[ aekhfig3 ] , we plot the ratio of the standard deviation to the mean of the peak location for @xmath59 ( just below critical ) against @xmath58 , and exhibit the minimum at @xmath74 , as quoted earlier in ( [ alpha ] ) . for a quantitative measure of
how good the gaussian approximation is , we calculate the absolute value of the overlap between @xmath75 and @xmath76 and find that it lies between 1 and 0.9795 , with an average of 0.9906 , for @xmath12 between 0 and 50 ( @xmath12 in units of 2@xmath77 ) .
to find @xmath22 , we examine the time development for different @xmath1 , to look for the onset of local collapse .
the signature of the collapse is the occurrence of a minimum in the wave function within a small distance from the origin as shown in fig .
[ aekhfig4 ] for @xmath74 and @xmath78 .
the initial wave function is stationary for a while , but begins to narrow after @xmath79 ( fig .
[ aekhfig4](a ) ) , and a dip occurs at @xmath80 ( fig .
[ aekhfig4](b ) ) , at @xmath81 , which describes the implosion of a shell of particles at that radius .
the first local minimum moves inwards with time and it eventually gets very close to @xmath48 , at @xmath82 , as shown in fig .
[ aekhfig4](c ) .
the density in the inner region can grow only by taking particles from the outside , since the number of particles is conserved .
an examination of the phase of the wave function , whose gradient gives the superfluid velocity , shows that the superfluid velocity has a peak at the dip , and is directed inward , as we expect from the continuity equation .
thus , qualitatively , the onset of collapse happens on a time scale an order of magnitude greater than that for the establishment of the initial wave function , and it is initiated by a sudden inward rush of particles from a finite distance from the center .
this signals the formation of a `` black hole '' , as suggested by ueda and huang @xcite . after the onset of collapse , the density in the black hole increases rapidly , as indicated in fig .
[ aekhfig4](c ) .
the increase is fueled by particles drawn from throughout the outer region , but the density outside does not decrease uniformly . instead , there are ripples of small implosions this is illustrated in the 3d plot of fig .
[ aekhfig5 ] , with @xmath72 plotted above the @xmath46-@xmath83 plane
. the black hole formation is indicated by the sudden rise of @xmath72 near the origin .
the terrain is smooth before the rise , but very rough after that .
the uneven bumps in the terrain are of course deterministic , but they are very sensitive to variations in the initial wave function , and in this sense they are random ( see fig . [ aekhfig6 ] ) .
the fact that the height of the plateau seems to remain constant in time is an artifact of the finite spatial grid size .
the height appears to increase without bound in the continuum limit , as illustrated in fig .
[ aekhfig8 ] . to determine @xmath22 more precisely , we plot the occurrence time of the first local minimum in the wave function within distance @xmath84 from the origin , as a function of @xmath85 .
this is done in fig .
[ aekhfig7 ] , for a gaussian initial wave function with @xmath86 , and for an optimal gaussian initial wave function with @xmath74 .
the results are as follows : @xmath87 which conforms to our expectation that the numbers are not very sensitive to initial conditions , because the time for the onset of collapse is long compared to the formation time of a quasi - stationary wave function . in fig .
[ aekhfig8 ] we show the dependence of some of our results on the spatial grid size .
the time for the onset of collapse appears to tend to a finite limit when the grid size approaches zero , but the height of a peak in the collapsed region seems to grow without bound , faster than ( grid size)@xmath88 .
this is consistent with the expectation that density fluctuations become divergent in the continuum limit .
we have ignored gain - loss mechanisms so far , because we wanted to understand various effects one at a time .
now we shall take them into account , and make the problem more realistic . in the experimental situation , the condensate in the trap can exchange atoms with an uncondensed `` thermal cloud '' , which contains far more atoms than the condensate .
equilibrium is reached when the chemical potentials equalize .
the equilibrium fraction of condensate to `` cloud '' atoms is determined by the temperature .
the kinetic equations governing the approach to equilibrium are quite intricate , and a subject of ongoing research @xcite .
we shall use a phenomenological gain equation based on a fit to data @xcite : @xmath89\ ] ] where @xmath90 is the equilibrium number of atoms in the condensate , with @xmath91 where @xmath92 is the elastic scattering rate , which is approximately equal to 1 s@xmath88 for the conditions of the experiment of ref.@xcite and @xmath93 .
@xmath94 is a phenomenological factor to reconcile the above theoretical prediction for @xmath95 with the experimental results for the repulsive case @xcite . in our runs
we used @xmath96 and @xmath97 nk , while @xmath98 was taken to be equal to the critical value of @xmath99 , i.e. about @xmath100 . assuming @xmath101 , we ignored the second term in the bracket .
the main loss mechanisms are two- and three - body collisions@xcite described by the following equation : @xmath102 where @xmath103 is the particle density in the condensate . adding the gain and loss terms , we have an extended nlse , which reads , in units such that @xmath104 , @xmath105 where @xmath106 in ref.@xcite , the value of @xmath107 is taken to be of order @xmath108 , which differs considerably from ours . with the new equation , we find that the initial conditions become quite irrelevant .
the number of particles quickly settle into a steady - state oscillation whose peak value defines the critical particle number in the new context .
this is shown in fig.9 , with : @xmath109 the mean number of atoms is approximately 735 , and the period of the oscillation is @xmath110 the long time scale of the oscillations explains why they are insensitive to initial conditions .
we have extended the calculations to the longest evolution time practicable , and found that , with a constant supply of particles , @xmath1 oscillates in a steady state for at least 1.44@xmath111 trap cycles , or 100 seconds .
these results are comparable with those obtained by hulet et al .
@xcite , except that , in contrast to our case , @xmath1 collapses to zero in that calculation .
this work is support in part by a doe cooperative agreement de - fc02 - 94er40818 .
m. h. anderson , j. r. ensher , m. r. mathews , c. e. wieman , and e. a. cornell , _ science _ * 269 * , 198 ( 1995 ) .
k. b. davis , m .- o .
mewes , m. r. andrews , n. j. van druten , d. s. durfee , d. m. kurn , and w. ketterle , _ phys .
rev . lett_. * 75 * , 3969 ( 1995 ) .
d. g. fried , t. c. killan , l. willmann , d. landhuis , s. c. moss , d. kleppner , and t. j. greytak , _ phys .
rev . lett_. * 81 * , 3811 ( 1998 ) .
c. c. bradley , c. a. sackett , j. j. tollett , and r. g. hulet , _ phys .
rev . lett_. * 75 * , 1687 ( 1995 ) ; c. c. bradley , c. a. sackett , and r. g. hulet , _ phys .
rev . lett_. * 78 * , 985 ( 1997 ) .
v. e. zakharov and v. s. synakh , _ zh . eksp .
fiz_. * 68 * , 940 ( 1975 ) .
* 41 * , 465 ( 1976 ) . ]
g. baym and c. j. pethick , _ phys .
lett . _ * 76 * , 6 ( 1996 ) .
p. a. ruprecht , m. j. holland , k. burnett , and m. edwards , phys .
rev . * a 51 * , 4704 ( 1995 ) .
f. dalfovo and s. stringari , _ phys .
rev . _ * a 53 * , 2477 ( 1996 ) .
m. ueda and a. j. leggett , _ phys .
rev . lett_.
* 80 * , 1576 ( 1998 ) .
y. e. kim and a. l. zubarev , _ phys .
lett . _ * a 246 * , 389 ( 1998 ) .
m. wadati and t. tsurumi , _ phys
. lett . _ * a 247 * , 287 ( 1998 ) .
kagan , g. shlyapnikov , and j. walraven , _ phys .
lett . _ * 76 * , 2670 ( 1996 ) .
e. shuryak , _ phys .
_ * a 54 * , 3151 ( 1996 ) .
h. t. c. stoof ,
_ j. stat . phys_. * 87 * , 1353 ( 1997 ) .
d. m. stamper - kurn , h .- j .
miesner , a. p. chikkatur , s. inouye , j. stenger , and w. ketterle , _ phys .
lett_. * 83 * , 661 ( 1999 ) reported observation of tunneling of two different condensates through each other .
m. ueda and k. huang , _ phys .
_ * a * ( in press ) ; cond - mat/9807359 .
c. a. sackett , h. t. c. stoof , and r. g. hulet , _ phys .
rev . lett_.
* 80 * , 2031 ( 1998 ) .
kagan , a. e. muryshev , and g. v. shlyapnikov , cond - mat/9801168 ( 16 jan 1998 ) .
see , for example , k.huang , _ quantum field theory , ( _ wiley , new york , 1998 ) , sec.15.7 .
a. goldberg , h. m. schey , and j. l. schwartz , _ am .
j. phys . _
* 35 * , 177 ( 1967 ) .
t. r. kirkpatric and j. r. dorfman , _ phys .
rev . _ * a 28 * , 2576 ( 1985 ) ; d. w. snoke and j. p. wolfe , _ phys .
rev . _ * b 39 * , 4030 ( 1989 ) ; d. v. semikoz and i. i. tkachev , _ phys .
rev . lett_. * 74 * , 3093 , ( 1995 ) ; c. w. gardiner , m. d. lee , r. j. ballagh , m. j. davis , and p. zoller , cond - mat/9806295 ( 24 jun 1998 ) ; t. nikuni , e. zaremba , and a. griffin , cond - mat/9812320 ( 18 dec 1998 ) .
miesner , d. m. stamper - kurn , m. r. andrews , d. s. durfee , s. inouye , w. ketterle , _ science _ , * 279 * , 1005 , ( 1998 ) .
r. j. dodd , m. edwards , c. j. williams , c. w. clark , m. j. holland , p. a. ruprecht , and k. burnett , _ phys .
rev_. * a 54 * , 661 ( 1996 ) .
|
we study the stability of a bose - einstein condensate of harmonically trapped atoms with negative scattering length , specifically @xmath0li .
our method is to solve the time - dependent nonlinear schrdinger equation numerically . for an isolated condensate , with no gain or loss , we find that the system is stable ( apart from quantum tunneling ) if the particle number @xmath1 is less than a critical number @xmath2 for @xmath3 , the system collapses to high - density clumps in a region near the center of the trap .
the time for the onset of collapse is on the order of 1 trap period . within numerical uncertainty ,
the results are consistent with the formation of a `` black hole '' of infinite density fluctuations , as predicted by ueda and huang @xcite .
we obtain numerically @xmath4 .
we then include gain - loss mechanisms , i.e. , the gain of atoms from a surrounding `` thermal cloud '' , and the loss due to two- and three - body collisions .
the number @xmath1 now oscillates in a steady state , with a period of about 145 trap periods .
we obtain @xmath5 as the maximum value in the oscillations
. boxedeps [ theorem]acknowledgement [ theorem]algorithm [ theorem]axiom [ theorem]claim [ theorem]conclusion [ theorem]condition [ theorem]conjecture [ theorem]corollary [ theorem]criterion [ theorem]definition [ theorem]example [ theorem]exercise [ theorem]lemma [ theorem]notation [ theorem]problem [ theorem]proposition [ theorem]remark [ theorem]solution [ theorem]summary
|
experiments using trapped ions often provide a superior system for making precision measurements in applications such as determining atomic structure parameters[[sherman ] ] , microwave or optimal atomic clocks[[warrington],[itano ] ] , searches for time variations of fundamental constants[[itano],[peik ] ] , or making precision measurements of atomic parity violation[[fortson - ionpnc],[schacht],[koerber ] ] .
the precision possible in such experiments is fundamentally due to the intrinsic properties of a trapped ion system through its relatively clean and isolated environment , but is finally determined as it is for all experiments by systematics and sensitivity .
systematic issues are commonly analyzed extensively , but sensitivity has not yet been considered in the detail common to many more traditional precision atomic measurements such as atomic parity violation[[vetter],[wood ] ] . a careful consideration of the choice of certain experimental parameters can yield important improvements in sensitivity for many kinds of experiments and a corresponding improvement in the precision possible .
trapped ion experiments typically consist of many trials of some pump , interaction , probe sequence .
the pump stage prepares the ion in a particular state . during the interaction stage
the ion is either actively driven by a set of applied fields , or just allowed to evolve in isolation , in both cases for the purposes of studying the dynamics of particular interest .
for example , the ion may be driven by another laser beam or a radio - frequency field to characterize a resonant coupling to another state , or allowed to evolve in the absence of all light in order to measure an excited state lifetime .
the probe stage detects whether a transition out of the initially prepared state occurred during this interaction period . by measuring the probe signal as a function of drive parameters such as drive rate , time or frequency , or wait time
, some quantity of interest relating to the dynamics of the interaction can be determined .
the probe is usually implemented using some variation of the electron shelving method[[dehmelt - shelving ] ] . with high cooling rates , and well isolated shelved states with sufficiently long lifetimes ,
state detection using shelving can be done with almost perfect efficiency for almost any combination of probe parameters .
the results of many trials can be interpreted directly as a transition probability . under less ideal circumstances the probe signal requires additional information to interpret , and in such cases even limited sensitivity for state detection may require parameters like probe time to be chosen from a particular , narrow range of values .
the sensitivity can be maximized by using specific optimal values of these parameters .
the improvements in sensitivity when using optimal values may be only a few@xmath210% , or a factor of 2 - 10 and more depending on the starting point .
determining these optimal values is straight - forward and for precision experiments that are primarily limited by statistics it is usually well worth the effort .
shelving makes use of a set of cycling transitions , the probe cycle , that includes at least two states , and another state that is not part of this cycle , the shelved state .
these cycling transitions and states are usually related to those that are used to doppler cool the ions .
ideally the shelved state is long - lived and well - isolated from the probe cycle , well isolated meaning that there is negligible rate or probability for transitions between this state and any part of the probe cycle when the beams that drive the cycle are applied . a generalized level diagram of this kind of system is shown in figure [ level - diagram ] .
this kind of level diagram is qualitatively similar to that of most alkali - like ions .
the @xmath3 are typically part of a @xmath4 cooling scheme .
@xmath5 may be a zeeman sublevel uncoupled by angular momentum selection rules by using a particular polarization , a hyperfine manifold not part of the nominal cooling cycle , or an entirely separate long - lived state .
for example , in singly - charged ytterbium ions , the doppler cooling cycle comprises the 370 nm transition from the @xmath6s@xmath7 ground state to the @xmath6p@xmath7 state with leakage to @xmath6d@xmath8 state ( corresponding to state c3 in figure [ level - diagram ] ) necessitating a repump laser beam at 935 nm .
the shelved state , s , in this instance can be a hyperfine sublevel of the @xmath6d@xmath8 state in the odd isotopes of yb or the long - lived @xmath6f@xmath9 state in any isotope .
an ion in a state that is part of the probe cycle when that cycle is driven will fluoresce at some rate @xmath10 that can be detected by a photo - multiplier tube ( pmt ) .
in addition there will be some background rate @xmath11 due to ambient light or light scattered from the laser beam driving the probe cycle .
this background rate is presumed to be independent of the ion s state , so the pmt signal for an ion in the cycle corresponds to a total rate @xmath12 . to shelve
the ion is to drive it to the shelved state .
an ion in the shelved state does not fluoresce when the probe cycle is driven , so the pmt signal will correspond to the background rate .
the probe cycle then provides a probe of the ion s state . to determine if an ion is in its shelved state the probe cycling transitions are driven for some probe time @xmath13 .
the resulting pmt counts collected , @xmath14 , will depend on the initial state of the ion .
if the cycling and background rates are constant during the course of the probe , the average number of photons collected will be @xmath15 for an ion that is not shelved and @xmath16 for a ion that is shelved .
if @xmath14 is large enough , uncertainties from poisson counting statistics will be small , the result of a single probe will give @xmath14 very close to either @xmath17 or @xmath18 , and the initial state of the ion can be determined with negligible uncertainty .
an ion was very likely in the shelved state if the probe results in @xmath19 , for some threshold @xmath20 , and likely not shelved if @xmath21 . with good counting statistics
@xmath20 can easily be chosen as almost any value between @xmath17 and @xmath18 and the particular value used is not important . as counting statistics become poor interpretation of the probe count @xmath14 becomes less straight - forward so that measurement precision will be affected by the choice of the threshold and an optimal value can be determined . in practice
, the complete probe stage may begin with a set of transitions that drive the ion to the shelved state from a particular state involved in the interaction stage .
the probability that the ion was in the state of interest at the beginning of the probe stage then corresponds directly to the probability that an ion is shelved after this probe preparation step . in practice the procedure that conditionally shelves the ion may not be perfectly efficient . generally the probabilities will be linearly related , but the details depend on the particular measurement and vary considerably . here only determining the resulting shelving probability , @xmath22 , will be considered . by repeating the probe step many times for an identically prepared ion , @xmath22
is given directly by the fraction of trials , @xmath23 , that resulted in @xmath24 giving @xmath25 where @xmath26 is the total number of trials .
the uncertainty in determining @xmath22 , @xmath27 , is then given simply by binomial statistics , @xmath28 the binomial distribution and its properties relevant to this material are reviewed in the appendix [ [ section - appendix ] ] . in some systems
a short shelved - state lifetime or losses from the cycling states during the probe mean that the cycling and background rates are not constant and this can limit the time during which the probe is effectively able to distinguish the initial state .
this will be considered in detail below , but for now note that if the cycling rate is too low relative to the background rate , the resulting @xmath14 may be too small to allow the results of a single probe to be used to infer the initial state of the ion .
figure [ figure - threshold - detection ] illustrates this by comparing the probability to collect some number of photons per probe time for both shelved and unshelved ions for various @xmath17 .
it is assumed that the cycling and background rates are constant and that the counts are given by a poisson distribution @xmath29 with @xmath30 for @xmath31 or @xmath32 which correspond to a shelved or unshelved ion , respectively . to easily compare
the various cases the amplitudes are scaled by @xmath33 , as the height of the peaks roughly decrease by this factor , and @xmath34 is plotted as a function of @xmath35 with @xmath36 fixed so that the peaks always appear in the same position .
this strictly corresponds to changing @xmath17 by varying the probe time @xmath13 , but the same qualitative behavior follows from changing @xmath17 by varying @xmath37 instead through the cycling rate @xmath10 for a fixed @xmath13 which would give a constant @xmath18 . with this
the functions explicitly plotted are @xmath38 and @xmath39 as a function of @xmath40 for selected @xmath41 .
these functions should strictly be defined only for integer values of @xmath42 but to more easily compare the behavior for different @xmath41 they are plotted for continuous @xmath40 as are subsequent similar plots . for large @xmath17 the distributions
are clearly distinguished and a wide range of thresholds could be used as a cutoff to reliably define the difference between shelved and unshelved ions in a single probe trial .
as @xmath17 is reduced the distributions get wider and the range of possible threshold values is reduced . for particularly small @xmath17
the distributions overlap considerably and there is no threshold that can be used to unambiguously determine the ion s state .
coarsely , the width of each distribution is given by @xmath43 so the total region occupied by the non - negligible parts of the distributions is roughly @xmath44 .
when this becomes comparable to the distance between the peaks , @xmath45 , the distributions start to overlap .
a threshold can then be defined only for @xmath46 which requires sufficiently large @xmath17 and relatively small @xmath18 .
@xmath47for @xmath48 probing an unshelved ion may yield a count less than any chosen threshold @xmath20 , in a sense a false negative , with non - negligible probability @xmath49 with @xmath50 and @xmath51 the complete and incomplete gamma functions , respectively [ [ arfkin ] ] .
there is a corresponding false positive probability to measure @xmath21 for a shelved ion @xmath52 as well as the correctly interpreted complements @xmath53 and @xmath54 .
the probability of detecting an ion to be shelved , @xmath55 will be given by the probability that the ion is actually shelved , @xmath22 , times the probability it is detected as shelved , @xmath56 , plus the probability that it is not shelved , @xmath57 , times the probability it is nevertheless detected as shelved , @xmath58 @xmath59 the corresponding @xmath60 .
@xmath61 is no longer given exactly by @xmath22 except in the limit of @xmath62 .
for all other values @xmath61 is linearly related to @xmath22 .
if @xmath22 must be determined directly from @xmath63 , then the @xmath64 must be known as well and may need to be determined experimentally .
binomial statistics now give the uncertainty in determining @xmath65 from @xmath26 trials @xmath66 simple error propagation using @xmath67 and @xmath68 gives an estimate for the uncertainty in determining @xmath22 , @xmath69 consider the ratio , @xmath70 , of @xmath27 for the binomial distribution to @xmath27 for this non - ideal case @xmath71 so that @xmath72 for @xmath73 , giving @xmath74 and @xmath75 , the limit of perfect detection of the ion s state , @xmath76 .
as @xmath58 and @xmath77 are increased , @xmath56 is always larger than @xmath58 since a shelved ion is always more likely to be detected as shelved than an unshelved ion , and @xmath78 $ ] so @xmath70 is always non - negative
. it also turns out that @xmath70 monotonically decreases with @xmath58 and @xmath77 so that @xmath70 can effectively be regarded as a detection efficiency .
when the ion s shelved state can be determined perfectly @xmath79 .
as detection becomes less reliable @xmath70 decreases until @xmath80 , at which point a probe gives no information about an ion s shelved state , @xmath81 and @xmath27 diverges .
@xmath70 depends on the threshold chosen , @xmath82 , and some choices of @xmath82 yield better sensitivity of @xmath61 to @xmath22 than others .
@xmath70 is also generally a function of @xmath22 , but to get a picture of its behavior take @xmath83 @xmath84 to first order in @xmath58 and @xmath77 this gives , @xmath85 , which is what might have been chosen intuitively as a qualitative measure of detection efficiency .
figure [ figure - threshold - detection - efficiency ] shows @xmath86 and its first order approximation as a function of @xmath87 for the parameters corresponding to figure [ figure - threshold - detection ] . for large @xmath17 many values for @xmath40
give @xmath88 , almost perfect detection efficiency .
as @xmath17 decreases this range is reduced , consistent with the qualitative observation that the distributions begin to overlap . at some point
no threshold allows perfectly reliable detection , though @xmath70 indicates that statistically using a threshold detection scheme still gives good sensitivity for determining the shelved state probability if the threshold is chosen to be the optimal value that maximizes @xmath70 , even though the distributions might suggest more significant ambiguity .
this optimal threshold value depends on details of the system including cycling and background rates , and probe time , through @xmath41 and @xmath89 , and the expected shelved state probability , @xmath22 , so finding a general result is not trivial , nor is it particularly illuminating .
however , it is easily determined numerically for any particular set of parameters .
figure [ optimal - threshold ] shows the optimal threshold , determined as the value of @xmath82 that maximizes the detection efficiency , @xmath70 , and the resulting maximal @xmath70 as a function of @xmath22 for the same @xmath90 and @xmath91 as used before .
a single threshold is not optimal for all @xmath22 , but where a range of @xmath22 is anticipated a fixed @xmath20 can be used that approximately optimizes detection over that range .
the dashed curves in figure [ optimal - threshold ] show the detection efficiency that results from using a fixed @xmath92 . for this fixed threshold detection efficiency
remains close to its maximum value over a range at least as large as @xmath93 $ ] .
this analysis might be generalized to a system using more than 1 ion where an array of thresholds could be used to determine for a particular trial how many of the ions were shelved .
the probe count distribution will now be a series of equally spaced peaks , as in figure [ figure - count - distribution - nions ] , corresponding to discrete multiples of the cycling rate that get broader with larger total counts .
as all the peaks get broader with decreasing @xmath94 the peaks corresponding to fewer shelved ions , with resulting relatively high probe counts , will begin to overlap first , while the peaks corresponding to many shelved ions , and few probe counts , may still be well separated . in this case
there would be an array of thresholds and results @xmath95 for the number of times @xmath96 ions are detected as shelved .
some kind of detection efficiency could be calculated again in terms of @xmath97 for @xmath22 as determined from the @xmath95 and with this measure it would then be possible to calculate an optimal set of thresholds @xmath98 .
but such a result is less straight - forward and less general , and this kind of measurement can also result in systematic shifts in the derived @xmath22 in addition to degraded sensitivity for poorly chosen @xmath98 .
instead such a system , as well as the original single ion case , is more easily analyzed using the multi - modal counting statistics considered presently .
for systems with poor probe counting statistics a threshold detection scheme may not provide sufficient information to confidently determine the state of a single ion in one trial or how many ions were shelved if using more than one ion .
instead consider using all the information in the full distribution of probe counts when determining the shelving probability @xmath22 .
take @xmath99 to be the probability of collecting @xmath14 photons during a single probe of an ion that has probability @xmath22 to be in the shelved state . for background , cycling and total rates @xmath11
, @xmath10 and @xmath100 , respectively , and a particular probe time @xmath13 giving @xmath30 , @xmath99 will be the probability that the ion is shelved , @xmath22 , times the probability of a shelved ion giving @xmath14 counts , plus the probability that the ion is not shelved , @xmath101 , times the probability of an unshelved ion giving @xmath14 counts .
if the counts for each pure state probe are given by a poisson distribution , @xmath99 will be @xmath102 @xmath103 is the usual poisson distribution as before .
familiar properties of the poisson distribution are reviewed in the appendix [ [ section - appendix ] ] .
figure [ figure - bimodal - distribution ] shows an example of @xmath99 as a function of @xmath14 , again using continuous values of @xmath14 for clarity , for various @xmath22 in the case of relatively poor counting statistics . again
the lack of a clear distinction between shelved and un - shelved ions is apparent .
denote an expectation value that explicitly indicates the distribution @xmath104 so that the expectation values of the composite distributions being considered can be calculated more compactly .
this allows for easily checking the normalization @xmath105 and determining the mean , @xmath106 derivations not shown here for these and following statistical results are also detailed in the appendix [ [ section - appendix ] ] . as might be expected @xmath107 is linearly related to @xmath22 , allowing @xmath107 to be easily used to measure the dependence of @xmath22 on drive parameters .
though as with the threshold probe , if the actual value of @xmath22 is needed , two other parameters must be known , in this case @xmath17 and @xmath94 . the variance requires a bit more work . with some algebra
it can be shown to be [ [ section - appendix ] ] @xmath108 the previous result for @xmath109 may be substituted into this result , but in this form it is easy to see that @xmath110 reduces to @xmath107 as in the poisson distribution only for the trivial cases of @xmath111 . in general
there is an extra contribution as large as @xmath112 which dominates as @xmath94 increases .
this is a consequence of the qualitative shape the distribution .
increased statistics will give increasingly narrow peaks , but the positions of the peaks do nt move and the variance eventually just reflects that geometry . further insight into this is gained by considering the full distribution for @xmath113 trials .
first consider the probability to measure a total of @xmath14 counts after @xmath114 trials , which will be denoted as @xmath115 .
this can be written as a composition of two single trial distributions .
a total of @xmath14 counts will be observed if the first trial results in @xmath96 counts , and the second in @xmath116 counts for any @xmath117 . since the trials are independent , the probability of a particular combination of two trials is the product of the probabilities of each single trial .
the resulting probability to observe @xmath14 counts after two trials is the sum of the probabilities of all these possible combinations that result in @xmath14 total counts , @xmath118 poisson distributions satisfy the composition property [ [ section - appendix ] ] @xmath119 allowing @xmath120 to be written as @xmath121 if the form for arbitrary @xmath26 is not already apparent , a further iteration to calculate @xmath122 again as detailed in the appendix , immediately suggests @xmath123 which can be verified by induction [ [ section - appendix ] ] .
here @xmath124 is the familiar binomial distribution .
computing expectation values of @xmath125 is straight - forward though tedious and yields [ [ section - appendix ] ] @xmath126 these @xmath113 results are just @xmath26 times the @xmath127 results as should be expected for @xmath26 independent trials . here
@xmath128 is the uncertainty in the determination of @xmath129 after @xmath26 trials .
the origin of the non - poisson contribution to @xmath110 is suggested by the large @xmath26 limit of @xmath130 . as @xmath26 and the argument of @xmath131 increases , the @xmath34 in each term in the sum in @xmath132 more closely resembles a @xmath133-function centered at @xmath134 so that @xmath135 .
the distribution is non - zero only for @xmath14 giving @xmath136 an integer , so we make @xmath96 the independent index giving @xmath137 reducing exactly to the binomial distribution considered for threshold detection . the variance can be computed directly [ [ section - appendix ] ] @xmath138 this is exactly the extra contribution to @xmath139 just directly determined . the general bimodal distribution then exactly accounts for both the poisson and binomial character of the probe .
figure [ figure - variance - comparison - s ] shows @xmath140 and @xmath141 as a function of @xmath22 for various @xmath94 .
the deviation from poisson statistics becomes significant for large @xmath94 .
this extra contribution has not been considered in previous experiments that use shelving in trapped ions and is usually missed if the variance is not determined from the data but instead assumed to be given by @xmath107 .
this will result in underestimating the errors of derived quantities and will result in fits being improperly weighted .
assuming poisson statistics will also result in choosing the wrong optimal values for the probe time as considered below .
for more than one ion this disparity may be reduced .
the count distribution for @xmath142 ions can be determined in the same way as for @xmath113 trials .
let @xmath143 now be the probability to measure @xmath14 counts from a single trial probing @xmath144 ions , each in the shelved state with independent probability @xmath22 .
composition and induction now give [ [ section - appendix ] ] @xmath145 where @xmath144 is temporarily taken to be @xmath26 for a more concise development .
expectation values are computed in the same way [ [ section - appendix ] ] @xmath146 if @xmath147 and @xmath148 are exactly related as indicated then there is no qualitatively different behavior since @xmath149 and @xmath150 reduces to the multi - trial result .
the non - poisson contribution to @xmath151 is enhanced as @xmath152 increases . but
both tuning and input alignment can be adjusted to maintain a particular lower cooling rate .
also , as the number of ions increases the contribution to the cooling signal from each additional ion may not be the same as ions towards the edges of a cloud may not be cooled as well , exhibit more micromotion , or are less well coupled to the detection system .
this latter complication should strictly require that the preceding analysis is done using some distribution of average cooling counts , but that will be disregarded for now . in either case the result is that @xmath26 and @xmath148 can be taken to be effectively independent and then in the case of large @xmath26 with fixed @xmath153 , @xmath154 reduces to @xmath155 as for a poisson distribution .
the large @xmath144 limit of the complete distribution may be computed as well . in this case , with or without fixed @xmath148 , the binomial coefficient of the terms approaches a @xmath133-function centered at its mean @xmath156 , and the sum then again reduces to one term @xmath157 exactly a poisson distribution for a mean count corresponding to an average of @xmath158 shelved ions , further justifying the behavior of the variance .
figure [ figure - multimodal - distribution ] shows @xmath125 for various @xmath26 .
the convergence to a poisson distribution for large @xmath26 is apparent .
the quantity of interest is the shelved state probability , @xmath22 , rather than the extrinsic probe count itself .
compare the precision of determining @xmath22 for a single ion using either the mean @xmath129 giving @xmath159 as just determined in section [ section - sigma - bimodal ] , to that using a threshold and @xmath160 having @xmath161 from section [ section - sigma - threshold ] .
figure [ figure - variance - comparison - n ] shows the ratio @xmath162 as a function of @xmath17 with various @xmath18 for @xmath83 . the threshold , @xmath20 is taken to be @xmath163 which is close to the optimal value for most values of @xmath17 . for very large @xmath94
there is no advantage in one method over the other for any @xmath89 as @xmath164 .
as @xmath94 decreases , the threshold method clearly yields a higher precision for @xmath94 above a particular value with an eventual transition to the opposite case as @xmath94 decreases further . the value of @xmath94 at which this transition occurs depends upon the absolute value of @xmath18 .
for very small @xmath94 , as shown in section [ section - sigma - threshold ] , it is not possible to unambiguously assign a threshold value , and one is better off using the mean , @xmath109 , as is clearly indicated in the figure .
note also that as the background increases , a threshold probe becomes less of an advantage , and only for larger @xmath147 .
these results for the variance favor using a threshold probe and @xmath165 in systems with good counting statistics .
threshold detection is also initially more robust to variations of the count rates .
for relatively large @xmath94 the peaks of the probe count distribution are well separated .
qualitatively , modest variations of the @xmath166 will not result in the peaks moving far enough that they begin to overlap whatever threshold @xmath20 is chosen .
an ion will be detected as shelved exactly as reliably in all cases .
more formally the @xmath64 resemble step functions as a function of @xmath82 and are very insensitive to changes in @xmath166 for @xmath166 large and @xmath167 .
then @xmath65 as given by @xmath168 is stable , that is it does nt change significantly as @xmath147 or @xmath89 are varied , or @xmath169 is small . as a result @xmath170 when determined from @xmath65 is also stable , @xmath171 is small .
conversely @xmath172 varies directly as the @xmath166 vary so that for @xmath173 , @xmath174 must increase due to @xmath175 from @xmath171 .
as @xmath166 decreases , the @xmath64 will start to become sensitive to variations of @xmath166 . at that point @xmath176
will also start to include contributions from these variations .
as @xmath166 decrease further @xmath165 actually becomes more sensitive to variations than @xmath177 .
figure [ figure - stability - comparison ] shows @xmath178 and @xmath179 as a function of @xmath17 for @xmath180 , @xmath83 .
@xmath177 becomes more robust to variations of @xmath17 than @xmath165 around the same point that it also becomes more precise .
the precision and stability of determining @xmath22 are always improved for larger @xmath181 whether using a threshold or mean count probe .
@xmath10 may be limited by intrinsic constraints such as the saturation rates of any of the cycling transition or practical constraints such as laser power or pmt collection efficiency .
@xmath13 may also be effectively limited by properties of the system .
a shelved ion may be able to make a transition to some state that is part of the probe cycle through natural decay , off - resonant couplings driven by the probe laser , or parasitic transitions driven by non - idealities of the probe laser such as impure polarization .
if this can happen in a time short compared to the probe time , the ion may make that transition during the probe after which point it will be driven through the probe cycle , and more counts than those coming only from the background rate will be collected .
similarly , the probe lasers may drive some transition into the shelved state so that an initially unshelved ion may become shelved during the probe resulting in fewer counts collected than @xmath182 .
together these couplings can effectively make the probe beams a weak pump .
if the total loss rate from the shelved state due to its natural lifetime and non - ideal couplings can be characterized by a single simple @xmath183 and the parasitic pump rate by @xmath184 , a familiar first order rate equation gives the probability for the ion to be shelved as a function of time to be @xmath185 with @xmath186 the initial probability to be shelved at the beginning of the probe , @xmath187 , and @xmath188 .
the conventional interpretation of shelving transitions as quantum jumps assumes that an ion is either shelved , having 100% probability to be in the shelved state , or unshelved with zero probability of being in the shelved state . in the former case ,
the ion does not fluoresce and the count rate will be given by @xmath189 , while in the latter case the count rate is given by @xmath190 .
the count distribution will therefore be given be the sum of two poisson distributions , corresponding to these two rates , weighted by the fraction of the time that the ion is shelved .
however , in this case @xmath22 is no longer constant , and the weight is instead the time average of @xmath22 : @xmath191 note that the results scale simply with @xmath0 through @xmath192 , and that @xmath193 for @xmath194 , and @xmath195 for @xmath196 .
an ion with time average probability @xmath197 to be in the shelved state is then understood to be shelved for @xmath197 fraction of the time .
this gives the same bimodal distribution as found in section [ section - sigma - bimodal ] with the same rates but with @xmath198 : @xmath199 in this case the interpretation of the distribution is slightly different as previously a single probe would follow a single poisson distribution corresponding to one of two different rates .
the composite bimodal distribution results from many trials where the rate for each trial is chosen to be one or the other of those rates with probability @xmath22 , and the possible results are summed .
now , since an ion can decay during a probe , a single trial will have contributions from distributions corresponding to both rates weighted by @xmath197 so that this composition happens during a single trial .
if this picture of the shelving dynamics is not correct then the observed distribution will be different than that determined here .
for example , if @xmath200 instead gives a constant scattering rate of @xmath201 then , in general , as @xmath22 evolves the distribution would be the time integral of a poisson distribution with mean @xmath202 , which does not generally reduce to any particular simple distribution .
such a distinction may provide a means of determining the degree to which a probe functions as an observation to collapse the ion to either a shelved or unshelved state .
for present purposes the usual quantum jump understanding of the shelving dynamics will be assumed and the corresponding simple modifications to the count distribution will be used . since the distribution is formally the same as that found in section [ section - sigma - bimodal ] , the expectation value and variance will be given by the same replacement of @xmath203 . in particular
@xmath204 which can then be rewritten in terms of @xmath22 as @xmath205 the @xmath206 are not themselves averages but derived parameters , effective renormalized counts as a consequence of @xmath207 varying over the probe time , and the notation indicates their correspondence to @xmath197 .
they change the relation between @xmath129 and @xmath22 but do not appear as actual rates or distribution parameters . for @xmath208 , @xmath193 , so that @xmath209 and @xmath210 thereby recovering the previous results .
for the complementary case of @xmath211 , @xmath195 and @xmath212 so that @xmath213 independent of @xmath22 and no information about the ion s state at the end of the interaction stage can be obtained in this limit .
hence , @xmath214 can be regarded as a probe coherence time . when the probe time exceeds the probe coherence time , sensitivity is reduced .
@xmath215 functions as another measure of detection efficiency in this case relative to the ideal shelved state system .
figure [ figure - probe - profile ] shows the average probe count and the effective rate @xmath216 as a function of @xmath217 using @xmath30 for various @xmath186 .
for @xmath218 the rate becomes insensitive to @xmath22 as just suggested . for long probe times @xmath219 , so that the effective rate @xmath220 goes to the same value for any @xmath186 .
the difference between initial states is only detected during the beginning of the probe during which a probe of an unshelved ion yields some fixed number of extra counts compared to a shelved ion . with longer probe times this difference becomes negligible compared to the total counts , and may become less than the variations due to counting statistics .
though @xmath107 appears most sensitive to @xmath22 for low @xmath13 , @xmath27 diverges as @xmath221 as counting statistics become poor , so that an optimal probe time that minimizes @xmath27 must be some finite , intermediate value .
this behavior , and the resulting optimal probe time can be determined directly from the explicit form of @xmath27 .
the usual procedure for error propagation applied to @xmath129 gives @xmath222 recall @xmath223 and @xmath224 . for @xmath113 trials
the net variance @xmath225 will include a factor of @xmath226 , @xmath227 .
this is more usefully written in terms of the total observation time , @xmath228 , and the time per trial , @xmath229 .
the trial time will depend on the probe time , so let @xmath230 where @xmath231 is the time required to do everything but the probe , such as pump and interaction steps or other probe time independent procedures . generally the probe time will be short compared to @xmath231 so that @xmath232 . in this limit
@xmath233 the general scale of the total sensitivity is determined by @xmath234 , and the precise value is given by the dimensionless @xmath27 .
this factor relating @xmath97 to @xmath235 is independent of the probe time so the functional dependence of @xmath235 on @xmath236 is the same as for @xmath97 and only @xmath97 needs to be considered in determining the significance of @xmath236 for @xmath235 .
consider first the case of relatively poor statistics , small @xmath237 or a large number of ions .
in both cases @xmath238 reduces to that for poisson statistics @xmath239 .
@xmath238 is @xmath240 dependent , but bounded by @xmath15 .
using that bound , writing @xmath13 in terms of @xmath241 , and expanding @xmath242 , @xmath243 , @xmath244 gives @xmath245 recall that @xmath215 is also a function of @xmath40 so that the first factor contains all the @xmath40 dependence of @xmath97 .
figure [ figure - probe - variance - profile - poisson ] shows the quantity @xmath246 as a function of @xmath40 .
this diverges for @xmath247 .
the @xmath194 limit corresponds to the familiar case of poor counting statistics for poisson processes with small @xmath129 . for @xmath196
the divergence corresponds to @xmath248 when the probe count becomes insensitive to @xmath22 as the probe time exceeds the probe coherence time .
the alpha dependent factors take on the minimum value of 1.57 for @xmath249 giving @xmath250 .
note that @xmath27 is fairly flat around its minimum .
further from that minimum it increases significantly for @xmath251 but less dramatically for @xmath252 .
so there is a large penalty for using too short a probe time , but less for using a longer probe time .
generally , taking @xmath13 to be within a factor of 2 of its optimal value gives @xmath27 to within about 10% of its minimum value .
note that for this situation @xmath253 improves quickly and simply as the probe coherence time , @xmath214 , increases so that improving @xmath214 in a particular system , if possible , or using a different system allowing for longer @xmath214 may be worthwhile . for the general case having arbitrary count rates and probe times , using the full form of @xmath238 in terms of @xmath240 , @xmath27 becomes @xmath254 where @xmath10 and @xmath37 are now the cooling and total probe count rates , corresponding to the previously denoted total counts @xmath148 and @xmath255 from the assembly of all the ions .
the @xmath256 term is the binomial statistics contribution , again the limiting value for the case of a threshold probe with perfect detection efficiency .
factoring this out gives @xmath257 which takes the form @xmath258 the actual @xmath27 then appears as a multiple @xmath259 of the binomial limit .
@xmath259 contains the explicitly @xmath40 dependent terms , remembering that @xmath260 .
the term @xmath261 is largely dependent only on system properties such as the count rates and probe coherence time .
the @xmath240 in @xmath261 are also @xmath40-dependent but are bounded : @xmath262\subseteq[0,1]$ ] for @xmath263 $ ] .
so the limiting behavior of @xmath264 with @xmath40 is determined largely by the explicit @xmath40 dependence and again the divergence of @xmath264 and @xmath97 for @xmath247 is apparent still suggesting the existence of an optimal @xmath236 that minimizes @xmath97 .
consider a probe stage for which the probe time is longer than the cycling period , @xmath265 or @xmath266 . then @xmath267 .
if in addition the probe coherence time is much longer than the probe time @xmath268 giving @xmath269 then @xmath270 , @xmath271 and @xmath97 becomes exactly the result for the binomial distribution @xmath272 note that @xmath268 and @xmath273 also implies @xmath274 and @xmath275 .
so for small @xmath261 , due to a sufficiently high cycling rate , or sufficiently long probe coherence time , it becomes possible to pick a probe time such that @xmath276 giving this limiting result which achieves essentially perfect threshold detection efficiency .
once at the point where @xmath277 and @xmath270 , further increasing @xmath40 by increasing the probe time does not improve sensitivity , so there is no point in using a probe time much longer than this minimal value @xmath278 .
this already suggests the characteristic size of the optimal probe time @xmath279 .
the previous complementary case is also simply contained in this result . for @xmath280 , @xmath281 , the @xmath282 in @xmath264 becomes negligible and the binomial like factors cancel in @xmath97 @xmath283 giving the previous result as @xmath197 can be neglected also as @xmath284 .
the quantity @xmath261 , or equivalently @xmath285 , then is a measure of the quality of the counting statistics .
small @xmath261 , allowing @xmath286 , implies good counting statistics and the variance becomes the binomial statistics result .
large @xmath261 corresponds to poor counting statistics , a poisson count distribution and increasing variance for @xmath22 .
in both cases @xmath97 is still dependent on the probe time through @xmath192 , and @xmath22 through @xmath197 and can be optimized by using a particular @xmath236 .
figure [ figure - probe - variance - profile - intermediate ] shows @xmath27 as a function of @xmath40 for various @xmath22 for a case corresponding to moderately good counting statistics .
the probe time which minimizes @xmath27 depends on @xmath22 .
there will usually be enough information that @xmath22 will be approximately known in advance so that an optimal probe time for that particular case could be chosen .
but generally a measurement will include trials with many different drive parameters that result in probing an ion having many different possible values for @xmath22 . in principal a different probe time
could be chosen to optimize each case individually , but there are enough uncertainties about the probe dynamics in an actual system that using a fixed probe time for all cases ensures against creating difficult to handle systematic shifts as well as enormously simplifying data collection and analysis .
so a probe time that approximately maximizes sensitivity for all @xmath22 is more useful .
\a ) b ) note that all the profiles in figure [ figure - probe - variance - profile - intermediate ] appear to be bounded by a qualitatively simple function with a well defined minimum in this case at @xmath287 .
a good approximation to this bound can be obtained easily from the expanded form of @xmath27 by taking @xmath288 in the first term in the root and noting that @xmath289 giving @xmath290 this approximate but hard upper bound is also shown in figure [ figure - probe - variance - profile - intermediate]a . for poor counting statistics @xmath291 as considered before , @xmath292 and @xmath293 , again the result previously obtained for the same limit . for the case of good counting statistics @xmath294
, @xmath295 this becomes @xmath296 which is the upper bound of the result for binomial statistics , and the general @xmath297 again appears as a multiple of this minimal value . for general @xmath298 ,
@xmath299 is minimized for a particular optimal @xmath300 . for the case shown in figure [ figure - probe - variance - profile - intermediate ] this is about @xmath301 or @xmath302 .
this is about the optimal value for the particular case of @xmath200 . for arbitrary @xmath22 ,
@xmath303 does not generally exactly minimize @xmath97 , but does give a variance close to the optimal result .
figure [ figure - probe - variance - profile - intermediate]b shows @xmath304 as a function of @xmath22 for these same system parameters .
it also shows the ratio of the value of @xmath97 that would be obtained by using @xmath300 to the value given if using a fixed @xmath303 . for most of the range of @xmath22 ,
the resulting @xmath97 is within a few percent of its optimal value . for this particular value of @xmath298
then , a probe time of @xmath305 gives a sensitivity close to the maximum value possible for the system , for any @xmath22 .
similar behavior is seen for arbitrary @xmath298 so that in general a probe time chosen to minimize @xmath299 gives an approximately minimal @xmath97 for all @xmath22 .
this optimal value is the one that minimizes @xmath306 this is the same form as @xmath264 with @xmath307 which is now explicitly @xmath22 independent .
figure [ figure - probe - variance - optimal ] shows this @xmath22 independent optimal probe time , @xmath308 and the relative sensitivity @xmath264 as a function of @xmath309 .
note that @xmath40 is bounded and approaches @xmath310 as @xmath311 as was previously determined to be the optimal probe time for the case of poor counting statistics .
\a ) b ) since @xmath304 is bounded it is worth considering the possibility of a universal approximately optimal probe time such as @xmath303 .
figure [ figure - probe - variance - profile - universal ] plots the ratio of the optimal value of @xmath264 , possible if the optimal probe time is used , to the @xmath264 obtained for various fixed @xmath40 .
note that when the probe time is more than about an order of magnitude different from the probe coherence time sensitivity can be very poor . within
that range the sensitivity is of the same order as the optimal value .
the particular value of @xmath312 gives the smallest range of variation over the entire range of @xmath261 where the minimum possible sensitivity is at most about 15% better than the value achieved .
a ) this gives a good universal optimal value .
for almost any system , probe sensitivity is maximized for @xmath313 .
this gives @xmath314 so that @xmath315 choosing this optimal probe time requires knowing the probe coherence time @xmath214 which may be non - trivial to calculate or may need to be determined experimentally .
this provides an @xmath22 independent upper bound for the variance that approximately minimizes @xmath97 as a function of the probe time for all @xmath22 and @xmath298 .
shelving provides a highly sensitive probe of atomic state population probabilities . in systems allowing good counting statistics analysis is simple , intuitively understood and the sensitivity is independent of modest instabilities in probe parameters . for less ideal systems the sensitivity can degrade considerably if these same probe parameters are not carefully chosen and stabilized . for a wide range of real systems , uncertainties in determining @xmath22 with a shelved state probe
are minimized by using a probe time on the order of half the probe coherence time , and deriving @xmath22 from the mean of a set of probe counts .
using this optimal probe time gives a sensitivity that is generally within 15% of the exact optimal value .
the importance of this , and the potential improvements , are especially significant in systems with probe coherence times short compared to the directly detected cycling time .
this is often the case when trying to isolate close hyperfine sublevels for use as shelved states where off - resonant couplings can result in non - negligible spurious transitions , or when trying to isolate individual zeeman levels using angular momentum selection rules where impurities in the polarization of probe beams also significantly limits discrimination between states in practice[[schacht ] ] .
the probe coherence time may be estimated to determine the optimal probe time if enough is known about all the dynamics of the system during the probe . in practice
it will often be simpler to measure the probe coherence time by measuring @xmath316 and @xmath317 as a function of probe time . even in the optimally probed case the resulting probe count distributions are still multi - modal and can have variances significantly different than that expected from poisson statistics .
neglecting such details can result in underestimation of the uncertainty of fit derived parameters or inaccurately weighting points in a fit resulting in less precise fits or systematic shifts in the fits .
the properties stated about the multi - modal distributions follow straight - forwardly from familiar properties of the binomial and poisson distributions .
a binomial distribution is defined by @xmath318 its normalization is easily verified @xmath319 for higher order moments temporarily take @xmath22 and @xmath320 to be independent , and use @xmath321 @xmath322 similarly @xmath323 giving the variance : @xmath324 the poisson distribution is given by : @xmath325 for the normalization @xmath326 for the mean note that @xmath327 @xmath328 similarly @xmath329 giving the variance @xmath330 poisson distributions have a simple composition property , @xmath331 the bimodal distribution was defined by @xmath332 expectation values will take the form @xmath333 where the expectation value again explicitly includes the distribution .
the normalization is easily determined @xmath334 and the mean , @xmath335 for the variance @xmath336 and @xmath337 giving the variance @xmath338 for more than one trial the distribution was stated to be @xmath340 which can be verified by induction .
adding one more trial , the net distribution will be given by a convolution of a single trial distribution with an @xmath26 trial distribution @xmath341 take the binomial coefficient @xmath342 to be zero for @xmath343 and @xmath344 .
the summation limits can then be extended and the terms combined @xmath345 writing the binomial distribution explicitly @xmath346 binomial coefficients have the recurrence relation @xmath347 giving @xmath348 regenerating the original form .
normalization and expectation values can be computed directly from this distribution @xmath349 mean , @xmath350 @xmath351 , @xmath352 expanding the square in the sum @xmath353 giving @xmath354 and the variance as @xmath355 as demonstrated previously , for large @xmath94 , the @xmath339 distribution becomes @xmath356 for integer @xmath96 . to compute the mean @xmath357 ,
@xmath14 is regarded as a function of @xmath96 @xmath358 similarly for @xmath357 @xmath359 giving directly @xmath360 for @xmath361 the distribution and expectation values can be determined in the same way . in this case
the background and cycling counts need to be considered differently .
consider @xmath362 and consider @xmath363 by induction .
adding another single ion gives a distribution that is a convolution of the @xmath26 trial distribution with a single ion distribution .
if the ion is shelved it will contribute a count distribution @xmath364 , if it is unshelved it will contribution the usual poisson distribution with a mean of the single ion cycling count @xmath94 @xmath365 shifting the summation in the first term and then extending the summation limits as before @xmath366 the coefficient is exactly that encountered in the @xmath339 case and @xmath367 again reduces to the proposed form @xmath368 expectation values for the @xmath361 distribution can be computed in the same way as for the @xmath339 .
the normalization is again trivial , @xmath369 and the mean similarly straight - forward @xmath370 in terms of the maximum possible total counts for all ions used previously @xmath371 , @xmath372 the previous results for @xmath351 become @xmath373 expanding the first term again @xmath374 with the variance @xmath375
[ sherman ] ja sherman , tw koerber , a markhotok , w nagourney , en fortson + `` precision measurement of light shifts in a single trapped ba+ ion '' + physical review letters 94 , 243001 ( 2005 ) 2 .
[ warrington ] rb warrington , pth fisk , mj wouters , ma lawn + `` a microwave frequency standard based on laser - cooled 171yb+ ions '' + precision electromagnetic measurements 2002 , conference digest 156 - 157 3 .
[ itano ] itano , w. m. et al .
+ `` optical frequency standards based on mercury and aluminum ions '' + proceedings of spie 6673 , 667303 - 11 ( 2007 ) 4 . [ peik ] peik , e. et al .
+ `` limit on the present temporal variation of the fine structure constant '' + physical review letters 93 , 15 - 18 ( 2004 ) 5 .
[ fortson - ionpnc ] n. fortson + `` possibility of measuring parity non - conservation with a single trapped atomic ion '' + physical review letters 70(16):2383 - 2386 ( 19 april 1993 ) 6 .
[ koerber ] tw koerber , m schacht , w nagourney , en fortson + `` radio frequency spectroscopy with a trapped ba+ ion : recent progress and prospects for measuring parity violation '' + journal of physics b : atomic , molecular and optical physics 36 ( 3 ) , 637 7 .
[ vetter ] p. a. vetter , d. m. meekhof , p. k. majumder , s. k. lamoreaux , and e. n. fortson + precise test of electroweak theory from a new measurement of parity nonconservation in atomic thallium " + physical review letters 74 , 2658 ( 1995 ) 8 .
[ schacht ] schacht , m + `` atomic parity violation in a single trapped ion '' 9 .
[ wood ] c. s. wood , s. c. bennett , d. cho , b. p. masterson , j. l. roberts , c. e. tanner , c. e. wieman + measurement of parity nonconservation and an anapole moment in cesium + science , vol .
1759 - 1763 ( 21 march 1997 ) + doi : 10.1126/science.275.5307.1759 10 .
[ dehmelt - shelving ] w nagourney , j sandberg , h dehmelt + `` shelved optical electron amplifier : observation of quantum jumps '' + physical review letters 56(26):2797 - 2799 ( 30 june 1995 ) 11 .
[ arfkin ] arfkin , g + mathematical methods for physicists
|
a generalized probe sequence typical of trapped ion experiments using shelving is studied .
detection efficiency is analyzed for finite shelved state lifetimes and using multi - modal count distributions .
multi - modal distributions are more appropriate for measurements that use a small number of ions than the simple poisson counting statistics usually considered and have a larger variance that may be significant in determining uncertainties and in making weighted fits .
optimal probe times and the resulting state detection efficiency and sensitivity are determined for arbitrary cooling rates , initial states and shelved state lifetimes , in terms of a probe coherance time @xmath0 . a universal optimal probe time of @xmath1 is shown to give an almost optimal probe sensitivity for most systems .
|
in flares and coronal mass ejections , coronal electrons can be accelerated above 100 kev @xcite , i.e. , more than three orders of magnitudes higher than the ambient coronal temperature of 100 ev .
the energy is ultimately believed to come from magnetic fields through magnetic reconnection , which occurs throughout the corona .
x - rays and microwaves produced by these energetic electrons are observed in both the chromosphere and the corona .
how energetic electrons are transported from a coronal acceleration site to the chromosphere is a key issue in understanding the dynamics of flares .
recent time - of - flight measurements of hard x - ray emission during solar flares @xcite report relative time delays of arrival at the chromosphere by electrons of different energies propagating from the coronal acceleration site .
the energy dependence of these time delays provides evidence that electrons can propagate along magnetic field lines by free - streaming .
however , hard x - ray observations from above - the - looptop ( alt ) sources @xcite suggest confinement of the energetic electrons above the bright soft x - ray loops for periods several orders of magnitude longer than the ( free - streaming ) transit time across the source . increased detection sensitivity indicates alt hard x - ray emission is likely a common feature of all flares @xcite although there have been fewer detections of alt emission than chromospheric emission because the higher density , and hence emission intensity , in the chromosphere overshadows alt emission . in this case , there must be mechanisms that inhibit the transport of electron heat flux .
it is puzzling why some electrons freely propagate to the chromosphere while some are trapped in the corona after being accelerated .
the physics of how energetic electrons propagate from the corona to the chromosphere , and the contrary , how confinement of electrons happens , remains poorly understood .
the transport of electron heat flux has previously been modeled as conduction mediated by classical collisions @xcite , convection at a subthermal characteristic speed observed in simulations of laser fusion studies @xcite , and turbulent transport limited by anomalous resistivity @xcite , also called anomalous conduction .
recent analytical and numerical studies of the transport of super - hot electron fluxes with energies greater than 10 kev show that the classical conduction model is inapplicable because the mean - free - path of electrons exceeds the characteristic length scales of the system @xcite . in the turbulent transport scenario ,
anomalous resistivity arises from electron scattering by turbulent wave fields excited by instabilities involving the interaction of the energetic electrons , ambient electrons and/or ions .
one commonly considered mode is the ion - acoustic instability @xcite .
when accelerated electrons propagate outwards , ambient electrons are drawn in as a return current .
their relative drift with ions excites ion - acoustic waves , which scatter the energetic electrons and anomalously enhance resistivity @xcite .
the heat flux carried by the energetic electrons is limited by the saturation of the instability due to ion trapping .
the formation of an anomalous conduction front due to ion acoustic turbulence was considered as a means to confine hot electrons for the production of hard x - rays @xcite .
the model , based on a 1d one - fluid code with a grid spacing of 100 km ( @xmath010@xmath1 ion inertial lengths ) , could not capture processes occurring at electron scales .
however , it was argued that the front propagates at approximately the ion acoustic speed @xmath2 and , with some heuristic analysis , predicted that the front would become thinner than the grid spacing , although this was artificially disallowed to avoid numerical instability .
more recently , the existence of a thermal front , literally defined as a region that links plasmas in thermal nonequilibrium and sustains the temperature difference for longer than the electron free - streaming time , was studied in 1d electrostatic vlasov simulations @xcite .
the main goal was to refute the results of an earlier particle - in - cell ( pic ) simulation @xcite in which a conduction front was not seen .
the authors observed the formation of a temperature difference propagating at a speed comparable to @xmath2 and identified the behavior as a conduction front .
the physics of the responsible mechanism was not identified or investigated .
we suspect that front to be an ion acoustic shock and the temperature difference to be a result of shock heating due to its extremely sharp transition .
we observe similar behavior in our simulation , to be presented in section [ result ] .
to gain a better understanding of the problem of energetic electron transport from the corona to the chromosphere , an important question that must be answered is whether thermal conduction is suppressed when the accelerated electrons start to propagate into the immediate surrounding plasma .
our goal in this paper is to answer this basic question . to study the physics at the electron scale and the nonlinear aspects of the problem
, we perform a 2d pic simulation .
we observe that the transport of energetic electrons is suppressed by a double layer that occurs at the shortest scale in a plasma , the debye length @xmath3 .
a double layer ( dl ) is a nonlinear electrostatic electric field localized within two adjacent layers of equal and opposite net charge @xcite .
it is highly localized to @xmath010 @xmath3 because quasi - neutrality in a plasma can only be violated on @xmath3 scales .
it can be understood as a capacitor in a plasma .
an ideal dl is a unipolar electric field , which corresponds to a monotonic drop in the potential across the structure . in most situations ,
however , the potential can contain dips and bumps at the high or low potential sides , but with a net potential drop across the entire layer . the potential drop @xmath4 can accelerate , decelerate and reflect particles entering from the two sides of the dl .
for example , electrons incoming from the high @xmath4 side will be decelerated ( or reflected ) if their kinetic energy is greater ( or less ) than @xmath4 .
dls can be classified into strong and weak depending on whether @xmath4 is much greater or comparable to the mean energy of the reflected particles on either side of the dl @xcite .
much research interest on dls has focused on their ability to directly accelerate particles in auroral ( ionospheric and magnetospheric ) and astrophysical plasmas @xcite . for particle acceleration ,
the potential across the dl is usually maintained by some external energy source .
mechanisms that have been used in simulations to produce dls include an imposed potential drop or current @xcite . in the latter case ,
dls occur as nonlinear waves arising from current - driven instabilities .
the dl observed in our simulation is generated by imposing a large field - aligned temperature jump in the initial state which results in strong currents that drive the dl .
parameters suitable for flare settings are used in our simulations and described in section [ sim ] . in section [ result ] , we present the results .
a brief discussion of our simulations and application to confining above - the - looptop hard x - ray sources are given in section [ dis_app ] .
we summarize in section [ con ] .
we are exploring the problem of energetic electron transport from the corona to the chromosphere by a two - dimensional electromagnetic particle - in - cell simulation using the p3d code @xcite .
the energetic electrons are pre - heated ( i.e. , present at the start of the simulation ) .
we consider it to be more natural that the hot population not have a preferred direction of propagation , so the simulated energetic electrons are not beamed in the initial state . fig .
[ box ] is a cartoon of the initial simulation setup .
our system represents a symmetric local segment of a flare loop with very hot electrons centered at the loop - top in contact with ambient cold electrons .
the computational size is @xmath5= 655.36 @xmath6 2.56 @xmath7 with a cell size of [email protected] @xmath7 .
there are 400 particles per cell .
@xmath8 is the direction parallel to the initial background magnetic field @xmath9 .
@xmath10=@xmath11 is the electron inertial length , where @xmath12=@xmath13 is the electron plasma frequency and @xmath14 the initial density , which is uniform .
the boundaries are periodic in both dimensions .
the simulation domain is of course far smaller than any realistic flare loop . a cartoon of the initial simulation setup . ]
the electrons are initially modeled by bi - maxwellian distributions and the ions by a maxwellian distribution . at the simulation s
start , one - third of the electrons , centered at the middle of the domain , have temperatures parallel to the background magnetic field @xmath15=0.5 ( orange in fig .
[ box ] ) while the remaining two - thirds have @xmath15=0.1 , corresponding to ratios of parallel plasma pressure to magnetic pressure of @xmath16=1 and 0.2 , respectively .
@xmath17=@xmath18 is the alfv@xmath19n speed based on @xmath9 .
the temperature perpendicular to @xmath9 is @xmath20=0.1=@xmath21 everywhere , so @xmath22=0.2=@xmath23 .
since it is easier to accelerate electrons along the field lines , the hot electrons are taken to be anisotropic .
ambient electrons and ions are taken to be isotropic .
note that the unity @xmath24 hot electron pressure is consistent with recent over - the - limb flare observations @xcite .
other parameters used are : mass ratio @xmath25=1/100 and speed of light @xmath26=100 .
electric fields are normalized to @xmath27=@xmath28 .
thermal speeds are defined as @xmath29=@xmath30 .
the debye length based on the initial hot electron temperature is @xmath3=@xmath31=0.1 @xmath10 .
we will use @xmath10 as the unit of length , but it can be conveniently converted to @xmath3 .
we will use @xmath32 to represent @xmath33 in the following for simplicity , since it is the parallel heat transport that is of most interest .
the total duration of the simulation is less than the transit time of the domain by electrons at 1.5 @xmath34 , so the majority of hot electrons will not encounter a boundary during a run .
evolution of electron temperature @xmath32 parallel to the local magnetic field near the transition between hot and cold electrons , at x= 0@xmath10 .
we begin with two electrons populations in contact ( black ) and let them evolve .
the initial gradient becomes a smooth diffusive profile at @xmath12t=100 ( blue ) .
the smooth transition is then interrupted by a jump which signifies transport suppression from the high to low @xmath32 sides . ]
the primary drive of the evolution of the system is the contact of hot electrons with cold electrons , so the dynamics starts in the region of contact .
since this is a symmetric system , it suffices to study either side of the contact .
we choose the right side , where hot electrons propagate outwards with positive velocities .
the figures are averaged in @xmath35 since there is no significant variation in that direction .
[ te ] shows the time evolution of electron temperature in the region where the hot and cold electrons come into contact . the initial profile ( black ) represents a high temperature region decreasing to a low temperature region through a narrow transition that has a length scale of several @xmath10 s
we do not expect the results to be sensitive to this transition scale since we find that the dl develops from a streaming instability after the temperature discontinuity has broadened . over a short period of time ( @xmath0100 @xmath36 , blue )
, the transition broadens as the hot electrons free - stream into the cold electron region .
if free - streaming were to continue , we would expect the profile to continue to broaden due to the mixing of the two populations .
however , a small jump develops ( see @xmath12t=300 ) within the smooth transition , meaning that the escape of hot electrons by free - streaming is inhibited . over time
, the jump grows in size and significantly suppresses mixing due to free - streaming . as a result
, the previously smooth transition is converted into a distinctive jump across two separate temperature regions .
this indicates that the transport of hot electrons into the surrounding plasma is being suppressed .
other features of fig .
[ te ] include langmuir waves propagating in the cold electron side ( see e.g. , the wiggles on the right side of the curve at @xmath12t=300 ) that are excited by the bump - on - tail instability as hot electrons free - stream into that region .
beam modes are excited on the hot electron side ( see , e.g. , longer scale waves on top of the curve at @xmath12t=1400 ) by the electron - electron streaming instability as cold electron beams that are accelerated by a dl enter the hot side and interact with the hot electrons .
@xmath37 ( black ) and @xmath32 ( green ) , overlaid with @xmath38 ( blue ) and the zero @xmath38 position ( dotted cyan ) , at @xmath12t=1200 .
the dl is characterized by a monotonic drop in the potential .
the potential barrier suppresses the transport of hot electrons and results in a drop in @xmath32 across the dl . ]
the heat flux suppression comes from a dl that arises within the transition region .
figure [ ex_timehistory ] shows the time history of the electric field parallel to the local magnetic field , @xmath38 , and [ phire ] the time evolution of the electric potential jump across the dl normalized to the instantaneous hot electron temperature measured at the center of the hot region , @xmath39 .
@xmath40 in @xmath39 thus represents the core temperature of the hot electron population .
@xmath40 drops over time as can be seen on the left side of fig .
[ te ] because hot electrons continuously leak out .
we normalize the potential jump of the dl to @xmath40 to more accurately reflect the suppression strength of the dl on the hot electrons that remain in the source region at any particular time . also shown in fig .
[ phire ] is the fraction of return current electrons that are reflected at the foot of the dl , normalized to the initial total electron number , @xmath41 .
the reflection mechanism and the reason for its importance are discussed later .
@xmath38 is averaged over an electron plasma period to eliminate initial fluctuations in the contact region that decay over time .
the dl emerges at @xmath12t@xmath0100 around x=0 . at @xmath12t@xmath0550
, it emits a large amplitude shock that moves in the positive x direction .
the dl is the more structured dark region in fig .
[ ex_timehistory ] that slowly drifts in the negative x direction .
the dl is constantly evolving over the entire course of the simulation . at @xmath12t=1200 ,
it reaches its maximum amplitude and extends over a scale of 4 @xmath10 , centered at x@xmath0 -6@xmath10 .
[ oplot1200 ] is a cut of fig .
[ ex_timehistory ] at this time in a blowup around its location . shown
are @xmath38 ( blue ) , the electric potential normalized to the core hot electron temperature @xmath37 ( black ) and the electron temperature @xmath32 ( green ) .
@xmath38 has a large positive peak at x@xmath0 -6@xmath10 that causes a large potential drop that reflects hot electrons and produces a sharp drop in the electron temperature .
we define the strength of the dl by its potential jump @xmath42 which is calculated as follows . to the left of the dl in fig .
[ oplot1200 ] , the potential increases , oscillates and eventually reaches a fairly constant value ( near x@xmath0 -13@xmath10 ) , the high end value . to the right of the dl , the potential drops and
oscillates about a roughly constant value ( near x@xmath0 -5@xmath10 ) , the low end value .
@xmath42 is the difference between the high and low end values and is shown as a function of time ( solid black ) in fig .
[ phire ] .
@xmath39 is a measure of the suppression capability of the dl on the hot electron transport . if the dl is large enough such that @xmath43 1 , the thermal bulk of the hot electrons will be reflected by the potential barrier of the dl
this will imply a significant reduction of the electron heat flux as well .
the electric field @xmath38 at @xmath12t=150 , overlaid with @xmath32 ( green ) . the large amplitude wave that forms around x@xmath440 arises from an ion / return - current - electron streaming instability . ] note also in fig . [ oplot1200 ] that the dl is not merely a unipolar electric field .
it has a small negative leg between x= -6 and -5@xmath10 , on the low @xmath32 side .
while the bigger positive part suppresses streaming hot electrons coming from the left , the negative leg can reflect cold electrons coming from the right ( low @xmath32 side ) .
the reflection of these electrons is the drive mechanism of the dl , which we now discuss .
[ oplot150 ] is a cut of fig .
[ ex_timehistory ] at @xmath12t=150 , shortly after the dl emergence .
multiple peaks in @xmath38 develop in the @xmath32 transition region . in the linear regime ( @xmath12t@xmath45150 ) , these waves have small amplitude , comparable to the background fluctuations .
they gradually grow into large - amplitude nonlinear structures at later times .
the peak which has the largest amplitude dominates the others and becomes the dl .
it then starts to hinder hot electron free - streaming and produces an obvious drop in @xmath32 as can be seen at @xmath12t=150 .
the two smaller peaks on either side of x=0 later turn into fairly symmetric waves , which produce no significant potential jump .
the wave turbulence from which the dl emerges is driven by the return current interacting with the background ions .
the dl is also driven by this streaming instability , whose origin is now described .
figure [ phase150 ] shows the electron phase space at @xmath12t=150 .
the hot and cold electron distributions are at the left and right sides , respectively .
note the narrower thermal spread of the cold electrons at the right , as well as the wider spread of hot electrons at the left .
the initial temperature transition is near x=0 .
a beam of cold electrons is entering the hot region . at @xmath12t=0 ,
both distributions are non - drifting and there is no beam . as the hot electrons free - stream into the cold electron region , ions can not follow and are left behind as a charge imbalance .
the resulting electric field drags a cold electron beam towards the hot electron region as a return current .
the relative drift between this return current and the ions excites the instability .
[ cut150 ] is a cut of the phase space at the position of the dl ( x@xmath0 -1@xmath10 ) .
the return current , with a negative beam speed , sits on top of the streaming hot distribution , which is moving to the right and has a wider thermal spread .
the drift of the return current electrons with respect to the ions ( red ) excites the ion / return - current - electron streaming instability .
the linear growth rate @xmath46 of the ion - electron streaming instability ( also known as the buneman instability @xcite ) for the most unstable wave calculated in fluid theory is given by @xmath47 using the parameters of our simulation , @xmath48 0.15 @xmath12 .
the growth time @xmath49 is thus @xmath507 @xmath36 .
the instability is sufficiently strong to produce the dl at @xmath12t@xmath0100 .
the time evolution of the electron phase space reveals that the growth of the dl depends on the number of return current ( rc ) electrons reflected at the foot of the dl .
one can see from fig .
[ oplot1200 ] that the electric field @xmath38 of the dl has a negative leg on the low @xmath32 side .
this shows up as a potential dip on the low potential side .
the rc electrons are reflected by the negative leg , which effectively reduces the rc speed and thus converts the kinetic energy of the rc to the electrostatic energy of the dl . by exchanging momentum with the reflected electrons ,
the dl gets amplified @xcite .
electron reflection also indicates when the instability enters into a nonlinear regime ( @xmath12t@xmath51150 ) .
the dl growth depends on the number of rc electrons reflected at the foot of the dl .
qualitatively , the dependence of dl growth on the number of reflected electrons can be seen by contrasting the phase space at a time when the dl is growing , e.g. , at @xmath12t=550 , with a time when it is not growing , e.g. , at @xmath12t=700 ( as @xmath39 plateaus in fig .
[ phire ] ) . at @xmath12t=700 ( fig .
[ phase700 ] ) to the right of the dl at x@xmath52 -4@xmath10 ( vertical black line ) , there are many fewer reflected electrons than to the right of the dl at @xmath12t=550 ( fig .
[ phase550 ] ) , x@xmath52 -2@xmath53 .
we estimated the number of reflected rc electrons , @xmath54 , by integrating the electron distribution to the right of the dl over a distance of one @xmath10 , which is the typical width of the dl , and for @xmath55(0 , 0.4 @xmath56 ) , which is approximately the velocity range covered by the reflected rc electrons as they turn around and acquire positive velocities .
@xmath54 is normalized to @xmath57 , the initial electron number contained in the same interval .
note that the integration for @xmath54 also includes a contribution from hot free - streaming population in that velocity range .
however , this contribution stays nearly constant and the variation in @xmath54 mainly comes from the change in the number of reflected cold rc electrons .
therefore , @xmath54 suffices as an estimate of the reflected rc electrons ( plus some constant offset ) .
@xmath41 ( green ) is plotted on top of the dl size , @xmath39 , in fig .
[ phire ] to show their correlation .
as @xmath41 increases during @xmath12t@xmath58550 and @xmath59950 , @xmath39 has a general increasing tendency as well . as @xmath54 drops abruptly at @xmath12t=550 and stays at a low level until @xmath12t=950 , @xmath39 plateaus .
hence , there is good correlation between the dl growth and the number of the reflected rc electrons .
an examination of the evolution of the electric field in fig .
[ ex_timehistory ] reveals that at @xmath12t@xmath0550 , the dl sheds an ion - acoustic shock wave , which shows up as an ion hole in ion phase space .
once it leaves the dl , the shock accelerates to the right while the dl continues to slowly move to the left . therefore , their separation increases .
the shock wave is bipolar and reflects most of the rc electrons with its large - amplitude negative leg .
one can see in fig .
[ phase700 ] , at @xmath12t=700 , most of the rc electrons are reflected behind the shock at x@xmath44 -2@xmath10 .
@xmath41 shows an abrupt drop at @xmath12t=550 because most of the rc electrons are stuck behind the shock and do not reach the dl .
the shock dissipates through ion heating .
the ion temperature increases by as much as a factor of two along the path traversed by the shock .
the shock gradually diminishes in amplitude .
similar dynamics ( acceleration and decay ) of an ion hole mode at the low potential side of a weak dl created in ion - acoustic turbulence is also observed in current - driven systems @xcite . electron phase space at @xmath12t=1200 with @xmath430.8 . to the left of the dl location , x@xmath0 -6@xmath10 , hot electron populations with @xmath60 0.9 ( cyan and green in color )
are reflected by the potential barrier .
they are trapped in the hot region . for higher velocity electrons ( at the tail ; blue in color ) , there is a drop in velocity as they are decelerated .
both contribute to suppression of the heat flux . ] during our simulation , @xmath39 reaches 0.8 at @xmath12t=1200 ( fig . [ oplot1200 ] ) .
this means that hot electrons with velocities @xmath61 0.9 will be reflected .
the electron phase space at @xmath12t=1200 in fig . [ phase1200 ] reveals that the population of hot electrons with @xmath62 0.9 wraps around from positive to negative velocities when encountering the dl .
so they are reflected and hence confined in the source region .
the suppression observed here is significant .
we calculate the heat flux carried by kinetic hot electrons that is positively directed as @xmath63=@xmath64 @xcite .
the mean flow speed is @xmath452% of the hot electron thermal speed and hence negligible in the calculation of @xmath63 .
similar to @xmath32 , @xmath65 has a sharp jump across the dl . towards the end of the simulation , @xmath65
is suppressed by @xmath040% to the right of the dl compared with the higher value to the left of the dl . to determine the dependence of the dl potential drop on the hot electron temperature
, we performed another run in which we raised the initial hot electron temperature @xmath66 from 0.5 to 1 , and @xmath9 from 1 to @xmath67 such that @xmath68 remains unity , while keeping all other parameters unchanged . slightly increasing @xmath9
should not have an effect on the result because the phenomena we observed here are dominantly electrostatic .
the size of the dl depends on the hot electron temperature .
the dl potential drop @xmath39 from another simulation with a higher initial hot electron temperature @xmath69=1 is plotted as a dotted line in fig .
[ phire ] .
this potential drop is normalized to the instantaneous hot electron temperature at the center of this new hot region .
the dl is on average stronger than that in the simulation with initial @xmath69=0.5 ( solid line ) .
the basic dynamics of the dls are the same in both cases . a phase of growth is followed by a period of pause , and the whole cycle repeats .
the dl evolves and grows faster in the higher @xmath69 run , probably because the return current electron response , which is the driver of the dl , is faster due to the increased thermal speed of the hot electrons ( relative to those of the ambient plasma ) in the @xmath69=1 run than in the @xmath69=0.5 run .
there is also a modest increase in the suppression of heat flux @xmath63 from @xmath040% in the @xmath69=0.5 run to @xmath050% in the higher @xmath69 run .
therefore , one might expect a higher ratio of hot to cold electron temperatures would lead to a stronger dl and associated suppression of transport .
we have studied the suppression of electron thermal transport due to double layer ( dl ) formation in 2d pic simulations . however , our system is effectively 1d since the transverse direction is only @xmath70 20@xmath3 wide .
the presence of the second dimension enables us to average the data for noise reduction .
2d effects on dls are not investigated in this study , but do not seem to be significant .
extended ( planar ) weak dls over 80 @xmath3 in transverse extent were observed in earlier 2d pic simulations where subthermal electron current was injected at the boundaries of the parallel direction @xcite .
such electron injection may correspond to the return current electrons being drawn in at the contact region of the two electron populations in the present simulations .
the planar 2d dls tend to develop substantial substructure across the magnetic field late in time , but the basic dynamics of the dls are the same as in 1d .
we expect the formation of dls , and therefore transport suppression of hot electrons , to persist in 2d
. however , 2d dls are unlikely to be planar over @xmath0 10@xmath71 , the scale of coronal looptops . on the other hand ,
a variation of the dl over a large transverse distance is not likely to strongly impact the largely 1d dynamics studied here .
the injection of a narrow band of hot electrons , however , might behave very differently .
this case remains to be studied .
we also note that the computational size of the pic simulations , which is 655@xmath72 66 m in the corona , is small compared to the actual size of flaring loops .
they , however , reveal important physics down to the kinetic electron scale , the debye length , where dls form .
we expect the dl to continue strengthening in a larger domain that allows the system to evolve for a longer period . applying our results to the above - the - looptop hard x - ray sources @xcite that show an exponential decay on a timescale of more than two orders of magnitude longer than the electron transit time through the source
, the formation of a dl provides a plausible mechanism to confine energetic electrons in the source region .
electrons that have kinetic energies less than the potential drop of the dl are reflected back to the source region . over the entire course of the simulations , the dl tends to amplify through multiple phases of growth .
there are no signs of decay .
since it is driven by the return current electrons , which exist for as long as some hot electrons are leaking out , the dl can persist until the eventual depletion of all hot electrons in the source region , which should take much longer than the electron transit time through the source due to the strong transport suppression caused by the dl .
the dl lifetime is therefore expected to be much longer than the transit time .
this can naturally explain the prolonged lifetimes ( @xmath51100 s ) of the energetic electrons from the above - the - looptop hard x - ray observations compared with the electron transit time ( @xmath01 s ) through the source .
results from pic simulations of a pre - accelerated hot electron source in contact with ambient plasma , which provide a basic model of coronal looptop sources , are presented .
the hot electrons free - stream along the magnetic field initially , but transport is suppressed by the formation of a double layer that is driven by an ion / return - current - electron streaming instability due to the drift of return current electrons with respect to the ions .
the dl subsequently reflects and decelerates hot electrons as they begin to propagate into the surrounding plasma . a significant reduction in electron heat flux through the ambient electron region is observed .
such a suppression mechanism differs from the conventional picture of electron scattering by turbulence .
our results are consistent with observations of both free - streaming propagation @xcite and the suggested confinement at looptop @xcite of hard x - ray producing electrons , corresponding to the escaping and reflected hot electrons , respectively . , r. p. , krucker , s. , hurford , g. j. , smith , d. m. , hudson , h. s. , holman , g. d. , schwartz , r. a. , dennis , b. r. , share , g. h. , murphy , r. j. , emslie , a. g. , johns - krull , c. , & vilmer , n. 2003 , , 595 , l69
|
during flares and coronal mass ejections , energetic electrons from coronal sources typically have very long lifetimes compared to the transit times across the systems , suggesting confinement in the source region .
particle - in - cell simulations are carried out to explore the mechanisms of energetic electron transport from the corona to the chromosphere and possible confinement .
we set up an initial system of pre - accelerated hot electrons in contact with ambient cold electrons along the local magnetic field , and let it evolve over time .
suppression of transport by a nonlinear , highly localized electrostatic electric field ( in the form of a double layer ) is observed after a short phase of free - streaming by hot electrons . the double layer ( dl )
emerges at the contact of the two electron populations .
it is driven by an ion - electron streaming instability due to the drift of the back - streaming return current electrons interacting with the ions .
the dl grows over time and supports a significant drop in temperature and hence reduces heat flux between the two regions that is sustained for the duration of the simulation .
this study shows transport suppression begins when the energetic electrons start to propagate away from a coronal acceleration site .
it also implies confinement of energetic electrons with kinetic energies less than the electrostatic energy of the dl for the dl lifetime , which is much longer than the electron transit time through the source region .
|
-5 mm the positive mass theorem provides a good example of _ `` the unreasonable effectiveness of physics in mathematics''_. the need to define mass in general relativity is motivated directly by the physics imperative to establish a correspondence between general relativity and classical newtonian gravity .
already difficulties arise : although the vacuum einstein equations @xmath0 for the lorentz metric @xmath1 suggest ( by analogy with the wave equation , for example ) that a mass ( energy ) which includes contributions from the gravitational field , should be built from the first derivatives of the field @xmath1 , it is clear that this is incompatible with coordinate invariance . the schwarzschild vacuum spacetime metric , for @xmath2
, @xmath3 provides an important clue , since the parameter @xmath4 governs the behaviour of timelike geodesics and may be regarded as the total mass . note that @xmath5 ensures the boundary @xmath6 is smooth and totally geodesic in the hypersurfaces @xmath7 .
a riemannian 3-manifold @xmath8 is said to be _ asymptotically flat _
if @xmath9 for some compact @xmath10 , and @xmath11 admits a metric @xmath12 which is flat outside @xmath10 , and the metric components @xmath13 in the induced rectangular coordinates satisfy @xmath14 the total mass of @xmath8 is defined informally by @xcite @xmath15 if the scalar curvature @xmath16 then @xmath17 is well - defined , independent of the choices of rectangular coordinates and of exhaustion of @xmath11 used to define @xmath18 see @xcite for weaker decay and smoothness assumptions . for simplicity ,
the discussion here is restricted to @xmath19 riemannian 3-dimensional geometry .
this corresponds to the case of time - symmetric initial data : @xmath8 is a totally geodesic spacelike hypersurface in a lorentzian manifold , and we can identify the local matter ( equivalently , energy ) density with the scalar curvature @xmath20 .
this simplification entails a small loss of generality : most , but not all , of the results we describe have been extended to general asymptotically flat space - time initial data @xmath21 , where @xmath22 is the second fundamental form of a spacelike hypersurface @xmath11 .
some results also generalize to the closely related bondi mass , which measures mass and gravitational radiation flux near null infinity , and to mass on asymptotically hyperbolic and anti - desitter spaces cf .
@xcite , but these involve additional complications which we will not discuss here . the positive mass theorem ( pmt ) in its simplest form
is suppose @xmath8 is a complete asymptotically flat 3-manifold with non - negative scalar curvature @xmath20 .
then @xmath23 , and @xmath24 iff @xmath25 .
the rigidity conclusion in the case @xmath24 shows that @xmath26 for @xmath8 scalar flat ( `` matter - free '' ) but non - flat , so @xmath17 does provide a measure of the gravitational field .
three distinct approaches have been successfully used to prove the pmt : with stable minimal surfaces @xcite ; with spinors @xcite and the schrdinger - lichnerowicz identity @xcite ; and using the geroch foliation condition @xcite .
a number of other appproaches have produced partial results : using spacetime geodesics @xcite ; a nonlinear elliptic system for a distinguished orthonormal frame @xcite ; and alternative foliation conditions @xcite .
the connection between these approachs remains mysterious ; the only discernable common thread is mean curvature , and this is quite tenuous .
the application of the positive mass theorem to resolve the yamabe conjecture @xcite is well known .
less well known is the proof of the uniqueness of the schwarzschild spacetime amongst static metrics with smooth black hole boundary @xcite , which we briefly outline .
a _ static spacetime _ is a lorentzian 4-manifold with a hypersurface - orthogonal timelike killing vector . with @xmath27 denoting the length of the killing vector , the metric @xmath28 on the spacelike hypersurface satisfies the static equations @xmath29 smoothness implies the boundary set @xmath30 is totally geodesic ; analyticity of @xmath31 can be used to show the asymptotic expansions @xmath32 as @xmath33 for some constant @xmath34 .
the metrics @xmath35 where @xmath36 both have @xmath37 , and @xmath38 is asymptotically flat with vanishing adm mass , and @xmath39 is a ( smooth ) metric on a compact manifold .
gluing two copies of @xmath8 along the totally geodesic boundary @xmath40 and conformally changing to @xmath41 where @xmath42 on the two copies of @xmath11 , gives a complete af manifold with @xmath43 and vanishing mass .
the pmt shows @xmath44 is flat and it follows without difficulty that @xmath8 is schwarzschild .
this extends previous results @xcite which required the boundary to be connected .
-5 mm a boundary component @xmath40 with mean curvature @xmath45 is called a _ black hole _ or _ horizon _ , since if @xmath8 is a totally geodesic hypersurface then @xmath40 is a trapped surface and hence , by the penrose singularity theorem @xcite , lies within an event horizon and is destined to encounter geodesic incompleteness in the predictable future . the spatial schwarzschild metric @xmath46 with @xmath47 shows that the completeness condition in the pmt is important , but it can be weakened to allow horizon boundary components of @xmath11 .
this follows immediately from the minimal surface argument @xcite ; or by an extension to the witten argument @xcite , imposing one of the boundary conditions @xmath48 on the spinor field @xmath49 , where @xmath50 satisfies @xmath51 .
an interesting extension is obtained by imposing the spectral boundary condition @xmath52 where @xmath53 is the projection onto the subspace of positive eigenspinors of the induced dirac operator @xmath54 . using the remarkable hijazi - br estimate @xcite @xmath55 for the eigenvalues of @xmath54 when @xmath56 , herzlich showed @xcite if @xmath8 is asymptotically flat with @xmath20 and boundary @xmath56 with mean curvature satisfying @xmath57 where @xmath58 , then @xmath59 , with equality iff @xmath60 .
the proof starts with the riemannian form of the schrdinger - lichnerowicz - witten identity @xcite @xmath61 where @xmath62 is the nester - witten form @xcite @xmath63 the boundary condition @xmath64 is elliptic and it can be shown @xcite there is a spinor on @xmath11 satisfying @xmath65 with boundary conditions @xmath66 as @xmath33 and ( [ p+psi ] ) on @xmath40 .
it follows from ( [ hijazi ] ) and ( [ p+psi ] ) that @xmath67 and the result follows .
observe that in each case , equality leads to flat @xmath68 .
an elegant physical argument lead penrose to conjecture an analogous inequality , but which distinguishes the schwarzschild metric instead @xcite , see also @xcite . if @xmath8 satisfies the conditions of the pmt , except that @xmath69 is compact with vanishing mean curvature and such that @xmath40 is the `` outermost '' closed minimal surface in @xmath11 , then @xmath70 with equality only for the schwarzschild metric .
a closed minimal surface is said to be an _
outermost horizon _ or _ outer - minimizing horizon _ if @xmath11 contains no least area surfaces homologous to @xmath40 in the asymptotic region exterior to @xmath40 .
the outermost condition is essential , since examples of non - negative scalar curvature manifolds can be constructed by forming the connected sum of @xmath11 and large spheres by arbitrarily small and large necks . the penrose conjecture has been established by huisken and ilmanen @xcite using a variational level set formulation of the inverse mean curvature flow @xcite , and by bray @xcite by a very interesting conformal deformation argument .
bray s proof is more general since it takes into account contributions from all the connected components of the boundary .
-5 mm thus it is natural to consider @xmath71 as the mass of a black hole ( minimal surface ) @xmath40 .
more generally , the correspondence with newtonian gravity suggests that any bounded region @xmath72 should have a _ quasi - local _ mass , which measures both the matter density ( represented in this case by the scalar curvature @xmath20 ) , and some contribution from the gravitational field .
the rather satisfactory positivity properties of the total mass , as established by the pmt , motivate the properties we might expect such a geometric mass to possess @xcite . 1 . * ( non - negativity ) * @xmath73 ; 2 . * ( rigidity / strict positivity ) * @xmath74 if and only if @xmath72 is flat ; 3 . *
( monotonicity ) * @xmath75 whenever @xmath76 , where it is understood that the inclusion is a metric isometry ; 4 . *
( spherical mass ) * @xmath77 should agree with the spherical mass , for spherically symmetric regions ; 5 . *
( adm limit ) * @xmath77 should be asymptotic to the adm mass ; 6 . *
( black hole limit ) * @xmath77 should agree with the black hole mass ( [ penrose ] ) .
many candidates have been proposed for quasi - local mass ( see for example @xcite for a comparison of some definitions ) , the most significant being that of hawking @xcite , @xmath78 where @xmath79 .
this equals @xmath80 for standard spheres in schwarschild .
although @xmath81 for surfaces in @xmath68 , it was shown in @xcite that @xmath82 for a stable constant mean curvature 2-sphere @xmath40 in a 3-manifold of non - negative scalar curvature . thus for such `` round '' spheres , @xmath83 is nonegative , and the black hole limit condition is trivially satisfied .
however the remaining properties , in particular rigidity and monotonicity , are rather problematic .
although the twistorially - defined penrose quasi - local mass @xcite is well - behaved in special cases @xcite , it is defined unambiguously only for surfaces arising from embedding into a conformally flat spacetime , and even then numerical experiments @xcite strongly suggest that monotonicity is violated . in fact , of the various proposals for @xmath77 , only the definitions of @xcite are known to satisfy positivity .
dougan and mason @xcite show that the integral @xmath84 of the nester - witten 2-form ( [ nwform ] ) is positive for spinor fields @xmath49 on @xmath40 which satisfy a certain elliptic system on @xmath40 .
however , bergqvist @xcite shows that positivity holds under much weaker conditions on @xmath49 , and there are many variant definitions with similar properties
. it would be useful to understand these dm - style definitions better , and in particular whether any satisfy monotonicity .
monotonicity and adm - compatibility imply @xmath85 for any region @xmath86 embedded isometrically in an @xmath8 satisfying ( as always ) the pmt conditions .
this motivates the following definition @xcite [ qlmass ] let @xmath87 denote the set of all asymptotically flat 3-manifolds @xmath8 of non - negative scalar curvature , with boundary which if non - empty , consists of compact outermost horizons , and such that @xmath8 has no other horizons . for any bounded open connected region @xmath72 , let @xmath88 be the set of @xmath89 such that @xmath86 embeds isometrically into @xmath11 , and define @xmath90 we say that @xmath11 satisfying these conditions is an _ admissible extension _ of @xmath86 .
the horizon condition serves to exclude examples which hide @xmath86 inside an arbitrarily small neck , which would force the infimum to zero .
this is a refinement @xcite of the original definition @xcite , which prohibited horizons altogether .
clearly @xmath91 is well - defined and finite , once the region @xmath86 admits just one admissible extension .
the pmt with horizon boundary implies non - negativity , and monotonicity follows directly .
strict positivity of @xmath77 was established in @xcite , with the slightly weaker rigidity conclusion that if @xmath74 then @xmath86 is locally flat .
agreement with the spherical mass , and the adm limit condition , follows also from @xcite .
bray s results imply that @xmath91 agrees with the black hole mass in the limit as @xmath86 shrinks down to a black hole .
in addition , @xmath92 for any admissible extension @xmath11 , so @xmath77 is the optimal quasi - local mass definition with respect to this condition .
the optimal form of the horizon condition remains conjectural .
bray has suggested an alternative condition , that @xmath86 be a `` strictly minimizing hull '' @xcite in @xmath11 , so @xmath93 has the least area amongst all enclosing surfaces in the exterior . in this case
we say @xmath40 is _ outer minimizing _ , and denote by @xmath94 the quasilocal mass function defined by restricting admissible extensions to those @xmath11 in which @xmath40 is outer minimizing . for this modified definition
the penrose inequality @xcite applies to show that if @xmath95 embeds into the schwarzschild 3-manifold with the same induced metric and mean curvature ( cf .
( [ 2ndvar ] ) , ( [ staticbndry ] ) ) and encloses the horizon , then @xmath96 .
it is not clear how to establish this natural result for the unmodified definition @xmath91 .
-5 mm although in many respects the definition of @xmath77 is quite satisfactory , it is not constructive , and thus it is important to determine computational methods .
the key is the following @xcite [ staticconjecture ] the infimum in @xmath77 is realised by a 3-metric agreeing with @xmath86 in the interior , _ static _ ( [ static ] ) in the exterior region , and such that the metric is lipschitz - continuous across the matching surface @xmath40 , and the mean curvatures of the two sides agree along @xmath40 . a similar conjecture for the space - time generalisation of the quasi - local mass , asserts that the exterior metric is _ stationary _ , ie .
admits a timelike killing field @xcite . as motivation for this conjecture ,
note first that if @xmath97 in some region , then a conformal factor @xmath98 can be found such that @xmath99 has less mass and @xmath100 .
thus a mass - minimizing metric for ( [ qlmass ] ) , if such a metric exists , must have vanishing scalar curvature . now
if the linearization @xmath101 is surjective then @xmath28 admits a variation which produces positive scalar curvature . the formal obstruction to surjectivity is non - trivial @xmath102 , which leads to the static metric equations ( [ static ] ) .
corvino @xcite shows that if @xmath102 is trivial in @xmath103 then there are compactly supported metric variations in @xmath104 which increase the scalar curvature .
this gives if @xmath8 realizes the infimum in definition [ qlmass ] , then there is a @xmath105 such that @xmath31 satisfy the static metric equations ( [ static ] ) in @xmath106 .
this suggests a computational algorithm for determining @xmath91 : find an asymptotically flat static metric with boundary geometry matching that of @xmath95 . to determine the appropriate boundary conditions , recall the second variation formula for the area of the leaves of a foliation labelled by @xmath107 : @xmath108 where @xmath109 are respectively the second fundamental form , mean curvature and gauss curvature of the leaves , @xmath110 is the lapse function , @xmath111 is the normal vector and @xmath112 is the laplacian on the leaves .
our conventions give @xmath113 where @xmath114 is the volume element of the leaves .
this shows that @xmath115 will be defined distributionally across a matching surface as a bounded function if @xmath116 h_{\partial\omega}&= & h_{\sigma } .
\end{array}\ ] ] [ staticexistence ] @xmath72 determines a unique static asymptotically flat manifold @xmath117 with boundary @xmath118 satisfying ( [ staticbndry ] ) .
if true , this would give a prime candidate for the minimal mass extension .
it is known ( pengzi miao , private communication ) that the boundary conditions ( [ staticbndry ] ) are elliptic for ( [ static ] ) .
it is tempting to conjecture that mass - minimizing sequences for @xmath77 should converge to a static metric .
for example , ( * ? ? ?
* theorem 5.2 ) shows that a sequence of metrics @xmath119 , close in the weighted sobolev space @xmath120 , @xmath121 , to the flat metric @xmath122 on @xmath68 and such that @xmath123 , converges strongly to @xmath122 in @xmath124 .
similar results , under rather different size conditions , are given in @xcite , and a discussion of the general `` weak compactness '' conjecture may be found in @xcite .
-5 mm to estimate @xmath77 from above , it suffices to construct admissible extensions metrics with non - negative scalar curvature and satisfying ( [ staticbndry ] ) .
these boundary conditions exclude the usual conformal method .
instead , metrics in _ quasi - spherical _
form @xcite @xmath125 satisfy a parabolic equation for @xmath126 on @xmath127 evolving in the radial direction , when @xmath128 , with @xmath129 freely specifiable . since the metric 2-spheres @xmath130 have mean curvature @xmath131 , ( [ qsmetric ] ) provides admissible extensions for @xmath132 with mean curvature @xmath133 .
the underlying parabolic equation derives from ( [ 2ndvar ] ) , and has been generalized to non - spherical foliations in @xcite . as an application ,
choosing @xmath134 we can show finding lower bounds for @xmath91 is more difficult .
bray s definition of _ inner mass _ ( * ? ? ? * p243 ) gives a lower bound , but for @xmath94 .
the difficulty here as above lies in showing that a horizon inside @xmath86 remains outermost when the inner region is glued to a general exterior region @xmath138 .
this follows easily when @xmath139 is outer - minimizing in @xmath140 , as guaranteed by the definition for @xmath94 . on physical grounds
one expects that if `` too much '' matter is compressed into region which is `` too small '' , then a black hole must be present .
the geometric challenge lies in making this heuristic statement precise , and the only result in this direction has been @xcite , which gives quantitative measures which guarantee the existence of a black hole .
an observation by walter simon ( private communication ) is thus very interesting : if @xmath141 ( say ) and @xmath86 embeds isometrically into a complete asymptotically flat manifold @xmath11 without boundary and with non - negative scalar curvature , and such that @xmath142 , then @xmath11 must have a horizon .
this reinforces the importance of finding good lower bounds for @xmath77 , since the existence of a horizon in a similar situation with @xmath143 does not follow .
p. t. chruciel .
boundary conditions at spatial infinity from a hamiltonian point of view . in p.
g. bergmann and v. de sabbata , editors , _ topological properties and global structure of space - time_. plenum , new york , 1986 .
g. w. gibbons .
the isoperimetric and bogomolny inequalities for black holes . in t.
willmore and n. hitchin , editors , _ global riemannian geometry _ , chapter 6 , pages 194202 .
ellis harwood , chichester , april 1984 .
j. m. nester .
the gravitational hamiltonian . in f.
j. flaherty , editor , _ asymptotic behaviour of mass and space - time geometry ( oregon 1983 ) _ , lecture notes in physics 212 , pages 155163 .
springer verlag , 1984 .
|
physicists believe , with some justification , that there should be a correspondence between familiar properties of newtonian gravity and properties of solutions of the einstein equations . the positive mass theorem ( pmt ) , first proved over twenty years ago @xcite , is a remarkable testament to this faith . however , fundamental mathematical questions concerning mass in general relativity remain , associated with the definition and properties of quasi - local mass .
central themes are the structure of metrics with non - negative scalar curvature , and the role played by minimal area 2-spheres ( black holes ) .
4.5 mm * 2000 mathematics subject classification : * 53c99 , 83c57 . * keywords and phrases : * quasi - local mass , einstein equations , scalar curvature .
|
the bursting activity from sgr 1900 + 14 is quite diverse .
the most common bursts observed are the recurrent , short bursts ( e.g. , aptekar et al .
they usually have durations of a few hundreds of ms and peak luminosities reaching @xmath010@xmath4 erg s@xmath5 ( for an assumed distance of 10 kpc ) .
the most rare but very luminous events are the so called ` giant ' flares : to now only one has been observed from sgr 1900 + 14 on 1998 august 27 ( feroci et al . 1999 , hurley et al .
1999b ) . the burst had a duration of about 300 s and its time evolution was characterized by a very intense , short and hard peak followed by a decaying tail that showed coherent pulsations at the frequency of the persistent pulsed emission .
the luminosity of this ` giant ' burst was estimated to be greater than 10@xmath6 erg s@xmath5 .
another very bright flare from sgr 1900 + 14 was observed on 18 april 2001 ( guidorzi et al .
2004 ) , shorter and less intense than that of 1998 , but it too showed periodicity . given that both its duration of @xmath040 s and peak luminosity of @xmath010@xmath7 erg s@xmath5 fell in between the common , recurrent bursts and the giant flare , this burst was classified as an ` intermediate ' flare .
spectral analysis of the sgr bursts has been generally performed at energies above @xmath015 - 20 kev , where the spectra are usually well represented by an optically thin thermal bremsstrahlung ( ottb ) law with temperatures ranging from 20 to 40 kev ( e.g. , aptekar et al . 2001 ) .
detailed spectral studies on uniform samples of short bursts over wide energy ranges are rare , and we are not aware of previous studies specifically for sgr 1900 + 14 .
laros et al .
( 1986 ) and fenimore et al .
( 1994 ) used the _ ice _
data from 5 kev to above 200 kev to study more than 100 bursts from sgr 1806@xmath820 .
they found that the photon number spectrum of the detected bursts was remarkably stable despite the large intensity spread ( a factor of 50 ) , and that the low energy ( below 15 kev ) data were _ inconsistent _ with the back extrapolation of an ottb model that provided a good fit to the high energy portion of the spectrum .
more recently , strohmayer and ibrahim ( 1998 ) confirmed this result using data from _
_ rossi__xte .
qualitatively similar spectral properties were measured during a bright burst from sgr 1900 + 14 using _ hete-2 _ ( olive et al . 2003 ) . for this one burst
, they noted that the ottb model overestimated the flux at low energies ( @xmath915 kev ) and the broad - band spectrum ( 7@xmath8150 kev ) was best fit by the sum of two blackbody ( bb ) laws . in this paper
, we describe the results of a spectral analysis of ten short bursts from sgr 1900 + 14 .
_ bepposax _ pointed at the source @xmath08 hours after the intermediate flare of april 18 and discovered an afterglow emission which decayed to the quiescent level in a few days ( feroci et al .
2003 , woods et al .
2003 ) . during the dimming afterglow
sgr 1900 + 14 emitted a series of bursts of short duration ( @xmath10 1 s ) , consistent with the ` normal ' short bursts .
here we report on the spectral analysis of these bursts in the energy range from 1.5 to 100 kev , and show that their spectra are _ inconsistent _ with an ottb law , due to a paucity of photons in the low energy range .
the _ bepposax _ observation started on 2001 april 18 at 15:10 ut and ended on april 19 at 19:38 ut .
net exposures times of the narrow field instruments ( nfis ) were 23.4 ks ( lecs ) , 57.2 ks ( mecs ) and 41.7 ks ( pds ) . a point - like source is clearly detected in the lecs and mecs images .
light curves and spectral data were accumulated from a circular region with 6@xmath11 radius , the optimal signal - to - noise region for which a response matrix is available .
the mecs ( 210 kev ) light curve ( fig . [ lc ] - middle panel ) shows many short bursts , many having peak count rates more than two orders of magnitude above the persistent x - ray emission level from sgr 1900 + 14 . in our analysis
we considered only the events with an integral number of counts in the mecs band larger than 100 , enough to obtain significant constraints on spectral fit parameters .
thus , we selected only ten bursts whose salient properties are given in table [ burstlist ] .
we searched the pds data for bursts coincident in time with those detected in the mecs light curve and found significant emission for all ten .
five of the bursts in table [ burstlist ] were also detected in the lecs data , but the number of counts in the energy range useful for spectral analysis ( 0.1 - 4 kev ) is small and does not further constrain the spectral fit even for the brightest event ( burst b ) .
thus , we limited our study only to the mecs and pds data , with an effective energy range between 1.5 and @xmath0100 kev .
the background counts in the mecs images within an annulus surrounding the source extraction circle ( of equal area ) , were found to be less than unity over the bursts time intervals .
hence , no background subtraction was applied .
pds background spectra were accumulated from simultaneous data acquired by off - source detectors .
publicly available matrices were used for all the instruments .
mecs and pds spectra were binned in order to have a minimum number of 20 counts in each bin .
we used xspec 11 for fitting the data .
the parameter uncertainties correspond to @xmath12% confidence for a single parameter , i.e. , @xmath13 .
mecs and pds spectra were simultaneously fitted leaving the normalization constant free to vary in the range 0.800.90 , according to the prescription given by fiore , guainazzi and grandi ( 1999 ) .
the bursts are often characterized by the presence of several well separated peaks .
when correlating the mecs and pds light curves , we detected a systematic time delay of about 50 ms between the two instruments .
this effect is possibly a consequence of the switch to the secondary clock onboard the _ bepposax _ satellite in the year 2000 . on the other hand ,
the similar time structure allowed for a post - facto empirical time re - alignment of the data , as shown in fig .
[ burst ] . based on this assumption ,
the time - integrated energy spectra were extracted for the two instruments on time intervals encompassing valid simultaneous data from both instruments .
we note that the two brightest bursts , b and g , had very strong initial spikes with high count rates ( larger than 4,000 and 12,000 count / s in the mecs and pds , respectively ) that saturated the satellite telemetry , effectively truncating the light curves in the two instruments1% ) . ] .
we conservatively included in the spectral analysis only the counts safely away from the saturation edge .
a critical parameter in the modeling of these sgr burst spectra is the column density of neutral absorbing gas along the line of sight .
the most precise measurement of this parameter comes from modeling of the persistent x - ray emission . in quiescence ,
the x - ray spectrum of sgr 1900 + 14 is best fit ( woods et al .
1999 ) by the sum of a blackbody ( @xmath14 0.5 kev ) and a power law ( photon index @xmath15 1.0 - 2.5 ) . for this model , the column density is consistently measured at n@xmath16 @xmath3 ( woods et al .
1999 ; woods et al .
2001 ; kouveliotou et al . 2001 ; feroci et al .
2003 ) . however , this particular observation where our bursts were recorded took place during a state of enhanced flux following the 2001 april 18 flare ( feroci et al .
2003 ) , a so - called ` afterglow state ' .
fit to a simple power law , the @xmath1 measured at this time is ( [email protected] ) @xmath18 10@xmath19 @xmath3 .
although not statistically required by the data , the addition of a blackbody component to the power law model provides a good fit to the persistent x - ray emission at this time , with an @xmath1 consistent with the quiescent state ( but more loosely constrained ) , but with a temperature close to 1 kev ( mereghetti et al . , in preparation ) .
indeed , similar variations of the blackbody temperature have been observed in this and other magnetar candidates ( e.g. , lenters et al . 2003 , woods et al .
2004 ) following intense burst activity .
thus , it can not be excluded that the @xmath1 for sgr 1900 + 14 at the time of this observation might have been consistent with the quiescent value of @xmath20 10@xmath19 @xmath3 . here
, we will adopt a conservative approach and allow @xmath1 to vary between 2 and 5 @xmath21 @xmath3 for our spectral fits .
one important conclusion from our analysis is that the ottb model fails to give acceptable fits to the spectral energy distributions of all bursts . using the combined data set ( mecs and pds ) for each burst and assuming @xmath1 = 2.5@xmath1810@xmath19 @xmath3 , the @xmath22 are all in the interval 27 ( with 12 to 46 degrees of freedom - d.o.f . ) , while for @xmath1 = 4.4@xmath1810@xmath19 @xmath3 the @xmath22 are lower but remain mostly statistically unacceptable ( @xmath22 between 1.4 and 4.8 , except for burst j for which @xmath22 = 1.0 , 22 d.o.f . ) .
when we limit the fit to the pds data only ( e @xmath23 15 kev ) we generally obtain very good results ( 0.5@xmath9 @xmath22@xmath9 1.2 ) , with the exception of burst c ( @xmath22=2.1 ) . the measured temperatures ( @xmath24 ) are in the range 1928 kev , in agreement with the values usually found for the short bursts in this energy range ( e.g. , aptekar et al .
2001 ) . to concisely show the inadequacy of the ottb model over the full band , we show a fit to the sum of the spectra of all bursts ( average spectrum , top panel of fig .
[ fit1 ] ) with @xmath1 fixed at the maximum value within our range , 5@xmath1810@xmath19 @xmath3 ( @xmath22 = 4.1 for 30 d.o.f . ) . if @xmath1 is left free to vary , a marginally good fit is found ( @xmath22 = 1.22 for 29 d.o.f . ) , but with an unrealistic value of @xmath1 ( 13.2@xmath25 @xmath21 @xmath3 , with kt=23.1@xmath26 kev ) . a back extrapolation of the fit to the pds data only severely overestimates the observed flux at lower photon energies ( fig .
[ fit1 ] , bottom panel ) .
we conclude that the ottb model fails to fit the broad - band burst spectra of sgr 1900 + 14 .
next , we attempted to find an acceptable model for the burst spectra that fits both the high and low energies
. a single bb spectrum , which is characterized by a greater curvature , does not give acceptable best fits for either the individual or the average _
bepposax _ burst spectra ( @xmath22 @xmath23 2.8 ) .
another possible thermal model is the sum of two bbs , as used by olive et al .
( 2003 ) to fit the 7150 kev fregate spectral data of the 2001 july 2 burst from this source . when applied either to the individual burst spectra or to the average spectrum it yields a statistically acceptable @xmath22 .
the best fit parameters are given in table [ fit2bb ] .
the fit with this model to the average spectrum is shown in the upper panel of figure [ fit2 ] .
the dispersion of the two bb temperatures is small , consistent within statistics : for the low temperature bb we found a mean @xmath27 value of 3.23 kev with a standard deviation of 0.56 kev , while the high temperature component has a mean @xmath28=9.65 kev and a standard deviation of 0.95 kev .
such a narrow distribution of spectral parameters raises the question of whether the 10 bursts are all consistent with the same spectral parameters .
we therefore performed a simultaneous fit to all the burst spectra linking the @xmath1 , @xmath29 and @xmath30 between the individual bursts .
the fit is satisfactory ( line `` joint fit '' in table [ fit2bb ] ) with parameters consistent with those derived from the fit to the average spectrum , as well as with the mean of the parameters derived from the fits to the individual spectra .
interestingly , the ratio of the bolometric luminosities of the two bbs is approximately constant for the different bursts ( table [ fit2bb ] and fig .
[ corr ] ) .
we note also that the best - fit parameters for the spectrum of the 2001 july 2 event obtained by olive et al .
( 2003 ) using the same model are nicely consistent with our temperature distributions ( @xmath31=4.15 kev and @xmath32=10.4 kev ) .
finally , we also tested a cut - off power law model .
the best - fit parameters for both individual and average burst spectra are given in table [ fitplc ] , for our restricted ranges of @xmath1 .
the fit results are satisfactory , especially for larger @xmath1 values near the maximum of our range . as was the case for the two bbs model
, we note that the spectral parameters are rather clustered .
the cut - off energy distribution is centered at @xmath33 = 15.8 kev , with a standard deviation of only 2.3 kev and the mean spectral index @xmath34 is 0.4 , with a dispersion of 0.2 . also for this case
, a joint fit of all the spectra with linked fit parameters ( except for the normalizations ) provided satisfactory results ( line joint fit " in table [ fitplc ] ) , consistent with the results on the average spectrum , as well as with the mean of the values found in the individual fits .
the common , short bursts from sgr 1900 + 14 are ordinarily detected by the wide field - of - view experiments typically working above 15 - 20 kev or more .
this led to a good empirical description of their spectra using the ottb model , however , often criticized as non - physical when applied in this context ( e.g. , fenimore et al . 1994 ) .
we have now demonstrated for the short bursts from sgr 1900 + 14 that the ottb law is not acceptable even at an empirical level when the energy spectrum is studied across a broader energy range .
a similar result was obtained by fenimore et al .
( 1994 ) for sgr 1806 - 20 .
we tested three alternative spectral models , still on phenomenological grounds , following an approach of simplicity .
a critical spectral parameter for all models is the neutral absorption column .
leaving @xmath1 free , the ottb model is marginally acceptable , but the column density required is much larger than what is found during the persistent x - ray emission seconds before and after the burst . as a conservative approach
, we therefore decided to constrain in all our fits an equivalent hydrogen column in the range of what has been measured for the persistent source . with this assumption in mind
, we found that our best - fit models are two bbs and a cut - off power law .
we have no strong statistical or physical argument to support one or the other .
however , we note that the goodness of the fits with the cut - off power - law model is more critically sensitive to the choice of @xmath1 . in particular ,
as shown in table [ fitplc ] , the fits to the burst spectra converge to the highest @xmath1 values allowed in the range .
similar to the ottb model , the @xmath22 increases as the @xmath1 decreases .
should the @xmath1 be constrained to small values from future measurements , this would possibly affect the validity of the cut - off power law as a spectral model for the bursts .
a striking property of both models is the clustering of the parameters governing the spectral shape ( i.e. the temperatures of the two bb model , and @xmath35 and @xmath36 for the power law with an exponential cut - off ) .
consistent with previous studies , this shows the uniformity of the sgr burst spectra with varying burst intensities .
interestingly , for the two bb model , the luminosity emitted by the two components is consistent with a constant ratio , showing that both components scale in intensity for brighter bursts .
as for the cut - off power law model , it has to be noted that the photon index obtained in all our fits is always significantly smaller than what is expected for an energy distribution of particles in thermal equilibrium ( @xmath36=1 ) .
it is very interesting to note that the short bursts from sgr 1806 - 20 also possess a similar spectral shape as the ones we analyzed here from sgr 1900 + 14 .
however , fenimore et al . ( 1994 ) measure an average column density of 1.1@xmath1810@xmath37 @xmath3 using an ottb model with @xmath3822 kev .
this column density is almost an order of magnitude larger than the value we measure for sgr 1900 + 14using the same model , with similar @xmath24 .
thus , it appears that the roll - over at low energies in sgr 1806@xmath820 is more severe than it is for sgr 1900 + 14 .
although perhaps quantitatively different , the spectral similarities for sgr 1900 + 14 and sgr 1806 - 20 may reflect a _ universal _ emission mechanism for short bursts independent of the specific source .
this assertion would need support from similar analysis of short bursts from the other sgr sources .
it is worth noting that such a universality is already suggested in the mechanism for the emission of the giant flares ( e.g. , mazets et al . 1999 , feroci et al .
2001 ) for the two cases known so far from sgr 0526 - 66 and sgr 1900 + 14 .
two conclusions that we can draw from our observations are : _ i _ ) the spectral shape of the individual bursts remains stable independent of the brightness , duration and light curve complexity of the events ; _ ii _ ) although we have identified two spectral models that provide a good description of the data , it remains to be seen whether these models truly reflect the physics behind the emission mechanism .
currently , the are no detailed predictions on the spectra of sgr bursts .
a discussion on the physical interpretation of our results is beyond the scope of this _ letter _ and we defer it to a future work .
we would like to thank enrico costa and dario trevese for useful discussions , and the referee ( dr .
patrick slane ) for useful comments .
* aptekar , r. , et al .
2001 , apjs , 137 , 227 * boella , g. , et al .
1997 , a@xmath39as , 122 , 327 * cline , t. , mazets , e. and golenetskii , s. 1998 , iau circ .
7002 * fenimore , e.e . ,
laros , j.g . , and ulmer , a. , 1994 , apj , 432 , 742 * feroci , m. , frontera , f. , costa , e. et al . 1999 ,
apj , 515 , l9 * feroci , m. , hurley , k. , duncan , r.c . , and thompson , c. , 2001 , apj , 549 , 1021 * feroci , m. , mereghetti , s. , woods , p. et al .
2003 , apj , 596 , 470 * fiore , f. , guainazzi , m. , and grandi , p. , 1999
, `` cookbook for bepposax nfi spectral analysis '' , available at `` http://www.asdc.asi.it '' * guidorzi , c. , et al . , 2004 , a&a , 416 , 297 * hurley , k. , et al .
1996 , apj , 463 , l13 * hurley , k. , et al . 1999a , apj , 510 , l107 * hurley , k. , cline , t. , mazets , e. et al . 1999b , nature , 397 , 41 * hurley , k. , et al .
1999c , apj , 510 , l111 * kouveliotou , c. et al .
1993 , nature , 362 , 728 * kouveliotou , c. et al . 1998 ,
6929 * kouveliotou , c. , strohmayer , t.e . , hurley , k. et al .
1999 , apj , 510 , l115 * kouveliotou , c. et al .
2001 , apj , 558 , l47 * laros , j.g . ,
1986 , nature , 322 , 152 * lenters , g.t .
2003 , apj , 587 , 761 * mazets , e.p .
, golenetskii , s.v . , guriyan yu.a .
1979 , sov .
lett . , 5(6 ) , 343 * mazets , e.p .
, et al . 1999 , astron .
25(10 ) , 635 * murakami , t. , et al .
1999 , apj , 510 , l119 * olive , j - f . , et al . , 2003 , in proc
`` gamma ray bursts and afterglow astronomy 2001 '' , g.r . ricker and r.k .
vanderspeck eds .
, aip 662 , p. 82 . * strohmayer , t.e . , and ibrahim , a. , 1998 , in proc .
`` 4th huntsville symp . on gamma ray bursts '' , c.a .
meegan , r.d .
preece and t.m .
koshut eds . , aip 428 , p. 947 .
* vasisht , g. , kulkarni , s.r . ,
frail , d.a . , and greiner , j. , 1994 , apj , 431 , l35 * woods , p.m. , et al . , 1999 , apjl , 518 , l103 * woods , p.m. , kouveliotou , c. , ggs , e. et al . , 2003 , apj , 596 , 464 * woods , p.m. , et al . , 2004 , apj , 605 , 378 lcccccccc a & 60328 & 20 & 153/151 & 289/270 & 0.67 & 199.6 & 55.60 & 5.13 + b & 65801 & 73/36 & @xmath23611/469 & @xmath231008/949 & 0.23 & 200.4 & 81.25 & @xmath2317.95 + c & 82558 & 14 & 119/107 & 208/185 & 1.16 & 430.8 & 21.65 & 3.32 + d & 91223 &
n / a & 293/284 & 621/607 & 1.45 & 682.3 & 45.71 & 10.63 + e & 95772 & n / a & 218/209 & 384/361 & 0.78 & 499.8 & 39.80 & 6.71 + f & 95794 & n / a & 339/337 & 707/680 & 0.14 & 250.1 & 80.11 & 12.42 + g & 102409 & n / a & @xmath23359/317 & @xmath231363/782 & 0.08 & 100.0 & @xmath23150.83 & @xmath2313.20 + h & 111159 & 17 & 175/171 & 441/409 & 0.07 & 110.8 & 85.74 & 7.05 + i & 134665 & 13 & 158/156 & 325/298 & 0.18 & 271.0 & 47.57 & 5.82 + j & 134687 & 22 & 243/239 & 394/379 & 0.25 & 381.0 & 34.40 & 7.47 + [ burstlist ] lccccccccc a & 2.0 & 3.2@xmath40 & 13.7 & 9.3@xmath41 & 1.8 & 0.95/14 & 1.15 & 0.90@xmath42 + b & 2.5 & 3.4@xmath43 & 23.5 & 9.9@xmath44 & 2.7 & 0.83/43 & 0.89 & 1.00@xmath45 + c & 2.1 & 2.1@xmath46 & 13.5 & 7.9@xmath44 & 1.5 & 0.64/9 & 0.71 & 0.83@xmath47 + d & 2.0 & 3.3@xmath40 & 9.9 & 10.2@xmath48 & 1.1 & 1.27/29 & 1.59 & 0.77@xmath49 + e & 2.0 & 2.7@xmath50 & 12.2 & 8.6@xmath51 & 1.6 & 1.52/18 & 1.80 & 0.82@xmath49 + f & 2.0 & 3.6@xmath52 & 16.5 & 10.3@xmath53 & 1.9 & 0.87/32 & 1.11 & 0.90@xmath54 + g & 2.0 & 3.7@xmath55 & 24.4 & 8.9@xmath56 & 4.2 & 0.83/31 & 1.02 & 0.83@xmath57 + h & 2.0 & 4.1@xmath58 & 15.7 & 10.5@xmath25 & 2.1 & 0.54/20 & 0.69 & 0.83@xmath54 + i & 2.5 & 3.3@xmath46 & 11.9 & 11.0@xmath59 & 1.2 & 1.04/13 & 1.09 & 0.76@xmath60 + j & 2.0 & 3.0@xmath61 & 13.9 & 9.9@xmath62 & 1.3 & 0.92/19 & 1.31 & 1.03@xmath63 + average spectrum & 2.0 & 3.4@xmath64 & 13.4 & 9.33@xmath65 & 1.9 & 0.62/27 & 0.92 & 0.85@xmath49 + joint fit & 2.0 & 3.34@xmath66 & n / a & 9.47@xmath67 & n / a & 1.02/255 & 1.19 &
n / a + [ fit2bb ] lccccccccccc a & 3.0 & 0.30@xmath68 & 14.9@xmath69 & 0.99/15 & & & & 4.30 & 0.31@xmath70 & 14.6@xmath71 & 0.89 + b & 3.0 & -0.01@xmath72 & 12.1@xmath73 & 1.60/44 & & & & 5.0 & 0.28@xmath74 & 14.1@xmath75 & 1.3 + c & 3.0 & 0.44@xmath76 & 16.2@xmath77 & 1.05/10 & & & & 4.0 & 0.43@xmath78 & 15.9@xmath79 & 1.06 + d & 3.0 & 0.38@xmath80 & 16.8@xmath81 & 0.97/30 & & & & 4.08 & 0.51@xmath82 & 18.0@xmath83 & 0.93 + e & 2.85 & 0.25@xmath84 & 14.5@xmath85 & 1.44/19 & & & & 4.0 & 0.27@xmath86 & 14.4@xmath87 & 1.48 + f & 3.0 & 0.22@xmath88 & 14.5@xmath89 & 1.00/33 & & & & 5.0 & 0.50@xmath90 & 16.6@xmath91 & 0.88 + g & 3.0 & 0.004@xmath92 & 12.2@xmath93 & 1.01/32 & & & & 4.27 & 0.04@xmath94 & 12.2@xmath95 & 0.94 + h & 3.0 & 0.14@xmath96 & 13.6@xmath97 & 0.81/22 & & & & 5.0 & 0.25@xmath98 & 13.9@xmath99 & 0.68 + i & 3.0 & 0.43@xmath100 & 19.1@xmath101 & 1.77/14 & & & & 5.0 & 0.46@xmath102 & 18.3@xmath103 & 1.37 + j & 3.0 & 0.58@xmath104 & 18.1@xmath105 & 0.87/20 & & & & 4.63 & 0.72@xmath106 & 19.5@xmath107 & 0.81 + average spectrum & 3.0 & 0.17@xmath108 & 13.5@xmath109 & 1.13/28 & & & & 5.0 & 0.42@xmath104 & 15.3@xmath110 & 0.82 + joint fit & 3.0 & 0.18@xmath111 & 13.9@xmath112 & 1.16/265 & & & & 5.0 & 0.40@xmath113 & 15.5@xmath65 & 1.04 + [ fitplc ]
|
we report on the x - ray spectral properties of 10 short bursts from sgr 1900 + 14 observed with the narrow field instruments onboard _
bepposax _ in the hours following the intermediate flare of 2001 april 18 .
burst durations are typically shorter than 1 s , and often show significant temporal structure on time scales as short as @xmath010 ms .
burst spectra from the mecs and pds instruments were fit across an energy range from 1.5 to above 100 kev .
we fit several spectral models and assumed @xmath1 values smaller than 5@xmath2 @xmath3 , as derived from observations in the persistent emission .
our results show that the widely used _ optically thin thermal bremsstrahlung _ law provides acceptable spectral fits for energies higher than 15 kev , but severely overestimated the flux at lower energies .
similar behavior had been observed several years ago in short bursts from sgr 1806 - 20 , suggesting that the rollover of the spectrum at low energies is a universal property of this class of sources .
alternative spectral models - such as two blackbodies or a cut - off power law - provide significantly better fits to the broad band spectral data , and show that all the ten bursts have spectra consistent with the same spectral shape .
|
fluctuation relations ( frs ) provide general results about the full work ( or entropy production ) distributions in non - equilibrium systems . however , in many cases the full work is not measurable and only partial work measurements are possible .
the latter do not fulfill a fr and can not be used to extract free energy differences from irreversible work measurements .
we propose a new application of frs to infer the full work distribution from partial work measurements .
we prove this new type of inference using dual - trap optical tweezers where two forces ( one per trap ) are measured , allowing us to derive full and partial work distributions .
we derive a set of results of direct interest to single molecule scientists and , more in general , to physicists and biophysicists .
fluctuation relations ( frs ) are mathematical equations connecting non equilibrium work measurements to equilibium free energy differences .
frs , such as the jarzynski equality ( je ) or the crooks fluctuation relation ( cfr ) have become a valuable tool in single - molecule biophysics where they are used to measure folding free energies from irreversible pulling experiments @xcite .
such measurements have been carried out with laser optical tweezers on different nucleic acid structures such as hairpins @xcite , g - quadruplexes @xcite and proteins @xcite , and with atomic force microscopes on proteins @xcite and
bi - molecular complexes @xcite . an important issue regarding frs is the correct definition of work , which rests on the correct identification of configurational variables and control parameters . in the single - trap optical tweezers configuration
this issue has been thoroughly discussed @xcite .
+ the situation about how to correctly measure work in small systems becomes subtle when there are different forces applied to the system . in this case theory
gives the prescription to correctly define the work ( @xmath0 ) for a given trajectory ( @xmath1 ) : integrate the generalized force ( @xmath2 ) ( conjugated to the control parameter , @xmath3 ) over @xmath3 along @xmath1 , @xmath4 .
however , in some cases one can not measure the proper generalized force or has limited experimental access to partial sources of the entropy production leading to what we call incorrect or partial work measurements .
a remarkable example of this situation are dual - trap setups , mostly used as the high - resolution tool for single molecule studies ( fig .
[ figure : opts]a ) . in this case
a dumbbell formed by a molecule tethered between two optically trapped beads is manipulated by moving one trap relative to the other . in this setup two different forces ( one per trap )
can be measured and at least two different work definitions are possible . in equilibrium conditions ,
i.e. when the traps are not moved both forces are equivalent : the forces acting on each bead have equal magnitude and opposite sign . on the contrary , in pulling experiments , where one trap is at rest ( with respect to water ) while the other is moved ,
the two forces become inequivalent .
this is so because the center of mass of the dumbbell drifts and the beads are affected by different viscous drags ( purple arrows in fig .
[ figure : opts]b ) . in such conditions
theory prescribes that the full thermodynamic work must be defined on the force measured at the moving trap whereas the force measured in the trap at rest ( with respect to water ) leads to a _ partial work measurement _ which , as we show below , entails a systematic error in free energy estimates .
the difference between both works equals the dissipation by the center of mass of the dumbbell , which is only correctly accounted for in the correct work definition .
+ in this paper we combine theory and experiments in a dual - trap setup to demonstrate several results .
first , we show that if the wrong work definition is used , free energy estimates will be flawed .
the error is especially severe in the case of unidirectional work estimates ( e.g. je ) while it influences bi - directional estimates to a lesser extent .
this fact is not purely academical : measuring the force in the trap at rest is experimentally easier in dual - trap setups and , in fact , many groups choose to do so @xcite .
for example if different lasers are used for trapping and detection , measuring the force in the moving trap poses the additional challenge of keeping the trapping and detection lasers aligned while moving them .
second , we show how it is possible , by using the cfr , to _ infer _ the the full work distribution from partial work measurements .
we demonstrate this new type of inference in our dual - trap setup by showing how to reconstruct the correct work distribution ( i.e the one we would have measured in the moving optical trap ) from partial work measurements in the wrong optical trap ( i.e. the one at rest with respect to water ) by using the cfr . in particular , for symmetric setups the correct work distribution can be directly inferred by simply shifting the partial work distribution . in asymmetric setups inference
is still possible in the framework of a gaussian approximation , but the knowledge of some equilibrium properties of the system is still required .
this type of inference should be seen an example of a more general application of frs which aims at extracting information about the total entropy production of a nonequilibrium system from partial entropy production measurements .
this allows us to determine the average power dissipated by the center of mass of the dumbbell , from which we can extract the corresponding hydrodynamic coefficient thereby avoiding direct hydrodynamic measurements .
moreover , we argue this type of inference might find applicability to future biophysical experiments where the sources of entropy production are not directly measurable , e.g. atp - dependent motor translocation where the hydrolysis reaction cycle can not be followed one atp at a time .
finally we show how , in symmetric setups , the work definition satisfying the cfr is not unique .
in particular the _ differential work _ , based on differential force measurements @xcite , still satisfies the cfr and leads to the least biased free energy estimates .
the distinguishing feature of this work definition is that it completely filters out the dissipation due to the motion of the center of mass of the dumbbell . throughout the paper , and for simplicity and pedagogical reasons , most of the derivations , calculations and experiments are shown for symmetric dual - trap setups whereas the asymmetric case is discussed towards the end of the paper .
in dual - trap setups a molecule is stretched by two optical traps , the control parameter @xmath3 being the trap - to - trap distance .
let a , b denote the two optical traps .
when one trap ( say trap a ) is moved with respect to the bath while trap b is at rest , suitable configurational variables are the positions of the beads , both measured from the center of the trap at rest ( trap b ) .
we shall denote these variables by @xmath5 and @xmath6 , fig .
[ figure : opts]b .
the total energy of the system is composed of three terms : @xmath7 where the quadratic terms model the potential of the optical trap and @xmath8 describes the properties of the tether .
however one could measure the positions of the beads and the trap - to - trap distance in the moving frame of trap a ( @xmath9 , @xmath10 and @xmath11 in fig .
[ figure : opts]b ) , and the potential energy in eq . would be written as : @xmath12 where @xmath13 ( @xmath9 and @xmath10 are negative ) .
central to our analysis will be the equation connecting the potential @xmath14 and the work performed on the system by changing the control parameter @xcite : @xmath15 from eq .
we get : @xmath16 with @xmath17 . inserting @xmath18 instead of @xmath14 in gives : @xmath19 despite of their similarity we will show that @xmath20 and @xmath21 are remarkably different .
in fact , from the reference frame of trap a , the bath is seen to flow with velocity @xmath22 ( fig . [
figure : opts]b ) . because of this flow an experiment in which trap a is moved is not galilean equivalent to one in which trap b is moved . in the presence of a flow the connection between potential and work , eq . , is not valid anymore .
in fact thermodynamic work measurements must be based on the force measured in the trap being moved .
this fact has been discussed in @xcite , yet the implications to single - molecule experiments have never been pointed out . summarizing , if trap a is moved and b is at rest with respect to water , using the work @xmath20 ( eq . ) in the je leads to correct free energy estimates whereas using @xmath21 ( eq . ) in the je leads to a systematic error .
below we quantify such error in detail .
the difference between @xmath20 and @xmath21 can be readily discussed in symmetric setups ( @xmath23 ) where calculations are much simpler . to do this
we switch to a new coordinate system : @xmath24 . here
@xmath25 is the position of the geometric center of the dumbbell , while @xmath26 is the differential coordinate @xcite . in this coordinate system the potential eq .
reads @xmath27 : @xmath28 the potential energy term @xmath29 associated to @xmath25 is that of a moving trap , a problem that has been addressed both with experiments and theory @xcite , while the potential energy term @xmath30 associated to @xmath26 corresponds to pulling experiments performed using a single trap and a fixed point .
the dumbbell is in contact with an isothermal bath where the equilibrium state is described by the boltzmann distribution and the corresponding partition function ( we assume a weak system - environment coupling , a situation satisfied in our experimental conditions , section s2 in the si ) .
consequently , the two degrees of freedom are uncoupled and the total partition function for the system factorizes : @xmath31 with @xmath32 , @xmath33 being the temperature and @xmath34 being boltzmann constant . as a consequence free energy changes in the system can be decomposed into two contributions : @xmath35 work can also be decomposed into two contributions , each regarding one of the subsystems : @xmath36 here @xmath37 contains the work done in stretching the molecule while @xmath38 is pure dissipation due to the movement of the center of mass of the dumbbell .
note that : @xmath39 which shows that the difference between @xmath20 and @xmath21 is entirely due to @xmath38 .
the je holds for @xmath20 , the standard work definition , so that @xmath40 inserting eqs . ,
in we get : @xmath41 .
in symmetric setups @xmath38 and @xmath37 are independently distributed random variables ( see sections s1 and s3 in the si ) and we can conclude that : @xmath42 using the je on both @xmath20 and @xmath21 we get two different free energy estimates : @xmath43 ( eq . ) and @xmath44 . the error @xmath45 committed by using @xmath21 instead of @xmath20 can be quantified as : @xmath46 from eqs . , and again using the fact @xmath38 and @xmath37
are independently distributed random variables we get @xmath47 and similarly for @xmath21 . as a consequence @xmath48 and eq .
is reduced to @xmath49 since @xmath25 is subject to a quadratic potential ( eq . ) , we expect @xmath38 to be a gaussian random variable .
this is not true in general for @xmath37 given that @xmath26 feels the nonlinear term @xmath8 . for gaussian random variables we have : @xmath50 where by @xmath51 we denote the variance of @xmath38 . moreover dragging a trapped bead in a fluid causes no free energy change ,
so that : @xmath52 or @xmath53 . inserting eqs . and in eq .
we get : @xmath54 equation gives @xmath55 showing that @xmath56 is lower than @xmath43 ( eq . ) .
interestingly enough , using @xmath21 instead of @xmath20 in the je leads to free energy estimates in apparent violation of the second law .
the error on free energy estimates obtained using @xmath21 instead of @xmath20 is proportional to the mean work performed on the center of the dumbbell .
this mean work @xmath57 is just the mean friction force times the total trap displacement @xmath58 : @xmath59 where @xmath60 is the friction coefficient of the drag force opposing the movement of the geometric center of the dumbbell .
the value of @xmath60 can be independently obtained from equilibrium measurements @xcite ( sections s4,s5 in the si ) .
equations and show that free energy estimates based on the standard work @xmath20 and the differential work @xmath37 are equivalent : @xmath61 we stress that this is only true for symmetric setups where @xmath38 and @xmath37 are independent random variables .
the case of asymmetric setups is discussed further below . therefore @xmath37 can be used for free energy determination , as it has been done in @xcite , although without discussion .
equation does only hold when the number of work measurements , @xmath62 , tends to infinity . in all practical cases we deal with finite @xmath62 and
the jarzynski estimator is biased @xcite .
the bias is strongly linked to the typical dissipation @xmath63 and a reliable estimate of free energy differences requires a number of work measurements which scales as @xmath64 @xcite , so that even a small reduction in @xmath63 entails a considerable improvement in the convergence of free energy estimators .
moreover the bias is superadditive .
let us consider for simplicity gaussian work distributions . in this case
the bias , @xmath65 , in the large @xmath62 limit is a function of the variance of the distribution @xmath66 and of @xmath62 @xcite : @xmath67 @xmath65 is a convex function of @xmath68 , and is superadditive i.e. @xmath69 .
this means that , should the work be the sum of two independent gaussian contributions , the bias on the sum is greater than the sum of the biases .
although eq .
was derived under strong assumptions , superadditivity does also hold for other theoretical expressions for the bias and has been checked in our experimental data ( see below ) .
let us introduce the following jarzynski estimators for finite @xmath62 : @xmath70 and the corresponding bias functions : @xmath71 since @xmath72 , superadditivity guarantees : @xmath73 because of eq .
differential work measurements always improve the convergence of free energy estimates in dual - trap setups .
this is especially important in all those cases in which bidirectional methods ( e.g. the cfr ) can not be used and one has to employ unidirectional methods .
the theory discussed so far has been put to test in a series of pulling experiments performed in a recently developed dual - trap optical tweezers setup which directly measures force in each trap @xcite .
the setup can move the two optical traps independently and measure their relative position with sub - nanometer accuracy , giving direct access to both @xmath20 and @xmath21 . in these experiments 3
kb ds - dna tethers ( @xmath74 @xmath75 m in contour length ) were stretched between 1 and 3 pn ( fig .
[ figure : opts]c ) in a symmetric dual - trap setup ( @xmath76pn / nm ) using 4 @xmath75 m silica beads as force probes .
the experiments were performed moving one of the two traps ( trap a ) with respect to the lab frame and leaving trap b at rest .
all experiments were performed in pbs buffer ( ph 7.4 , 1 m nacl ) .
we chose cyclical protocols ( @xmath77 , where @xmath78 is the total duration of the cyclic protocol ) .
the excursion of the control parameter , @xmath79 , was varied between 200 , 400 and 600 nm , while the pulling speed was varied between [email protected] , [email protected] and 7.2@xmath801 @xmath75 m/s . given the force - distance curves the total dissipation
along cycles was measured : @xmath81 the cfr @xcite is a symmetry relation between the work distribution associated to the forward ( @xmath82 ) and time reversed ( @xmath83 ) protocols : @xmath84 in the case of cyclic protocols @xmath85 and @xmath86 so that the cfr takes the form : @xmath87 such symmetry of the probability distribution for @xmath88 can be directly tested in cases where negative dissipation events are observed . in fig . [
figure : multi]a we show measured work histograms ( solid points , left hand side of eq . ) and reconstructed histograms ( open points , right hand side of eq . ) . if eq .
is fulfilled then the measured and reconstructed histograms match each other . a quantitative measure of the deviation from eq .
can be obtained from the ratio @xmath89 , as shown in fig .
[ figure : multi]b .
experimental data shows that @xmath88 fulfills the fr whereas @xmath90 does not . in fig .
[ figure : multi]c we show that the pdfs of @xmath91 and @xmath92 , with : @xmath93 are experimentally found to satisfy a fr as in eq . .
@xmath92 is just the differential work , @xmath37 , eq .
evaluated on a cyclic protocol , whereas @xmath91 is the dissipation due to the movement of the center of mass of the dumbbell .
summarizing , although in general @xmath20 is the only observable we expect to fulfill a fr , in symmetric setups two new frs emerge , for @xmath37 and @xmath38 . in fig .
[ figure : seifert]a , b we compare the predictions of eqs . , with experimental results for different pulling speeds , @xmath94 , and different displacements @xmath58 .
equation must be used to correct free energy estimates obtained in all those dual - trap setups which do not measure the force applied by the trap which is being moved ( as in @xcite ) .
the advantages of using @xmath37 in free energy estimates are shown in fig .
[ figure : seifert]c .
there we show the convergence of the jarzynski estimator with sample size for the cycles in fig . [
figure : opts]c .
being evaluated over cycles , the expected free energy change is zero . the convergence of the estimator is faster for @xmath37 than for @xmath20 in the three cases ( eq . ) .
the effect is enhanced in our experiments by the high pulling speed ( in the range 1 - 7 @xmath75 m ) and by the large bead radius ( 2 @xmath75 m ) .
let us note that due to the finite lifetime of molecular tethers and unavoidable drift effects , raising the pulling speed is a convenient strategy to improve the quality of free energy estimates .
similar results have been found also at low pulling speeds where , again , @xmath21 does not satisfy the cfr ( section s7 in the si ) .
fluctuation theorems are used to extract folding free energies for nucleic acid secondary structures or proteins .
we further tested the different work definitions by performing pulling experiments on a 20bp dna hairpin ( fig .
4a ) at a @xmath95 @xmath75 m/s pulling speed in the same dual - trap setup as in the previous dsdna experiments . in this case
the work performed during the unfolding and refolding of the molecule were considered separately , as it is customary for free energy determination . in fig .
4b we present forward and reverse work histograms for @xmath96 and @xmath38 . again
@xmath20 and @xmath37 both fulfill the cfr but @xmath20 shows higher dissipation than @xmath37 , resulting in slower convergence of unidirectional free energy estimators ( fig .
the difference between unidirectional free energy estimates based on @xmath20 and @xmath37 is in this case @xmath97 @xmath98 . as previously discussed for double - stranded dna ( fig .
2a ) , @xmath21 does not fulfill the cfr and , as a consequence , unidirectional free energy estimates based on @xmath21 are flawed . in our experimental conditions the error committed by using the wrong work definition is again positive and equal to @xmath99 @xmath100 .
as previously discussed this leads to a negative average dissipated work , apparently violating the second law .
it must be noted that the difference in free energy estimates based on @xmath37 and @xmath20 is a finite - size effect , whose magnitude decreases when an increasing number of work measurements is considered . on the contrary
the error committed by using @xmath21 does not vanish by increasing the number of work measurements .
it can be noted from fig 4b that , although @xmath21 does not fulfill the cfr and gives wrong unidirectional estimates , its forward and reverse distributions apparently cross at @xmath101 within the experimental error .
although this could be used for free energy determination the result should be taken with caution as we have no general proof that this should happen in all cases .
the je and cfr are statements on the statistics of the total dissipation in irreversible thermodynamic transformations . the need to measure the total dissipation limits the range of applicability of these and other frs .
for example testing frs concerning the dynamics of molecular motors would need the simultaneous measurement of both the work performed by the motor and the number of hydrolyzed atps . here
we demonstrate that , at least in some cases , a different approach is possible .
let us start by considering the simple case of symmetric setups as developed in the previous sections . in the experiments discussed
so far there are two sources of dissipation that we were able to measure and characterize separately : the motion of the dumbbell and the dissipation of the differential coordinate.we already learnt that @xmath20 satisfies a fr while @xmath21 does not
. moreover we know that @xmath102 and @xmath103 , and that , being @xmath38 and @xmath37 uncorrelated random variables , @xmath20 and @xmath21 have the same variance .
imagine now to have only partial information on the system .
for example , one could be able to measure force only in the trap at rest , as many experimental setups do . with this information , and in absence of any guiding principle ,
no statement about the total entropy production is possible .
we will find such guiding principle if we assume the fr to hold for @xmath20 .
knowing that @xmath20 and @xmath21 have the same variance , we just have to shift the work distribution @xmath104 by @xmath105 to get a new distribution that satisfies the cfr . in practice
this is done starting from the set of @xmath21 values and tuning the value of @xmath106 ( fig .
[ figure5]a ) until @xmath107 fulfills the cfr ( fig .
[ figure5]b ) . in the case of the hairpin
, this same shifting procedure is operated for both forward and reverse work distributions . again
the value of @xmath106 is tuned ( fig .
[ figure5]c ) until the cfr symmetry is recovered ( fig .
[ figure5]d ) . the unique value of @xmath106 that restores the validity of the cfr equals the average work dissipated by the motion of center of mass of the dumbbell , giving the hydrodynamic coefficient @xmath60 via eq .
let us note that , once the work distribution in the correct trap ( i.e. the moving trap ) has been recovered , then we could also extract the correct free energy difference ( the value of @xmath43 , fig .
[ figure5]e ) and infer the distribution for the differential work @xmath108 by deconvolution .
+ the extension of this analysis to the asymmetric case is more complex but equally interesting ( fig .
the decomposition of @xmath96 in @xmath38 and @xmath37 ( eq . ) is still possible although @xmath109 are not uncorrelated variables anymore and neither @xmath38 nor @xmath37 satisfy a fr . in this general case
only @xmath20 satisfies a fr ( but not @xmath21 , @xmath38 , @xmath37 ) .
remarkably enough , in the framework of a gaussian approximation , it is still possible to infer the correct work distribution @xmath110 out of partial work @xmath21 measurements .
the analysis is presented in section s6 of the si . in this case
it is enough to know the trap and molecular stiffnesses @xmath111 i.e. some equilibrium properties of the system , for a successful inference . to reconstruct @xmath110
both the mean and the variance of @xmath104 must be changed , which can be achieved by doing a convolution between the @xmath104 and a normal distribution : @xmath112 where @xmath113 denotes the convolution operator and @xmath114 is a normal distribution with mean @xmath106 and standard deviation @xmath115 .
starting from a distribution @xmath104 there are infinitely many choices of @xmath106 and @xmath115 which yield a @xmath116 satisfying the cfr .
indeed , let us suppose that the pair @xmath117 is such that @xmath118 satisfies the cfr .
then it is easy to check that @xmath119 will also satisfy the cfr for any @xmath120 ( fig .
in this situation the inference can not rest on the cfr alone .
explicit calculations in the gaussian case ( section s6 in si ) show that variances ( @xmath66 ) and means ( @xmath121 ) of @xmath110 and @xmath104 are related by an asymmetry factor ( af ) , @xmath122 which only depends on _ equilibrium properties _ such as the stiffnesses of the different elements ( section s6.3 in the si ) . knowing the af allows us to select the unique pair @xmath123 such that @xmath124 with @xmath116 satisfying the cfr .
the inference procedure can be described with a very simple formula ( section s6 in the si ) .
the key idea is to proceed as previously done in the case of symmetric setups by just shifting the mean of @xmath104 by a parameter @xmath125 until the cfr is satisfied , i.e. @xmath126 . from the values of @xmath127 and @xmath125 we can reconstruct @xmath110 by using the formulae , @xmath128 the inference procedure for an asymmetric setup is shown in fig .
6c,6d for cyclic ds - dna pulling experiments . for non - cyclic pulls ( @xmath129 )
the procedure can be easily generalized in the line of what has been shown for the case of hairpin in the symmetric setup ( fig .
frs are among the few general exact results in non - equilibrium statistical mechanics .
their validity has been already extensively tested in different systems , ranging from single molecules to single electron transistors , and in different conditions ( steady state dynamics , irreversible transformations between steady states , transient nonequilibrium states ) . at the present stage ,
the main widespread application of fr is free energy recovery from non - equilibrium pulling experiments in the single molecule field .
what we are presenting here is a new application of fr for inference .
all frs are statements about the statistics of the total entropy production in a system plus the environment .
if some part of the entropy production is missed or inadequately considered frs will in general not hold .
this is why , for irreversible transformations between equilibrium states , we have a fr for the dissipated work ( which is the total entropy production ) but not for the dissipated heat ( which is just the entropy production in the environment ) .
the main tenet is now that the violation of frs in a given setting provides useful information : it is an evidence that some contribution to the total entropy production is being missed .
we have given rigorous examples in which the violation of frs can be used to characterize the missing entropy production .
remarkably , in our model system , one could even replace the moving trap by a moving micropipette , an object lacking any measurement capability , and still infer the work distribution exerted by that object on the molecular system ( this extremely asymmetric setup would still be described by eq.([eqaf ] ) , with @xmath130 and @xmath131 ) .
these results open the exciting prospect of extending and applying these ideas to steady state systems , such as molecular motors , to extract useful information about their mechanochemical cycle .
in order to give it a clear and definite meaning to free energy inference we have discussed irreversible transformations between equilibrium states performed with dual - trap optical tweezers . in these experiments a molecular tether
is attached between two beads which are manipulated with two optical traps .
the irreversible transformation is performed by increasing the trap - to - trap distance at a finite speed . in this kind of transformations
the dissipated work equals the total entropy production leading to our first result : in pulling experiments work @xmath20 must be defined on the force measured in the trap which is moved with respect to the thermal bath .
the force measured in the trap at rest gives rise to a work definition , @xmath21 , which does not satisfy the fr and is unsuitable to extract free energy differences .
we have called @xmath21 a partial work measurement because it misses part of the total dissipation .
this result is of direct interest to experimentalists : many optical tweezers setups are designed so that they can only measure @xmath21 .
we have thus imagined a situation in which @xmath21 is measurable while @xmath20 is not and asked the question : can we infer the distribution of @xmath20 from that of @xmath21 ? if the question is asked in full generality , without any system - specific information , the answer is probably negative . knowing only the extent of violation of the fr will be of little use , in general some additional system specific information will be needed for a successful inference .
here we discussed free energy inference in the framework of a gaussian approximation , the extent to which such inference is generally possible should be the subject of future studies .
let us summarize our main results : * a symmetry of the system can be crucial for the inference . for left - right symmetric systems ( as exemplified in our symmetric dual - trap setup )
the @xmath110 can be inferred from @xmath104 just by imposing that the former satisfies the cfr .
when symmetry considerations can not be used , the knowledge of some equilibrium properties of the system may suffice to successfully guide the inference ( such as the stiffnesses for the asymmetric setup ) .
* the inference process can be used both to recover the full dissipation spectrum plus additional information about the _ hidden _ entropy source . in our specific dual - trap example @xmath21 does not account for the dissipation due to the movement of the center - of - mass of the dumbbell and the inference procedure can be seen as a method to measure the associated hydrodynamic drag .
* we stress the benefits of using symmetric dumbbells in single molecule manipulation . in this case an alternative work definition , the _ differential work _
@xmath37 , fulfills the cfr and is thus suitable for free energy measurements .
being @xmath37 less influenced by dissipation than @xmath20 , switching from @xmath20 to @xmath37 ensures faster convergence of unidirectional free energy estimates . for asymmetric setups @xmath37 does not satisfy a fr anymore ( only @xmath20 does ) and
can not be used to extract free energy differences .
a deep understanding of how to correctly define and measure thermodynamic work in small systems ( a long debated question in the past 20 years ) is not just a fine detail for experimentalists and theorists working in the single molecule field , but an essential question pertaining to all areas of modern science interested in energy transfer processes at the nanoscale .
the new added feature of free energy inference discovered in this paper paves the way to apply frs to new problems and contexts .
this remains among the most interesting open problems in this exciting field .
dual - trap setups are preferred in single - molecule applications due to their higher resolution . at equilibrium
the forces exerted by the two traps are statistically equivalent . in non - equilibrium pulling experiments , due to the movement of the center of mass of the dumbbell , the two forces become inequivalent . it has been shown that thermodynamic work must be defined using the force measured in the trap which is being moved with respect to water during the experiment .
work defined on the force measured in the trap at rest does not satisfy the crooks relation , the jarzinsky equality or even the second law .
this has important consequences in free energy determination , as many dual - trap setups can not measure the force in the moving trap and therefore the thermodynamic work . in all these cases
unidirectional free energy estimates are subject to an error that increases with pulling speed .
such errors can be estimated and the free energy corrected when the hydrodynamic parameters of the dumbbell are known .
one of the salient features of equilibrium measurements in dual - trap setups is the possibility of differential detection . in equilibrium experiments
this allows to filter out the fluctuations due to the movement of the center of mass of the dumbbell .
differential detection proves to be useful in irreversible pulling experiments : differential work @xmath132 measurements fulfill fluctuation relations and are less influenced by dissipation .
this is achieved by filtering out the dissipation due to the movement of the center of mass of the dumbbell .
the decrease in dissipation obtained by switching from @xmath20 to @xmath37 ensures faster convergence of unidirectional free energy estimates and @xmath37 should be preferentially used to extract free energy differences in single molecule pulling experiments .
finally , a deep understanding of how to correctly define and measure thermodynamic work in small systems ( a long debated question in the past 20 years ) is not just a fine detail for experimentalists and theorists working in the single molecule field , but an essential issue in all areas of modern science interested in energy transfer processes at the nanoscale .
* buffers and dna substrates . *
all experiments were performed in pbs buffer 1 m nacl at 25@xmath133 ; 1 mg / ml bsa was added to passivate the surfaces and avoid nonspecific interactions .
the dsdna tether was obtained ligating a 1 kb segment to a biotin - labeled oligo at one end and a dig - labeled oligo at the other end .
the dna hairpin used in the experiments has short ( 20bp ) molecular handles and was synthesized by hybridization and ligating three different oligos .
one oligo is biotin - labeled and a second is dig - labeled .
details of the systhesis procedure are given in @xcite .
* optical tweezers assay .
* measurements were performed with a highly stable miniaturized laser tweezers in the dual trap mode @xcite .
this instrument directly measures forces by linear momentum conservation . in all experiments we used silica beads with 4 @xmath75 m diameter , which give a maximum trapping force around @xmath134 pn .
data is acquired at 1 khz .
we thank a. alemany and m. palassini for a critical reading of the manuscript .
fr is supported by icrea academia 2008 grant .
the research leading to these results has received funding from the european union seventh framework programme ( fp7/2007 - 2013 ) under grant agreement @xmath135 308850 infernos .
10 hummer g , szabo a ( 2001 ) free energy reconstruction from nonequilibrium single - molecule pulling experiments .
_ proceedings of the national academy of sciences _ 98:36583661 .
hummer g , szabo a ( 2010 ) free energy profiles from single - molecule pulling experiments .
_ proceedings of the national academy of sciences _ 107:2144121446 .
liphardt j , dumont s , smith sb , tinoco i , bustamante c ( 2002 ) equilibrium information from nonequilibrium measurements in an experimental test of jarzynski s equality . _ science _ 296:18321835 .
, et al .
( 2005 ) verification of the crooks fluctuation theorem and recovery of rna folding free energies . _
nature _ 437:231234 .
gupta a , et al . ( 2011 ) experimental validation of free energy - landscape reconstruction from non - equilibrium single - molecule force spectroscopy measurements .
_ nature physics _ 7:631634 .
alemany a , mossa a , junier i , ritort f ( 2012 ) experimental free energy measurements of kinetic molecular states using fluctuation theorems . _ nature physics _
dhakal s , et al .
( 2010 ) coexistence of an ilpr i - motif and a partially folded structure with comparable mechanical stability revealed at the single - molecule level .
_ journal of the american chemical society _ 132:89918997 .
dhakal s , et al .
( 2013 ) structural and mechanical properties of individual human telomeric g - quadruplexes in molecularly crowded solutions . _ nucleic acids research _ 41:39153923 .
cecconi c , shank e , bustamante c , marqusee s ( 2005 ) direct observation of the three - state folding of a single protein molecule . _ science _ 309:20572060
. shank e , cecconi c , dill j , marqusee s , bustamante c ( 2010 ) the folding cooperativity of a protein is controlled by its chain topology .
_ nature _ 465:637640 .
gebhardt j , bornschlgl t , rief m ( 2010 ) full distance - resolved folding energy landscape of one single protein molecule .
_ proceedings of the national academy of sciences _ 107:20132018 .
yu h , et al .
( 2012 ) energy landscape analysis of native folding of the prion protein yields the diffusion constant , transition path time , and rates .
_ proceedings of the national academy of sciences _ 109:1445214457 .
harris n , song y , kiang c ( 2007 ) experimental free energy surface reconstruction from single - molecule force spectroscopy using jarzynski s equality .
_ physical review letters _ 99:6810168104 .
bizzarri a , cannistraro s ( 2010 ) free energy evaluation of the p53-mdm2 complex from unbinding work measured by dynamic force spectroscopy . _ phys .
. jarzynski c ( 1997 ) nonequilibrium equality for free energy differences .
_ physical review letters _ 78:26902693 .
crooks ge ( 1999 ) entropy production fluctuation theorem and the nonequilibrium work relation for free energy differences .
_ physical review e _ 60:27212726 .
douarche f , ciliberto s , petrosyan a ( 2005 ) estimate of the free energy difference in mechanical systems from work fluctuations : experiments and models .
_ journal of statistical mechanics : theory and experiment _ 09:p09011 mossa a , de lorenzo s , huguet jm , ritort f ( 2009 ) measurement of work in single - molecule pulling experiments . _ the journal of chemical physics _ 130:234116234125 .
alemany a , ribezzi - crivellari m , ritort f ( 2011 ) nonequilibrium statistical physics of small systems : fluctuation relations and beyond , eds klages r , just w , jarzynski c ( wiley - vch , weinheim ) pp 155179 .
van mameren j , et al . (
2008 ) counting rad51 proteins disassembling from nucleoprotein filaments under tension .
_ nature _ 457:745748 .
cisse i , mangeol p , bockelmann u ( 2011 ) in single molecule analysis , eds peterman ejg , wuite gjl ( humana press , new york ) pp 45 - 61 .
moffitt j , chemla y , izhaky d , bustamante c ( 2006 ) differential detection of dual traps improves the spatial resolution of optical tweezers . _
proceedings of the national academy of sciences _ 103:90069011 .
speck t , mehl j , seifert u ( 2008 ) role of external flow and frame invariance in stochastic thermodynamics .
_ physical review letters _ 100:178302178305 .
mazonka o , jarzynski c ( 1999 ) exactly solvable model illustrating far - from - equilibrium predictions .
_ arxiv preprint cond - mat/9912121_. wang g , sevick em , mittag e , searles dj , evans dj ( 2002 ) experimental demonstration of violations of the second law of thermodynamics for small systems and short time scales .
_ physical review letters _ 89:5060150604 .
meiners jc , quake sr ( 1999 ) direct measurement of hydrodynamic cross correlations between two particles in an external potential . _ physical review letters _ 82:22112214 .
gore j , ritort f , bustamante c ( 2003 ) bias and error in estimates of equilibrium free energy differences from nonequilibrium measurements .
_ proceedings of the national academy of sciences _ 100:1256412569 .
palassini m , ritort f ( 2011 ) improving free energy estimates from unidirectional work measurements : theory and experiment .
_ physical review letters _ 107:60601 .
ritort f , bustamante c , tinoco i ( 2002 ) a two - state kinetic model for the unfolding of single molecules by mechanical force . _
proceedings of the national academy of sciences _ 99:1354413548 .
jarzynski c ( 2006 ) rare events and the convergence of exponentially averaged work values .
e _ 73:046105046114 .
ribezzi - crivellari m , ritort f ( 2012 ) force spectroscopy with dual - trap optical tweezers : molecular stiffness measurements and coupled fluctuations analysis .
_ biophysical journal _ 103:19191928 .
ribezzi - crivellari m , huguet jm , ritort f ( 2013 ) counter - propagating dual - trap optical tweezers based on linear momentum conservation .
_ review of scientific instruments _ 84:043104043104 .
forns n , de lorenzo s , manosas m , hayashi k , huguet jm , ritort f ( 2011 ) improving signal / noise resolution in single - molecule experiments using molecular constructs with short handles . _ biophysical journal _ 100:17651774 .
] * pulling experiments with dual - trap optical tweezers . * a ) force - distance curves in a pulling experiment on a 20 bp hairpin with our dual - trap setup .
a molecular tether is attached between two trapped beads . by increasing the distance , @xmath3 , between the traps the tether
is stretched or released until some thermally activated reaction is triggered , e.g. the unfolding or folding of a dna hairpin , and detected as a force jump ( black arrows ) . the small force jump ( 0.2 pn )
is due to the low - trap stiffness of our dual - trap setup ( @xmath1360.02 pn / nm ) .
inset : scheme of the hairpin with color - coded sequence ( a / t : yellow / green , g / c : red / blue ) .
b ) pulling experiments in a dual - trap setup where trap a is moved at speed @xmath137 and trap b is at rest with respect to water .
@xmath3 is the control parameter , @xmath5 and @xmath6 are the configurational variables with respect to the moving trap a while @xmath9 and @xmath10 are the configurational variables with respect to the trap at rest ( trap b ) .
c ) pulling curves ( red stretching , black releasing ) for a 3 kb ds - dna tether in a dual - trap setup .
inset : the cyclic pulling protocol used in the experiments .
] * work distributions . * work statistics obtained on cyclic protocols with a 200 nm excursion with different pulling speeds ( columns ) .
four different work observables are considered . in each case
the solid points are direct work measurements ( l.h.s . of eq . ) . in order to improve statistics these distributions
are calculated as the convolution of the distribution of the work performed while stretching with that performed while releasing .
for that we took all forward and reverse work values @xmath138 and combined them @xmath139 in order to get a joint work distribution for the cycle .
open symbols are the reconstructed histogram ( r.h.s . of eq . ) .
different columns refer to different pulling speeds,@xmath94 , as shown on top . a ) comparison of the measured and reconstructed distributions according to the two definitions of eq .
( @xmath140 ) .
the distribution for @xmath88 satisfies the cfr eq .
, i.e. the measured and reconstructed distributions superimpose . the distribution for @xmath90
does not satisfy it .
horizontal error bars represent the systematic error in work measurements , while vertical error bars denote statistical errors .
b ) the cfr , eq . is satisfied within the experimental error for @xmath141 and @xmath92 but not for @xmath90 .
c ) comparison between the measured and reconstructed distributions for @xmath92 and @xmath91 ( eq . and eq . ) both of which satisfy the cfr .
horizontal error bars represent the systematic error in work measurements , while vertical error bars denote statistical errors .
, scaledwidth=60.0% ] * bias in unidirectional free energy estimators . *
a ) error ( @xmath45 ) on free energy estimates eq . committed by using je for @xmath21 .
circles , diamonds and squares refer to excursions of 200 nm , 400 nm and 600 nm .
three different pulling speeds were considered in each case .
note that in this case work is evaluated over closed cycles ( @xmath86 ) and the error is defined as : @xmath142 .
b ) @xmath143 displays a bilinear dependence on pulling speed @xmath137 ( main figure ) and @xmath58 ( inset ) as expected from eq . .
continuous lines are linear fits to the experimental results which contains a single fitting parameter .
the shaded area in the inset corresponds to the region within one standard deviation from the expected value of @xmath57 based on equilibrium measurements of @xmath60 ( see section 4,5 in the si ) .
c ) experimental bias measurements from the cycles shown in fig .
[ figure : opts]c .
the plots show the bias ( eqs .
, ) as a function of the number of work measurements , @xmath62 .
the three plots correspond to different pulling speeds ( 7.2 @xmath75 m/s , 4.3 @xmath75 m/s , 1.35 @xmath75 m/s ) .
interestingly @xmath144 which guarantees faster convergence of free energy estimates .
moreover @xmath145 is also larger than the sum @xmath146 ( dashed line ) i.e. the bias is superadditive ( cf .
the error bars represent the statistical error on free energy determination , not including systematic calibration errors in force and distance .
continuous lines show the theoretical predictions from ref .
@xcite for gaussian work distributions .
note that these are not fits but predictions which only use the mean dissipation as input parameter .
] * work measurements on a dna hairpin .
* a)scheme of the experimental setup ( beads and hairpin not to scale ) .
the hairpin is presented with color - coded sequence ( a / t : yellow / green , g / c : red / blue ) .
b ) work measurements upon unfolding and refolding according to three different work definitions : @xmath20 ( upper panel ) , @xmath21 ( central panel ) and @xmath37 ( lower panel ) . in this experiment
pulling speed was @xmath95 @xmath75 m/s .
open symbols show an estimate of the refolding work distributions reconstructed from the unfolding distribution via the cfr ( eq . ) . @xmath20 and @xmath37 fulfill the fluctuation theorem while @xmath21 does not .
horizontal error bars represent the systematic error on work measurements , while vertical error bars denote statistical errors .
the contribution of trap , handles and single stranded dna have been removed as detailed in @xcite .
c ) unidirectional estimates for the free energy from the unfolding work distribution .
the optimal estimator based on @xmath37 ( red ) converges to the correct value @xmath147 @xmath34 t as measured from bi - directional estimates .
the estimator based on @xmath20 ( green ) shows a larger bias and overestimates the free energy by @xmath148 .
the estimator based on @xmath21 ( blue ) converges to a wrong free energy difference ( @xmath149 ) which is @xmath150 @xmath98 below the correct value , against the second law .
note that the error committed by using @xmath20 is due to finite - size effects and decreases when more unfolding curves are measured .
in contrast the error committed by using @xmath21 remains finite for all sample sizes .
the error bars represent the statistical error on free energy determination and do not include the systematic error due to force and distance calibrations .
, scaledwidth=99.0% ] * inference of @xmath110 from partial work measurements in the symmetric case .
* a ) the distribution @xmath104 in the case of the dsdna tether , for the work measured in the wrong trap , @xmath21 .
the distribution does not fulfill the cfr . to recover the correct work distribution @xmath110 ,
@xmath21 is shifted by a constant amount @xmath106 .
the shifted distribution is tested for the cfr by defining the function : @xmath151 .
the prediction by the cfr is @xmath152 which can be tested by determining the @xmath153 function .
b ) evolution of @xmath154 as a function of @xmath21 for different values of @xmath106 .
the value @xmath155 for which the slope of @xmath154 is equal to one ( work being measured in @xmath98 units ) determines the correct work distribution @xmath110 ( @xmath156 , inset ) .
c ) in the case of bidirectional measurements both the forward and the reverse work distributions @xmath157 are shifted by an amount @xmath106 in opposite directions .
d ) evolution of the function @xmath158 as a function of @xmath106 .
again the cfr predicts @xmath153 should be linear in @xmath21 with slope one .
e ) inference of the correct work distributions and @xmath43 measurement . for each value of @xmath106 a linear fit @xmath159 to @xmath153 is performed .
the value @xmath155 for which @xmath160 ( @xmath161 in the figure ) , is the shift needed in order to recover the full work distribution @xmath110 from partial work measurements in the wrong trap .
moreover the cfr implies @xmath162 ( @xmath163 ) .
, scaledwidth=70.0% ] * inference of @xmath110 from partial work measurements in the asymmetric case . * ( a ) equilibrium force distributions at @xmath164pn for the 3 kb ds - dna tether measured in an asymmetric dual - trap setup ( @xmath165 pn / nm,@xmath166
pn / nm,@xmath167 pn / nm ) .
( b ) convolution of @xmath104 with different gaussian distributions corresponding to different pairs @xmath123.we show @xmath168 ( blue filled circles ) , @xmath169 ( blue empty circles ) , @xmath110 ( green filled circles ) . among all different
@xmath170 only one matches the correct work distribution @xmath110 , i.e. only one reconstructed distribution is physically correct ( rightmost graph , with @xmath171 @xmath34 t , @xmath172 ( @xmath34t@xmath173 ) .
in this situation the inference can not rest on the cfr alone , and additional information is required to infer @xmath110 .
( c ) asymmetry factor ( af ) as a function of @xmath174 for different values of @xmath175 .
the blue ( red ) circles indicate the symmetric ( @xmath176 ) and asymmetric ( @xmath177 ) cases respectively .
( d ) the af defined by @xmath178 ( red line ) and the cfr invariance @xmath179 for any @xmath120 ( blue line ) , do select a narrow range of possible pairs ( @xmath106,@xmath180 ) at the intersection between the blue and red lines .
the intersection region is compatible with the parameters ( @xmath106 = 8 @xmath34 t , @xmath181 = 7.2 ( @xmath34t)@xmath182 ) describing the true correct work distribution @xmath110 ( black point ) .
|
fluctuation relations ( frs ) are among the few existing general results in non - equilibrium systems .
their verification requires the measurement of the total work ( or entropy production ) performed on a system .
nevertheless in many cases only a partial measurement of the work is possible .
here we consider frs in dual - trap optical tweezers where two different forces ( one per trap ) are measured . with this setup
we perform pulling experiments on single molecules by moving one trap relative to the other .
we demonstrate that work should be measured using the force exerted by the trap that is moved .
the force that is measured in the trap at rest fails to provide the full dissipation in the system leading to a ( incorrect ) work definition that does not satisfy the fr .
the implications to single - molecule experiments and free energy measurements are discussed . in the case of symmetric setups a new work definition , based on differential force measurements ,
is introduced .
this definition is best suited to measure free energies as it shows faster convergence of estimators .
we discuss measurements using the ( incorrect ) work definition as an example of partial work measurement .
we show how to infer the full work distribution from the partial one via the fr .
the inference process does also yield quantitative information , e.g. the hydrodynamic drag on the dumbbell .
results are also obtained for asymmetric dual - trap setups .
we suggest that this kind of inference could represent a new and general application of frs to extract information about irreversible processes in small systems .
|
observations of the hubble deep field , and other surveys of high redshift galaxies described in these proceedings , are contributing to a new picture of how large galaxies such as the milky way were assembled .
one interpretation of the data so far is that large galaxies collapse out of what appears as several star - forming proto - galactic fragments at @xmath3 ( pascarelle et al .
1996 ; haehnelt , steinmetz @xmath4 rauch 1998 ) . here
we describe our search for h@xmath5 emission from the star - forming regions in high redshift galaxies , which is redshifted into the near - infrared .
we focus on one of the outstanding questions about these objects : is the burst of star - formation seen in the uv continuum producing the dominant stellar population ( by mass ) ?
we show that for most high redshift objects , the data available to date provide only weak constraints on the age of the dominant stellar population ( see also the discussion by sawicki @xmath4 yee 1998 ) .
an exception is our observation of the observed mid - ir ( rest - frame near - ir ) continuum of the @xmath62.7 galaxy ms1512-cb58 using @xmath7 . for this object
, we can put a strong limit on the fraction of the mass comprised of an old population of stars .
to illustrate the ambiguity in the age of the dominant stellar population , in this section we describe observations of the @xmath62.515 galaxy which is lensed by abell 2218 .
abell 2218 is a rich cluster of galaxies at @xmath8 ( abell et al .
1989 ; kristian et al .
it was one of the first to to be observed to have arcs caused by the lensing of background galaxies ( lynds @xmath4 petrosian 1986 ; pello - descayre et al .
1988 ) , and has been the subject of many studies . of particular interest is the discovery by ebbels and collaborators ( e96 ) that one of the brightest blue arcs is a star - forming galaxy at @xmath62.515 ( arc a , or galaxy @xmath9384 ) .
the appearance of the optical spectrum of the arc , as well as the optical / ir spectral energy distribution , imply that arc a is typical of the star - forming galaxies discovered by lyman dropout techniques ( steidel et al .
1996a ) or in the hubble deep field ( williams et al . 1996 ,
lowenthal et al .
1997 , cohen et al 1996 , steidel et al .
1996b ) , except that its apparent magnitude is boosted by about a factor of 14.5 by lensing ( saranti , petrosian @xmath4 lynds 1996 ) .
arc a is similar in apparent properties to ms1512-cb58 , described below .
we obtained a narrow - band image of h@xmath5 emission from a2218 s arc a using the irtf nsfcam and tunable narrow - band filter , or cvf ( r@xmath1090 , figure 1 ) .
further details are provided in bechtold @xmath4 elston ( 1998 , in preparation ) . for arc
a , the excess emission in the cvf image compared to the @xmath11 image implies a rest - frame equivalent width of @xmath1085 @xmath12 and a narrow band flux of 7.35 @xmath13 10@xmath14 ergs @xmath15 sec@xmath16 . for @xmath17
= 2.515 , this corresponds to a luminosity in the excess narrow - band emission of 3.25@xmath1310@xmath18 @xmath19@xmath20 ergs sec@xmath16 for @xmath21 = 0.1 .
the observed h@xmath5 luminosity implies a total observed star - formation rate ( sfr ) of 290 @xmath19@xmath20 m@xmath22 yr@xmath16 if we adopt the calibration of kennicutt ( 1983 ) , that is , sfr ( m@xmath22 yr@xmath16 ) = l(h@xmath5)/1.12@xmath1310@xmath23 ergs s@xmath16 . for @xmath24=0.5 , sfr = 135 @xmath19@xmath20 m@xmath22 yr@xmath16 .
this assumes an imf consistent with salpeter ( 1955 ) , an upper mass cut - off of 100 m@xmath22 , and 1.1 magnitudes of extinction at h@xmath5 .
this value for the extinction is consistent with the reddening derived from the spectral energy distribution ( figure 2 ) .
if we assume that the galaxy seen as arc a is magnified by lensing by a factor of 14.5 , then the star - formation rate for arc a is 20 m@xmath22 yr@xmath16 for @xmath24=0.1 , and 9.3 m@xmath22 yr@xmath16 for @xmath24=0.5 .
this is very typical of the star - formation rates estimated from the uv continuum luminosity for lyman dropout galaxies or hubble deep field galaxies ( madau et al 1996 ; steidel et al .
1996a ) moreover , arc a in the cvf image appears bright on its eastern edge compared to the @xmath11 image .
the seeing for the irtf data was about 0.4 @xmath25 , compared to a total linear extent of about 5 @xmath25 for the arc , so it is well - resolved spatially .
we also show the wfpc2 image of a2218 through the f702w filter from the hst archive ( the wfpc2 data was presented by kneib et al .
the bulk of the excess emission appears to be concentrated in a region along the arc which is fairly faint in the f702w stellar continuum ( rest frame @xmath262000@xmath12 ) .
thus the effect of the lensing is to stretch out what is probably a small , compact galaxy and allow us to see that star - formation is not happening in a single , co - eval and co - spatial burst .
we refit the broad - band spectral energy distribution ( sed ) of arc a , which was discussed by e96 .
the theoretical curves shown by e96 did not fit the sed very well , and new models have subsequently become available ( leitherer et al .
1996 and references therein ) .
we tried several models , chosen to span the important parameters , and found a few good fits to the sed : see figure 2 .
instantaneous single burst models , e.g. those of bruzual @xmath4 charlot ( 1996 , denoted gissell in the figures ) or fioc @xmath4 rocca - volmerange ( 1997 , denoted rvf in the figures ) , fit better than continuous star - formation models , and an age of 40 myrs gives the best fit , with e(b - v)=0.18 , assuming an lmc - like reddening law .
however , this model predicts a very small equivalent width for h@xmath5 emission , w(h@xmath5)@xmath10 0.1 @xmath12 ( leitherer @xmath4 heckman 1995 ) , in clear conflict with the observations . for the reddening law for starbursts given by calzetti ( 1997 ) ,
which one expects to be more appropriate , the fit is somewhat worse , and the age derived quite different : 3.8 myrs , and e(b - v)=0.35 gives the best fit , and the predicted w(h@xmath5 ) @xmath10 100@xmath12 , in better agreement with the h@xmath5 observation .
to2.6 in ' '' '' thus the large reddening correction needed in the uv makes conclusions regarding the age and star - formation rate of arc a uncertain . moreover , the data points shown in figure 2 are for the broad band fluxes integrated over the entire galaxy .
the h@xmath5 picture indicates already that the star - formation is not coeval over the whole galaxy .
an alternative possibility is that star - formation has been episodic . in figure 2
we show a second model , where we fit two bursts , one fixed at an age of 2 gyr ( with lmc reddening ; the age of the universe at z@xmath27 2.5 is not much older than 2 gyrs ) and a 5 myr old burst ( and calzetti reddening ) .
this two burst model fits the data very well , and the older burst contains @xmath2895 percent of the mass of the galaxy .
this is qualitatively a very different star - formation history for this galaxy than the single burst model , one which the current data can not rule out .
an older population could be detected by rest - frame near - ir photometry , but for z@xmath102 - 3 , the rest - frame near - ir is redshifted into the mid - ir , and is beyond the sensitivity of current instruments for most objects . in the next section , we describe @xmath7 observations of ms1512-cb58 , at @xmath29 , which we obtained in order to limit the presence of an older stellar population .
ms 1512-cb58 is a galaxy which was discovered serendipitiously ( yee et al .
1996 ) during the course of the cnoc-1 redshift survey of moderate redshift galaxy clusters ( yee , ellingson @xmath4 carlberg 1996 ) .
it is at @xmath62.7 with apparent @xmath3020.5 .
its observed optical ( rest - frame uv ) spectrum resembles that of local starburst galaxies ( yee et al .
1996 , steidel et al . 1996a
; see also conti , leitherer & vacca 1996 ) . the lack of a 4000 break in the rest frame ( ellingson et al .
1996 ) conveniently redshifted so that the broad - band @xmath31 and @xmath32 filters straddle it , implies a very young age , as does the appearance of the c iv @xmath261550 p - cygni profile caused by winds associated with the hot stars .
the best fit to the sed is a continuous star - formation model of about 10 - 20 myr , with lmc reddening of @xmath33 @xmath10 0.3 .
bechtold et al . ( 1997 ) detected h@xmath5 emission , confirming a high star - formation rate for ms 1512-cb58 .
however , as in the case of the a2218 arc , the limits on the presence of an older stellar population are weak .
ellingson et al . ( 1996 ) showed that if a two burst model is fit to the spectral energy distribution , an older , 2 gyr old burst could comprise as much as 85@xmath34 of the galaxy mass .
ms1512-cb58 has the brightest apparent magnitude of the high redshift star - forming galaxies discovered so far , which is in part a result of the relatively young age of its starburst , and in part because of magnification by gravitational lensing by a foreground @xmath35 cluster of galaxies ( seitz et al .
1998 , ellingson et al . 1998 ) .
the magnification factor is probably as high as 25@xmath13 .
thus , ms1512-cb58 is an ideal candidate for mid - ir observations .
we used the long wavelength camera of the @xmath36 instrument ( cesarsky et al .
1996 ) aboard the _ infrared space observatory _ ( _ iso _ , kessler et al .
1996 ) to image ms1512-cb58 through two broad band filters , lw2 ( centered on 6.7@xmath37 m ) and lw10 ( centered on 11.5@xmath37 m ) , covering rest - frame 1 - 4@xmath37 for ms1512-cb58 .
details of the observations and data reduction are given in bechtold , yee @xmath4 ellingson ( 1998 , in preparation ) .
the data were spatially oversampled in order to resolve ms1512-cb58 from the central cd galaxy of the foreground cluster , which lies only 6 @xmath25 away .
the noise in deep @xmath7 images is primarily from unresolved point sources , and so we evaluated the significance of all sources from the variation from point to point in the images themselves .
we require sources to be detected at @xmath283@xmath38 in both filters , with the result that ms1512-cb58 , the central cd and several other members of the foreground cluster are detected .
to2.6 in ' '' '' the spectral energy distribution for ms1512-cb58 , including the @xmath7 flux points , is shown in figure 3 .
the @xmath7 flux points fall on the extrapolation of the young burst model which best fits the optical and near - ir data points , with very little room for the addition of an older population . less than 10@xmath34 of the mass of ms1512-cb58 can be in a 2 gyr population .
thus , ms1512-cb58 may well be a true proto - galaxy , undergoing its very first burst of star - formation .
futher mid - ir observations of ms1512-cb58 and other high redshift galaxies will be possible with sirtf and ngst .
for many years , quasar absorption lines have been used to study the evolution of galaxies and the intergalactic medium at high redshift .
the damped ly@xmath5 systems in particular are generally thought to be the progenitors of present - day galaxies ( wolfe et al .
1986 ) and should thus be sign posts for distant normal galaxies in an early stage of evolution .
prochaska @xmath4 wolfe ( 1996 , 1997 ) have suggested that the asymmetric line profiles seen in the metal lines associated with damped systems support the view that damped ly-@xmath5 systems are large rotating disks . on the other hand , haehnelt , steinmetz @xmath4 rauch ( 1998 )
suggest that such profiles can occur naturally in systems composed of smaller protogalactic fragments .
direct detection of the galaxies causing damped lyman-@xmath5 absorbers will help interprete the large body of spectroscopic observations of them in the context of the formation of galaxies and structure .
deep imaging of the fields of damped lyman-@xmath5 systems does appear to show a statistical excess of luminous galaxies near the quasar line of sight ( e.g. aragon - salamanca , ellis @xmath4 obrien 1996 ) .
but without redshifts , one can not actually associate any particular galaxy with a given absorption line system .
we have used the irtf nsfcam and palomar 5 m near - ir cass camera to search for h@xmath5 redshifted into the k band , associated with damped ly@xmath5 quasar absorption line systems .
details of the observations are given in elston , bechtold @xmath4 cutri ( 1998 , in preparation ) .
we obtain images through a 1@xmath34 narrow band filter ( cvf ) tuned to h@xmath5 at the redshift of the damped ly@xmath5 absorbers .
so far , this technique has proven very powerful : we have detected h@xmath5 companions for most of the systems observed , with separation of between 2 and 12 arcsec from the quasar line - of - sight .
often , multiple faint , very compact objects are found in a single field .
we interprete these h@xmath5 emitters to be clumps of star - forming gas in the same sheet or filament which contains the damped ly@xmath5 absorber .
we are in the process of confirming these with slit spectra at the kpno 4 m with crsp , and ukirt .
using the standard conversion between h@xmath5 flux and star - formation rate our typical 3@xmath38 sensitivity is about 5 m@xmath39 yr@xmath16 .
if the h@xmath5 fluxes are attributed to photo - ionization by young stars it appears that star formation rates of 10 - 20m@xmath40 yr@xmath16 are pervasive at redshifts of 2.5 in agreement with the lyman - dropout population found by steidel et al .
( 1996 ) suggesting they are similar populations .
if these observations of individual systems are representative , we can combine the star - formation rates with the statistics of the incidence of damped ly@xmath5 absorbers in quasar lines - of - sight ( wolfe et al . 1995 ) , to make a preliminary estimate of the volume - averaged star - formation rate ( pei @xmath4 fall 1995 ; fall , charlot , @xmath4 pei 1996 ; madau et al . 1996 ) at @xmath02.5 ( not all candidates have been spectroscopically confirmed yet ) .
figure 4 shows the result .
our results are consistent , within large uncertainties , with the results derived from the uv continuum luminosity by madau et al .
( 1996 ) . however , our data so far suggest that the peak in star - formation may have taken place at somewhat higher redshift than indicated by the madau et al . analysis . combining our data with observations of the h@xmath5 luminosity of low and moderate redshift galaxies will allow an important check on the star - formation rates in the universe derived from other means .
we are very grateful to the staff of the @xmath41 , @xmath42 and kpno for providing the instrumentation which made this work possible , and for help at the telescope .
we are indebted to ken ganga , dave van buren and the staff at ipac for help with the planning and analysis of the @xmath7 observations .
this work was supported by nsf grants ast-9058510 and ast-9617060 , nasa hst grants ar057850194 and go074480296 , and nasa iso grant nag5 - 3359 .
abell , g. o. , corwin , h. g. @xmath4 olowin , r. p. 1989
, , 70 , 1 aragon - salamanca , a. , ellis , r. s. , obrien , k. s. 1996 , , 281 , 945 bechtold , j. , yee , h.k.c . , elston , r. , ellingson , e. 1997 , , 477 , l29 bruzual , g. @xmath4 charlot , s. 1993 , , 405 , 53 calzetti , d. 1997 , preprint cesarsky , c. j. , et al .
1996 , , 315 , l32 cohen , j. g. et al .
1996 , 471 , l5 conti , p. s. , leitherer , c. , @xmath4 vacca , w. d. 1996 , , 461 , l87 ebbels , t.m.d .
1996 , , 281 , l75 ( e96 ) ellingson , e. e. , yee , h. k. c. , bechtold , j. , @xmath4 elston , r. 1996 , , 466 , l71 fall , s. m. , charlot , s. , pei , y. c. 1996 , , 464 , l43 fioc , m. @xmath4 rocca - volmerange , b. 1997 , , 326 , 950 gallego , j. et al .
1996 , @xmath43 , 120 , 323 haehnelt , m. g. , steinmetz , m. and rauch , m. 1998 , preprint .
kennicutt , r. c. , jr .
1983 , , 272 , 54 kessler , m. f. et al . 1996 , , 315 , l27 kneib , j .-
, et al . , 1996 , 471 , 643 kristian , j. , sandage , a. , @xmath4 westphal , j. a. , 1978 , , 221 , 383 leitherer , c. @xmath4 heckman , t. m. 1995 , , 96 , 9 leitherer , c. et al . 1996 , , 108 , 996 lilly , s. et al .
1996 , , 460 , l1 lowenthal , j. et al .
1997 , , 481 , 673 lynds , r. @xmath4 petrosian , v. 1986 , baas , 18 , 1014 madau , p. et al .
1996 , , 283 , 1288 pascarelle , s. m. , et al .
1996 nature 383 , 45 pei , y. c. , fall , s. m. , 1995 , , 454 , 69 pello - descayre , r. et al .
1988 , , 190 , l11 prochaska , j. x. @xmath4 wolfe , a. m. 1996 , , 470 , 403 prochaska , j. x. @xmath4 wolfe , a. m. , 1997 , , in press , astro - ph/9704169 salpeter , e. e. 1955 , , 121 , 161 saranti , d. w. , petrosian , v. @xmath4 lynds , r. 1996 , , 458 , 57 sawicki , m. , yee , h.k.c . , 1998 , in press , astro - ph/9712216 seitz , s. et al . 1998 , ,
submitted , astro - ph/9706023 steidel , c. c. et al .
1996a , , 462 , l17 steidel , c. c. et al . 1996b , , 112 , 352 tresse , l. @xmath4 maddox , s. j. , 1997 preprint , astro - ph/9712061 williams , r. e. et al .
1996 , , 112 , 1335 wolfe , a. m. et al .
1986 , , 61 , 249 wolfe , a. m. et al .
1995 , , 454 , 698 yee , h. k. c. , ellingson , e. , @xmath4 carlberg , r. g. , 1996 , , 102 , 269 yee , h. k. c. , ellingson , e. , bechtold , j. , carlberg , r. g. , @xmath4 cuillandre , j .- c .
1996 , , 111 , 1783
|
observations of the high redshift universe , interpreted in the context of a new generation of computer simulated model universes , are providing new insights into the processes by which galaxies and quasars form and evolve , as well as the relationship between the formation of virialized , star - forming systems and the evolution of the intergalactic medium .
we describe our recent measurements of the star - formation rates , stellar populations , and structure of galaxies and protogalactic fragments at @xmath02.5 , including narrow - band imaging in the near - ir , ir spectroscopy , and deep imaging from the ground and from space , using @xmath1 and @xmath2 .
|
( teuben 1994 ) and are traditional programming environments with which n - body simulations can be setup , run and analyzed .
nemo also has a number of tools to import and export data in tables , ccd type images , fits files and a large number of other n - body formats .
nemo is more geared towards collisionless stellar dynamics , while starlab has more sophisticated programs to deal with close encounters , and can now also incorporate stellar evolution through the seba package ( portegies zwart et al . 2001 ) .
nemo and starlab present themselves to a user as a large set of programs , often glued together using pipes in shell scripts to set up and run complex simulations . for the programmer ,
a large set of classes and functions are available to construct new integrators and analysis programs .
for example , in the following starlab example an anisotropic king model with 2048 particles has been evolved with 50% binaries ( i.e. 3096 actual stars ) and stellar evolution : .... mk_aniso_king -i -n 2048 -u -w 4 -f 3 |\ mkmass -i -u 100 -l 0.1 -f 3 |\ mksecondary -f 0.5 -l 0.1 |\ addstar -q 0.5 -r 2.5 |\ scale -m 1 -e -0.25 -q 0.5 |\ mkbinary -f 2 -l 1 -u 1000000 -o 2 |\ kira -a 0.1 -d 1 -d 25 -n 25 -t 4000 -q -g 2 -u
-b -z 1 > run001 .... the grape special purpose hardware ( hut and makino , 1999 ) , now running at 100 terraflops speed , has been successfully interfaced with starlab , and now is starting to produce massive datasets .
analysis and visualization techniques of those dataset are becoming increasingly challenging .
the american museum for natural history ( amnh ) in new york city has recently renovated its planetarium , and converted it into a state - of - the - art digital planetarium with capabilities for scientific visualization .
their computer system consists of an onyx2 , with 28 cpus , 14 gb of memory , 2 tb diskspace and 7 graphics pipes .
each graphics pipe controls one of 7 projectors which illuminate the dome in a dodecahedral pattern .
the software that drives most visualization is an ncsa product called ( virdir ) , that we have now been using during a number of night sessions in the dome , much like optical observers ( during daytime the planetarium is of course used for public viewing ) .
it allows us to `` fly '' through the data , in space and time . by adding complete orbital information for a select number of stars we have started fully interactive data mining of our 4d spacetime histories of these star cluster simulations runs . in order for us to test new visualization techniques , algorithms and interfaces with the starlab environment , we used an existing program partiview , which had been derived from virdir , and which can be run on a normal workstation or laptop .
it uses the fltk and mesa / opengl libraries for its user interface and fast graphics .
a screenshot of partiview in action can be seen in figure [ p1 - 39-fig-1 ] .
we have modified partiview to understand our starlab simulatation data , and added interfaces that allow this workstation version to animate and move in time and space .
partiview comes with a small but powerful set of commands with which dataselections and viewing can be made , and we hope to expand this into a more mature scripting language .
it is also fairly straightforward for other packages to benefit from using partiview .
in the spirit of the federated model of archiving observational data , recently proposed by the nvo ( national virtual observatory ) initiative , we will develop a starlab - based archive .
a simulation of a globular cluster with a million stars stars for ten billion years will generate 100 tbytes of raw data , of which we would like to store at least 1 tbyte , and preferably more , for 4d visualization of the full history of the evolution of a star cluster . although our main goal will be to enable rapid and intelligent access to our simulation output files , we will simultaneously develop a flexible and transparent interface with the nvo databace and protocols . our starlab policy will be to make all simulation results freely and publicly available to ` guest observers ' .
|
in dense clusters a bewildering variety of interactions between stars can be observed , ranging from simple encounters to collisions and other mass - transfer encounters . with faster and special - purpose computers like grape
, the amount of data per simulation is now exceeding 1 tb . visualization of such data has now become a complex 4d data - mining problem , combining space and time , and finding interesting events in these large datasets .
we have recently starting using the virtual reality simulator , installed in the hayden planetarium in the american museum for natural history , to tackle some of these problem .
reports on our first `` observations '' , modifications needed for our specific experiments , and perhaps field ideas for other fields in science which can benefit from such immersion .
we also discuss how our normal analysis programs can be interfaced with this kind of visualization .
|
the standard cosmology based on general relativity and inflation has been remarkably successful .
but there remain deep puzzles left for theorists to resolve what is the cause of the late - time acceleration of the universe ( the dark energy " problem ) ?
how is the classical big bang singularity removed by quantum gravity effects ?
one approach to start tackling these problems is via the brane - world scenario , which is motivated by string theory .
most brane - world models , including those of randall - sundrum type @xcite , produce ultra - violet modifications to general relativity , with extra - dimensional gravity dominating at high energies .
however it is also possible for extra - dimensional gravity to dominate at low energies , leading to infra - red modifications of general relativity @xcite .
the dvali - gabadadze - porrati ( dgp ) models @xcite ( see also @xcite ) achieve this via a brane induced gravity effect .
the generalization of the dgp models to cosmology lead to late - accelerating cosmologies , even in the absence of a dark energy field @xcite .
this exciting feature of self - acceleration " may help towards a new resolution to the dark energy problem .
but the models suffer from the short - coming that they do not modify 4d gravity at high energies , where we expect stringy corrections to start having an effect .
how can we generalize the dgp models so that they also show ultra - violet modifications to general relativity ?
one possibility is to introduce into the gravitational action a term that is associated with higher - energy stringy corrections the gauss - bonnet term @xcite . indeed , in certain realizations of string theory
, the ghost - free gb term in the bulk action may naturally lead to a dgp induced gravity term on the brane boundary @xcite .
we investigate what happens when a gauss - bonnet term is introduced in the 5d minkowski bulk containing a friedmann brane with dgp induced gravity .
as we will show , this combination of infra - red and ultra - violet modifications leads to intriguing cosmological models .
the gravitational action contains the gauss - bonnet ( gb ) term in the bulk , as a correction to the einstein - hilbert term , and the induced gravity ( ig ) term on the brane , @xmath0\right\ } \nonumber\\&+&\frac{r}{2\kappa^2_5 } \int_{\rm brane}d^4x\sqrt{- g^{(4)}}\,r^{(4)}\,,\end{aligned}\ ] ] where @xmath1 is the gb coupling constant @xcite and @xmath2 is the ig cross - over " scale , which marks the transition from 4d to 5d gravity .
the dgp models are the special case @xmath3 , and in this case the cross - over scale defines an effective 4d gravitational constant via @xmath4 we assume mirror ( @xmath5 ) symmetry about the brane .
the standard energy conservation equation holds on the brane , @xmath6 the modified friedmann equation was found in the most general case ( where the bulk contains a black hole and a cosmological constant , and the brane has tension ) in ref .
@xcite . for a spatially flat brane without tension , in a minkowski bulk , @xmath7 ^ 2 \left(h^2-\phi \right ) = \left[rh^2 -\frac{\kappa^2_5}{3}\rho \right]^2\!,\end{aligned}\ ] ] where @xmath8 is determined by @xmath9 ( in the most general case ,
the right - hand side of this condition is nonzero @xcite . )
equation ( [ phi ] ) has solutions @xmath10 , but here we only consider @xmath11 , since the second solution has no ig limit and thus does not include the dgp model @xcite . vs @xmath12 ) for a minkowski bulk .
( arrows indicate the flow of brane proper time @xmath13 , with @xmath14 at @xmath15.),title="fig:",width=264,height=288 ] ( -1.2,-0.85)@xmath16 ( -1.2,-2.95)@xmath16 ( -1.2,-3.8)@xmath16 the dgp models have @xmath17 and the friedmann equation ( [ f ] ) reduces to a quadratic in @xmath18 , with solutions @xmath19 there are two branches dgp(@xmath20 ) for the two signs on the right ( corresponding to different embeddings of the brane in the minkowski bulk ) . both branches have a 4d limit at high energies , @xmath21 while at low energies , @xmath22 dgp( ) has a non - standard ( and non - accelerating ) late universe .
the self - accelerating dgp(+ ) branch is of most interest for cosmology , and we focus here on this model and its generalization via a gb term
. the pure gb model with a minkowski bulk has @xmath23 and the friedmann equation ( [ f ] ) reduces to @xmath24 which is a cubic in @xmath18 .
this gb friedmann equation has no 4d limit : @xmath25 the friedmann equations for pure dgp and pure gb models with a minkowski bulk are compared in fig .
[ gbdgp ] .
the dgp(+ ) models are attractive for cosmology since they accelerate at late times , without the need for dark energy , when gravity begins to leak off the brane , i.e. , when the 5d ricci term in eq .
( [ acgbig ] ) begins to dominate over the 4d ricci term . at early times
, the 4d term dominates and general relativity is recovered ( in the background
note that the perturbations are not general relativistic @xcite ) .
the dgp models are in some sense unbalanced " , since they do not include ultra - violet modifications to cosmological dynamics . in order to modify 4d gravity at high energies as well as low energies
, we can include a gb term in the action .
the combined gauss - bonnet induced gravity ( gbig ) model has a dgp brane in a minkowski bulk with einstein - gauss - bonnet gravity .
the gbig friedmann equation follows from putting @xmath11 in eq .
( [ f ] ) . defining dimensionless variables , @xmath26 the gbig friedmann equation becomes @xmath27 while the conservation equation becomes @xmath28 where a dash denotes @xmath29 , and @xmath30 .
combining eqs .
( [ dgpf2 ] ) and ( [ ec2 ] ) , we find the modified raychaudhuri equation , @xmath31 the acceleration @xmath32 is then given by @xmath33 \over 4(\gamma h^2 + 1)(3\gamma h^2 + 1)-2(h^2-\mu)}.\ ] ] the gb correction , via a non - zero value of @xmath34 , introduces significant complexity to the friedmann equation , which becomes cubic in @xmath35 , as opposed to the quadratic dgp(@xmath20 ) case , @xmath36 , for which @xmath37 this additional complexity has a dramatic effect on the dynamics of the dgp(+ ) model , as shown in fig .
[ gbigdgp ] .
the contribution of gb gravity at early times removes the infinite density big bang , and the universe starts at finite maximum density and finite pressure ( but , as we show below , with infinite curvature ) .
furthermore , there are two such solutions , each with late - time self - acceleration , marked gbig1 and 2 on the plots .
since gbig2 is accelerating throughout its evolution ( actually super - inflating , @xmath38 ) , the physically relevant self - accelerating solution is gbig1 . vs @xmath39 ) for the dgp(+ ) model and its gauss - bonnet corrections , gbig1 and gbig2 .
the curves are independent of the equation of state @xmath40 .
arrows indicate the flow of normalized brane proper time @xmath41 , with @xmath42 at @xmath43 .
( here @xmath44.),title="fig:",width=264,height=288 ] ( -4,-1.6)@xmath16 ( -4,-4)@xmath16 ( -1.2,-4.8)@xmath16 ( -1.2,-0.9)@xmath16 ( -1.2,-1.7)@xmath16 the cubic in @xmath35 , eq .
( [ dgpf2 ] ) , has three real roots when @xmath45 ( see below ) .
two of these roots correspond to gbig12 , which are modifications of the dgp(+ ) model .
the third root gbig3 is a modification of the dgp( ) model , as illustrated in fig .
[ gbigdgp2 ] .
note that the curves in these figures are independent of the equation of state @xmath40 of the matter content of the universe
@xmath40 will determine the time evolution of the universe along the curves , via the conservation equation ( [ ec2 ] ) .
the plots show that gbig3 starts with a standard big bang , @xmath46 , in common with the dgp(@xmath20 ) and gb models in fig .
[ gbdgp ] .
by contrast , gbig12 have a finite - temperature big bang , since the density is bounded above , @xmath47 where @xmath48 ( which is positive only for @xmath49 ) , is found below , in eq .
( [ rhomax ] ) .
the finite - density beginning was pointed out in ref .
@xcite , where the cubic for the general case ( i.e. , with tension , bulk cosmological constant and bulk black hole ) was qualitatively analyzed . here
, we focus on the simplest generalization of the dgp models , and give a detailed quantitative analysis of the cosmological dynamics .
in particular , our analysis shows that one solution is not bounded , which was not noticed in ref .
the numerical plots of the friedmann equation in figs . [ gbigdgp ] and [ gbigdgp2 ] are crucial to a proper understanding of the algebraic analysis of the cubic roots .
a detailed analysis @xcite of the cubic equation ( [ dgpf2 ] ) confirms the numerical results , and shows that ( for @xmath50 ) : @xmath51 the real roots are given as follows : + @xmath52 for @xmath45 : the roots gbig12 are @xmath53 where @xmath54 for gbig1 , @xmath55 for gbig2 , and the root gbig3 is @xmath56 @xmath52 for @xmath57 : the only real root gbig3 is @xmath58 + in the above , @xmath59 are defined by @xmath60^{1/3},\\ & & q = { 8\gamma^2 \over 3}(\mu+2)-{1\over
9}(1 - 8\gamma)^2\,,\\ & & r = 8\gamma^4\mu^2- { 4\gamma^2 \over 3}(1 - 8\gamma)(\mu+2 ) + { ( 1 - 8\gamma)^3\over 27}\ ! , \label{r } \\ & & \cos 3\theta = r/ \sqrt{-q^{3}}\ , .
\label{theta}\end{aligned}\ ] ] the explicit form of the solutions can be used to confirm the features in figs .
[ gbigdgp ] and [ gbigdgp2 ] .
equations ( [ roots2])([r ] ) show that gbig3 starts with a big bang , @xmath61 , with @xmath62 near the big bang .
this is the same as the high - energy behaviour of the pure gb model , eq .
( [ bbgb ] ) the gb effect dominates at high energies in gbig3 .
this is not the case for gbig12 .
the maximum density feature of gbig12 is more easily confirmed by analysing the turning points of @xmath39 as a function of @xmath35 .
the friedmann equation ( [ dgpf2 ] ) gives @xmath63 substituting @xmath64 into eq .
( [ dgpf2 ] ) , we find that @xmath65 the second equation shows that positive maximum density only arises for the upper sign and with @xmath49 , in agreement with the cubic analysis .
thus the initial hubble rate and density for gbig12 are @xmath66 if @xmath67
, then gbig12 reduce to dgp(+ ) , and @xmath68 .
note that @xmath69 the case @xmath70 corresponds to a vacuum brane with de sitter expansion , and @xmath71 , generalizing the dgp(+ ) vacuum de sitter solution @xcite .
the late - time asymptotic value of the expansion rate , as @xmath72 , is @xmath73^{1/2},\ ] ] where the minus sign corresponds to gbig1 and the plus sign to gbig2 . in the limit @xmath74
, gbig1 recovers the dgp(+ ) case , @xmath75 , while for gbig2 , @xmath76 ; the parabolic gbig12 curve in fig .
[ gbigdgp ] unwraps " and transforms into the dgp(+ ) curve .
equations ( [ class ] ) and ( [ gbighinfin ] ) show that @xmath77 while @xmath78 for gbig2 . the behaviour of the key gbig12 parameters is illustrated in figs .
[ himui2 ] and [ hinfin ] . .
, width=264,height=288 ] . , title="fig:",width=264,height=288 ] ( -7.4,-3.2)@xmath79
the gbig1 model , which is the physically relevant generalization of the dgp(+ ) model , exists if eq .
( [ class ] ) holds . by eq .
( [ dv ] ) , this means that the gb length scale @xmath80 must be below a maximum threshold determined by the ig cross - over scale : @xmath81 if the gb term is taken as the correction term in certain string theories , then @xmath82 , while @xmath83 , so that this bound is easily satisfied . when eq .
( [ gam2 ] ) holds , the universe starts with a maximum density @xmath84 and maximum hubble rate @xmath85 , and evolves to an asymptotic vacuum de sitter state : @xmath86 at any epoch @xmath87 , the proper time back to the beginning is @xmath88 since @xmath89 and @xmath90 are nonzero on the interval of integration , the time back to the beginning is finite .
the current hubble rate can be approximated by the final de sitter hubble rate , @xmath91 , so that by eq .
( [ cross ] ) , the cross - over scale obeys @xmath92 in the dgp(+ ) limit , @xmath93 .
the effect of gb gravity is to increase @xmath94 but not beyond @xmath95 .
however , there is a uv - ir bootstrap " operating to severely limit the gb effect at late times .
the key point is that appreciable late - time gb effects require an increase in @xmath34 , whereas the primordial hubble rate @xmath85 is suppressed by an increase in @xmath34 as shown in figs .
[ himui2 ] and [ hinfin ] .
equations ( [ hrhomax ] ) and ( [ gbighinfin ] ) imply that @xmath96 thus the gbig1 model does not alleviate the dgp(+ ) fine - tuning problem of a very large cross - over scale , @xmath97 .
the gbig1 friedman equation ( [ roots1 ] ) gives @xmath98 \label{root}\end{aligned}\ ] ] where @xmath99^{3/2 } } .\end{aligned}\ ] ] a more convenient form of the friedmann equation follows from solving eq .
( [ dgpf2 ] ) for @xmath39 , @xmath100 by expanding to first order in @xmath101 , we find that at late times , @xmath102 taking the dgp(+ ) limit @xmath103 , and comparing with eqs .
( [ fd ] ) and ( [ dd ] ) , we find that the effective newton constant in gbig1 is @xmath104 where @xmath105 is the fundamental , 5d gravitational constant . in the dgp(+ ) case @xmath106 .
equation ( [ newt ] ) gives a relation for the fundamental planck scale @xmath107 @xmath108 where @xmath109 is the effective 4d planck scale , and we used @xmath110 .
as @xmath111 ( its upper limit ) , so @xmath107 increases .
this is very different from the dgp(+ ) case , where @xmath112 , so that @xmath107 is constrained to be very low , @xmath113mev . in principle , gb gravity allows us to solve the problem of a very low fundamental planck scale but in practice the uv - ir bootstrap , eq .
( [ boot ] ) , means that @xmath114 so that @xmath107 is effectively the same as in the dgp(+ ) case . what is the nature of the beginning of the universe in gbig1 ?
we can use eq .
( [ in ] ) in eq .
( [ ray ] ) , for matter with @xmath115 , to analyze the initial state , @xmath116 .
we find that @xmath117 , i.e. , infinite deceleration , @xmath118 ( for gbig2 , with @xmath119 , we have @xmath120 . ) the initial state has no infinite - temperature big bang , but it has infinite deceleration , and thus infinite ricci curvature . the brane universe is born in a quiescent " singularity .
( although similar singularities may be found in induced gravity models @xcite , they arise from the special extra effect of a bulk black hole or a negative brane tension . ) the key point is that neither the dgp(+ ) model nor the gb model avoid the standard big bang , as shown in eqs . ( [ bbdgp ] ) and ( [ bbgb ] ) . but
together , the ig and gb effects combine in a nonlinear " way to produce entirely new behaviour .
if we switch off either of these effects , the infinite - temperature big bang reappears .
this singularity is reminiscent of the sudden " ( future ) singularities in general relativity @xcite but unlike those singularities , the gbig12 singularity has finite pressure .
the initial curvature singularity signals a breakdown of the brane spacetime .
the ( minkowski ) bulk remains regular , but the imbedding of the brane becomes singular .
higher - order quantum - gravity effects will be needed to cure this singularity .
the fact that the matter is regular at the singularity indicates that the singularity is weaker than a standard big bang singularity , and may be easier to cure " with quantum corrections . by performing an expansion near the initial state , using eq .
( [ fnew ] ) , we find that the primordial hubble rate in gbig1 , after the infinite deceleration at the birth of the universe , is given by @xmath121^{1/2}(\rho_{\rm i}-\rho)^{1/2}.\ ] ] this is independent of the equation of state @xmath40 . if there is primordial inflation in the gbig1 universe , then the acceleration @xmath122 will become positive . for a realistic model ( satisfying nucleosynthesis and other constraints )
, @xmath122 must subsequently become negative again , so that the universe decelerates during radiation- and early matter - domination .
finally , @xmath122 will become positive again as the late universe self - accelerates .
vs @xmath123 , for a gbig1 cosmology with inflation , followed by radiation domination , followed by late - time self - acceleration .
arrows indicate the flow of brane proper time . here
@xmath44 , and @xmath124 in eq . ( [ eos ] ) . , title="fig:",width=264,height=288 ] ( -3.45,-2.45)@xmath16 ( -0.8,-6)@xmath16 we can simplify the expression ( [ acc ] ) for the acceleration in gbig1 via eq .
( [ fnew ] ) , @xmath125}{2(3\gamma x+1-\sqrt{x})}\!,\ ] ] where @xmath126 . for a given @xmath127 , we can plot @xmath128 .
we show an example in fig .
[ cos ] of a simple model , with primordial inflation followed by radiation domination , followed by late - time self - acceleration .
we have used the effective equation of state @xmath129 here @xmath130 is a parameter determining the time of reheating ( with @xmath131 corresponding to no inflation and @xmath132 to no reheating/ radiation ) .
in summary , the gbig1 model provides an intriguing generalization of the dgp(+ ) model the gauss - bonnet ( ultra - violet ) correction to the ( infra - red ) induced gravity preserves the late - time self - acceleration of the universe , but leads to striking new behaviour in the early universe . although there is still a curvature singularity at the beginning , the density , pressure and temperature are finite .
this model deserves further investigation as a viable cosmological model .
future work will impose constraints on the model parameters from nucleosynthesis and supernova redshifts ( compare ref .
we expect that these constraints will not differ appreciably from the dgp(+ ) case , given the very small value of the gb parameter @xmath34 that is imposed by the uv - ir bootstrap .
however , a nonzero @xmath34 , no matter how small , leads to dramatic and nonperturbative changes at high energies in the primordial universe . from a theoretical viewpoint , it will also be important to investigate how the gb term affects the issues of strong coupling and ghosts in the dgp(+ ) cosmological model @xcite .
can the gb term provide a lowest - order ultra - violet completion of the dgp(+ ) theory ?
the analysis of perturbations about a minkowski brane with gb and ig terms @xcite gives a starting point for tackling the friedmann brane case .
@xmath133*acknowledgements : * rb and rm are supported by pparc , ep and vz by the ntua research program protagoras and by ( epeaek ii)-pythagoras ( co - funded by the european social fund and national resources ) .
we thank mariam bouhmadi - lopez , georgios kofinas , kazuya koyama and alexey toporensky for very helpful discussions .
see , e.g. , p. brax , c. van de bruck and a. c. davis , rept .
phys . * 67 * , 2183 ( 2004 ) [ arxiv : hep - th/0404011 ] .
see , e.g. , r. gregory , v. a. rubakov and s. m. sibiryakov , phys .
lett . * 84 * , 5928 ( 2000 ) [ arxiv : hep - th/0002072 ] ; i. i. kogan , arxiv : astro - ph/0108220 ; a. papazoglou , arxiv : hep - ph/0112159 ; a. padilla , class .
* 22 * , 681 ( 2005 ) [ arxiv : hep - th/0406157 ] ; a. padilla , class .
grav . * 22 * , 1087 ( 2005 ) [ arxiv : hep - th/0410033 ] .
g. r. dvali , g. gabadadze and m. porrati , phys .
b * 485 * , 208 ( 2000 ) [ arxiv : hep - th/0005016 ] .
h. collins and b. holdom , phys .
d * 62 * , 105009 ( 2000 ) [ arxiv : hep - ph/0003173 ] ; y. v. shtanov , arxiv : hep - th/0005193 . c. deffayet , phys .
b * 502 * , 199 ( 2001 ) [ arxiv : hep - th/0010186 ] .
see , e.g. , c. charmousis and j. f. dufaux , class .
* 19 * , 4671 ( 2002 ) [ arxiv : hep - th/0202107 ] ; s. nojiri , s. d. odintsov and s. ogushi , int .
j. mod .
a * 17 * , 4809 ( 2002 ) [ arxiv : hep - th/0205187 ] ; s. c. davis , phys .
d * 67 * , 024030 ( 2003 ) [ arxiv : hep - th/0208205 ] ; e. gravanis and s. willison , phys .
b * 562 * , 118 ( 2003 ) [ arxiv : hep - th/0209076 ] ; j. e. lidsey and n. j. nunes , phys . rev .
d * 67 * , 103510 ( 2003 ) [ arxiv : astro - ph/0303168 ] ; k. i. maeda and t. torii , phys .
d * 69 * , 024002 ( 2004 ) [ arxiv : hep - th/0309152 ] ; j. f. dufaux , j. e. lidsey , r. maartens and m. sami , phys .
d * 70 * , 083525 ( 2004 ) [ arxiv : hep - th/0404161 ] ; t. g. rizzo , jhep * 0501 * , 028 ( 2005 ) [ arxiv : hep - ph/0412087 ] .
n. e. mavromatos and e. papantonopoulos , arxiv : hep - th/0503243 .
the assumption that @xmath134 is non - negative is motivated by string theory , where typically @xmath135 .
g. kofinas , r. maartens and e. papantonopoulos , jhep * 0310 * , 066 ( 2003 ) [ arxiv : hep - th/0307138 ] .
note that the @xmath136 branch corresponds to an anti de sitter bulk , even though there is no cosmological constant .
t. tanaka , phys .
d * 69 * , 024001 ( 2004 ) [ arxiv : gr - qc/0305031 ] ; c. deffayet , phys .
d * 71 * , 103501 ( 2005 ) [ arxiv : gr - qc/0412114 ] ; k. koyama and k. koyama , phys .
d * 72 * ( 2005 ) 043511 [ arxiv : hep - th/0501232 ] .
m. bouhmadi - lopez , unpublished notes .
y. shtanov and v. sahni , class .
* 19 * , l101 ( 2002 ) [ arxiv : gr - qc/0204040 ] .
j. d. barrow , class .
* 21 * , l79 ( 2004 ) [ arxiv : gr - qc/0403084 ] .
r. g. cai , h. s. zhang and a. wang , arxiv : hep - th/0505186 .
c. deffayet , g. r. dvali , g. gabadadze and a. i. vainshtein , phys .
d * 65 * ( 2002 ) 044026 [ arxiv : hep - th/0106001 ] ; m. a. luty , m. porrati and r. rattazzi , jhep * 0309 * , 029 ( 2003 ) [ arxiv : hep - th/0303116 ] ; v. a. rubakov , arxiv : hep - th/0303125 ; g. dvali , arxiv : hep - th/0402130 ; a. nicolis and r. rattazzi , jhep * 0406 * , 059 ( 2004 ) [ arxiv : hep - th/0404159 ] ; n. kaloper , phys .
d * 71 * , 086003 ( 2005 ) [ erratum ibid .
d * 71 * , 129905 ( 2005 ) ] [ arxiv : hep - th/0502035 ] ; k. koyama , arxiv : hep - th/0503191 ; c. deffayet and j. w. rombouts , arxiv : gr - qc/0505134 . c. charmousis and j. f. dufaux , phys .
d * 70 * , 106002 ( 2004 ) [ arxiv : hep - th/0311267 ] ; p. brax , n. chatillon and d. a. steer , phys .
b * 608 * , 130 ( 2005 ) [ arxiv : hep - th/0411058 ] .
|
brane - world models offer the possibility of explaining the late acceleration of the universe via infra - red modifications to general relativity , rather than a dark energy field .
however , one also expects ultra - violet modifications to general relativity , when high - energy stringy effects in the early universe begin to grow .
we generalize the dgp brane - world model via an ultra - violet modification , in the form of a gauss - bonnet term in the bulk action .
the combination of infra - red and ultra - violet modifications produces an intriguing cosmology .
the dgp feature of late - time acceleration without dark energy is preserved , but there is an entirely new feature there is no infinite - temperature big bang in the early universe .
the universe starts with finite density and pressure , from a sudden " curvature singularity .
|
the nuclear physics data accumulated up to now indicate that there is an essential difference between spherical and nonspherical nuclei .
this difference does not reduce merely to nuclear shape effects " as such , but intimately involves the internal structure of the nucleus .
the shell oscillations of nuclear masses [ 1 ] can serve as an example of the phenomena associated with the difference between the structures of spherical and nonspherical nuclei .
the shell oscillations are due to the ordered grouping of one - quasiparticle levels in spherical nuclei that arises because of the existence of the orbital angular momentum quantum number @xmath0 of the quasiparticle [ 1,2 ] .
there is no such ordered grouping in nonspherical nuclei , and the trend of the mass of such nuclei as a function of the number of nucleons is entirely different .
this difference is manifested on a plot of nuclear masses versus number of nucleons as a break in the curve corresponding to a phase transformation from the normal phase ( nonspherical nuclei ) to the magic phase " ( spherical nuclei ) [ 3 ] .
the difference between the internal structures of spherical and nonspherical nuclei does not affect only the ground states : the thermodynamic properties of the two phases are also different .
this difference is manifested , for example , in the dependence of the form of the spectra of the @xmath1-ray cascades emitted by heated nuclei on phase of the compound nucleus [ 4,5 ] .
the energy spectrum of a nonspherical nucleus is not distorted by the residual interaction and is essentially of the fermi - liquid type .
spherical nuclei , on the other hand , apparently can not exist without the residual interaction ( this was proved in [ 6 ] for a very simple nuclear model ) , and this interaction has an important effect on their thermodynamic behavior . as was explained in [ 2 ] , the residual interaction must have a certain macroscopic structure , the only interactions between the quasiparticles that are compatible with the observed pattern of the shell oscillations of nuclear masses being those that lead to a quadratic dependence of the width of the diffuse region of the fermi distribution on the orbital angular momentum @xmath0 of the quasiparticle : @xmath2 in this paper we consider the thermodynamics of spherical nuclei on the basis of the simplest model of the residual interaction that has the required macroscopic structure .
the racah - mottelson model was used in [ 2 ] to describe the residual interaction : the hamiltonian for the interaction between quasiparticles on the same @xmath3 level is given by @xmath4 where @xmath5 and @xmath6 are the creation and annihilation operators for a quasiparticle whose angular momentum has the @xmath7 projection @xmath8 .
the eigenvalues of this hamiltonian ( see , e.g. , [ 7 ] ) are @xmath9 where @xmath10 , @xmath11 is the number of interacting pairs on level @xmath3 , and @xmath12 is the seniority quantum number . to construct the thermodynamics of spherical nuclei , we need a formula for the degree of degeneracy @xmath13 ( we shall drop the subscript @xmath3 in the rest of this section ) of the racah - mottelson energy levels ( 3 ) .
the symmetry of the zero - seniority ground - state wave function ( see , e.g. , [ 8 ] ) is such that the wave function can be constructed from single - particle wave functions in only one way .
hence the @xmath14 ground state is not degenerate : @xmath15 .
one more quasiparticle can be added to the @xmath14 state in @xmath16 ways ( @xmath17 is the number of combinations of @xmath18 things taken @xmath19 at a time ) , and we thereby obtain an @xmath20 state , so that @xmath21 .
two quasiparticles can be added to the @xmath14 state with @xmath22 quasiparticles in @xmath23 ways , but among them there is one that describes the ground state of @xmath24 quasiparticles , so that @xmath25 .
arguing further in the same way we can prove the following general formula for the degree of degeneracy of the racah - mottelson energy levels by mathematical induction : @xmath26 for the case @xmath27 , @xmath28 , we can use stirling s formula @xmath29 to rewrite eq .
( 4 ) in the form @xmath30.\ ] ] now let us find the number @xmath31 of states with seniority @xmath19 and angular momentum projection @xmath32 .
it is clear that the sum of @xmath31 over all possible values of @xmath32 is @xmath13 . in order to find a way of evaluating @xmath31
let us consider the matrix of hamiltonian ( 2 ) in more detail .
the dimension of this matrix is the number @xmath33 of ways in which @xmath34 fermions can be distributed among @xmath18 cells .
( it is easily seen that the sum of @xmath35 over all seniorities @xmath19 possible for the given @xmath34 is @xmath33 . )
the hamiltonian matrix for the @xmath33 basis functions breaks up into blocks corresponding to different values of the quantum number @xmath19 .
however , the block for the states with a given seniority breaks up in turn into still smaller blocks , since the hamiltonian conserves the angular momentum projection @xmath32 , so that different values correspond to different blocks .
finally , the block for states with given fixed values of @xmath19 and @xmath32 also consists of smaller blocks , since different sets of unpaired particles can give rise to the same angular momentum projection @xmath32 .
( here the term pairing " is used in a purely kinematic sense , namely , as an expression of the fact that two quasiparticles lie in cells corresponding to the same absolute value but opposite signs of the @xmath7 projection of the angular momentum . ) for definiteness let is consider the case of even @xmath19 ( and hence even @xmath34 ) .
then the smallest " blocks in the matrix for hamiltonian ( 2 ) describe states in which there occur some number @xmath36 of unpaired quasiparticles giving a total angular momentum projection @xmath32 , and @xmath37 paired quasiparticles , which have @xmath38 and seniority @xmath39 .
the number of such blocks is equal to the number @xmath40 of ways in which the angular momentum projection @xmath32 can be expressed as the sum of @xmath36 single - particle projections taken from the set @xmath41 ( no two single - particle projections equal in modulus but opposite in sign occurring in the sum ) , while the dimension of each block is equal to the degree of degeneracy @xmath42 of a state of paired quasiparticles that are distributed among @xmath43 free cells and have seniority @xmath39 .
thus , the sum of the dimensions of all the blocks for states with angular momentum projection @xmath32 and seniority @xmath19 is @xmath44 a similar expression can also be written for odd @xmath19 .
direct diagonalization of the hamiltonian matrix for small values of @xmath45 and @xmath34 leads to values of @xmath46 that can be expressed by the formula @xmath47 by definition , @xmath46 is meaningful only for even @xmath19 .
if we assume that formula ( 7 ) is valid for all even seniorities @xmath48 , we can prove that it is valid for seniority @xmath49 in much the same way as eq .
( 4 ) was proved : hence it follows by mathematical induction that ( 7 ) is valid for all even seniorities @xmath50 .
now let us seek the number @xmath51 of ways in which @xmath19 unpaired particles can give a total angular momentum projection of @xmath32 . since the formula for @xmath31 will be used later to describe macroscopic phenomena for which it is essential that @xmath52 , we can set @xmath53 ( also see [ 2 ] ) , so that the single - particle angular momentum projections are the elements of the set @xmath54 .
now we shall construct a generating function for the unknown numbers @xmath55 . to do this we consider the sum @xmath56 taken over the values @xmath57 under the condition that @xmath58 the coefficient of @xmath59 in the sum ( 8) is @xmath60 . to take condition ( 10 ) into account , we make use of the identity @xmath61 in which the integration contour goes around the origin . if we multiply ( 11 ) by ( 8) , we can sum over all the values ( 9 ) independently , without regard for condition ( 10 ) . as a result
we obtain the following expression for the generating function : @xmath62 if we make the substitution @xmath63 in ( 12 ) , the sum over @xmath32 becomes a finite fourier series . using the orthogonality relation @xmath64
we obtain the following expression for the coefficients : @xmath65 now we expand the product of the expressions in parentheses in powers of @xmath66 and extend the integration over the entire real axis , obtaining : @xmath67\exp\left[-\frac{3(z+2 ) } { 4\omega^3}m^2\right].\ ] ] this operation is legitimate since the integral over @xmath66 converges for @xmath68 and the characteristic values of @xmath7 ( see below ) are such that @xmath69 , so that @xmath70 .
we evaluate the remaining integral over @xmath7 by the saddle - point method , choosing the integration contour so that the convergence of the integral will be due mainly to the function @xmath71=z^s\exp[\omega\ln(1 + 2/z)]=\exp[(s-\omega)\ln z+\omega\ln(z+2)].\ ] ] at the saddle - point we have @xmath72 and we must therefore integrate over the contour @xmath73 . as a result we obtain @xmath74\exp(-3m^2/2s\omega^2 ) .
\end{split}\ ] ] it is easy to show that the condition for the validity of eq .
( 18 ) is that @xmath28 . to calculate @xmath31
it remains to substitute expressions ( 7 ) and ( 18 ) into ( 6 ) and perform the summation .
approximating the right - hand side of ( 7 ) with the aid of stirling s formula and replacing the summation by integration , we obtain @xmath75\exp(-3m^2/4k\omega^2)dk}{k(\omega - s/2-k)\sqrt{(s/2-k)(\omega - s/2-k)}},\ ] ] where @xmath76 the function @xmath77 has a maximum at the point @xmath78 and at that point @xmath79 thus , if @xmath80 then @xmath77 will have a sharp maximum within the integration range , so we can replace the other slowly varying functions by their values at @xmath81 and integrate over @xmath82 from @xmath83 to @xmath84 . as a result we obtain the following final expression : @xmath85\\ & \times \exp[-3m^2/\omega s(2\omega - s ) ] .
\end{split}\ ] ] condition ( 23 ) for the validity of eq .
( 24 ) is satisfied when @xmath86 .
it should be noted , however , that the exponential factors in ( 24 ) , which are essential for the thermodynamic applications , are determined only by the position ( 21 ) of the maximum , so that the condition @xmath86 is essentially just a condition for the validity of the preexponential factor in ( 24 ) ; the exponential factor , on the other hand , is valid for @xmath28 .
on integrating ( 24 ) over @xmath32 from @xmath83 to @xmath84 we again obtain eq .
( 5 ) . with
that we conclude our study of an individual @xmath3 level of the racah - mottelson model and turn to the application of the model to the nucleus as a whole .
let us calculate the function describing the distribution of quasiparticles in a nucleus heated to the temperature @xmath87 . the preexponential factor in ( 5 ) can be neglected in calculating the entropy of a single @xmath3 level with the formula @xmath88 .
then we obtain the following expression for the entropy of the nucleus as a whole : @xmath89.\ ] ] to find the function describing the distribution of the quasiparticles among the levels , we must minimize the free energy @xmath90 ( here @xmath91 is the energy of a quasiparticle on level @xmath3 without the residual interaction ) under the condition that the number of quasiparticles , @xmath92 remains constant . in view of the macroscopic nature of the problem under study
we have neglected the 1 " in eq .
( 3 ) for the interaction energy ( see [ 2 ] ) . condition ( 27 ) can be easily taken into account by the lagrange multiplier method . by varying the function @xmath93
( @xmath94 is the chemical potential ) with respect to @xmath11 and @xmath12 we obtain the set of equations @xmath95^{-1 } , \end{split}\ ] ] for distribution functions @xmath96 and @xmath97 in the region @xmath98 @xmath99 ( region ii on fig . 1 ) . on substituting the expression for @xmath100 from the first of eqs
( 28 ) into the second we obtain the following equation for the distribution function @xmath101 : @xmath102 + 1\right\}^{-1 } \quad \text{(region ii).}\ ] ] on increasing @xmath91 , we pass from region ii to region iii , where there are no paired quasiparticles , i.e. , @xmath103 . on taking this additional condition into account in the @xmath12 variation we obtain the following expression for the distribution function @xmath101 : @xmath104^{-1 } \quad \text{(region iii),}\ ] ] i.e. , an ordinary fermi distribution . on decreasing @xmath91 we pass from region ii to region i , where all the vacancies permitted by the pauli principle are filled , i.e. , @xmath105 .
this auxiliary condition gives a fermi distribution for the holes : @xmath106^{-1 } \quad \text{(region i).}\ ] ] the equations for the boundaries between regions i , ii , and iii are obtained by equating the expressions for the distribution functions on the two sides of the boundary .
this gives the equation @xmath107 for the boundary between regions ii and iii .
we denote the solution of this equation by @xmath108 for the boundary between regions i and ii we obtain @xmath109 since the solution of eq .
( 29 ) is independent of @xmath91 , its value within region ii is equal to its value on the boundary ( for constant @xmath110 ) or to the values of functions ( 30 ) and ( 31 ) on the corresponding boundaries ( 33 ) and ( 34 ) , so that @xmath111 + 1\right\}^{-1 } \quad \text{(region ii).}\ ] ] a graph of the function @xmath112 , defined implicitly by the equation @xmath113 is shown in fig . 2 .
we have @xmath114 as @xmath115 , and @xmath116 as @xmath117 .
equations ( 30)-(36 ) completely determine the distribution function @xmath101 and make it possible to calculate the entropy of a spherical nucleus from eq .
( 25 ) .
it is convenient to do the calculation in the variables @xmath118 and @xmath119 introduced in [ 6,2 ] ( @xmath120 , @xmath82 is the wave number of the quasiparticle , and @xmath121 is the radius of the nucleus ) : @xmath122=\pi(n+3/4 ) , \\ & \widetilde{\beta}=\arcsin(\tilde{l}/\rho),\quad \tilde{l}=l+1/2,\qquad dnd\rho=(\rho d\rho/\pi)\cos^2\widetilde{\beta } d\widetilde{\beta } , \end{split}\ ] ] where @xmath123 is the principal quantum number . to the macroscopic accuracy that we are confining ourselves to ,
we can replace the total angular momentum @xmath3 of the quasiparticle by its orbital angular momentum @xmath0 , so that @xmath124 .
taking also into account the fact that in our theory small values @xmath119 and @xmath125 are important at low temperatures ( @xmath126 , where @xmath127 is the wave number at the fermi boundary ) we can put @xmath128 in order to maintain the required macroscopic structure ( see the introduction ) in this same limit , it is necessary , as was shown in [ 2 ] , that @xmath129 the distribution function @xmath130 assumes the following form in the variables @xmath118 and @xmath119 ( see fig .
3 ) : @xmath131^{-1},\qquad\qquad \rho-\rho_f < -\tau{\gamma}\left(\frac{g\rho_f\widetilde{\beta}^2}{2\tau}\right),\\ \left[\exp\left({\gamma}\left(\frac{g\rho_f\widetilde{\beta}^2}{2\tau}\right)\right)+1\right]^{-1 } , \quad -\tau{\gamma}\left(\frac{g\rho_f\widetilde{\beta}^2}{2\tau}\right)<\rho-\rho_f < \tau{\gamma}\left(\frac{g\rho_f\widetilde{\beta}^2}{2\tau}\right),\\ \left[\exp\left(\frac{\rho-\rho_f}\tau\right)+1\right]^{-1},\qquad\qquad \rho-\rho_f > \tau{\gamma}\left(\frac{g\rho_f\widetilde{\beta}^2}{2\tau}\right ) .
\end{array } \right.\ ] ] we also transform to the variables @xmath118 and @xmath119 in eq .
( 25 ) , and using the jacobian @xmath132 ( see eq .
( 37 ) ) to replace the summation over the levels by an integration , we obtain @xmath133 \frac{\rho_f}\pi d\rho d \widetilde{\beta}.\ ] ] substituting ( 40 ) into ( 41 ) and integrating , we obtain the following expression for the entropy of one nucleon component of a spherical nucleus : @xmath134 where @xmath135 is the effective mass of a quasiparticle in the nucleus . taking into account the fact that nuclear matter is a two - component system , we find @xmath136 thus , in spherical nuclei ( at low enough temperatures
see below ) the entropy depends quadratically on the temperature , in contrast to the frequently assumed linear dependence .
the latter is characteristic of the energy spectrum of a normal fermi liquid1 [ 9,10 ] , but in nuclear physics it can apparently be regarded as fairly reliably established only for nonspherical nuclei .
the thermodynamic relations @xmath137 follow from eqs .
( 43 ) , so that the level density in a spherical nucleus depends as follows on the excitation energy : @xmath138=\exp\left(\frac{3^{2/3}}2b^{1/2}e^{2/3}\right).\ ] ] now let us calculate the moment of inertia of a spherical nucleus . to do this we shall examine the racah - mottelson states that have not only a definite seniority @xmath12 , but also a fixed angular momentum projection @xmath139 .
then the entropy of level @xmath3 will be @xmath140 , where the number of states is given by eq .
( 24 ) ( as before , we neglect the preexponential factor ) .
the distribution functions for the quasiparticles and the angular momentum @xmath139 are determined from the condition that the free energy @xmath141 be minimum , where @xmath142 is the angular velocity and @xmath32 is the total angular momentum projection .
thus , the @xmath139 dependent terms @xmath143\ ] ] are added to the terms determined by formulas ( 25)(27 ) .
varying @xmath139 yields the angular momentum projection @xmath144 of the @xmath3 level , so that the total angular momentum projection is given by @xmath145 in this case the distribution function @xmath101 depends on @xmath142 , but for small @xmath142 we can neglect this dependence in eq .
( 48 ) and use eq .
the coefficient of @xmath142 in ( 48 ) is the moment of inertia : in the usual units ( @xmath32 in erg@xmath146sec and @xmath142 in sec@xmath147 ) it takes the form @xmath148 now we transform ( 49 ) to the variables @xmath149 @xmath119 and substitute expression ( 40 ) for the distribution function ; then after integrating over regions i iii we obtain @xmath150 in our calculation , however , we have not taken into account the effect of the second functional derivative of the energy with respect to the distribution function [ 9,10 ] , which leads to an additional constant factor in eq .
this factor can be found by calculating the moment of inertia of a normal fermi liquid by a method similar to that used above .
a simple calculation shows that the degree of degeneracy of a state with @xmath19 fermions on level @xmath3 having the angular momentum projection @xmath32 differs from ( 24 ) only in the preexponential factor , which is not important for thermodynamic applications .
this means that in the thermodynamic sense the seniorities @xmath19 are fermion excitations with a dispersion law leading to distribution function ( 40 ) .
the moment of inertia of a fermi liquid is therefore given by eq .
( 49 ) with the ordinary fermi distribution substituted for @xmath130 .
now we easily find @xmath151 here , too , we have neglected the second functional derivative .
requiring that taking this into account should lead to the rigid - body value @xmath152 , where @xmath153 is the nucleon mass , we find @xmath154 i.e. , @xmath155 should have the fermi - gas value .
hence one of the factors @xmath155 in eq .
( 50 ) should also have the fermi - gas value . taking this into account
, as well as the fact that nuclear matter is a two - component system , we obtain the expression @xmath156 for the moment of inertia of a spherical nucleus .
the function describing the distribution of the nuclear levels over the angular momentum @xmath157 at fixed temperature has the form ( see [ 11 ] ) @xmath158 so that the mean and mean square values of the angular momentum are @xmath159 thus we have @xmath160 and this relation can be used to check the correctness and completeness of the experimental data on level spins .
we also note the equation @xmath161 which follows from eqs .
( 55 ) , ( 53 ) , and ( 44 ) .
as we already mentioned , the formulas derived here are valid only at low temperatures . in order to find out how low the temperature must be , we take account of the fact that eq . (
39 ) was established only for angles satisfying the condition @xmath162 ( see [ 2 ] ) .
hence the quantity @xmath163 which is characteristic of our theory , marking , as it does , the point at which region ii begins to have a finite extent along the @xmath164 or @xmath118 axis ( see figs . 1 and 3 and eq .
( 40 ) ) , should satisfy condition ( 58 ) .
this gives the criterion @xmath165 where @xmath166 is the characteristic width of the diffuse region of the fermi distribution caused by the residual reaction ( see [ 2 ] ) .
the values of @xmath167 found in [ 2 ] range from 1.9 mev ( heavy nuclei ) to 5.2 mev ( light nuclei ) for neutron magic nuclei and from 1.6 mev ( heavy nuclei ) to 3.8 mev ( light nuclei ) for proton magic nuclei .
however , it is difficult at present to judge the part played by another criterion , @xmath168 implied by the theory , where @xmath169 is the curie temperature at which the spherical nucleus passes from the magic phase to the normal phase , i.e. , becomes nonspherical ( see [ 3 ] ) .
we can only assert that condition ( 61 ) becomes more important , and may become dominant , for spherical nuclei that lie close as regards number of nucleons to the phase transformation point .
the @xmath1-ray spectra and radiative widths of nuclei that are spherical in the ground state confirm this ( see [ 5,12 ] ) .
we also note that all the equations we have derived relate the average values of the thermodynamic quantities to one another , i.e. , they are valid only when the fluctuations are small .
however , the temperature fluctuations [ 10 ] @xmath170 amount to @xmath171 mev for @xmath172 , so that the thermodynamic formulas can only serve as rough estimates at the excitation energies now being investigated for spherical nuclei .
for example , it follows from eq .
( 57 ) that @xmath173 for @xmath174pb at an excitation energy of 5 mev , whereas eqs .
( 55 ) with the rigid - body moment of inertia and temperature @xmath175 mev ( which is apparently characteristic of nonspherical nuclides ; see [ 12 ] ) give @xmath176 . as well as
can be judged from the available data [ 13 ] on levels of @xmath174pb , however , the experimental value is @xmath177 , which is in qualitative agreement with the theory under discussion . although there is no thermal phase transformation in the present theory , the macroscopic structure of the residual interaction as a function of the orbital quantum number @xmath0 together with the fact ( a macroscopic quantum effect ) that the spherical nuclear configuration is unstable in the absence of interaction between the quasiparticles ( see [ 6 ] ) enables us to suggest a possible reason for such a transformation .
in fact , as the temperature rises region ii shifts to the right on the @xmath0 axis and more and more quasiparticles fall into regions i and iii , where they conform to the ordinary fermi distribution .
hence a phase transformation from the magic phase to the normal phase should take place when there are no longer enough quasiparticles left in region ii to stabilize the spherical shape of the nucleus .
l. groshev , a. demidov , v. lutsenko , and v. pelekhov , atlas spektrov @xmath1-luchej radiatsionnogo zakhvata teplovykh neitronov [ atlas of @xmath1-ray spectra from radiative capture of thermal neutrons ] , atomizdat , 1958 [ english translation , pergamon press , london , 1959 ] .
|
the effect of the residual interaction between the nucleons ( quasiparticles ) on the thermodynamic behavior of spherical nuclei is considered .
thermodynamic relations are derived for the simplest model of the residual interaction that has the necessary macroscopic structure previously established .
the range of applicability of the theory is discussed .
|
massive star clusters are an important mode of star formation , having an impact on a wide range of galaxy properties .
however , the earliest stages of these clusters are notoriously challenging to study since the youngest clusters are still enshrouded in remnants of their gaseous and dusty birth cocoons . properly accounting for the effects of gas and dust on observations of young massive clusters ( ymcs )
is nontrivial , rendering it difficult to understand their formation and earliest phases of evolution .
some of these difficulties , such as interstellar extinction , are commonly dealt with in studies of young star clusters . while it is often not possible to completely disentangle the effects of extinction , the general effects are well - known and accounted for ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
. on the other hand , the treatment of ionized gas emission can be quite complicated and the effects of nebular emission on broad - band observations of ymcs are not always obvious .
emission lines detected via spectroscopy or narrow - band imaging , however , have proved useful for both identifying young ( @xmath7 myr ) clusters ( e.g. h@xmath8 ) and serving as valuable diagnostics on their physical conditions ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
if these lines are strong enough , as is typically the case for massive star forming regions , the composite emission can even affect the integrated broad - band photometry ( e.g. * ? ? ?
* ; * ? ? ?
. however , nebular emission also takes on a more subtle form via continuum emission ( e.g. free - free , free - bound ) .
in fact , discontinuities or `` jumps '' in the nebular continuum ( e.g. the balmer and paschen jumps ) can be strong enough to serve as diagnostics in their own right ( see * ? ? ? * and references therein ) .
this nebular continuum emission can also severely affect the broad - band photometry of very young clusters ( e.g. * ? ? ?
* ; * ? ? ?
* ) . in this paper
we investigate the impact of nebular continuum and line emission on observations of very young massive star clusters .
we carry out a detailed multiwavelength study of two ymcs that are still partially embedded in their birth material .
the clusters were targeted for spectroscopic follow - up after @xcite found an _ hst _
f814w @xmath0-band excess in their photometric study of radio - detected clusters in the nearby dwarf starburst galaxy ngc 4449 .
@xcite considered several possible explanations for this @xmath0-band excess , including red supergiants , emission lines , thermal emission from hot dust , continuum emission from a sub - population of deeply embedded stars , and extended red emission ( ere ; * ? ? ?
* ) . with the observations presented here
, we have determined that the _ combination _ of nebular continuum and line emission can account for the @xmath0-band excess found by @xcite .
consequently , the chief goal of this study is to gain a better empirical understanding of the effects of nebular emission ( both continuum and line ) associated with extremely young massive star clusters .
this paper is organized as follows : the data are described in [ sec_data ] . in [ sec_lines ] we estimate the physical properties of the clusters from their h@xmath8 emission . we model the spectral energy distributions ( seds ) in [ sec_model ] and investigate the impact of nebular emission ( line and continuum ) on broad - band photometry in [ sec_broad - band ] . a summary of our main conclusions is given in
[ sec_summary ] .
a three - color sdss image of ngc 4449 is shown in figure [ image_n4449 ] with our target ymcs labeled .
source 26 ( according to the notation of * ? ? ?
* ) was selected for its brightness ( to get a high signal - to - noise spectrum ) and source 30 was targeted for its large @xmath0-band excess
the largest of all of the clusters studied in @xcite .
we adopt a distance of 3.9 mpc to ngc 4449 as in @xcite , consistent with the results of @xcite . at this distance , @xmath9 pc .
spectra of the two ymcs were obtained using the 3.5 m apo telescope .
source 26 ( ra = 12@xmath10 28@xmath11 13.86@xmath12 , dec = + 44@xmath13 07 10.4 ) was observed on the night of 2008 april 13 under clear conditions and source 30 ( ra = 12@xmath10 28@xmath11 16.02@xmath12 , dec = + 44@xmath13 06 29.3 ) was observed on 2009 may 24 through intermittent clouds .
the observations were made using the red and blue channels of the dual imaging spectrograph ( dis ) in the low - resolution mode with the b400 and r300 gratings .
the linear dispersions for the blue and red channels are 1.83 and 2.31 pix@xmath14 , respectively , and the central wavelengths are @xmath34400 and @xmath37500 .
we used the @xmath15 slit oriented along the parallactic angle and centered on the ymcs .
the resulting spectra have a wavelength range of @xmath33800 - 9800 and a resolution of @xmath37 .
the total exposure time on source 26 was 30 minutes ( 2 @xmath16 15 minutes ) , resulting in signal - to - noise ratios of @xmath360 and @xmath345 in the blue and red continua , respectively .
source 30 was observed for a total of 2.5 hours ( 5 @xmath16 30 minutes ) and the signal - to - noise ratio of both the blue and red continua is @xmath35 .
multiple exposures of each cluster were taken to allow for the removal of cosmic rays .
the white dwarf primary spectrophotometric standard star gd 153 @xcite was observed for flux calibration and a he - ne - ar arc lamp spectrum was taken for wavelength calibration .
the data were reduced using a combination of iraf and custom idl routines .
the two - dimensional images were processed using standard procedures in iraf including bias subtraction and flat - fielding @xcite .
iraf was also used to extract the one - dimensional spectra , subtract the background , and apply the wavelength calibration @xcite .
the spectra were extracted from an aperture window of 11 pixels ( 44 ) in the spatial dimension .
the remaining data reduction and analysis was performed in idl .
flux calibration was obtained by comparing the observed spectrum of gd 153 to the model spectrum obtained from the space telescope science institute s calibration data base system .
atmospheric extinction was corrected for using the standard extinction curve for apo .
finally , the blue and red spectral regions were combined and put on a common wavelength scale of 2 pix@xmath14 .
the blue and red spectra were averaged in an overlap region approximately 50 pixels wide and centered at @xmath17 .
after a comparison with the photometry ( see below ) , the flux calibrated spectra were multiplied by scale factors of 1.94 ( source 26 ) and 1.84 ( source 30 ) to account for loss of light in the slit .
we scaled the spectra to match the _ hst _
f550 m ( narrow @xmath1-band ) data since this filter has a negligible contribution from emission lines .
figure [ plot_spectra ] shows the final reduced spectra of sources 26 and 30 .
the spectra exhibit many strong emission lines and nebular continuum emission as is evident by the paschen jump at 8207 . the wolf - rayet blended emission feature at @xmath18 known as the `` w - r bump '' ( e.g. * ? ? ? * ; * ? ? ?
* ) is also clearly visible in the spectrum of source 26 .
cccccc f170w & wfpc2 & uv & 400 @xmath16 2 & 6716 & t. stecher + f336w & wfpc2 & wfpc2 u & 260 @xmath16 2 & 6716 & t. stecher + f435w & acs / wfc & johnson b & 3660 & 10585 & a. aloisi + f550
m & acs / wfc & narrow v & 1200 & 10522 & d. calzetti + f658n & acs / wfc & h@xmath8 + [ nii]@xmath19 & 1539 & 10585 & a. aloisi + f660n & acs / wfc & [ nii]@xmath19 & 1860 & 10522 & d. calzetti + f814w & acs / wfc & broad i & 2060 & 10585 & a. aloisi _ hst _
observations of ngc 4449 taken with the advanced camera for surveys ( acs ) and the wide field and planetary camera 2 ( wfpc2 ) were obtained from the archive .
these data are described in detail in @xcite and summarized in table [ tab_hstobs ] .
here we use wfpc2 f170w and f336w images , as well as acs wide field channel ( wfc ) images through the f435w , f550 m , f658n , f660n and f814w filters .
the broad - band filters ( f170w , f336w , f435w , f550 m and f814w ) span ultraviolet through near - ir wavelengths . the narrow - band filters , f658n and f660n , capture h@xmath8 + [ n ii ] @xmath19 and [ n ii ] @xmath19 emission , respectively .
we have also obtained @xmath20 imaging of ngc 4449 from the sdss data release 6 @xcite .
a three - color ( rgb=@xmath21 ) image of the galaxy is shown in figure [ image_n4449 ] with our target ymcs labeled . ) , f814w ( broad @xmath0 ) , and f550 m ( narrow @xmath1 ) filters .
the circles indicate the photometric apertures used on the _ hst _ and sdss imaging ( see [ sec_phot ] ) .
[ hstim ] ] photometry of the ymcs was performed using surphot @xcite , a custom idl program allowing for consistent apertures and background annuli across multiple wavebands .
circular apertures of radii 33 and 17 were used for sources 26 and 30 , respectively .
background levels were determined in annuli with inner and outer radii equal to @xmath22 and @xmath23 the aperture radii .
aperture sizes were chosen to include essentially all of the light from the clusters in all bands , and were mostly influenced by the ( seeing - limited ) resolution of the ground - based sdss images and the intrinsic cluster sizes .
figure [ hstim ] shows a three - color _ hst _ image of the clusters with the photometric apertures overlaid .
flux densities for the ymcs are given in table [ tab_phot ] and plotted in figure [ plot_spectra ] along with the spectra .
cccc f170w & 1821 & @xmath24 & @xmath25 + f336w & 3359 & @xmath26 & @xmath27 + f435w & 4317 & @xmath28 & @xmath29 + f550
m & 5581 & @xmath30 & @xmath31 + f658n & 6584 & @xmath32 & @xmath33 + f660n & 6599 & @xmath34 & @xmath35 + f814w & 8061 & @xmath36 & @xmath31 + @xmath37 & 3543 & @xmath38 & @xmath39 + @xmath40 & 4770 & @xmath41 & @xmath42 + @xmath43 & 6231 & @xmath44 & @xmath45 + @xmath46 & 7625 & @xmath47 & @xmath48 + @xmath49 & 9134 & @xmath50 & @xmath51 in order to compare our spectroscopic observations with the broad - band imaging data , we simulate `` dis photometry '' by convolving the ymc spectra with the total system throughput curves for the @xmath40 , f550 m , @xmath43 , @xmath46 , and f814w filters . filters . ]
figure [ plot_phots ] shows the broad - band flux densities obtained from the imaging data ( listed in table [ tab_phot ] ) , as well as the flux densities obtained from the spectroscopy .
figure [ plot_phots ] illustrates that the flux densities obtained from convolving the spectra with the filter throughput curves match the the photometry quite well , especially for source 26 which has a high signal - to - noise spectrum .
however , the sdss @xmath40 flux densities for source 30 are clearly discrepant , with the spectroscopic value significantly higher than the photometric value .
we believe this is due to an incorrect absolute flux level in the blue part of the spectrum for two reasons .
first , we expect the f435w photometric data point to lie above the continuum in figure [ plot_spectra ] since this filter contains strong emission lines ( e.g. h@xmath52 and h@xmath53 ) .
this effect can be seen for source 26 in figure [ plot_spectra ] ( i.e. the f435w data point lies above the continuum in the spectrum ) , and the emission lines are even stronger relative to the continuum for source 30 .
the second reason we think the blue end of the spectrum of source 30 has an incorrect flux level is that no single model sed can reproduce the shape of the continuum in the spectrum .
there is no combination of age and extinction that can simultaneously produce such a steep blue spectrum and flattened red spectrum .
also , we show later in [ sec_model ] that the best - fitting sed to the broad - band _ photometric _ data has a much flatter slope than the blue part of the spectrum of source 30 .
flux densities for source 30 are most likely due to an incorrect absolute flux level in the blue part of the spectrum ( see [ sec_fluxden ] ) . ]
it is not absolutely clear why the spectrum of source 30 has an incorrect absolute flux level in the blue , although it is most likely due to the bad weather ( clouds ) during the observations .
we note that besides the @xmath40-band , the flux densities derived from the spectroscopy are consistent with the _ hst _ and sdss photometry , and that the error in the absolute flux level in the blue part of the spectrum of source 30 does not affect our analysis or conclusions .
ccccccc source 26 & & & & & & + photometry & @xmath54 & @xmath55 & 1600(370 ) & 230(40 ) & ( 0.1)2.9(0.8 ) & 65(12 ) + spectroscopy & @xmath56 & @xmath57 & 1500(50 ) & 200(30 ) & ( 0.1)3.0(0.1 ) & 62(9 ) + & & & & & & + source 30 & & & & & & + photometry & @xmath58 & @xmath59 & 3160(730 ) & 17(3 ) & @xmath60 & 4(1 ) + spectroscopy & @xmath61 & @xmath59 & 3260(120 ) & 17(3 ) & @xmath62 & 4(1 ) + in studies of extragalactic massive star forming regions , narrow - band imaging is commonly used to estimate the emission of h@xmath8 , which in turn is used to infer physical properties such as ionizing luminosity , age and mass .
since we have spectroscopy and narrow - band imaging of our target clusters , we compare the flux and equivalent width of h@xmath8 from both data sets to see how well the measurements agree . using the _ hst _ photometry , the h@xmath8 flux , @xmath63 , and equivalent width , @xmath64 ,
are given by @xmath65 - \\
\left[\left(f_{\rm f660n } - f_{\rm cont}^{({\rm nii})}\right ) \delta\lambda_{\rm f660n}\right ] , \nonumber\end{aligned}\ ] ] and @xmath66 where @xmath63 has units of erg s@xmath14 @xmath67 , @xmath64 has units of and @xmath68 is the continuum flux density at 6563 in erg s@xmath14 @xmath67 @xmath14 . in equation [ haflux ] ,
the first term in square brackets is the total h@xmath8 + [ n ii ] @xmath19 flux and the second term in square brackets is the total [ n ii ] @xmath19 flux.line at 6548 is a relatively weak contaminant to the h@xmath8 flux and the filter throughput drops at this wavelength .
] the measured flux densities and widths of the f658n and f660n filters are given by @xmath69 , @xmath70 and @xmath71 , @xmath72 , respectively .
the flux densities of the continuum values , @xmath73 and @xmath74 , are found by interpolating between the f550 m and @xmath46-band data points .
table [ tab_ha ] lists the measured h@xmath8 fluxes and equivalent widths obtained from the photometry and from direct measurements of the h@xmath8 line in the spectra using the iraf task ` splot ` .
these values agree within the errors , confirming that the appropriate imaging data can be used to obtain reliable h@xmath8measurements .
table [ tab_ha ] also lists extinction corrected h@xmath8 fluxes using internal extinctions of a@xmath75=0.4 for source 26 and a@xmath75=0.5 for source 30 ( see [ sec_sb99 ] ) .
galactic foreground extinction is also accounted for ( @xmath76 , * ? ? ?
a comparison of the h@xmath8 measurements with radio continuum observations @xcite suggest that these a@xmath75 may actually underestimate the true extinctions of these infant clusters since optical observations may not probe the deepest embedded regions of the ymcs .
nonetheless , we use the optically derived extinctions here .
the flux and equivalent width of the h@xmath8 emission line can be used to infer properties of a ymc such as the number of ionizing photons produced by the massive stars , the cluster s age , and its mass ( e.g. * ? ? ?
* ; * ? ? ?
we calculate ionizing luminosities , n@xmath77 , using the extinction corrected h@xmath8 luminosities ( table [ tab_ha ] ) in the following equation derived from @xcite for a @xmath78 k gas : @xmath79 adopting a distance of 3.9 mpc to ngc 4449 , the production rates of ionizing photons are approximately @xmath80 and @xmath81 s@xmath14 , or the equivalent of @xmath3200 and @xmath317 o7.5 v stars @xcite for sources 26 and 30 , respectively .
we note that the ionizing luminosity of source 30 derived here is lower than that in @xcite ( @xmath82 s@xmath14 ) . in that paper , the ionizing luminosity
was derived from radio continuum measurements which probe higher extinction regions than the h@xmath8line used here .
emission ( see [ sec_sb99 ] for a description of the model ) .
the measured h@xmath8 equivalent widths of sources 26 and 30 , from both the photometry and spectroscopy , are shown indicating their young ages .
uncertainties are reflected in the sizes of the shaded regions . ]
ages of the ymcs are estimated from their h@xmath8 equivalent widths .
the equivalent widths from both the photometry and spectroscopy are plotted in figure [ plot_haew ] with a starburst99 model as a function of age .
we note that the model equivalent widths include a nebular component in the continuum ( as do our measurements ) .
the h@xmath8 equivalent width of source 26 suggests the cluster is @xmath83 3 myr old .
source 30 has an h@xmath8 equivalent width consistent with the maximum value , and therefore the youngest ages , of the model evolutionary track .
source 30 is certainly younger than source 26 and is likely @xmath84 myr old .
@xmath85{f6a.pdf } } & { \includegraphics[width=3.3in]{f6b.pdf } } \\ { \includegraphics[width=3.3in]{f6c.pdf } } & { \includegraphics[width=3.3in]{f6d.pdf } } \end{array}$ ] assuming mass scales with ionizing luminosity , masses of the clusters can be estimated by multiplying an input model mass by the ratios of the measured ionizing luminosities to the model ionizing luminosities at their respective ages . using the starburst99 model described in [ sec_sb99 ] , we infer masses of @xmath86 and @xmath87 m@xmath88 for sources 26 and 30 , respectively . again , the mass of source 30 is lower than that presented in @xcite ( @xmath89 m@xmath88 ) which used the radio derived ionizing luminosity .
ionizing luminosities , ages , and masses of sources 26 and 30 , based on the h@xmath8 measurements presented in [ sec_ha ] , are summarized in table [ tab_ha ] .
it is noted that caution must be applied when interpreting the physical properties of observed clusters derived from population synthesis models , since an underlying assumption of every simple stellar population ( ssp ) model is a well - populated initial mass function ( imf ) and stochastic fluctuations can bias the results .
in addition to using hydrogen recombination lines such as h@xmath8 , broad - band photometric observations of extragalactic massive star clusters are commonly compared to models of ssps in order to estimate their physical properties such as age , extinction and mass . here
we perform sed fitting to our _ photometric _ data using the starburst99 @xcite and galev @xcite evolutionary synthesis models , and investigate how well the model seds fit the continua in our observed _
spectra_. in addition to the stellar light , both sets of models include nebular continuum ( although these are computed in slightly different ways ) .
the galev models also have the option to include a set of metallicity - dependent emission lines @xcite .
physical properties of the clusters are estimated by comparing a grid of model seds ( starburst99 and galev separately ) to measured photometric flux densities .
the grid includes seds with ages @xmath90 myr in steps of 0.1 myr , and visual extinctions @xmath91 in increments of 0.1 magnitudes .
we apply a 30 doradus extinction curve adopted from ( * ? ? ?
* table 3 ) and ( * ? ? ?
* table 6 ) , using the parameterization given by @xcite .
galactic foreground extinction towards ngc 4449is also accounted for ( @xmath76 , * ? ? ?
* ) using the extinction curve of @xcite .
each model sed , of a given age and a@xmath75 , is then convolved with the total system throughput curves for the photometric filters to obtain synthetic flux densities . the best - fit model sed is determined by minimizing a goodness - of - fit parameter , @xmath92 , equal to the standard deviation of the logarithmic residuals of the observed and synthetic flux densities ( weighted by the errors of the observed flux densities ) .
a mass estimate is obtained by scaling the model mass by the mean logarithmic offset between the observed and best - fit model flux densities . in all , we ran our sed fitting routine four times for each cluster : 1 ) using starburst99 and all available broad - band photometric data , 2 ) using starburst99 and only photometric data with minimal line emission , 3 ) using galev ( with emission lines ) and all available photometric data , and 4 ) using galev ( without emission lines ) and only photometric data with minimal line emission . stellar and nebular continuum emission
is included in the models for each case .
filters with minimal line emission were identified using our figure [ plot_spectra ] and table 1 in @xcite ( for the short - wavelength filters not covered by our spectra ) .
these include the f170w , f336w , @xmath37 , f550 m , and @xmath46 filters .
details about the models and the results of sed fitting are described below .
we use the latest starburst99 models ( version 5.1 ) with a metallicity of @xmath93 ( as appropriate for ngc 4449 , * ? ? ?
* ) , an instantaneous burst of @xmath78 m@xmath88 with a kroupa imf ( @xmath94 m@xmath88 ) , the geneva evolutionary tracks with high mass loss , and the pauldrach / hillier atmospheres .
the models include stellar and nebular continuum but no emission lines .
sed fitting is first performed using all of the available broad - band photometric data in this study ( f170w , f336w , @xmath37 , f435w , @xmath40 , f550 m , @xmath43 , @xmath46 , f814w , and @xmath49 ) .
the best - fitting model seds for sources 26 and 30 are shown at the top of figure [ plot_seds_all ] .
it is clear that the model seds do a poor job fitting the spectral continua .
this is because the spectra contain strong emission lines which are not accounted for in the models .
the photometric bands containing significant line emission cause the model seds to be higher than the observed continua .
@xmath85{f7a.pdf } } & { \includegraphics[width=3.3in]{f7b.pdf } } \\ { \includegraphics[width=3.3in]{f7c.pdf } } & { \includegraphics[width=3.3in]{f7d.pdf } } \end{array}$ ] next , we perform sed fitting using only those filters with minimal line emission ( f170w , f336w , @xmath37 , f550 m , and @xmath46 ) . in this case , the best - fit models do a very good job fitting both the photometry included in the fitting process and the observed continua of the spectra ) .
however , this does not affect our results since the photometry is being fit . ]
( see figure [ plot_seds ] ) .
also , the inferred physical properties of the clusters ( shown in figure [ plot_seds ] ) are consistent with those derived from the h@xmath8 emission ( see table [ tab_ha ] ) .
comparing the best - fit starburst99 model parameters from the two fitting processes ( all bands versus bands with minimal line emission ) , we find that mass estimates can vary significantly . for source 30 , the mass derived from fitting all of the photometry is a factor of 3 times too high due to strong lines raising the overall sed .
the extinction estimate for source 30 is also higher by 0.7 mag when all of the bands are fit since the model sed needs to be flatter in the red to better match the f814w and @xmath49-band data ( which contain significant line emission ) .
the impact of nebular continuum on the total seds is striking in both clusters ( see figure [ plot_seds ] ) , especially in the @xmath4- and @xmath0-band regions of the spectra where the balmer ( 3646 ) and paschen ( 8207 ) discontinuities are located .
these discontinuities , or `` jumps '' , in the nebular continuum emission are due to hydrogen and helium free - bound recombination processes .
the nebular continuum actually dominates the total continuum in these wavelength regions for the younger ymc , source 30 . to further illustrate the importance of nebular continuum in young stellar populations , we plot the ratio of nebular continuum to the total continuum as a function of wavelength for starburst99 model seds ( @xmath93 ) at various ages ( figure [ plot_gasfrac ] ) .
it is clear that , at young ages , nebular continuum is a significant component of the total continuum , even at optical wavelengths .
galev models were kindly provided to us by ralf kotulla and the galev team .
these models were run using the geneva evolutionary tracks with a minimum age and time resolution of 0.1 myr , unlike the models available on the galev website which use the padova isochrones and have a minimum age and time step of 4 myr .
the models also use a `` fraction of visible mass '' equal to 1 rather than 0.5 , which is used for the standard models available on the web ( ralf kotulla , private communication ) .
the models were run with a metallicity of @xmath93 and a kroupa imf ( @xmath94 m@xmath88 ) .
these models are very similar to the starburst99 models described above , with the notable exception that the galev models include emission lines from nebular gas @xcite .
the flux of the hydrogen lines are computed using atomic physics and the production rate of ionizing photons , whereas non - hydrogen line strengths are computed using metallicity - dependent line ratios relative to h@xmath95 . in the case of low - metallicity gas ( including @xmath93 ) , observed line ratios are taken from the large sample of giant h ii regions and blue compact dwarf ( bcd ) galaxies presented in @xcite and @xcite . ) at various ages . at young ages ,
nebular continuum is a significant component of the total continuum at optical wavelengths , especially shortward of the balmer ( 3646 ) and paschen jumps ( 8207 ) located in the @xmath4- and @xmath0-band wavelength regions .
the impact of emission lines is not included here . ] we perform sed fitting with the galev models in the same way as the starburst99 models .
first we fit all of the photometric filters and include emission lines in the models ( the option to include lines is not available for starburst99 ) .
the best - fitting model seds for sources 26 and 30 are shown in figure [ plot_seds_all ] .
it is clear that the galev models including emission lines do a much better job matching the continua in the observed spectra than the starburst99 models do when we fit all of the photometry .
however , the galev model seds shown in figure [ plot_seds_all ] are still a bit too high in the red to near - infrared and too low in the ultraviolet .
this is likely due to mismatched line intensities between the models and the observed spectra .
we note that currently the galev models are missing the higher level paschen lines , although they do include pa@xmath8 through pa@xmath96 .
overall , however , the galev models including emission lines perform much better than the starburst99 models when we fit all of the photometry including bands with significant line contamination .
we also only fit the photometry with minimal line emission with galev seds as we did with the starburst99 models . here
we exclude emission lines in the models so they only contain components from the stellar and nebular continua .
the best - fitting seds are shown in figure [ plot_seds ] .
these models perform equally well as the starburst99 models fit only to the photometry with minimal line emission ( also shown in figure [ plot_seds ] ) . however , it is curious that the best - fit model parameters from the two families of models are as different as they are .
the differences may be due to the ways in which the nebular continuum is calculated .
the best - fit galev seds shown in figure [ plot_seds ] also do a better job fitting the continua of the observed spectra than the galev models ( including lines ) fit to all of the photometry ( figure [ plot_seds_all ] ) .
a number of authors have discussed the impact of nebular emission on broad - band photometry of young stellar populations based on synthesis modeling ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
our spectrophotometic observations of infant massive star clusters clearly support the conclusion that nebular emission , both in lines and continuum , significantly affects broad - band photometry of these objects . in an effort to provide a benchmark for estimating the impact of nebular emission on photometric observations of young massive stellar populations , we compute the relative contributions of the stellar continuum , nebular continuum , and emission lines to the total flux of source 26 through various _ hst _ filter / instrument combinations ( including wfc3 ) and the sdss @xmath97 filters . and
[ sec_fluxden ] ) . ] the contribution of nebular emission in the _
wfc f814w filter is of particular interest since , in their study of radio - detected clusters in ngc 4449 , @xcite found an excess in this band relative to model seds for a pure stellar continuum .
is shown as a solid blue line .
the stellar and nebular continua are shown as dashed and dotted lines , respectively .
the relative contributions of the stellar continuum , nebular continuum and emission lines to the total flux of this filter are approximately 40% , 40% and 20% . ] the spectrum of source 26 multiplied by the _
wfc f814w total system throughput curve is shown in figure [ plot_f814w ] .
the best - fitting starburst99 model sed ( from figure [ plot_seds ] ) is also shown .
in addition to stellar and nebular continuum , the spectrum contains several prominent emission lines in this band .
the strongest are [ s iii ] @xmath989532 and [ s iii ] @xmath989069
. a number of other lines , including the paschen series , are detected and labeled in figure [ plot_f814w ] .
ccccc f845 m & wfc3/uvis & 0.49 & 0.41 & 0.09 + f814w & wfc3/uvis & 0.39 & 0.38 & 0.23 + f814w & acs / wfc & 0.41 & 0.39 & 0.20 + f814w & acs / hrc & 0.38 & 0.36 & 0.25 + f814w & wfpc2 & 0.41 & 0.40 & 0.20 + f791w & wfpc2 & 0.46 & 0.45 & 0.10 + f775w & wfc3/uvis & 0.46 & 0.45 & 0.09 + f775w & acs / wfc & 0.46 & 0.46 & 0.08 + f775w & acs / hrc & 0.46 & 0.46 & 0.08 + f763 m & wfc3/uvis & 0.47 & 0.50 & 0.03 + f555w & wfc3/uvis & 0.31 & 0.12 & 0.56 + f555w & acs / wfc & 0.27 & 0.11 & 0.63 + f555w & acs / hrc & 0.26 & 0.11 & 0.63 + f555w & wfpc2 & 0.32 & 0.13 & 0.54 + f550 m & acs / wfc & 0.70 & 0.31 & 0.00 + f550 m & acs / hrc & 0.70 & 0.31 & 0.00 + f547 m & wfc3/uvis & 0.71 & 0.30 & 0.00 + f547 m & wfpc2 & 0.68 & 0.29 & 0.03 + @xmath40 & sdss & 0.32 & 0.10 & 0.58 + @xmath43 & sdss & 0.27 & 0.16 & 0.57 + @xmath46 & sdss & 0.46 & 0.46 & 0.07 + @xmath49 & sdss & 0.34 & 0.26 & 0.40 + @xmath99{f10a.pdf } } & { \includegraphics[width=3in]{f10b.pdf } } \\ { \includegraphics[width=3in]{f10c.pdf } } & { \includegraphics[width=3in]{f10d.pdf } } \\ { \includegraphics[width=3in]{f10e.pdf } } & { \includegraphics[width=3in]{f10f.pdf } } \end{array}$ ] we determine the relative contributions of stellar continuum , nebular continuum and emission lines to the total observed flux in the f814w filter as follows .
the contributions from the stellar and nebular continua are estimated by multiplying the models for these components with the filter throughput curve and integrating over the bandpass .
the total flux from emission lines in this filter is determined by multiplying the observed spectrum with the f814w filter throughput curve and measuring the flux of each line in the resulting spectrum .
the relative contributions of each component are found by dividing by the total observed flux in the f814w filter ( obtained by convolving the observed spectrum with the throughput curve ) .
we find that the relative contributions of the stellar continuum , nebular continuum and emission lines to the total observed flux of source 26 through the f814w filter are approximately 40% , 40% , and 20% , respectively .
_ clearly , nebular emission contributes a significant fraction ( @xmath3 60% ) of the total flux in the f814w filter and should not be neglected .
_ the combination of nebular continuum and emission lines can account for the f814w @xmath0-band excess found by @xcite .
this result is consistent with their prediction that the origin of the excess should not affect clusters older than @xmath3 5 myr .
relative contributions to the total flux of source 26 through other _
filter / instrument combinations are listed in table [ tab_contributions ] .
throughput curves were obtained using synphot , the synthetic photometry package distributed by the space telescope science data analysis system ( stsdas ) .
the total flux and the flux of the stellar and nebular continua are found in the same way as described above .
the line contributions , however , are simply estimated by subtracting the continuum contributions from the total observed flux .
plots of the observed spectrum of source 26 multiplied by the _ hst _
wfc3 filters listed in table [ tab_contributions ] are shown in figure [ plot_wfc3 ] .
we find that the relative contribution of the nebular gas ( continuum plus lines ) to the total flux of source 26 in any of the _ hst _
@xmath0-band filters listed in table [ tab_contributions ] is _ at least 50%_. in the @xmath1-band , roughly 70% of the f555w flux comes from ionized gas .
younger clusters will have an even larger contribution from ionized gas .
relative contributions to the sdss fluxes of source 26 are also listed in table [ tab_contributions ] using the throughput curves available online .
we emphasize that the quantitative results listed in table [ tab_contributions ] are specific to source 26 .
relative contributions will vary from source to source , depending on ( at a minimum ) age and metallicity ( e.g. our figure [ plot_gasfrac ] and * ? ? ?
however , it is evident that the impact of nebular emission on broad - band fluxes can be quite large and should not be readily dismissed .
the ionized gas associated with a young stellar population can provide a significant contribution to the total observed flux in a given filter ( [ sec_contributions ] ) .
the nebular emission , therefore , can have large impact on the physical properties derived from magnitudes and broad - band colors .
@xmath100{f11a.pdf } } & { \includegraphics[width=3.4in]{f11b.pdf } } \end{array}$ ] using the galev models described in section [ sec_galev ] , we have computed the magnitude difference as a function of age between a pure stellar population and one of the same mass but including ionized gas emission ( continuum plus lines ) for the _ hst_/wfc3 f814w ( broad @xmath0 ) , f555w ( wide @xmath1 ) , and f547 m ( narrow @xmath1 ) filters .
the results are shown in figure [ plot_magcolor ] .
for a 3 myr old cluster , including the gaseous emission results in magnitudes that are brighter by 1.0 ( f814w ) , 1.5 ( f555w ) , and 0.4 ( f547 m ) mag .
neglecting nebular emission and assuming all of the flux comes from the stars would therefore lead to overestimates of the stellar mass by factors of 2.5 ( f814w ) , 4.0 ( f555w ) , and 1.4 ( f547 m ) . from figure [ plot_wfc3 ]
, we can see that significant contributions from both nebular continuum and lines are present in the f814w filter , whereas the f555w flux is most affected by strong lines ( e.g. h@xmath95 and [ o iii ] ) .
the f547 m flux is affected by the nebular continuum but does not contain strong line emission .
we obtain similar results from modeling the sed of source 26 and determining the relative contributions of the stellar continuum , nebular continuum and emission lines to the total flux through various filters ( [ sec_contributions ] ) .
mass ( over)estimates of this ymc ( age @xmath101 myr , @xmath102 ) , assuming all of the flux in a given filter is due only to stars , can be obtained by taking the inverse of the relative contribution of the stellar continuum to the total flux ( the third column in table [ tab_contributions ] ) .
for example , neglecting the effects of nebular emission in the acs / wfc f555w filter would result in a mass @xmath103 larger than the mass obtained by properly accounting for the gas contribution in this filter .
since the magnitudes of young stellar populations can be considerably brightened by the presence of ionized gas , colors can also be significantly affected .
for example , the work of @xcite found an apparently anomalous color - color diagram in which the radio - detected ymcs in ngc 4449 with the bluest _
wfc [ f435w@xmath104f550 m ] colors had the reddest [ f550m@xmath104f814w ] colors , contradictory to the starburst99 model evolutionary track for a pure stellar continuum .
we have reproduced this color - color diagram but have now included a starburst99 model evolutionary track including nebular continuum , and a galev track including nebular continuum plus emission lines ( figure [ plot_magcolor ] ) .
only the galev model is able to reproduce the trend seen in the data , signifying that emission lines have a large effect on these colors at young ages ( @xmath6 myr ) .
although the f550 m filter is essentially line - free , the f814w and f435w filters contain a number of emission lines .
the forbidden [ s iii ] lines at 9532 and 9069 are the most prominent in the f814w filter , but many others are also present ( see figure [ plot_f814w ] ) .
the f435w filter contains line emission from h@xmath52 , h@xmath53 , h@xmath105[n iii ] @xmath106 , h@xmath107[he i ] @xmath108 , [ n iii ] @xmath109 , and [ o ii ] @xmath110 ( see our figure [ plot_spectra ] and * ? ? ?
it is worth noting that the [ f435w@xmath104f550 m ] versus [ f550m@xmath104f814w ] color - color diagram may prove to be a useful diagnostic for identifying ymcs with ages @xmath6 myr .
it is also interesting that the extinction vector in figure [ plot_magcolor ] is perpendicular to the galev model track for ages @xmath6 myr . in other words
, there is no degeneracy in this color space between age and extinction at young ages .
therefore , rough estimates of both age and extinction could be obtained from this color - color plot .
however , we emphasize that the line strengths of any given source may vary from the ones input into the galev models which could potentially lead to large uncertainties in the derived properties . as a check , we look at sources 26 and 30 ( indicated in figure [ plot_magcolor ] by purple and orange dots respectively ) .
the ages inferred from their colors are consistent with those obtained from their h@xmath8 equivalent widths and from sed fitting ; approximately 3 myr for source 26 and 1 - 2 myr for source 30 .
the visual extinction of source 26 also appears to be consistent with our previous estimates of @xmath3 0.4 magnitudes .
the extinction of source 30 inferred from figure [ plot_magcolor ] ( @xmath83 1 mag ) , on the other hand , is half a magnitude larger than that obtained from sed fitting to the photometry with minimal line emission ( figure [ plot_seds ] ) .
s are accurate measures of the extinction at such young ages , since a significant fraction of the cluster may be heavily embedded and visibly obscured . ]
this difference is likely the result of different line intensities between the models and the data .
the galev models , however , at least qualitatively explain these colors of very young star clusters , which neither of the starburst99 models can do .
this paper examines the importance of nebular continuum and line emission in observations of young massive stellar populations .
we have obtained spectroscopy ( @xmath111 ) of two young ( @xmath112 myr ) star clusters in the nearby starburst galaxy ngc 4449 and supplement these data with archival _ hst _ and sdss photometry of the clusters .
we compare our data to the starburst99 and galev evolutionary synthesis models to estimate the physical properties of the clusters and determine the impact of ionized gas emission ( line and continuum ) on broad - band photometry .
our main results are summarized below : 1 .
the _ combination _ of nebular continuum and line emission can account for the f814w @xmath0-band excess found by @xcite in their study of radio - detected ymcs in ngc 4449 .
this result is consistent with their prediction that the origin of the excess should not affect clusters older than @xmath3 5 myr .
2 . at very young ages ( @xmath112 myr ) ,
nebular continuum emission from ionized gas can _ rival _ the stellar luminosity of ymcs at optical wavelengths .
the relative contribution from the nebular continuum emission is largest in the @xmath4- and @xmath0-bands , where the balmer ( 3646 ) and paschen jumps ( 8207 ) are located .
nebular line emission is significant in many commonly used broad - band _ hst _ filters including the f814w @xmath0-band , the f555w @xmath1-band and the f435w @xmath2-band .
emission lines ( in addition to nebular continuum ) are required to explain the distribution of radio - detected ymcs in ngc 4449 in the [ f435w@xmath104f550 m ] versus [ f550m@xmath104f814w ] color - color diagram first presented in @xcite .
4 . sed fitting to broad - band photometry of ymcs is most successful when the photometric bands do not contain significant line emission . in this case ,
evolutionary synthesis models only containing _ continuum _ emission from stars and ionized gas are adequate .
if the photometric bands include significant line emission , however , models including emission lines perform much better than those that do not . 5 . the impact of ionized gas emission on broad - band fluxes , magnitudes and colors can be large and should not be readily dismissed .
we urge caution when comparing observations of young ( @xmath6 myr ) clusters to synthesis models since nebular continuum and line emission can significantly affect inferred properties such as ages , masses and extinctions .
we thank ralf kotulla for providing us with the custom - run galev models used in this work .
we also thank the anonymous referee for numerous useful comments and suggestions which led to the overall improvement of this paper . a.e.r .
appreciates helpful discussions with adolf witt , yuri izotov and crystal brogan . a.e.r .
also gratefully acknowledges support from a nasa earth and space science fellowship , the virginia space grant consortium and the university of virginia through a governor s fellowship .
k.e.j . gratefully acknowledges support provided in part by nsf through career award 0548103 and the david and lucile packard foundation .
support for program # ar09934 was provided in part 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 - 26555 .
|
in this spectroscopic study of infant massive star clusters , we find that continuum emission from ionized gas rivals the stellar luminosity at optical wavelengths .
in addition , we find that nebular line emission is significant in many commonly used broad - band _ hubble space telescope ( hst ) _ filters including the f814w @xmath0-band , the f555w @xmath1-band and the f435w @xmath2-band . two young massive clusters ( ymcs ) in the nearby starburst galaxy ngc 4449 were targeted for follow - up spectroscopic observations after @xcite discovered an f814w @xmath0-band excess in their photometric study of radio - detected clusters in the galaxy . the spectra were obtained with the dual imaging spectrograph ( dis ) on the 3.5 m apache point observatory ( apo ) telescope and have a spectral range of @xmath33800 - 9800 .
we supplement these data with _ hst _ and sloan digital sky survey ( sdss ) photometry of the clusters . by comparing our data to the starburst99 and galev evolutionary synthesis models , we find that nebular continuum emission competes with the stellar light in our observations and that the relative contribution from the nebular continuum is largest in the @xmath4- and @xmath0-bands , where the balmer ( 3646 ) and paschen jumps ( 8207 ) are located .
the spectra also exhibit strong line emission including the [ s iii ] @xmath5 lines in the _ hst _
f814w @xmath0-band .
we find that the _ combination _ of nebular continuum and line emission can account for the f814w @xmath0-band excess previously found by @xcite . in an effort to provide a benchmark for estimating the impact of ionized gas emission on photometric observations of young massive stellar populations , we compute the relative contributions of the stellar continuum , nebular continuum , and emission lines to the total observed flux of a 3 myr - old cluster through various _ hst _ filter / instrument combinations , including filters in the wide field camera 3 ( wfc3 ) .
we urge caution when comparing observations of ymcs to evolutionary synthesis models since nebular continuum and line emission can have a large impact on magnitudes and colors of young ( @xmath6 myr ) clusters , significantly affecting inferred properties such as ages , masses and extinctions .
|
1 . see for example b. batlogg in _ high temperature superconductivity_.
the los alamos symposium - 1989 , edited by k. s. bedell et al .
( addison wesley , reading , ma ; 1989 ) .
2 . p. w. anderson and z. zhou , phys .
rev . lett . * 60 * , 2557 ( 1988 ) .
3 . g. kotliar and c. m. varma , physica * a 167 * , 288 ( 1990 ) ; g. kotliar et al .
, europhysics lett . *
63 * , 1996 ( 1989 ) .
4 . y. ando et al . , phys .
rev . lett . * 75 * , 4662 ( 1995 ) . 5 .
g. boebinger et al .
lett , * 77 * , ( 1996 ) .
s. martin and c. m. varma , unpublished analysis of results in s. martin et al .
b * 41 * , 846 ( 1990 ) . 7 . c. m. varma et al . , phys . rev .
lett . * 63 * , 1996 ( 1989 )
p. b. littlewood and c. m. varma , phys .
b. * 45 * , 12636 ( 1992 ) .
9 . c. m. varma , phys
( submitted ) , available on http://www.lanl.gov as cond - mat/9607105 .
see for example , h. ikeda and k. miyake , j. phys .
japan * 6 * , ( 1996 ) . 11 .
o. betbeder - matibet and p. nozieres , annals of physics , * 37 * , 17 ( 1966 ) . 12
. j. s. langer , phys . rev . * 128 * , 110 ( 1962 )
. 13 . if impurities couple to conserved quantities such as _ total _ charge or spin density ( in the absence of spin - orbit scattering ) , eq .
( 6 ) is multiplied by @xmath119 or @xmath120 where @xmath121 are the lowest spin - symmetric and anti - symmetric landau parameters .
impurity scattering rates can then be singular only if compressibility or spin - susceptibility in the pure system diverges .
we do not consider such a situation here .
y. maeno et al . ,
nature * 372 * , 532 ( 1994 ) . 15 .
h. von lohneysen , j. phys .
* 8 * , 9689 ( 1996 ) 16 .
b. i. halperin , p. a. lee and n. read , phys .
b * 47 * , 7312 ( 1993 ) .
* graphs describing the calculation of the impurity self - energy : + ( a ) the impurity self - energy for a given configuration ( i , j , ... ) of impurities .
the line with arrows is the single particle propagator including renormalization due to electron - electron interaction and the circle connected to the dashed lines is the impurity vertex renormalized for the effect of interactions .
+ ( b ) vertex corrections due to interactions neglected in ( a ) .
the wavy line represents electron - electron interactions .
+ ( c ) impurity self - energy ( after configuration averaging ) in the self - consistent born approximation .
the thick line includes the self - energy due to interactions as well as due to impurities .
+ ( d ) the @xmath122-matrix with renormalized vertices of interaction from an impurity at site @xmath123 .
+ ( e ) the crossed - graphs for interference effects between two impurities at site @xmath123 and at site @xmath124 .
|
any singular deviation from landau fermi - liquid theory appears to lead , for arbitrarily small concentration of impurities coupling to a non - conserved quantity , to a vanishing density of states at the chemical potential and infinite resistivity as temperature approaches zero .
applications to copper - oxide metals including the temperature dependence of the anisotropy in resistivity , and to other cases of non fermi - liquids are discussed .
the resistivity in the c - direction , @xmath0 , in the normal phase of most copper - oxide ( cuo ) compounds increases with decreasing temperature while the inplane resistivity @xmath1 has the opposite behavior , which for compositions near those for the highest @xmath2 is proportional to @xmath3 down to @xmath4.@xmath5 one may be led to suppose that , if superconductivity were not to intervene , @xmath6 while @xmath7 finite value as @xmath8 . for any finite quantum - mechanical transfer matrix element @xmath9 between adjacent planes ,
the asymptotic low temperature dependence in different directions must be identical for @xmath10 , where @xmath11 here @xmath12 is the inelastic scattering rate .
anderson and zhou@xmath13 conjectured that the renormalized matrix elements @xmath14 in cuo compounds at low temperatures due to orthogonality effects .
an alternative conjecture@xmath15 put forward to resolve the issue is that @xmath16 also @xmath17 as @xmath8 due to impurity scattering in a non fermi - liquid .
recent experiments@xmath18 measuring the resistivity at low temperatures by suppressing @xmath2 in a large magnetic field support this conjecture and find @xmath19 at low temperature .
similar behavior is also found ( without applying a magnetic field ) in samples of the single layer bi compound.@xmath20 here theoretical support for the conjecture that the resistivity of a non - fermi - liquid is infinity for @xmath8 for any finite concentration of impurities as well as the logarithm temperature dependence are obtained . a landau fermi - liquid has the property that the real part of the single particle self - energy @xmath21 here @xmath22 for @xmath23 and @xmath24 for @xmath25 . for @xmath26 ,
the pole in the single particle green s function , the quasiparticle , vanishes at the chemical potential and is replaced by a branch - cut .
@xmath26 may be used to characterize a non fermi - liquid . the gentlest departure from a landau fermi - liquid with @xmath27
has been termed a marginal fermi - liquid.@xmath28 the observed linear temperature dependence of the resistivity and corresponding behavior of the frequency dependent conductivity implies directly that the momentum scattering rate , @xmath29 , in cu - o compounds is proportional to @xmath30 .
the imaginary part of the single particle self - energy @xmath31 can not have a higher power dependence on @xmath32 than the momentum scattering rate . through kramers - kronig transformation , this implies @xmath26 .
the marginal case , eq . ( 3 ) is consistent with the measured tunneling conductance@xmath33 as well as the deduced electronic heat capacity,@xmath34 as is the low temperature resistivity derived here .
i discuss below the @xmath35 dependence of the scattering rate from _ s - wave _ scattering off impurities for the non - fermi - liquids characterized in the pure limit by the marginal case as well as other @xmath26 .
this can be done to a considerable extent without reference to any microscopic theory of the non - fermi - liquid .
also , for a given @xmath36 , the dimensionality @xmath37 will not be important as long as @xmath38 .
of course @xmath36 may depend on @xmath37 .
consider impurities which in the absence of electron - electron interactions have finite s - wave scattering amplitude.@xmath39 the vertex correction due to the interactions will in general introduce scattering into higher partial waves .
but there is no reason why the s - wave scattering should vanish .
consider only this part for which the forward scattering limit characterizes the properties . in this limit
, a ward identity provides the vertex including the effect of electron - electron interactions .
let the bare impurity potential be @xmath40 where @xmath41 are the position of the impurities assumed randomly distributed .
we will assume also that @xmath42 and the concentration of impurities @xmath43 is low enough so that the mean - free path @xmath44 calculated without electronic renormalizations satisfies @xmath45 .
the s - wave part of the renormalized scattering from a given impurity is @xmath46 where @xmath47 is the irreducible vertex due to electron - electron interactions for elastic scattering with momentum @xmath48 .
if an impurity couples to a non - conserved quantity , ( for instance for cu - o compounds , it can change the equilibrium charge difference between cu and o in a unit cell or alter the local kinetic energy in a cu - o bond ) a ward - identity gives that@xmath49 ( at @xmath50 ) , @xmath51 where @xmath52 the self - energy due to impurities from such a vertex for a given impurity configuration illustrated in fig .
( 1a ) is @xmath53 where @xmath54 and @xmath55 is the exact single particle green s function including the effect of impurities . in eq .
( 8) , vertex renormalization at a given impurity alone has been considered .
vertex renormalization which spans across impurities , such as in fig .
( 1b ) are proportional to @xmath56 so that ( 8) may be considered asymptotically exact .
the strategy adopted here is to first evaluate a subset of the processes in the expression ( 8) , which are expected to be the most singular .
the result so obtained is used to calculate corrections which show that the expectation is borne out .
the most singular processes for the impurity averaged self - energy are expected to be given by the self - consistent born approximation represented by fig .
( 1c ) : @xmath57 where we include in @xmath58 the self - energy @xmath59 which includes @xmath60 as well as @xmath61 . assuming a constant density of states @xmath62 over a bandwidth @xmath63 in the pure limit , ( other assumptions about the pure density of states do not produce any essential difference ) , eq .
( 9 ) gives @xmath64 where @xmath65 . for @xmath66
evaluated from the marginal self - energy and for @xmath67 , @xmath68 in the opposite limit @xmath69 the above results are valid for @xmath70 . for @xmath71 should be substituted for @xmath72 in eqs .
( 10 - 11 ) .
the cross - over from ( 11 ) to ( 12 ) occurs for @xmath73 or at @xmath74 at @xmath75 note that below @xmath76 the density of states near the fermi - energy : @xmath77 which approaches zero at the chemical potential .
the two major omissions in evaluating ( 7 ) in the self - consistent born approximation are ( i ) possible strong renormalized scattering from a given impurity and ( ii ) interference effects between scattering at different impurities . consider the former .
for fermi - liquids the singular scattering from a given impurity in the born - approximation is remedied by evaluating the t - matrix , fig .
( 1d ) : @xmath78 the @xmath79 matrix obeys the unitary limit @xmath80 as @xmath66 becomes very small ( but not zero ) , as in heavy fermions .
this is because for fermi - liquids @xmath81 , the bare density of states , independent of the value of @xmath66 .
we may check the validity of the self - consistent born approximation by evaluating the corrections to the self - energy due to ( 16 ) by using @xmath82 in ( 16 ) .
now we note that on impurity averaging , one gets correction terms @xmath83\ ] ] using ( 15 ) , we see that the singularity due to @xmath84 is cancelled out to all orders leaving an analytic correction of @xmath85
. we may similarly consider correction due to crossed graphs , fig .
( 1e ) .
again the renormalized density of state cancels the singularity in @xmath86 and the corrections are successive powers of @xmath87 .
since interference between scatterers is not important , the wavefunctions are not localized ; the important effect is the vanishing of the density of states at the chemical potential , as in eq .
( 15 ) . for @xmath26 , eq .
( 12 ) is modified to @xmath88 so the effects of impurities are more singular .
the most important effect neglected above is the renormalization by impurities of the fluctuation which produced the non - fermi - liquid state in the pure limit .
this can not be discussed without a microscopic theory for such fluctuations .
a theory of such fluctuations has been constructed@xmath34 .
the corrections to the fluctuations due to impurities have been briefly examined which appear not to change the results here in an essential way .
this matter is however far from being settled . at this point
it is best to leave this as an assumption and to point out that the results thus obtained appear to agree with experiments , as discussed below . a prediction following from eq .
( 15 ) is that the specific heat @xmath89 and the magnetic susceptibility @xmath90 have the forms @xmath91 , for @xmath92 . for s - wave scattering by impurities , there are no backward scattering ( vertex ) corrections for the calculation of resistivity .
the resistivity can then be calculated from the single particle green s function alone and is proportional to @xmath93 for @xmath92 .
consider now the anisotropic resistivity of quasi - two - dimensional materials .
two distinct temperature scales need to be defined to discuss the anisotropic resistivity of non - fermi - liquids .
one of them is @xmath94 defined in eq .
( 1 ) , and the other is @xmath95 defined in eq . (
14 ) .
@xmath94 is enough to discuss fermi - liquids . for @xmath96 ,
the layers are mutually phase incoherent and momentum in is not conserved in the process of electron transfer between the planes .
one can calculate @xmath97 by a tunneling rate calculation .
the tunneling rate ( if the tunneling matrix element has some momentum dependence ) is proportional to the ( in - plane ) inelastic scattering rate . then @xmath98 . for @xmath99
a coherent propagation must prevail in a fermi - liquid and @xmath100 .
this behavior is indeed observed , for example in @xmath101.@xmath102 for non - fermi - liquids two limiting cases can easily be distinguished .
( i ) @xmath103 : which occurs for highly anisotropic and fairly dirty materials . the resistivity in the c - direction increases with decreasing temperature for all temperatures but below @xmath94 . both @xmath16 and @xmath104
have similar temperature dependences below @xmath94 but for @xmath105 .
very high quality samples of @xmath106 appear to fall in this class if we assume that in this low anisotropy material @xmath94 is above the measured temperature range .
the more anisotropic dirty materials are consistent with class ( i ) but in the high quality samples studied @xmath95 is quite low , not too far from @xmath94 .
for instance , in the measurements of boebinger et al.@xmath107 on @xmath108 , @xmath109 is estimated to be about 15 which with @xmath110 and @xmath111 , estimated from the slope of the linear restivity and the optical conductivity@xmath28 gives @xmath112 from eq .
( 17 ) .
this is consistent with the temperature at which @xmath1 has a minima in the experiments and asymptotically below which a logarithmic temperature dependence is observed .
systematic estimations of @xmath109 are not yet available to test eq .
( 14 ) .
the results here have possible applications to other situations where interactions lead to non - fermi - liquid properties in the pure limit . these include the quantum critical points in itinerant ferromagnets and antiferromagnets as well as the mysterious transitions in @xmath113.@xmath114 it would appear from the results here that disorder is strongly relevant at metal - insulator transitions otherwise driven by electronic correlations . the effect of disorder in driving @xmath115 appears stronger than the localization effect which sets in only for @xmath116 .
the pseudo - particle green s function at the @xmath117 quantum hall effect also have a marginal fermi - liquid form.@xmath118 but since in this case the green s function is a gauge dependent object , a separate investigation is required for physical quantities .
i wish to thank y. ando , g. boebinger , g. kotliar , m. r. li , s. martin , and k. miyake for useful discussions .
|
the spin - orbit interaction ( soi ) in low - dimensional structures attracts a great deal of interest since it opens up the possibility to manipulate the electron spin in nonmagnetic structures using electrical means .
@xcite in this view , semiconductor heterostructures with 2d electrons are very promising since the rashba soi is effectively controlled @xcite by varying applied bias or gate voltages .
in recent years , predominant interest was paid to effects appearing when the soi modifies propagating electron modes with energy above the conduction band bottom .
suffice it to mention the spin - hall effect , @xcite or spin manipulation in strained semiconductors @xcite . in this paper
we show that interesting effects of the soi arise also when the electron energy is lower than or near to the conduction band bottom and evanescent states are involved .
these states determine electron tunneling .
they are important in 2d structures with laterally inhomogeneous potential landscape .
we find that such structures can effectively polarize the transmitted electron current .
3d tunnel structures , in which spin polarization arises due to the soi , were considered in a number of recent works .
zakharova et al @xcite studied the interband tunneling , voskoboynikov et al @xcite considered a tunnel structures with the rashba soi at the interfaces . in these cases
the electron flow acquires a spin polarization if the structure is asymmetric . in symmetric tunnel structures
the spin polarization arises , if the barrier material is noncentrosymmetrical .
@xcite the polarization mechanism , proposed by perel et al @xcite , consists in a spin - dependent renormalization of the electron effective mass owing to the dresselhaus soi in the barrier .
however , all these structures have a common property restricting their capability to generate spin polarization .
the polarization is absent if the electron current is perpendicular to the barrier .
in other words , for the spin polarization to appear the momentum distribution function of incident electrons must be asymmetric with respect to the momentum component parallel to the barrier .
the effective mass renormalization occurs if the hamiltonian of the soi is quadratic in longitudinal momentum .
however , the dispersion relation of electrons in the presence of the soi is generally much more complicated and therefore a more careful analysis of the complex band structure and evanescent states should be carried out to study the spin - dependent tunneling . in the 3d case ,
such calculations were recently carried out for some specific materials and qualitatively new features were found .
@xcite 2d tunnel structures are scantily studied to date . in particular ,
as far as we know , even the complex band structure of 2d electrons was not explored .
though the presence of evanescent states is obvious , only a few of works touched upon these modes .
usaj , reynoso and balseiro @xcite attracted evanescent states to study the electron scattering at the edges of 2d samples , but the total spectrum of evanescent states was not considered .
the importance of evanescent modes in quasi - one - dimensional systems in the presence of the soi was pointed out in a number of works .
@xcite khodas , shekter , and finkelstein @xcite studied the electron beam propagation in 2d electron gas with spatially inhomogeneous soi .
they considered the transmission through a strip , in which the soi strength differs from that in the rest of the 2d electron gas , to show that an initially unpolarized beam splits into two beams with different spin polarizations propagating in different directions .
the consideration was restricted by propagating states since only the case of uniform potential landscape was studied .
spin - dependent reflection of electrons from a lateral barrier in 2d system was observed in insb / inalsb heterostuctures @xcite and described theoretically in ref . .
silvestrov and mishchenko @xcite demonstrated the possibility to achieve spin - polarized currents in a 2d system with smooth potential barrier and spatially - uniform soi by considering propagating modes within semiclassical approach . in this paper
we show that strong polarization effect appears in 2d structures when electrons pass through a lateral potential barrier , in which the soi is stronger than in the outside 2d electron gas regions ( reservoirs ) .
the polarization arises even if the electric current is normal to the barrier .
the highest polarization is attained when the electron energy is close to the conduction band bottom in the barrier . in this case
the fact becomes important that the soi effectively splits the barrier height so that some part of electrons passes through the barrier via propagating states while others do this via the evanescent modes .
since the spin and orbital degrees of freedom are coupled , rather strong spin filtering occurs .
we study the total spectrum of electron states in 2d bounded system with soi to find that there are two type of evanescent states .
the first - type states are characterized by an imaginary longitudinal wavevector .
they fill in a gap in the propagating state spectrum .
the states of the second type lie in the forbidden gap .
they are described by a complex wavevector .
the electron tunneling through a lateral barrier with soi via these evanescent states exhibits unusual features , such as an oscillatory behavior of the transmission coefficient with the barrier width .
but of most interest is the spin polarization of the electron current .
the polarization efficiency is high enough even if the distribution function of incident electrons is symmetric with respect to the transverse momentum .
we explore the polarization efficiency in a wide range of electron energy to find that most effective spin filtering occurs if the fermi energy is close to the barrier height .
the paper is organized as follows . in sec .
[ complexstructure ] we describe the complex band structure and total spectrum of electron states . sec .
[ tunneling ] is devoted to the tunneling through a barrier with soi . in sec .
[ polarization ] the electric and spin currents through the barrier with soi are considered for a wide range of the fermi energy in the reservoirs .
the results obtained for the rashba soi are generalized to the dresselhaus soi in sec .
[ dresselhaus ] .
we summarize and conclude in sec .
[ conclude ] .
we start by considering the total spectrum of propagating and evanescent electron states in 2d electron gas with the rashba soi .
the hamiltonian is @xcite @xmath0 where @xmath1 is the soi constant , @xmath2 are pauli matrices .
the eigenfunctions are @xmath3 where @xmath4 stands for spin states , @xmath5 , @xmath6 is a constant .
the eigenenergy @xmath7 and the spin function @xmath8 are defined by following equations : @xmath9 where @xmath10 is a characteristic wavevector of soi , @xmath11 and @xmath12 is the normalized eigenenergy : @xmath13 using eqs ( [ zeta - chi ] ) one obtains the dispersion equation @xmath14 and the spin function @xmath15 let us analyze the dispersion equation
. to be specific , assume that the system under consideration is infinite in @xmath16 direction and has a boundary in @xmath17 direction , i.e. the system is semi - infinite or finite in the @xmath17 direction .
in this case , the @xmath16 component of the wavevector @xmath18 is real , while @xmath19 is generally complex , @xmath20 . dividing the real and imaginary parts of eq .
( [ zeta ] ) we obtain an equation set determining the energy @xmath21 as a function of @xmath22 and trajectories in the ( @xmath23 ) plane , along which @xmath24 is real : @xmath25 where @xmath26 . and
@xmath27 are spin - splitted propagating modes of branch 1 ; lines @xmath28 and @xmath29 are evanescent states , branch 2 ; lines @xmath30 , @xmath31 are evanescent states in the forbidden gap , branch 3 . dashed lines @xmath32 , @xmath33 and @xmath34 , @xmath35 are the trajectories corresponding to these branches on the complex plane ( @xmath23 ) . ]
( [ dis2 ] ) possesses a solution in the following cases : i ) @xmath36 , ii ) @xmath37 and iii ) @xmath38 for @xmath39 .
the first case contradicts to eq .
( [ dis1 ] ) and hence can not be realized .
the second case generates two branches of the solution , which can be found after dividing the real and imaginary parts of eq .
( [ dis1 ] ) : \1 .
@xmath40 , @xmath41 \2 .
@xmath42 , @xmath43 , @xmath44 the third case gives one further branch : \3 .
this branch is defined for @xmath23 belonging to the folowing trajectory @xmath45 in the complex plane ( @xmath23 ) .
the eigenenergy and the spin function are @xmath46 the total spectrum is shown schematically in fig . [ f_spectrum ] where the energy @xmath7 is presented as a function of @xmath23 for given transverse momentum @xmath18 .
the form of all three branches is different in the cases @xmath47 and @xmath48 .
branch 1 describes the propagating states with real @xmath19 .
this branch is splitted by spin .
the energy gap , which opens at @xmath49 for @xmath50 , depends on @xmath51 .
branch 2 describes purely decaying evanescent states defined on the imaginary @xmath19 axis .
this branch connects the propagating state branches along the imaginary axis .
the spectrum of branch 2 is also splitted by spin .
the energy minimum @xmath52 is common for branches 1 and 2 .
the minimum is attained on branch 1 , if @xmath47 , or on branch 2 in the opposed case .
branch 3 corresponds to evanescent states in the forbidden gap , @xmath53 .
they are described by a complex longitudinal wavevector and therefore can be named `` oscillating '' evanescent states .
the trajectories , along which these states are defined , obey the following equation : @xmath54 they are shown in fig . [ f_trajectories ] .
there are trajectories of two types . if @xmath47 , the trajectories ( lines 1 , 2 ) intersect the imaginary axis @xmath55 . in the vicinity of the intersection point ,
@xmath56 and hence the wavefunction oscillates with the distance faster than decreases .
if @xmath48 the trajectories ( lines 4 - 6 ) intersect the @xmath57 axis . is fixed for each line : lines _ 1 - 6 _ correspond to @xmath58 . ]
it is seen that there are four states for any energy , in accordance with the number of the degrees of freedom of the system . at a given energy
all the states are distinguished by the wavevector components @xmath59 and the spin function .
the propagating states and the evanescent states of branch 2 have two subbranches divided by the energy , while the evanescent states of branch 3 are splitted in the complex momentum plane .
though the wavefunctions of the third branch are complex , they do not carry the current . using the hamiltonian ( [ hamiltr ] ) one obtains the following expression for the particle current in the state @xmath60 @xmath61 the wavefunctions of branch 3 , calculated with using eqs ( [ psi ] ) and ( [ branch3 ] ) , are easily seen to turn the @xmath17 component of the current ( [ current ] ) to zero while @xmath16 component is nonzero .
in this section we study the electron tunneling via the `` oscillating '' evanescent states of branch 3 . consider a 2d electron gas without soi divided into two semiplanes ( reservoirs ) by a rectangular barrier , in which the soi is present .
the barrier height is @xmath62 and the width is @xmath63 . let us calculate the transmission probability for electrons incident on the barrier from the left reservoir .
we use the presentation of the wavefunctions in the reservoirs in the basis of eigenstates @xmath64 of the @xmath65 matrix .
the basis functions are @xmath66 , where @xmath19 and @xmath18 are the wavevector components in the reservoirs , @xmath67 . in the left reservoir ( @xmath68 )
the wavefunction is @xmath69 where @xmath66 is the state vector of incident electrons .
the wavefunction of transmitted electrons ( @xmath70 ) is @xmath71 here @xmath72 and @xmath73 are reflection and transmission matrices .
the wavefunction in the barrier is expanded in the eigenstates ( [ psi ] ) of the hamiltonian ( [ hamiltr ] ) @xmath74 where @xmath75 denotes the complex wavevector in the barrier ; @xmath76 are indexes labeling all four evanescent states in the barrier .
they are described by the spinors ( [ psi ] ) , in which the real and imaginary parts of @xmath77 should be taken with different signs . the matrices @xmath72 , @xmath73 and @xmath78 are determined by an equation set , which follows from the boundary conditions at the interfaces of the barrier and 2d electron reservoirs .
boundary conditions for wavefunctions are obtained in a standard way by integrating the schrdinger equation over the infinitesimal vicinity of the boundary .
these conditions are well known for a boundary between regions with different strength of soi @xcite . in our case
it is necessary to take into account that the lateral potential step at the boundary also contributes to the soi . in the transition region , where the potential @xmath79 varies with @xmath17 , the following additional term
should be added to the hamiltonian ( [ hamiltr ] ) @xmath80 where @xmath81 is soi constant connected with @xmath1 by @xmath82 , @xmath83 being the electric field perpendicular to the 2d layer .
this term having been integrated over the transition region gives a finite contribution to the boundary conditions .
finally one obtains the following equations for the spinor ( [ psi_spinor ] ) components : @xmath84_{-0}^{+0}=0\\ \label{boundary_cond3 } & \dfrac{1}{m(x)}\left[\dfrac{\partial \psi_2}{\partial x } \mp \beta(x)k_y \psi_2 + a(x ) \psi_1\right]_{-0}^{+0}=0\,,\end{aligned}\ ] ] where the parameter @xmath85 describes the rashba soi caused by the in - plane field .
the upper and lower signs in eqs ( [ boundary_cond2]),([boundary_cond3 ] ) correspond to the boundaries at which @xmath79 increases or decreases with @xmath17 . applying these boundary conditions to the system under consideration we put @xmath86 in the 2d electron reservoirs and
keep @xmath87 in the barrier . for simplicity , we ignore the difference in the effective masses of electrons in the barrier and reservoirs .
in addition , the wavevectors in the barrier and the reservoirs should be matched .
the tangential components @xmath18 must be equal .
the relation of the normal components is determined by equaling the energy @xmath88 of an incident electron to the electron energy in the barrier @xmath89 where @xmath90 is defined by eqs ( [ branch1]),([branch2]),([branch3 ] ) in accordance with the energy spectrum branch ( 1 , 2 or 3 ) , which is considered . in the present section we restrict ourselves by branch 3 .
finally one obtains two equation sets for the cases of the incident spin directed up and down .
each equation set contains eight equations . dropping elementary calculations and combersome expressions for matrices @xmath73 , @xmath72 and @xmath78 , we turn directly to main results .
first , note that the @xmath73 matrix obeys the following symmetry relations : @xmath91 the particle current through the barrier is @xmath92 where @xmath93 is the transmission coefficient , @xmath94 , which has a meaning of the probability for an electron to tunnel with any spin in the final state .
the peculiarity of the tunneling in the presence of the soi consists in the involvement of four interfering states with different spin structure .
one of unusual consequences of this fact is an oscillatory dependence of the tunneling coefficient on the barrier width .
this feature is demonstrated in fig .
[ tunnel_oscillat ] .
the oscillations exist when the electron energy is close to the top of the third branch : @xmath95 , where @xmath96 is a characteristic energy scale of soi .
the oscillations fade away as the energy decreases deep into the forbidden band , because @xmath97 exceeds @xmath98 .
the oscillations disappear also if the tangential momentum @xmath18 goes to zero . , @xmath99 , @xmath100 and @xmath101 .
] fig .
[ tunnel_oscillat ] shows clearly that the barrier filters incident electrons by spin .
this process depends evidently on the incident angle and the energy of electrons .
consider the spin polarization of transmitted electrons in more detail .
let an unpolarized electron flow with the wavevector @xmath102 is incident on the barrier .
the flow consists of two components @xmath103 and @xmath104 with opposed spins .
the transmitted flow acquires spin polarization .
the spin density of transmitted electrons is @xmath105 with using eq .
( [ rwave ] ) , the spin density takes the form : @xmath106 , \label{mu_k - s1}\ ] ] where @xmath107 is a normalization constant .
the components of the spin polarization are shown in fig .
[ tunnel_polariz ] as functions of the transverse momentum @xmath18 for a given energy of incident electrons .
the main property of the acquired polarization is that the @xmath17 and @xmath108 components of the spin polarization are odd functions of @xmath18 , while the @xmath16 component is an even function of @xmath18 .
this property is independent of the electron energy , the barrier height , the soi strength , and thus is universal for the rashba soi .
two consequences follow from this fact .
first , the unpolarized electron follow acquires spin polarization even if the distribution function of the incident electrons is symmetric with respect to the tangential momentum direction , the polarization being directed parallel to the barrier .
second , if the distribution function is not symmetric about @xmath18 , the spin polarization arises also in the @xmath17 and @xmath108 directions . .
the curves are marked by letters corresponding to the polarization components . the incident electron energy @xmath109 , the barrier height @xmath110 , the barrier width @xmath111 , the lateral soi parameter @xmath101 . ]
detail analysis shows that the spin polarization is caused mainly by the soi in the barrier , as it is described by the hamiltonian ( [ hamiltr ] ) .
the soi at the boundaries of the barrier ( described by eq .
( [ boundarysoi ] ) ) does not essentially affect the results if the parameter @xmath112 defined by eq .
( [ beta ] ) is small .
this case corresponds to realistic situation in experiments .
numerical estimations for inas quantum well ( @xmath113 evcm , @xmath114 v / cm , @xmath115 mev ) give @xmath116 .
if @xmath117 , the boundary soi changes the value of polarization , but main features ( such as the symmetry relations , the oscillatory behavior of the tunneling coefficients ) remain qualitatively similar .
in this section we turn from the separate electron states to the total spin polarization produced by all electron states contributing to the current through a barrier with soi .
in addition , we consider a wide range of the incident electron energy to include all three branches of the electron spectrum .
this allows one to find out conditions under which the electron current is polarized most effectively .
consider an electron current directed normally to the barrier . for definiteness ,
let the current be caused by a voltage @xmath118 applied across the 2d electron reservoirs , as it is shown in fig .
[ scheme ] .
the incident electron states , which contribute to the current , occupy an energy layer near the fermi level .
they are located in a semi - ring in the @xmath119 plane , shown in the insertion .
the particle and spin currents through the barrier are determined by summing over all these states .
we carry out this calculation for various positions of the fermi level @xmath120 relative to the barrier height @xmath62 to find the spin - polarization efficiency as a function of @xmath120 .
space , in which the shaded region shows the electron states contributing to the current . ] for simplicity suppose that the voltage is small compared to other energies , @xmath121 .
this simplification allows one to restrict the summation by the integration in @xmath122 space over the azimuthal angle at a given energy @xmath88 .
the integration is conveniently to carry out over @xmath18 , but in doing this the fact should be taken into account that the set of branches , which must be used at given @xmath18 and @xmath88 , can change with varying @xmath18 .
this happens because the gap @xmath123 between spin - splitted subbranches of propagating modes ( curves 1 + and 1- in fig .
[ f_spectrum ] ) increases with @xmath18 ( @xmath124 ) and furthermore the form of the dispersion curves describing the propagating ( 1 + , 1- ) and evanescent ( 2 + , 2- ) modes changes .
the switches between the actual branches occur when any extremum of the dispersion curves ( which is a function of @xmath18 ) coinsides with the energy @xmath88 .
physically this means that electrons incident on the barrier at different angles and in different spin states feel different effective barrier height .
[ diagram ] presents a diagram showing which branches of the dispersion relation are accessible for electrons with given @xmath88 and @xmath18 . within each region
bounded by thick lines , there are two different branches with two fundamental eigenfunctions on each ones or one branch with four solutions . and tangential momentum @xmath18 .
thick lines divide the @xmath125 plane into 6 regions , in which corresponding branches are denoted by numbers 1,2,3 and chirality indexes @xmath126 in accordance with fig .
[ f_spectrum ] .
line @xmath127 is defined by equation @xmath128 for each region of the diagram the wavefunctions are determined in the same way as in sec .
[ tunneling ] .
the only difference is that the eigenfunctions of the branches specified in the diagram are to be used in eq .
( [ bwave ] ) instead of the wavefunctions of branch 3 . as a result of these calculations the transmission @xmath129 and reflection
@xmath130 matrices are obtained for the whole @xmath125 plane .
using @xmath129 we find the particle and spin currents in the right electron reservoir .
the particle current is found by the summation of eq .
( [ current_k ] ) over all states of incident electrons : @xmath131 where the integration symbol implies also the summation over all regions of the @xmath125 plane which fall within the @xmath132 interval , with @xmath133 being the fermi wavevector in the 2d reservoirs .
the transmitted spin current in the left reservoir is defined in a standard way @xcite using the following expression for the current in a state @xmath134 : @xmath135 where @xmath136 denotes the velocity components in the plane , @xmath137 denotes the spin components in 3d space . in the case under consideration ,
the @xmath17 component of the total spin current is @xmath138 where three lines in the rhs correspond to the @xmath139 components of the spin polarization for the spin current directed along @xmath17 axis .
the efficiency of spin polarization is characterized by the ratio of the spin current to the particle current @xmath140 using the symmetry relations ( [ symmetry ] ) and eq .
( [ spin_current ] ) one finds that @xmath17 and @xmath108 components of the spin current are absent in the case of the rashba soi , and only the @xmath16 component is nonzero .
of course , this is a consequence of the symmetry of the distribution function with respect to the sign of @xmath18 . if the current had not be perpendicular to the barrier
, the polarization would appear also in the @xmath17 and @xmath108 directions .
the polarization efficiency turns out to be sufficiently high .
the dependencies of the polarization on the fermi energy and the barrier width are nontrivial because they reflect a complex structure of the electron spectrum .
below two most significant results are considered .
the dependence of @xmath141 on the barrier width is shown in fig .
[ polar - d_3 ] for energies below the top of the energy band where the evanescent states of branch 3 exist in the barrier , @xmath142 .
the polarization efficiency is seen to oscillate with @xmath63 because of the interference of four oscillating evanescent modes .
the effect is rather strong and becomes the stronger the closer is the energy to the band top .
the oscillation period is of the order of @xmath143 . for tunneling through branch 3 .
@xmath144 , @xmath145 . ] the dependence of @xmath141 on the fermi energy is presented in fig .
[ polar - e ] for a wide energy range including the energy both below and above the barrier .
it is seen that there is a critical energy @xmath146 , which coincides with the top energy of the third branch band , @xmath147 , in the barrier . at energies well below @xmath148 , the polarization efficiency @xmath141 increases with the energy . in the vicinity of the threshold
@xmath148 an oscillatory behavior appears as a result of the interference of four slowly decaying waves . above this energy
the transmission process goes via the propagating states of branch 1 and the evanescent states of branch 2 . at @xmath120
slightly higher than @xmath148 the polarization efficiency attains the highest value . with further increasing the energy the efficiency @xmath141 decreases .
it is worth noting that the highest efficiency of the spin polarization slightly depends on the soi strength , but the barrier width , at which this high polarization is attained , varies inversely with the soi constant .
[ polar - e ] demonstrates also how @xmath141 changes with the barrier width .
increasing @xmath63 above @xmath149 leads to more pronounced oscillatory dependence of @xmath141 on the energy caused by the four - wave interference . for @xmath150 ( solid line 1 ) and @xmath151 ( dashed line 2 ) ;
vertical dashed line marks the threshold energy @xmath147 , which divides the evanescent states of branch 3 ( on the left of @xmath148 ) from the states of branches 2 and 1 . ]
the physical mechanism , owing to which the polarization of normally incident electron current appears , is connected with the splitting of electron waves in the barrier because of the soi .
let us consider first a simplified case of semiinfinite barrier region with soi .
let the energy is high enough so that electrons occupy propagating states ( branches 1+,1- ) .
the incident electron states can be represented in terms of two chiral modes : @xmath152 in the barrier region each electron beam splits into two beams , which propagate at different angles and have different chiralities : @xmath153 where @xmath154 is @xmath17-component of the wavevector of the upper mode with positive chirality @xmath155 and @xmath156 corresponds to the lower mode with negative chirality @xmath157 .
it is essential that @xmath158 .
the incident and refracted beams as well as their spin polarizations are illustrated in fig .
[ birefringence ] .
the transmitted beam amplitudes are @xmath159 and @xmath160 for the incident beam with positive chirality , and @xmath161 and @xmath162 for negative chirality .
the @xmath17 and @xmath16 components of the spin polarization in the barrier are estimated as @xmath163 where the dots denote spatially dependent terms originating from the interference . and @xmath164 are amplitudes for incident beam with @xmath165 .
the primed letters denote the same amplitudes for @xmath165 . ]
the total spin polarization is determined by the sum over all incident angles . to estimate this sum ,
consider the dependence of the spin components on @xmath18 .
it is clear that @xmath166 and @xmath167 are even functions of @xmath18 .
real and imaginary parts of @xmath168 are seen from eq .
( [ chi_12 ] ) to be correspondingly odd and even functions of @xmath18 .
therefore @xmath169 is an even function of @xmath18 and @xmath170 is an odd function , as it is illustrated in fig .
[ birefringence ] .
thus the total @xmath169 component vanishes while @xmath170 is nonzero .
it is easy to find the direction of the total spin for the energy close to the barrier height .
if @xmath171 , the amplitudes @xmath172 and @xmath173 are equal , @xmath174 and @xmath175 .
hence , the total spin density is zero because the spins of opposed chiralities cancel each other . at small @xmath18
, the spin polarization appears .
since the amplitudes are functions of @xmath176 , they remain unchanged in the first approximation , so that the resulting spin is determined by the difference @xmath177 and @xmath178 since @xmath158 , the @xmath16-component of the spin density is negative , i.e. the sign of the spin polarization is determined by the lower - energy branch of propagating mode . if the barrier width is finite , there are four modes in the barrier , but the above property remains unchanged .
it is also kept for other branches of the electron spectrum .
all the above results are generalized to the case of dresselhaus soi . for a 2d system oriented along [ 001 ]
crystallographic direction , the soi hamiltonian is @xcite @xmath179 it is well known that dresselhaus and rashba hamiltonians are unitary equivalent .
@xcite an unitary matrix @xmath180 transforms the rashba hamiltonian ( [ hamiltr ] ) to the dresselhaus one : @xmath181 .
therefore , the dresselhaus soi case does not require separate calculations .
it is enough to carry out this transformation .
taking into account that matrix ( [ r - d_matrix ] ) transforms the spin matrices as follows : @xmath182 , @xmath183 , and @xmath184 , we arrive at the conclusions : ( i ) the dispersion equation is the same as in the rashba case ( [ zeta ] ) ; \(ii ) the spin functions differ from those defined by eq .
( [ chi ] ) by a simple substitution @xmath185 ; \(iii ) the spin components of transmitted electrons in the incident states @xmath103 and @xmath104 differ from those of sec.[tunneling ] by replacements : @xmath186 , @xmath187 and @xmath188 ; \(iv ) the spin polarization of the current transmitted through a barrier with soi has only @xmath17 component , if the distribution function is even with respect to the transverse momentum .
in particular , figs [ polar - d_3 ] , [ polar - e ] are valid for the spin polarization normal to the barrier , @xmath189 .
of course , if the rashba and dresselhaus mechanisms act simultaneously the results change qualitatively .
we have found the total spectrum of electron states in a bounded 2d electron gas with soi .
it addition to well known propagating states it contains two branches of evanescent states .
their wavefunctions decay with the distance in the direction @xmath17 perpendicular to the boundary . one branch ( `` purely decaying '' evanescent mode ) is described by an imaginary wavevector .
the energy of this state is splitted by spin , so that there are two subbranches .
they fill in the gap , which opens in the propagating state spectrum at @xmath190 .
other branch ( `` oscillating '' evanescent mode ) is characterized by a complex wavevector @xmath77 .
these states lie in the forbidden gap .
we have studied the electron transmission through a lateral potential barrier with the soi . in the energy range , where electrons tunnel via the oscillating evanescent states
, the tunneling reveals unusual features , such as an oscillatory dependence of the transmission coefficients on the barrier width and the energy .
but of most importance is the spin polarization of the electron current .
the value and direction of the polarization depend on the angle of incidence and the energy of incident electrons .
the polarization appears even if the distribution function of incident electrons is symmetric with respect to the transverse momentum . in this case
the polarization is directed parallel to the barrier ( in the rashba soi case ) or perpendicular to it ( for dresselhaus soi ) .
the highest efficiency of the spin polarization is attained when the fermi energy is close to the barrier height . in this case , electrons pass through the barrier partially via the propagating states and partially via the purely decaying evanescent states . under this condition
the most effective spin filtering occurs .
the maximal polarization efficiency depends on the barrier height and can exceed 0.5 if the barrier width is on the order of @xmath143 .
this work was supported by russian foundation for basic research ( project no .
05 - 02 - 16854 ) , russian academy of sciences ( programs `` quantum nanostructures''and `` strongly correlated electrons in semiconductors , metals , superconductors , and magnetic materials '' ) , rf ministry of education and science .
rashba , fiz .
tela ( leningrad ) * 2 * , 1224 ( 1960 ) [ sov .
solid state * 2 * , 1109 ( 1960 ) ] ; yu.a .
bychkov and e.i .
rashba , pisma zh .
. fiz . * 39 * , 66 ( 1984 ) [ sov .
jetp lett .
* 39 * , 78 ( 1984 ) ] .
|
we find that the total spectrum of electron states in a bounded 2d electron gas with spin - orbit interaction contains two types of evanescent states lying in different energy ranges .
the first - type states fill in a gap , which opens in the band of propagating spin - splitted states if tangential momentum is nonzero .
they are described by a pure imaginary wavevector .
the states of second type lie in the forbidden band .
they are described by a complex wavevector .
these states give rise to unusual features of the electron transmission through a lateral potential barrier with spin - orbit interaction , such as an oscillatory dependence of the tunneling coefficient on the barrier width and electron energy .
but of most interest is the spin polarization of an unpolarized incident electron flow .
particularly , the transmitted electron current acquires spin polarization even if the distribution function of incident electrons is symmetric with respect to the transverse momentum .
the polarization efficiency is an oscillatory function of the barrier width .
spin filtering is most effective , if the fermi energy is close to the barrier height .
|
the exchange bias phenomenon is associated with the interfacial exchange coupling between ferromagnetic and antiferromagnetic spin structures , resulting in an unidirectional magnetic anisotropy that causes a shift of the hysteresis loop to negative field values as one cools the system through the nel temperature of the antiferromagnet(af ) in a positive magnetic field @xcite .
the exchange bias effect is essential for the development of magneto - electronic switching devices ( spin - valves ) and for random access magnetic storage units . for these applications a predictable , robust , and tunable exchange bias effect
is required .
extensive data have been collected on the exchange bias field @xmath0 , and the coercivity field @xmath1 , for a large number of bilayer systems , which are reviewed in ref .
the details of the eb effect depend crucially on the af / f combination chosen and on the structure and thickness of the films .
however , some characteristic features apply to most systems : ( 1 ) @xmath0 and @xmath1 increase as the system is cooled in an applied magnetic field below the blocking temperature @xmath2 of the af layer , where @xmath3 is the nel temperature of the af layer ; ( 2 ) the magnetization reversal can be different for the ascending and descending part of the hysteresis loop @xcite , as was first pointed out in reference @xcite , and recently also observed by lee et al .
@xcite and gierlings et al .
@xcite ; ( 3 ) thermal relaxation effects of @xmath0 and @xmath1 indicate that a stable magnetic state is reached only at very low temperatures @xcite . furthermore
, a positive @xmath0 has been observed after cooling an af / f system in very high magnetic fields @xcite , and a perpendicular exchange bias was realized by using a ferromagnetic layer with perpendicular anisotropy @xcite .
several theoretical models have been developed for describing possible mechanisms of the eb effect , including domain formation in the af layer with domain walls perpendicular to the af / f interface @xcite , creation of uncompensated excess af spins at the interface @xcite , or the formation of a domain wall in the af layer parallel to the interface @xcite .
another approach is the consideration of diluted antiferromagnets in a field . in the work of milt@xmath4nyi et al .
, keller et al .
@xcite , and nowak et al.@xcite the discussion about compensated versus uncompensated interfacial spins is replaced by a discussion of net magnetic moments within the antiferromagnetic layer@xcite .
they show that perpendicular magnetic domain walls in the af layer can effectively be pinned at defect sites , providing an enhanced exchange bias field .
recent overviews of theoretical models on the exchange bias effect are provided by stamps @xcite and by kiwi @xcite .
however , so far there is still not a satisfactory agreement between the different theoretical approaches and experimental observations of the macroscopic and microscopic response of exchange bias systems . one step forward in describing the exchange field behavior , as we will demonstrate in the present report , could come from a detailed analysis and understanding of the asymmetric behavior of the magnetization reversal and relaxation processes of the magnetization close to the coercive fields . to this end
, we have chosen to investigate the archetypal coo / co(111 ) af / f exchange bias system because it offers a number of advantages over similar systems , which show also the eb effect .
first , co has very good growth properties as a thin film and very flat interfaces can be prepared by sputtering and molecular beam epitaxial techniques .
second , the nel temperature of the oxide is at a convenient temperature of @xmath5 k. third , coo provides a large eb even for very thin layers , which can easily be prepared by thermal oxidation @xcite .
furthermore , coo / co bilayers exhibit straightforward and pronounced eb properties , which qualifies them as a model system for detailed investigations of the magnetization reversal process @xcite .
this is the case although the spin structure of the antiferromagnetic coo is rather complex . in coo
the spin - orbit interaction and therefore the crystal anisotropy is strong . for bulk
coo the easy axis is reported to be along the [ 117 ] directions @xcite . however , in very thin layers because of missing neighbors the easy axis has still to be determined .
training effects , which have been observed for coo / co , add an additional difficulty
. however , training effects can also be useful for the understanding of the metastability of interfaces , as we will show further below . in the present paper
we investigate the thickness dependence of the exchange bias effect by using a wedge type coo / co(111 ) bilayer with a co thickness varying from 27 to 400 .
previous studies by gruyters and riegel @xcite have shown that the shape of the hysteresis changes dramatically with the co thickness , indicating that aside from film orientation and exchange coupling the film thickness plays an important role for the type of domains formed and magnetization reversal observed . here we report extensive hysteresis measurements of coo / co(111 ) bilayers , including a detailed analysis of the interfacial domain state by investigating the specular and off - specular neutron spin flip cross section at the coercive fields for descending and ascending magnetic fields .
a series of nine samples were prepared by rf - sputtering in 99.99% ar ( samples s1-s9 in table .
[ table1 ] ) .
the a - plane al@xmath6o@xmath7 substrates were kept at @xmath8 c during the co deposition , which is the optimized growth temperature as concerns mosaicity , structural coherence length , and surface morphology , as determined in ref .
the substrate for sample s9 was placed in the middle of the sample holder while the other 8 substrates were placed at equal distances away from each other .
the design of the wedge type coo / co(111 ) sample grown on a sapphire substrate . before growth ,
the al@xmath6o@xmath9 substrate is cut in 9 pieces and placed on the sample holder at equal distances away from each other .
the gradient growth rate provides samples s1 to s9 of different thickness . ]
taking advantage of the natural gradient in the sputtering rate , the samples acquire an increasing thickness from 30 ( s1 ) to 400 ( s9 ) ( see fig . [ sample ] ) .
subsequently , the samples were exposed to air at room temperature which results in a 25 thick coo layer on top of the co film .
the characterization of the surface morphology by atomic force microscopy ( afm ) shows an exceptionally low roughness of 1 for all samples , which has also been confirmed by x - ray reflectivity measurements performed at the w1.1 beam - line of the hasylab . as a representative example in fig .
[ xrays5 ] the reflectivity is shown for the thickest sample ( s9 ) together with a best fit to the data points .
the thickness parameters obtained for all samples from the fits are listed in table [ table1 ] .
high angle x - ray diffraction measurements ( fig .
[ xrays5 ] ) show that the co layer consists mainly of the cubic fcc phase with the ( 111 ) growth direction yielding a strong peak at @xmath10 @xmath11 , and a minority hcp phase with a tilted @xmath12 growth responsible for the peak at @xmath13 @xmath11 .
the observed fcc peak position is closer to the ideal fcc position at @xmath14 @xmath11 than to the ideal hcp position at @xmath15 @xmath11 , which is in good agreement with earlier measurements of co films on sapphire substrates grown at the same substrate temperature and similar top panel : reflectivity measurement of sample s9 ( coo(25)/co(400)).using synchrotron radiation ( open dots ) .
the solid line shows a best fit to the data points .
bottom panel : x - ray radial bragg scan parallel to the surface normal of coo(25)/co(150 ) .
the main peak at @xmath163.08 @xmath11 is close to the ideal fcc co(111 ) position , indicating a majority fcc phase .
the second peak is from a minority tilted hcp component with [ 10@xmath171 ] orientation .
the inset shows a transverse scan through the co(111 ) peak . ]
the full width at half maximum ( fwhm ) of the rocking curve of the pseudo co(111 ) bragg peak is 0.01 degrees for an incoming wavelength of 1.393 , confirming a very good texture of the co film .
natural oxidation of the co(111 ) films leads to a preferential growth of coo with ( 111 ) orientation .
we have studied , in addition , epitaxial co(111 ) layers , which have been thermally oxidized _ in - situ_. synchrotron diffraction studies of the in - plane coo structure showed that [ 111 ] is the surface normal of the 2.5 nm thick coo film @xcite . by analogy
we assume that for the highly textured coo / co bilayer used for the present study , the same orientational relationship holds .
if the interface were ideal , this orientation would provide uncompensated spins and therefore would exhibit an exchange bias effect .
this has been proven in the work of gkemeijer et al.@xcite , who showed that for coo / co single crystalline films an exchange bias occurs only in the ( 111 ) orientation but not in the ( 110 ) and ( 100 ) orientations .
the presence of uncompensated spins at the interface is one of the key requirements for achieving a shift of the hysteresis loop @xcite , other important factors are af exchange constant and af magneto - crystalline anisotropy @xcite .
we first show magnetic characterizations of the co films in the unbiased state at room temperature by taking hysteresis loops with the magneto - optical kerr effect ( moke ) .
several hysteresis loops were taken with the field parallel to the film plane but with different azimuth angles of the sample .
a plot of the ratio between the remanent magnetization and saturation magnetization @xmath18 , versus the rotation angle @xmath19 , reveals a two - fold in - plane anisotropy , which is observed for all samples and which is induced by the sapphire substrate .
in addition a small contribution from a four - fold anisotropy can be recognized by closer inspection .
these results are in complete agreement with earlier hysteresis measurements of co(111 ) films on al@xmath6o@xmath20 substrates @xcite . a typical example for an unbiased hysteresis loop recorded at 270 k is plotted in fig .
[ 5hys ] for sample s4 .
it is characterized by a completely symmetric shape and a small coercive field of 60 oe .
additional temperature dependent studies of the coercive field showed that the nel temperature of a 25 thick coo layer is about the same as in the bulk ( @xmath21291 k ) .
in fact , it might even be slightly higher than in the bulk as shown in ref.@xcite .
remanent magnetization divided by the saturation magnetization versus the rotation angle of the sample coo(25)/co(26 ) ( full circles ) and coo(25)/co(400 ) ( open triangles ) showing .
the two fold anisotropy was observed for all samples . ]
the temperature dependence of the hysteresis loops for samples s1-s9 was measured by a superconducting quantum interference device ( squid ) magnetometer .
the exchange bias field @xmath0 , the coercive fields @xmath22 and @xmath23 , and the half width of the total coercive field @xmath24 ) were extracted from the hysteresis loops measured at temperatures between @xmath25 and @xmath26 .
the samples were cooled from above the nel temperature to below different target temperatures in a magnetic field of @xmath27 = + 2000 oe applied closely along the easy axis of the sample magnetization .
the procedure was repeated for each target temperature . in fig . [
5hys ] typical hysteresis loops are shown for the sample s4 with a co thickness of @xmath28 and for the target temperatures 270 k , 160 k , 80 k , and 5 k. the hysteresis curves were taken directly after cooling and thus before training , unless otherwise stated .
the hysteresis loops show the following typical and general features : @xmath22 increases strongly with decreasing temperature , while @xmath23 remains almost constant at a field value of about 300 oe . the slope of the hysteresis loops at @xmath22 is steeper than at @xmath23 on the return path .
hysteresis loops for different temperatures of sample s4 ( coo(25)/co(120 ) ) . for each hysteresis
the bilayer was cooled in a field of + 2000 oe from 320k to the respective temperature .
the hysteresis curves were taken after cooling and before training . ]
[ poseb ] summarizes the analysis of the temperature dependence of @xmath22 , @xmath23 , @xmath29 , and @xmath0 for the samples s2 and s3 .
the top panel reproduces the coercive field of the first reversal @xmath22 and of the second reversal @xmath23 for the samples s3 .
both coercive fields start to slightly increase just below the nel temperature with the same rate of 0.21 oe / k .
top panel : coercive fields @xmath22 and @xmath23 of sample s3 .
the line is a fit to the linear region of @xmath23 .
it intersects the abscissa at the blocking temperature @xmath30 = 186 k. + bottom panel : temperature dependence of the coercive field and the exchange bias field for samples s2 and s3 . in the inset
the positive exchange bias occurring close to the blocking temperature is shown .
the lines are guides to the eye . ] at the blocking temperature of about @xmath31 186 k , the slope increases drastically .
below @xmath30 a bifurcation for the temperature dependence of @xmath22 and @xmath23 develops .
while @xmath22 keeps rising with a rate of 11.1 oe / k , @xmath23 levels off and reaches saturation at the lowest temperature .
thus , there are three distinguishable temperature regions : ( 1 ) from @xmath3 to @xmath30 the coercive fields are equal and increase slowly ; ( 2 ) close to @xmath30 the slopes increase drastically and @xmath22 is slightly smaller than @xmath23 ; ( 3 ) below @xmath30 both coercive fields develop linearly but with different slopes such that the absolute value of @xmath22 is bigger than @xmath23 . only in this last region
a strong negative eb is observed .
the different temperature dependencies of both coercive fields are consistent with the different magnetization reversal processes , which are recognized via polarized neutron reflectivity measurements at these points to be discussed further below .
previous studies of coo / co thin film systems have shown an inverse proportionality between the exchange bias field and the co layer thickness @xcite . our experiments completely confirm these results and extend the linearity down to a thickness of 27 ( see table .
[ table1 ] and fig .
[ ddep ] ) . in fig .
[ poseb ] the exchange bias and coercive fields for samples s2 and s3 are shown . at @xmath30=186
k the @xmath0 becomes positive and reaches a maximum of @xmath0 = + 20 oe . the temperature region of the positive eb is shown on an enlarged scale in the insert of fig .
[ poseb ] .
this surprising feature has been observed for all our samples . after changing sign , @xmath0 decreases steadily as the temperature is lowered .
because of the sign change we define the blocking temperature @xmath30 as the temperature where @xmath0 first deviates from zero .
this definition coincides with a linear extrapolation of @xmath22 to zero , shown by the solid line in the bottom panel .
the linear dependence of @xmath29 and @xmath0 on the temperature is in good agreement with reports from other coo / co bilayers @xcite .
a positive @xmath0 close to @xmath30 is seen in our bilayers for the first time @xcite .
positive exchange bias fields have been reported in the literature , but only for high cooling fields , when the external field exceeds the interfacial coupling , breaking the parallel alignment between the f and af layer @xcite .
recently , three further papers have reported about the observation of a positive exchange bias effect @xcite .
in contrast to those papers we observe a weakening of the positive exchange bias after training .
this is shown in fig .
[ pextr ] .
first ( solid circles ) and second hysteresis ( open squares ) loop taken at 175 k. for the second hysteresis loop h@xmath32 decreased while h@xmath33 remained constant , implying a reduction of the positive exchange bias after training . ]
the positive exchange bias is considered as a proof for an antiparallel alignment of the spins at the af / f interface in moderate cooling fields @xcite . for parallel alignment of the magnetic moments in the f and
af layers the bias field would not change sign even in high fields .
therefore , the change of sign from negative to positive exchange bias as a function of applied field hints to antiferromagnet interfacial coupling requiring a superexchange mechanism working at the f / af interface@xcite . in
order get more insight into the origin of the positive exchange bias , we have measured its field cooling dependence shown in fig .
[ fcdep ] . the sample s2(see table .
[ table1 ] ) was cooled down from t=320 k to 175 k in different applied fields , ranging from -20 oe to 40 koe ( see the inset of ( fig .
[ fcdep ] ) .
first observation is that the exchange bias is positive for all cooling fields .
second the bias field is constant and about + 20 oe between 1 koe and 40 koe , but below 1 koe the bias field increases dramatically to twice its former value .
this increase is solely due to a shift of @xmath22 to smaller values with decreasing cooling field , while @xmath23 remains constant .
we also note that the increasing eb field correlates with a decreasing remanence at 320 k as the cooling field is lowered .
thus for small cooling fields the initial magnetic state of the sample is not fully saturated . at these small cooling fields ( @xmath34 1000 oe ) we observe a surprisingly strong increase of the eb field .
@xmath35 ( open squares ) , @xmath23 ( full squares ) , @xmath36 ( full circles ) , @xmath29 ( open circles ) are plotted as a function of the co thickness .
the samples were cooled down through the nel temperature of coo to 10 k in an applied magnetic field of + 2000 oe .
the lines are linear fits to the data points . ]
this behavior can be explained assuming two contributions to the interfacial exchange coupling .
normally the interface exchange coupling is ferromagnetic between coo and co as already mentioned above , being the result of a direct exchange of co monolayers on either side of the ferro- and antiferromagnetic layer@xcite . for a positive exchange bias effect we have to assume that some parts at the interface are , however , antiparallel coupled .
then , an antiparallel alignment may be caused by a superexchange type of mechanism mediated by an oxygen monolayer at the interface .
it is reasonable to assume that due to roughness or thickness fluctuations , small fractions of the coo film may be oxygen terminated instead of metal terminated at the interface .
next we discuss the training effect in the region below the blocking temperature @xcite , which is different from the training for the positive exchange bias fig.[pextr ] . after closing a complete loop
the hysteresis becomes symmetric and assumes an s - like shape .
the @xmath22 field for all subsequent loops and the eb - field is smaller than for the first reversal .
thus the trained hysteresis has a different shape and a smaller eb field than the hysteresis of the virgin sample .
obviously the first reversal is dominated by a different process than the following reversals , transforming the sample from a unique and irreversible state to a different state with reproducible and symmetric branches of the descending and top : the field cooling dependence of the positive exchange bias for sample s2 ( coo(25)/co(56 ) ) .
the bilayer was cooled in different fields from above the nel temperature of coo to 175 k , where hysteresis loops were measured .
the inset shows a hysteresis loop at t=320 k ( full dots ) , together with the cooling field values ( down triangles ) as reference of the initial state of the sample .
bottom : the dependence of the coercive field values ( @xmath35 , @xmath23 and @xmath37 ) as a function of the applied cooling field .
the lines are guides to the eyes . ] ascending parts of the hysteresis loop .
when the sample is first cooled in a positive field , the co layer consists of a single ferromagnetic domain . upon cooling below the blocking temperature ,
the orientation of antiferromagnetic domains in coo is affected by the exchange coupling at the af / f interface . to optimize the interfacial energy and to avoid spin frustration
it is favorable @xmath22[oe ] & @xmath23[oe ] & @xmath0[oe]&@xmath29 + s1&27&-6004&1362&-2321&3683 + s2&56&-2825&677&-1073&1751 + s3&87&-1990&640&-675&1315 + s4&119&-1458&343&-557&901 + s5&153&-1233&346&-443&789 + s7&260&-678&176&-251&427 + s8&320&-549&144&-202&346 + s9&398&-464&115&-174&290 + for the coo film to develop a multi - domain state .
however , after cooling in the saturation field of the ferromagnetic layer , the frustrated spins at the interface may very well be preserved , which would yield a coo layer in a single domain state .
the lower the interface roughness , the higher will be the probability for a single domain state .
the first reversal at @xmath22 changes the magnetic state of the interface , leaving behind a more disordered spin structure .
a more precise analysis of the processes which occur at the coo / co interface is possible with polarized neutron reflectivity measurements , to be discussed next .
in order to gain more insight into the reversal mechanism in decreasing and increasing magnetic fields , we have performed polarized neutron reflectivity ( pnr ) measurements on co / coo bilayers , using the angle dispersive neutron reflectometer adam at the institut laue - langevin , grenoble @xcite with a fixed wavelength of 4.41 .
preliminary results are presented in ref .
the sample has been optimized for the needs of neutron reflectivity and consists of a coo(25)/co(200 ) bilayer grown under the same conditions as explained above , but on a ti(2000 )/cu(1000 )/al@xmath6o@xmath9 neutron resonator template .
this configuration of the sample allows to observe with high sensitivity the ferromagnetic domains and the spin misalignment at the interfaces@xcite .
the method we used to study the magnetization reversal by pnr includes the following concepts : ( 1 ) the dependence of spin asymmetry measured at one specific value of the wave vector transfer depends only on the cosine of the angle ( @xmath38 ) between the magnetic induction in the magnetic layer and the neutron polarization direction ; ( b ) the spin - flip reflectivity is proportional to @xmath39 ; ( c ) the off - specular scattering shows the existence of magnetic domains within the magnetic layer ; ( d ) the enhancement of the neutron density achieved at the magnetic interface , using the neutron resonator , increases the off - specular scattering from the interfacial magnetic domain walls . in order to find the optimized scattering vectors for measuring neutron hysteresis loops , we have first recorded a complete set of polarized neutron reflectivities for all four cross - sections .
the raw data without experimental corrections ( spin flip efficiency and footprint ) is shown in fig .
[ refl4 ] .
the measurements were carried out at room temperature and close to the coercive field . since at room temperature without exchange bias ,
reversal is dominated by coherent rotation , the resonant peaks in the spin - flip reflectivities are strong and well defined .
we used this information to choose the proper wavevector ( @xmath40=0.016 @xmath41 ) for recording the spin - flip intensity as a function of the applied magnetic field at low temperatures .
similar , for the non - spin flip intensities we have fixed the scattering vector just above the critical edge for total external reflection of the unpolarized beam ( @xmath42=0.019 @xmath41 ) . @xmath43 and @xmath44 refer to spin flip and non - spin flip scattering , respectively .
we have chosen this @xmath42 value , because it provides the highest contrast for non - spin flip reflectivity from the magnetic layer combined with high intensity , whereas @xmath40 provides the maximum intensity . polarized neutron reflectivity measurements of a coo / co / ti / cu / al@xmath6o@xmath9 resonator sample taken at room temperature in a magnetic field of about 60 oe ( = h@xmath32 ) . the arrows for @xmath42 and @xmath40 denote the points where non - spin flip and spin flip intensities were recorded at low temperature after field cooling . ]
the magnetic hysteresis loop can be measured by pnr using the spin asymmetry ( sa ) defined as : @xmath45 where @xmath46 , @xmath47 , @xmath48 , and @xmath49 are the four reflectivities from a magnetic multilayer system , @xmath50 , @xmath51 , and @xmath38 is the angle between the neutron polarization axis and the direction of the magnetic induction @xmath52 , with @xmath53 being the magnetization of the ferromagnetic layer .
@xmath54 is the perpendicular component of the incoming wave vector .
the spin asymmetry , depends obviously on the optical parameters and the thickness of the layers .
they have been omitted in the variable list because they are constant . for a single magnetic layer embedded in a non - magnetic matrix
the sa is expressed as : @xmath55 where @xmath56 is the component of the magnetic induction parallel to the polarization axes . for a simple case
we derive this expression in the appendix .
the spin - flip reflectivities are : @xmath57 where @xmath58 is the perpendicular ( to the neutron polarization direction ) component of the magnetic induction . the equations [ sa2 ] and [ sf1 ] are directly related to the perpendicular and parallel components of the magnetic induction of a ferromagnetic layer .
moreover , through the spin flip scattering one can decide whether the magnetization reversal is dominated by rotation or domain wall movement . the magnetic hysteresis @xmath59 loop measured by pnr
is shown in fig .
[ nhl1 ] together with the magnetic hysteresis obtained by moke from the same sample .
the field dependencies of the non - spin flip ( nsf ) reflectivities , from which the magnetic hysteresis is derived , are shown in panels ( b ) and ( c ) .
the non - spin flip reflectivities were measured at a wave vector transfer @xmath60 corresponding to the inflection point of the non - polarized neutron reflectivity ( near the critical edge for total external reflection ) and by sweeping the magnetic field in the usual manner . note that for technical reasons the moke hysteresis curve was taken at 50 k while the neutron measurements were performed at 10 k. both curves show the same asymmetric shape at @xmath22 and @xmath23 as discussed before .
the different @xmath22 values for the moke and pnr measurements are solely due to the different sample temperatures .
nonetheless , we find very good agreement between both loops during the first reversal . on the return path through @xmath23
the agreement is by far not as good as one would expect if no specular intensity loss were occurring due to off - specular diffuse scattering . on the other hand ,
this discrepancy between moke and pnr is an indication for the presence of magnetic domains in the ferromagnetic layer on the return path .
this will be discussed in more detail in the appendix ( section iv.2 ) .
in addition to the standard magnetization curves , pnr is capable of distinguishing between different magnetization reversal processes and to provide information on magnetic domains by analyzing the specular and the off - specular spin flip ( sf ) intensities @xmath61 and @xmath62 .
specular sf scattering is sensitive to magnetization components in the sample plane , which are perpendicular to the applied field direction .
the off - specular sf signal reveals the presence of magnetic domains @xcite .
we first discuss the specular spin flip intensities shown by triangles in panels ( b ) and ( c ) of fig .
they were measured at the scattering vector @xmath60 corresponding to the resonance peak near the critical edge .
the magnetization reversal at @xmath23 exhibits strong spin - flip intensities @xmath61 and @xmath62 .
this is always observed for rounded or trained hysteresis loops and is characteristic for a magnetization reversal via domain rotation @xcite .
magnetization reversal by rotation provides a large magnetization component perpendicular to the field or polarization axis , giving rise to neutron spin - flip process .
vice versa , the rather low spin - flip intensities , which are observed during the first magnetization reversal in the virgin state at @xmath22 are indicative of pure 180@xmath63 domain wall movement .
the step like intensity change at @xmath22 = -750 oe is followed by a steady decrease as the system approaches saturation ( see fig .
[ nhl1]e ) .
the slightly enhanced specular spin - flip scattering takes place in a very narrow field range of not more than 15 oe [ fwhm=9 oe ] . from these features it is obvious that domain wall nucleation and propagation let the magnetic spins at the interface canted away from the applied field direction .
next we discuss the off - specular spin flip scattering close to @xmath22 , which is reproduced in fig .
the off - specular diffuse intensities taken at @xmath22 = -750 oe , at @xmath64= -1400 oe , and upon return to @xmath64= -230 oe are rather strong .
their asymmetric shape is attributed to the specific wave - vector where they have been measured@xcite .
note that the off - specular intensity appears only after the magnetization reversal at @xmath22 has taken place for the first time and is negligible before .
the solid line in fig .
[ nhl1]d is the sf scattering taking in a descending field at -530 oe ( before first magnetization reversal ) , representing the instrumental resolution .
after reversal the diffuse scattering is large , then decreases towards saturation in negative fields and increases again on the way back towards h@xmath33 .
the diffuse spin flip scattering reveals the existence of magnetic domains walls@xcite .
we argue that those domain walls are located at the interface between the ferromagnetic and antiferromagnetic layer and that they have to be distinguished from ferromagnetic domains in the co layer .
if the diffuse scattering were originating from domain walls inside of the co - layer , then the off - specular scattering should vanish in saturation , which is clearly not the case here .
the creation of interfacial domains is crucial for the change of the reversal character from domain wall motion at @xmath22 to domain rotation at @xmath23 .
the breakdown of the ferromagnetic film in domains along the trained hysteresis has also been observed by velthuis et al .
here we show that the breakdown occurs already during the very first reversal and does not heal anymore even in saturation .
opposite to velthuis et al .
@xcite we argue that the observed domain formation is the reason for the trained hysteresis curve , instead of training being the cause for a partial breakdown of the alignment of the ferromagnetic domains .
( a ) moke hysteresis loop ( full squares ) and neutron hysteresis loop ( open squares ) of a coo / co bilayer after field cooling to 50k ( moke ) and to 10 k ( neutrons ) in an external field of + 2000 oe .
( b ) and ( c ) the specular non spin - flip intensities @xmath65 and @xmath66 and the specular spin - flip intensities @xmath67 and @xmath68 are plotted as a function of external magnetic field .
the intensities are measured at special scattering vector values of the reflectivity curves ( see text ) ; ( d ) off - specular diffuse spin - flip scattering taken before magnetization reversal at -530 oe ( line ) , at @xmath22=-750 oe ( full squares ) , in saturation at -1400 oe ( down full triangles ) and before the second magnetization reversal at -230 oe ( up full triangles ) ; e ) a zoom in from panel d ) showing the @xmath67 intensity at the first magnetization reversal . ]
the pnr results lead us to the following interpretation of the magnetization reversal in an eb - field . at @xmath22
the magnetization reversal takes place via nucleation and domain wall movement and leaves the interface in a ( partial ) magnetic domain state .
the spins in the interfacial domains are strongly coupled to the antiferromagnet and even in saturation they are not aligned along the external field direction .
those interface spins give rise to the strong off - specular diffuse spin - flip scattering . upon returning from saturation in negative fields we notice an increase of the diffuse spin flip scattering , which suggests that a torque is acting on the ferromagnetic spins trying to reverse them back to the direction of the cooling field long before the magnetization reversal actually takes place .
it seems that those interfacial magnetic domains are the seeds of the second magnetization reversal at @xmath23 , which proceeds via domain rotation .
the domain rotation at @xmath23 is characterized by a large specular spin - flip scattering ( panel c of fig .
[ nhl1 ] ) and off - specular diffuse scattering ( not shown ) .
therefore the second and all other reversals are dominated by domain rotation processes .
the behavior of the interface layer as revealed by pnr is in good qualitative agreement with the observation of uncompensated spins at the coo / co interface by gruyters and riegel @xcite and with the model put forward by mauri and co - workers @xcite .
the latter authors suggest the formation of a domain wall parallel to the interface during reversal of the fm orientation , which can effectively lower the interfacial energy cost of reversing the fm layer without removing the condition of strong interfacial af / f coupling . the interface may be considered as a diluted antiferromagnet .
this should be distinguished from the work of miltenyi et al.@xcite , who purposely keep the defects away from the interface .
it would be desirable to image the interfacial domains by real space methods such as done recently via polarized electron emission microscopy @xcite .
this methods can , however , only be applied in remanence .
ferromagnetic domains as they develop at the interface and in high fields can presently only be followed via polarized neutron reflectivity .
in summary , using a wedge type sample with a co thickness gradient we have determined the coercive fields and the exchange bias field after field cooling for different co layer thicknesses .
the exchange bias field at low temperature follows the @xmath69 behavior .
in addition , the temperature dependence of the exchange bias field exhibits a small positive value just below the blocking temperature , which is most likely due to a superexchange mechanism with antiferromagnetic exchange coupling at the af / f interface . with polarized neutron reflectivity measurements
we have shown that the first magnetization reversal at @xmath22 is dominated by nucleation and domain wall motion , while the second magnetization reversal at @xmath23 can be described by magnetization rotation .
all subsequent reversals are similar to the second , yielding a symmetric hysteresis , which , however , remains shifted .
furthermore diffuse spin - flip scattering reveals the formation of magnetic domains at the interface , induced by the first magnetization reversal .
we have shown that these magnetic domains are preserved even in saturation and that they serve as seeds for the second magnetization reversal .
overall , our data reveal clear differences between the coercive fields of the very first reversal @xmath22 and the second reversal at @xmath23
. the mechanisms of the magnetization reversal ( domain wall motion versus domain rotation ) and their temperature dependencies are clearly distinct .
for the sake of clarity we shall demonstrate the formulae [ sa2 ] and [ sf1 ] for a simple case .
let us consider a semi - infinite magnetic media for which we shall obtain the analytical formulae of reflectivities .
we start from the general formalism presented in ref .
@xcite . using the expression @xmath70 and
the well known properties of the pauli operator @xmath71 one can write the expression of the wave vector operator as : @xmath72 where @xmath73 are the eigenvalues of the operator @xmath74 and @xmath75 is a unit vector which defines the direction of the magnetic induction in the @xmath76 media , @xmath77 is the nuclear scattering length density of the i - the layer , where @xmath78 is the nuclear density and @xmath79 is the coherent scattering length.the matrix representation of @xmath80 reads : @xmath81 where @xmath38 and @xmath19 are the polar and the axial angle in the spherical coordinates with the polar axis being the quantization axis . choosing the quantization axis to be parallel to the polarization axis , the matrix representation of the reflection operator @xmath82 becomes : @xmath83 with : @xmath84 it is easy to check that the expressions above are the same as ones obtained by pleshanov @xcite . in order to understand how the reflectivities depend on the magnetization reversal in an ferromagnetic media
let us consider the case when @xmath88 .
choosing just a single value for the incident wave vector and sweeping the applied magnetic field is a key ingredient for measuring a complete hysteresis loop via pnr .
we neglect the neutron absorption into the sample and consider the case when the applied magnetic field is much smaller then the internal fields into the ferromagnetic media . with these assumptions and for @xmath89 the reflectivities become:@xmath90 @xmath91 @xmath92 where @xmath93 is the ratio of the magnetic to the nuclear potential .
the conclusion from the formulae above are that at the critical scattering vector for non - polarized neutron , the reflectivities : + - depend on the absolute value of the magnetization induction ; + - depend on the relative orientation between the polarization axis and the direction of the magnetic induction into the media ; + -do not depend on the polar angle @xmath19 .
+ we consider below two cases : one when the magnetization reversal proceeds by rotation and another case where the magnetization changes through domain wall movement .
let us consider that the magnetic moments will rotate during the magnetization reversal in such a way that @xmath94 with the components @xmath95 and @xmath96 . the easiest way to imagine such a rotation is to rotate the sample instead of rotating the magnetic moment , keeping the remanent magnetization equal to the saturation magnetization ( @xmath97 )
. with this assumption the spin asymmetry @xmath98 becomes:@xmath99 @xmath100 the spin flip reflectivity is:@xmath101 @xmath102 measurements of @xmath103 and @xmath104 at any @xmath54 value as a function of the applied magnetic field provide the same hysteresis loop as obtained by moke or squid measurements .
moreover , fitting the non - spin flip reflectivity curves taken in saturation ( r@xmath105 or r@xmath106 or both ) one can accurately evaluate the absolute magnetic moment . in comparison , for moke one would have to make a calibration for the kerr angles and for squid one needs to determine the volume of the magnetic layer .
now we discuss a magnetization reversal by domain wall movement . for simplicity
we consider a situation where a magnetic film consists of ferromagnetic domains with the following configuration : a fraction of @xmath107 domains have their spins oriented parallel to the neutron polarization and a fraction of @xmath108 domains have the spin oriented antiparallel .
the measured reflectivities would not be simply @xmath103 or @xmath104 but given by @xmath109 and @xmath110 .
the measured spin asymmetry becomes : @xmath111 where @xmath112 with @xmath113 being the magnetic induction into the magnetic domains .
thus , the shape of the hysteresis loop is completely defined .
the spin flip scattering will result only from domain walls . in a more general case
when for instance the magnetic domains are rotated with respect to the neutron polarization axis , there will also be spin flip scattering .
for this case the sa expression will still be valid and , in addition , from the spin - flip scattering their angle relative to the neutron polarization axis can be measured .
we mention that the formulae above are only correct in the absence of off - specular scattering .
the values of sa and the spin - flip reflectivity will be affected by the loss of the specular signal .
this can be seen when comparing squid or moke based magnetic hysteresis loops to the one extracted from the sa .
regions where the loops do not overlap indicate the presence of magnetic domains . this can be clearly seen in fig . [ nhl1](a ) .
a different approach for evaluating the average magnetization vectors inside of a film in a domain state was taken by lee et al .
@xcite . instead of measuring the magnetization reversal at one specific wave vector ,
complete reflectivity curves are recorded and the average magnetization as well as the mean square dispersion of the domains as function of @xmath54 is evaluated .
they use similar reflectivity formulae as shown above , but do not derive magnetization loops . for a quantitative comparison of hysteresis loops
the method presented here is faster and more effective .
we gratefully acknowledge support through the sonderforschungsbereiche 491 `` magnetische heteroschichten : struktur und elektronischer transport '' of the deutsche forschungsgemeinschaft .
the neutron scattering experiments were performed at the adam reflectometer of the ill , which is supported by the bmbf grant no . 03zae8bo .
special thanks are to erik a. verduijn for critical reading of the manuscript .
m. r. fitzsimmons , c .
leighton , a. hoffmann , p. c. yashar , j. nogues , k. liu , c. f. majkrzak , j. a. dura , h. fritzsche , and i. k. schuller , phys . rev .
b * 64 * , 104415 ( 2001 ) . c. schlenker , phys .
status solidi , * 28 * , 507 1968 .
f. radu , m. etzkorn , t. schmitte , v. leiner , a. schreyer , k. westerholt , h. zabel , annual report 2000 , page 70 ( http://www.ep4.ruhr-uni-bochum.de/fk/publikationen/03_contributions-e.pdf ) .
the _ domain rotation _ applies to the magnetization reversal process which shows both magnetic spin rotation revealed by neutron spin - flip scattering and presence of magnetic domains seen from spin - flip off - specular scattering .
|
we have carried out detailed experimental studies of the exchange bias effect of a series of coo / co(111 ) textured bilayers with different co layer thickness , using the magneto - optical kerr effect , squid magnetometry , polarized neutron reflectivity , x - ray diffraction , and atomic force microscopy .
all samples exhibit a pronounced asymmetry of the magnetic hysteresis at the first magnetization reversal as compared to the second reversal .
polarized neutron reflectivity measurements show that the first reversal occurs via nucleation and domain wall motion , while the second reversal is characterized by magnetization rotation .
off - specular diffuse spin - flip scattering indicates the existence of interfacial magnetic domains .
all samples feature a small positive exchange bias just below the blocking temperature , followed by a dominating negative exchange bias field with decreasing temperature .
|
the discovery of high temperature superconductivity in fe - based materials attaining t@xmath6 as large as 109k @xcite , has lead to a huge up surge of research for further discovery of such new materials @xcite .
seven years after its discovery , while a clear consensus on the mechanism of superconductivity has not yet been reached , understanding on the structural , magnetic transitions and their mutual influences on superconductivity remain central issue of frontier research @xcite .
a large number of undoped fe - based materials show spin density wave ( sdw ) magnetic state whose transition temperature coincides with that of the structural transition ( which gets separated through doping as well as pressure ) .
both the transitions being second order in nature , can have a conflict with landau theory of phase transition unless there would be a precursor transition at higher temperatures . according to landau theory , occurrence of two simultaneous transitions
may be purely coincidental , mutually independent , or one of the transitions be first order type or there must be a precursor to one of the transitions at a higher temperature .
what is that precursor ?
origin of the structural transition is not purely lattice driven but an electronic one ; the orbital ordering of fe d@xmath0 and d@xmath1 orbitals @xcite is the primary reason for structural transition . among some of the normal state properties ,
transport in `` preferred '' direction has been observed unambiguously in many experiments
inelastic neutron scattering ( ins ) @xcite , scanning tunnelling microscope , impurity @xcite , resistivity @xcite , optical conductivity @xcite , angle resolved photo electron spectroscopy @xcite and so on .
overall , origin of such phenomena is related to the breaking of four - fold rotational ( c@xmath7 ) symmetry of the tetragonal phase known as nematicity the precursor . whether the origin of nematic phase is spin driven or orbital driven
is far from being settled .
the nematic phase is observed in fese materials , that has structural transition at 90 k but no trace of long - range magnetic order @xcite indicating nematicity is orbital fluctuation driven @xcite .
however , observation of an additional c@xmath7 phase deep inside the orthorhombic ( c@xmath8 ) phase in ba@xmath9na@xmath10fe@xmath8as@xmath8 close to the suppression of magnetic spin density wave ( sdw ) order favours magnetically driven nematic order @xcite .
role of nematic phase as regards to the mechanism of superconductivity or symmetry of cooper pair wave function is not straight forward @xcite but the fact that spin fluctuation leads to s@xmath11 superconductivity @xcite whereas the orbital fluctuation leads to s@xmath12 superconductivity in fe - pnictides @xcite are established and magnetism competes with superconductivity @xcite .
there are also evidences of nematic order in the pseudogap phase of the other class of high temperature cuprates superconductors which are known to be strongly correlated materials @xcite . on the other hand , 122 family of fe - based superconductors
are generally considered as weakly correlated systems .
therefore , the study of nematicity in fe - based superconductors is of fundamental importance .
possible origins of nematic phase are well described in @xcite as ( a ) structural distortion , ( b ) charge / orbital order , ( c ) spin order . whatever be the nematic order parameter it must couple linearly to the orthorhombic distortion @xcite .
nematicity introduces electronic anisotropy leading to two different nesting vectors which in turn leads to two competing spin density wave ( sdw ) instabilities ( z@xmath13 o(3 ) symmetry breaking ) @xcite .
coupling of the orbital order to sdw and vice versa has been used as inputs in ginzburg - landau formalism which provides qualitative understanding of nematic phase . a clear first principles understanding on whether there is any direct coupling between the magnetic ( sdw ) and orbital order in fe - pnictides is absent till to date
is the main aim of this work .
we show through electronic structure calculation that the electronic orbital ordering locally breaks the in - plane symmetry and generate two non - equivalent @xmath2 , @xmath3 directions .
in particular , we show that below structural transition ( orthorhombic phase ) there is a strong orbital anisotropy along the @xmath14 and @xmath15 polarizations ; the band at x is dominantly fe - d@xmath0 derived whereas that at y , fe - d@xmath1 derived respectively .
this feature reproduces correctly experimental angle resolved photo electron spectroscopy ( arpes ) observation @xcite .
this is the root cause of orbital ordering in bafe@xmath16ru@xmath10as@xmath8 we show that the temperature dependence of the orbital ordering at x(y ) point reproduces exactly that of orthorhombicity parameter ( hence structural transition ) . thus structural transition is primarily electronic in origin and phonons can not be a primary order parameter for nematicity . whereas the temperature dependence of the same at @xmath17 point is very weak ( nearly independent of temperature )
as observed experimentally @xcite .
interestingly , zhang et al . , @xcite observed that the orbital splitting ( ordering ) at @xmath17 point ( in case of fese ) persists beyond structural transition temperature and argued that as against ferro - orbital ordering . in order to have a complementary first principles understanding over the experimental and other studies @xcite we introduce magnetic interaction through tuning integrated spin density defined as , i@xmath18=@xmath19(n@xmath20 ( r ) -n@xmath21 ( r))d@xmath22r . in presence of finite integrated spin density i@xmath18 , spin selective orbital ordering
are observed .
due to electronic orbital anisotropy ( see fig . [ bs ] ) the sdw state may be viewed as a superposition of two sdw states .
this is because of two reasons , ( i ) nesting wave vector that connects nested parts of fermi arcs along the @xmath23 and @xmath24 directions are different ; ( ii ) overlap of fe - d@xmath25 with d@xmath1 and d@xmath0 is different ( below structural transition ) causing anisotropic charge and spin density fluctuations . tuning i@xmath18 causes further perturbation to the underlying sdw as well as splits the spin degeneracy of energy bands .
this magnetic interaction couple with orbital ( charge ) fluctuations causing further orbital anisotropy .
remarkably , this later effect is observable only in the orthorhombic phase and _ not _ in the tetragonal phase
. this would be experimentally verifiable by arpes in presence of weak zeeman field . in presence of i@xmath18
we evaluate thermal variations of orbital ordering at different high symmetry points .
we show that the magnetic interaction couples to the zone centre orbital ordering very strongly where as it has a substantial effect on the corner zone orbital fluctuation .
these observations support the claim by zhang @xcite , fernandes @xcite of magnetic origin of nematic phase .
finally , we show that the orbital occupancies of all the five d - orbitals of fe show temperature dependencies below structural transition .
this indicates to the fact that nematic order parameter may not simply be defined as ( n@xmath4-n@xmath5 ) but charge fluctuations from other orbitals also need to be considered , for example , the thermal variation of ( n@xmath4-n@xmath5)+(n@xmath26-n@xmath27 ) also follows orthorhombicity ( see inset fig .
[ oosc]c ) .
this will put constraints on many theoretical and experimental works so far .
first principles density functional theories can produce exact solutions of the many electron schrdinger equation if exact electronic density is being used as input .
various modern x - ray diffraction techniques _
e.g_. , synchrotrons radiation source etc . that determines crystallographic information at different external perturbations are essentially result of diffraction from various atomic charge densities ( bragg s diffraction ) .
considering experimentally determined structural parameters at different temperatures as input thus in turn provides temperature dependent densities in our first principles calculation .
these input structural parameters are kept fixed through out the calculation for a fixed temperature .
this is how we use a t=0 dft formalism to bring out temperature dependent observable with the help of experimental input ( energy being functional of electron density e @xmath28 e[@xmath29\equiv e[\rho \{a(t ) , b(t),c(t)\}$ ] ] .
the main effect on the electronic structure from finite temperature is the underlying crystal structure , and the average crystal structure at finite temperature can usually be reliably determined from the diffraction experiment at a given temperature .
this method is somewhat superior to other similar methodology , like molecular dynamics simulation ( md ) . through md simulation
one finds temperature dependent lattice parameters and then use standard t=0 dft method using gga exchange potential ( see for example , @xcite ) .
however , experimentally determined temperature dependent lattice parameters can be obtained with an accuracy better than 0.001 a which may not be possible in md .
using temperature and doping dependent experimental lattice parameters @xmath2(t,@xmath30 ) , @xmath3(t,@xmath30 ) , @xmath31(t,@xmath30 ) and @xmath32(t,@xmath30 ) @xcite , we obtain electronic structure as a function of temperature as well as doping , to explain the experimentally observed anomalies microscopically .
one of the shortcomings of the density functional theory ( dft ) under generalized gradient approximation ( gga ) for calculating electronic structures of fe - based scs is that it fails to reproduce accurate experimental z@xmath33 @xcite .
this conflict arises due to strong magnetic fluctuation associated with fe based compounds @xcite .
this insist us to take experimental z@xmath33 instead of relaxed z@xmath33 obtained by total energy minimization , as one of our input parameters .
we simulate electronic structures for both the phases , low temperature orthorhombic phase with anti - ferromagnetic ( afm ) as well as spin density wave ( sdw ) ordering and high temperature paramagnetic tetragonal phase .
in low temperature orthorhombic phase , along with non magnetic structures various spin configurations have been employed among which the lowest energy configuration is considered for electronic structure calculation .
our first principles electronic structure calculations are carried out implementing plane - wave pseudopotential method within the framework of density functional theory @xcite . in all of our temperature and doping dependent calculations
the electronic exchange correlation energy is treated under the generalized gradient approximation ( gga ) using perdew - burke - enzerhof ( pbe ) functional @xcite . tackling small fraction of ru substitution in place of fe
is accomplished by considering both , the virtual crystal approximation ( vca ) as well as super - cell method for convenience .
super - cell method is a computationally expensive method adopted to mimic finite percentage of doping at a particular site .
let s say e.g , for 5% doping at the fe site one needs to build a super - cell ( bigger unit cell ) that contains 20 fe atoms ; then 1 of the fe atoms are replaced by ru atom ( say ) . in the present case
however , a super - cell containing 16 fe atoms ( total 40 atoms ) are taken out of which one is replaced by a ru ( shown in fig.[sc ] ) .
this corresponds to 6% ru doping which is close to the experimental situation .
note the size of the unit cell in the given symmetry is such that it does not allow exactly a super - cell with 20 fe atoms .
as@xmath8 , which contains 16 fe atoms and one ru atom corresponding to about 6@xmath34 doping .
different colours are used to indicate different atoms.,width=302,height=330 ] spin polarized single point energy calculations are performed using afm and sdw configuration @xcite ( see inset fig.[oosc ] ) for the low temperature orthorhombic phase with space group symmetry fmmm ( no.69 ) using ultrasoft pseudopotentials .
plane wave basis set with energy cut off 500 ev and self consistent field ( scf ) tolerance @xmath35 ev / atom has been opted for all calculations .
brillouin zone is sampled in the k space within monkhorst - pack scheme and grid size for scf calculation is @xmath36 for electronic density of state calculation in primitive cell for orthorhombic phase .
band structure calculations are performed along various k - paths ( x , @xmath17 and y ) with k point separation 10@xmath37 .
standard rotationally invariant approach due to matteo cococcioni and stefano de gironcoli @xcite and v. i. anisimov @xcite is used to treat the hubbard on site repulsion effect by post - dft lsda+u method which calculates and uses the total spin - projected occupation of the localized manifold , as this is essential to treat the hubbard term .
therefore , the method of calculation of occupation probabilities of u effected orbitals remain same as that of the reference given above which is also valid even in the limit u tending to zero .
first , we have calculated band structures of bafe@xmath8as@xmath8 system for anti - ferromagnetic ( afm ) spin configuration ( total spin zero ) using experimental lattice parameters at 20k as well as 300k along some specified k - path to probe orbital anisotropy .
our calculated band structures of bafe@xmath8as@xmath8 along the k path @xmath38 at two different temperatures corresponding to orthorhombic and tetragonal ( 20k and 300k ) phases respectively are shown in fig .
circular envelopes are drawn around x , y as@xmath8 along @xmath38 direction at 20k ( red ) in orthorhombic phase and 300k ( magenta ) in tetragonal phase .
orbital anisotropy along @xmath14 and @xmath15 direction in the orthorhombic phase is worth noticing.,width=302,height=330 ] as@xmath8 along @xmath38 direction at 20k ( upper ) and 300k ( lower ) indicating various d orbital ( d@xmath0 , d@xmath1 and d@xmath25 ) using different colours .
orbital ordering in the orthorhombic phase is shown in the inset figure .
electronic orbital anisotropy at the x and y points in the orthorhombic phase is the root cause of orbital order leading to structural transition.,width=302,height=255 ] points which are then shown in fig.[bs ] where splitting of @xmath39/@xmath40 at @xmath17 point has also been highlighted .
it is very clear from fig.[bs](a ) that at 20k ( orthorhombic phase ) , band dispersion along @xmath14 direction is quite different compared to that in the @xmath15 direction . in fig .
[ bs ] the splitting of @xmath39/@xmath40 at @xmath17 point has also been highlighted in the inset .
the same for the room temperature is then compared with .
fig.[bs ] demonstrate that the orbital ordering locally breaks the in plane symmetry and generates two non - equivalent @xmath41 directions @xmath42 to @xmath31 .
this results in two different nesting wave vectors along q@xmath10 = ( @xmath43 and q@xmath44 = ( @xmath45 ) directions - that is spins are parallel to each other along x - direction and anti - parallel along y - direction ( o@xmath46 ) .
we would also like to mention that at lower temperatures there exists two fe - fe distances ( z@xmath8 ) @xcite and this makes the system anisotropic both magnetically as well as in terms of band motion .
this situation resembles to that of the nematic phase where the bilinear combination of the order parameter ( o@xmath47z@xmath8 ) breaks the tetragonal symmetry and is invariant under symmetry transformation .
because of the anisotropy of the bands along x and y directions , in general , ( overlap of the d@xmath25 band with d@xmath1 and d@xmath0 bands are specially different ) causes inter - band charge and spin fluctuations , which may cause for example , coupling between them resulting in different amplitudes of the sdw along q@xmath10 and q@xmath44 directions .
energy orderings of the non - degenerate @xmath39/@xmath40 bands sets in orbital ordering @xcite , ( red ) points as a function of temperature for 5@xmath34 ru doped bafe@xmath8as@xmath8 .
brillouin zone of orthorhombic bafe@xmath8as@xmath8 has been shown in the inset of the figure indicating various k points ( x , y and @xmath17 ) .
the temperature dependence of orbital order is same as that of orthorhombicity ( @xmath48 ) @xcite.,width=245,height=188 ] temperature dependence of which defines structural transition .
this is depicted in fig.[oo ] .
it should be clearly noted that the structural distortion is predominantly determined by the orbital ordering at x(y ) point ; it is very weakly influenced by the orbital ordering at @xmath17 point which has very weak temperature dependence .
this clearly shows that the orbital ordering is very anisotropic .
now , question is why is that the orbital ordering at the ( zone centre ) @xmath17 point is so weak but finite and independent of temperature ?
this is also observed experimentally by zhang et al . ,
@xcite and modelled as bond - order .
we argue below this as manifestations of orbital anisotropy in presence of zone folding due to magnetic order .
it is easy to envisage from the band structure in fig.[bs ] that because of interband nesting e@xmath49 , but e@xmath50 and e@xmath51 , but e@xmath52 .
these would make the nematic order parameter ( n@xmath4-n@xmath5 ) null if the nesting wave vectors q@xmath10 , q@xmath44 were equivalent , but as it is not , it results in some small but finite quantity which is nearly independent of temperature .
this feature is indicative of the fact that the orbital ordering gap would form a density wave and this along with the sdw state is interband in nature @xcite .
-n@xmath5 ) and ( n@xmath4-n@xmath5)+(n@xmath26-n@xmath27 ) ( inset ) as a function of temperature for afm ( blue ) and sdw ( red ) spin structures .
( d ) thermal variation of ( n@xmath4-n@xmath5 ) for sdw spin configuration considering correlation ( u=1 ) using gga+u formalism and with out correlation.,width=302,height=255 ] by now through above discussions it is nearly evident as to what is the source of temperature dependence of the orbital order parameter and may relate to the same of the nematic order parameter .
the fe band energies presented in figs .
[ bs ] may be written as , e(k)=@xmath53 and the corresponding eigenstates involving orbitals are @xmath54 .
@xmath55 and n@xmath56(k)s are the band energies and occupation probabilities of ith orbital @xmath57 respectively .
the @xmath58(r)s being the kohn - sham orbit energies and the corresponding fourier transformed @xmath58(k)s are independent of temperature ( and magnetic interaction introduced later through i@xmath18 ) , whereas the orbital occupancies ( or densities ) n@xmath56(k ) are function of temperature .
therefore , lifting of degeneracy of the d@xmath1 , d@xmath0 bands at the @xmath17 and x points , as the temperature is lowered below structural transition temperature is a consequence of the fact that their occupation probabilities become different ( _ i.e _ , partial densities become unequal ) .
note , the energy gap between the d@xmath1 , d@xmath0 bands at the @xmath17 and x points is a function of temperature ( see fig .
the temperature difference essentially originates from the temperature dependencies of n@xmath4 , n@xmath5 and is proportional to n@xmath4-n@xmath5 ( see fig .
[ oosc]c ) .
therefore , it is desirable to calculate the temperature dependencies of the occupation probabilities of all the five d - orbitals of fe from first principles calculation .
such a rare calculation is presented in fig .
this quantity ( n@xmath4-n@xmath5 ) represents inter orbital charge fluctuation or orbital fluctuation in short , one of the important contender for nematic phase and is also called nematic order parameter . using super - cell of orthorhombic bafe@xmath8as@xmath8 structure corresponding to @xmath59 6 @xmath34 ru doping ( cf .
fig.1 ) and two types of spin arrangements afm and sdw ( shown in the inset of fig .
[ oosc ] ) first principles simulations of orbital occupancies are presented .
^2+i^{ru}\times[n^{ru}(e_f)]^2 $ ] ) as a function of temperature for 5@xmath34 ru doped bafe@xmath8as@xmath8 systems and ( b ) thermal variation of stoner factor for different u values as indicated in the figure.,width=302,height=143 ] why dope bafe@xmath8as@xmath8 with ru ?
like hole doped 122 systems iso - electronic ru doped 122 system also have inseparably same structural and magnetic transitions and the nematic phase in this system remain unexplored .
this is unlike other iso - electronic p doping in place of as .
it is particularly an interesting case , it is not clear as to where does the charge carrier go in case of ru doping in place of fe . both the hole and electron fermi pockets either remain unaltered or expands at an equal rate @xcite ) .
furthermore , both the structural and magnetic transitions are 2nd order in nature in case of underdoped ru-122 system .
therefore , study of temperature dependence of orbital fluctuation from first principles is of genuine interest . in fig.[oosc ] orbital occupancies of d@xmath1 orbital ( n@xmath4 ) modifies significantly with temperature compared to the other d orbitals specially d@xmath0 and d@xmath25 ( but they also do show substantial temperature dependence ) .
we have also calculated the difference in the occupancies of d@xmath1 and d@xmath0 orbitals _ i.e _ , n@xmath4-n@xmath5 ( nematicity ) as a function of temperature for both afm and sdw spin configuration ( see fig . [
since , above structural transition there is no splitting between the d@xmath1 and d@xmath0 bands ( _ i.e , _ @xmath60 temperature dependence of the nematic order parameter n@xmath4-n@xmath5 is proportional to that of the orbital order . in other words ,
the nematic order essentially grow as orbital order which is responsible for orthorhombic transition .
therefore , this result may be interpreted as the first principles evidence of the fact that if orbital fluctuation is the primary order responsible for nematicity , then it is proportional to the orthorhombicity parameter @xcite .
this is one of the inputs of all the theories involving ginzburg landau formalism . as already mentioned towards the end of theoretical method section that the method of calculation of occupation probabilities of u effected orbitals remains same as that of the ref .
@xcite which is also valid even in the limit u tending to zero .
therefore , occupation probabilities are obtained for a very small u@xmath61ev ( not exactly equal to zero ) which makes up / down spin states different even in the tetragonal phase . as a result n@xmath4-n@xmath5 become very small but non - zero and has hardly any temperature dependence .
also , the difference in the nematic order parameter for sdw and afm clearly indicates that it is a spin nematicity .
furthermore , in the inset of fig .
[ oosc]c , we depict the thermal variation of ( n@xmath4-n@xmath5)+(n@xmath26-n@xmath27 ) which also reproduces thermal behaviour of orthorhombicity corresponding to orbital fluctuation involving all four d - orbitals .
thus , in contrast to the usual belief in literature , nematic order parameter which is defined as n@xmath4-n@xmath5 , perhaps involve all other @xmath62-orbitals as well .
ru doped bafe@xmath8as@xmath8 systems at 20 k with i@xmath18 = 1 for down spin ( left ) and up spin ( right ) electrons around x , y and @xmath17 points .
we assign red green and blue colours to d@xmath25 , d@xmath0 and d@xmath1 orbitals respectively.,width=302,height=332 ] the nematic order parameters show substantial modifications near the temperature where magnetic fluctuation is very strong ( see fig.[oofsm ] ) .
this provides a distinct evidence to the fact that probably magnetic fluctuation , orbital fluctuation , the nematicity are interdependent ( we discuss this issue below ) . also it should be noted that the orbital occupancies of d@xmath1 orbital is always greater than that of the d@xmath0 orbital _ i.e. , _
this propound ferro - orbital ordering .
partial densities n@xmath1 , n@xmath0 being unequal would correspond to different bonding along @xmath30 and @xmath64 directions , a mark of nematicity due to orbital anisotropy .
unlike cuprates it is experimentally well established that 122 family of fe - based scs are weakly correlated . in fig .
[ oosc](d ) temperature dependence of n@xmath4-n@xmath5 has been shown with small on site correlation ( u=1 ) and with out correlation .
it is clear that correlation reduces orbital order and so the magnetic fluctuation ( see fig . [
stoner]b ) .
magnetic fluctuations play an important role in these family of fe - based scs .
stoner factor is the measure of these magnetic fluctuation .
stoner factor of this compound can be defined as @xmath65 ^ 2+i^{ru}\times[n^{ru}(e_f)]^2 $ ] , where @xmath66 and @xmath67 are the density of states at the fermi level from fe and ru atoms respectively @xcite .
the value of stoner parameters @xmath68 and @xmath69 are taken from ref @xcite .
we have calculated the stoner factor as a function of temperature and displayed in fig.[stoner]a . as temperature changes ,
partial density of states of fe and ru at the fermi level ( e@xmath70 ) get modified due to substantial moderation of fe - as hybridization .
this is the root cause of temperature dependence of stoner factor .
this observation is very much consistent with recent experimental findings @xcite . in fig .
[ stoner]b thermal variation of stoner factor has been depicted with different values of u ( strength of electron correlation ) .
it is clear that with increasing electron repulsion ( system would prefer stable antiferromagnetic configuration ) magnetic fluctuation is actually decreasing ( so is the nematic order ) which enhances further our suspect that magnetic fluctuation triggers orbital fluctuation and nematic phase . + for this purpose , we tune magnetic interaction manually by introducing integrated spin density parameter as explained above and see its influence on electronic structure . while keeping initial afm spin structure we provide constraint through the integrated spin density parameter for spin polarized calculations which induces some magnetic moment in the system
this method is a simple extension of standard lsda formalism where total energy as a function of moment can be obtained .
after complete spin polarized calculation being performed , the main idea is to perform a single point energy calculation keeping the total moment constrained to some fixed small but finite non zero values .
this is equivalent to generating a weak zeeman field ( see also the form of the integrated spin density parameter ) . in this way by keeping the afm spin configuration , magnetic interaction can be introduced to the system by simply fixing the total ( difference in up and down spin ) magnetic moment of the system through integrated spin density parameter .
actually , from electronic structure calculation we calculate orbital ordering again in presence of magnetic interaction ( see below ) and see that orbital anisotropy is enhanced , and hence the nematicity .
( red ) points as a function of temperature for ( a ) i@xmath18 = 1 and ( b ) i@xmath18 = 2 for 5@xmath34 ru doped bafe@xmath8as@xmath8 systems .
hollow symbols represent data from fig .
[ oo].,width=321,height=151 ] in presence of finite integrated spin density , the band of up spin electrons and band of down spin electrons split . we observed that for i@xmath18 = 1 and 2 , one of the bands ( up spin ) goes deep below the fermi level .
so only one of the spin electronic bands are contributing significantly at the fermi level . in fig .
[ bsfsm ] we have shown the band structures of 5@xmath34 ru doped bafe@xmath8as@xmath8 systems at 20 k , i@xmath18 = 1 for down spin ( left ) and up spin ( right ) electrons around different k points ( x , y and @xmath17 points ) .
notably , around x , y and @xmath17 points all the orbitals are ordered differently in contrast to that in fig .
[ bs ] . around x as well as around @xmath17
point the energy ordering of d@xmath1 and d@xmath0 orbitals are exactly opposite to each other for up spin and down spin bands .
this leads to spin - polarized orbital orderings ( possibly orbital density wave ) .
we do all the same exercises ( in all the figures above ) for k doped as well as
p doped 122 systems and found that the orbital ordering is a common phenomena ( there are differences in details and will be reported elsewhere ) . in fig .
[ oofsm ] we have presented temperature variation of orbital order around x , y and @xmath17 points for i@xmath18 = 1 ( fig . [ oofsm]a ) and i@xmath18 = 2 respectively ( fig . [
oofsm]b ) after extracting the required information from fig .
[ bsfsm ] . in case of i@xmath18
= 1 the orbital order around @xmath17 increases to about 3 fold compared to the case where we optimized the total spin of the system ( represented by hollow symbols ) .
however , the orbital ordering around the zone corners x(y ) are less affected .
when the integrated spin density is increased to 1 from 0 , it stabilizes the underlying sdw , but when i@xmath18 is further increased to 2 , the underlying sdw ordering will be strongly suppressed due to ferromagnetic nature of the i@xmath18 . hence , orbital ordering is strongly coupled to the underlying magnetic fluctuation ( it enhances orbital fluctuation or nematic order parameter ) and our study thus is complementary to recent studies @xcite on nematic phase .
we establish the microscopic relationship between the orbital order , structural transition and nematic order in 122 family of fe - based superconductors .
while electronic orbital anisotropy gives rise to orbital order , temperature dependence of the orbital order is found to be exactly same as that of the orthorhombicity , indicating orbital ordering is responsible for structural transition .
temperature dependence of the orbital order is proportional to the temperature dependence of the nematic order ( n@xmath4-n@xmath5 ) .
this indicates that the nematic order grows as orbital order in the orthorhombic phase .
we have explicitly evaluated the temperature dependencies of orbital occupancies of all fe - d - orbitals .
almost all the d - orbitals show substantial charge fluctuations in the orthrhombic phase , indicating that the actual definition of nematic order parameter may be more complicated . the nematic order parameter is found to show temperature dependence close to the onset of magnetic fluctuation , obtained rigorously through evaluation of stoner factor . when magnetic fluctuations are enhanced , the orbital fluctuations are also enhanced and vice versa establishing their couplings .
spin - polarised orbital ordering revealed from this work would be experimentally observable .
finally , our work supports magnetic origin of nematicity in 122 family of fe - based superconductors .
we believe this work would generate further interest in theoretical as well as experimental studies . + * acknowledgements * + we thank dr .
a. bharathi and dr .
a. k. sinha for discussion on experimental aspects .
we thank dr .
p. a. naik and dr .
p. d. gupta for their encouragement in this work .
one of us ( ss ) acknowledges the hbni , rrcat for financial support and encouragements .
+ * additional information * + competing financial interests : the authors declare no competing financial interests .
+ jian - feng ge , zhi - long liu , canhua liu , chun - lei gao , dong qian , qi - kun xue , ying liu , jin - feng jia , superconductivity above 100k in single - layer fese films on doped srtio@xmath46 , nat .
mater 14 ( 2015 ) , 285 - 289 .
hideo hosono , keiichi tanabe , eiji takayama - muromachi , hiroshi kageyama , shoji yamanaka , hiroshi kumakura , minoru nahara , hidenori hiramatsu , satoru fujitsu , exploration of new superconductors and functional materials , and fabrication of superconducting tapes and wires of iron pnictides , sci .
technol . adv .
mater 16 ( 2015 ) , 033503 - 033589 .
j. j. wua , jung - fu lin , x. c. wang , q. q. liu , j. l. zhu , y. m. xiao , p. chow , changqing jin , pressure - decoupled magnetic and structural transitions of the parent compound of iron - based 122 superconductors bafe@xmath8as@xmath8 , proc .
usa 110(43)(2013 ) , 17263 - 17266 .
ming yia , donghui lu , jiun - haw chua , james g. analytis , adam p. sorini , alexander f. kemper , brian moritz , sung - kwan mod , rob g. moore , makoto hashimoto , wei - sheng lee , zahid hussain , thomas p. devereaux , ian r. fisher , zhi - xun shen , symmetry - breaking orbital anisotropy observed for detwinned ba(fe@xmath9co@xmath10)@xmath8as@xmath8 above the spin density wave transition , proc . natl .
usa 108(17 ) ( 2011 ) , 6878 - 6883 .
smritijit sen , haranath ghosh , a. k. sinha , a. bharathi , origin of structural and magnetic transitions in bafe@xmath16ru@xmath10as@xmath8 materials , supercond .
( fast track communication ) 27 ( 2014 ) , 122003 - 12009 .
jun zhao , dao - xin yao , shiliang li , tao hong , y. chen , s. chang , w. ratcliff , ii , j. w. lynn , h. a. mook , g. f. chen , j. l. luo , n. l. wang , e. w. carlson , jiangping hu , pengcheng dai , low energy spin waves and magnetic interactions in srfe@xmath8as@xmath8 , phys .
( 2008 ) , 167203 - 167206 .
m. p. allan , t - m .
chuang , f. massee , yang xie , ni ni , s. l. budko , g. s. boebinger , q. wang , d. s. dessau , p. c. canfield , m. s. golden , j. c. davis , anisotropic impurity states , quasiparticle scattering and nematic transport in underdoped ca(fe@xmath9co@xmath10)@xmath8as@xmath8 , nature phys . 9
( 2013 ) , 220 - 224 .
jiun - haw chu , james g. analytis , kristiaan de greve , peter l. mcmahon , zahirul islam , yoshihisa yamamoto , ian r. fisher , in - plane resistivity anisotropy in an underdoped iron arsenide superconductor , science 329 ( 2010 ) , 824 - 826 .
a dusza , a lucarelli , a sanna , s massidda , j - h chu , i r fisher , l degiorgi , anisotropic in - plane optical conductivity in detwinned ba(fe@xmath9co@xmath10)@xmath8as@xmath8 , new j. phys .
14 ( 2012 ) , 023020 - 023039
. can - li song , yi - lin wang , peng cheng , ye - ping jiang , wei li , tong zhang , zhi li , ke he , lili wang , jin - feng jia , hsiang - hsuan hung , congjun wu , xucun ma , xi chen , qi - kun xue , direct observation of nodes and twofold symmetry in fese superconductor , science 332 ( 2011 ) , 1410 - 1413 .
s. avci , o. chmaissem , j.m . allred , s. rosenkranz , i. eremin , a.v .
chubukov , d.e .
bugaris , d.y .
chung , m.g .
kanatzidis , j .-
p castellan , j.a .
schlueter , h. claus , d.d .
khalyavin , p. manuel , a. daoud - aladine , r. osborn , magnetically driven suppression of nematic order in an iron - based superconductor , nat .
commun . 5 ( 2014 )
, 3845 - 3850 .
t. shimojima , f. sakaguchi , k. ishizaka , y. ishida , t. kiss , m. okawa , t. togashi , c .- t .
chen , s. watanabe , m. arita , k. shimada , h. namatame , m. taniguchi , k. ohgushi , s. kasahara , t. terashima , t. shibauchi , y. matsuda , a. chainani , s. shin , orbital - independent superconducting gaps in iron pnictides , science 332 ( 2011 ) , 564 - 567 .
y. suzuki , t .
shimojima , t .
sonobe , a .
nakamura , m .
sakano , h .
tsuji , j .
omachi , k .
yoshioka , m. kuwata - gonokami , t .
watashige , r. kobayashi , s. kasahara , t .
shibauchi , y. matsuda , y. yamakawa , h. kontani , and k. ishizak , momentum - dependent sign inversion of orbital order in superconducting fese , phys .
b 92,(2015 ) 205117 i. i. mazin , d. j. singh , m. d. johannes , m. h. du , unconventional superconductivity with a sign reversal in the order parameter of lafeaso@xmath71f@xmath10 , phys . rev . lett . 101 ( 2008 ) , 057003 - 057006 .
r. s. dhaka , s. e. hahn , e. razzoli , rui jiang , m. shi , b. n. harmon , a. thaler , s. l. budko , p. c. canfield , adam kaminski , unusual temperature dependence of band dispersion in ba(fe@xmath9ru@xmath10)@xmath8as@xmath8 and its consequences for antiferromagnetic ordering , phys .
( 2013 ) , 067002 - 067006 .
s. sharma , a. bharathi , k. vinod , c. s. sundar , v. srihari , s. sen , h. ghosh , a. k. sinha and s. k. deb , structural investigations in bafe@xmath73ru@xmath10as@xmath8 as a function of ru and temperature , acta cryst .
b 71 ( 2015 ) , 61 - 67 . i. i. mazin , m. d. johannes , l. boeri , k. koepernik , d. j. singh , problems with reconciling density functional theory calculations with experiment in ferropnictides , phys .
b 78 ( 2008 ) , 085104 - 085110 .
t. shimojima , k. ishizaka , y. ishida , n. katayama , k. ohgushi , t. kiss , m. okawa , t. togashi , x .- y .
wang , c .- t .
chen , s. watanabe , r. kadota , t. oguchi , a. chainani , and s. shin , orbital - dependent modifications of electronic structure across the magnetostructural transition in bafe@xmath8as@xmath8 , phys .
104 ( 2010 ) , 057002 - 057005 .
o. k. andersen , j. madsen , u. k. poulsen , o. jepsen , j. kollr , physica c 86 - 88 ( 1977 ) , 249 - 256 ; e. c. stoner , collective electron ferromagnetism , proceedings of the royal society a : mathematical , physical and engineering sciences 165 ( 1938 ) , 372 - 414 .
yan , a. kreyssig , s. nandi , n. ni , s. l. budko , a. kracher , r. j. mcqueeney , r. w. mccallum , t. a. lograsso , a. i. goldman , and p. c. canfield , structural transition and anisotropic properties of single - crystalline srfe@xmath8as@xmath8 , phys .
b 78 ( 2008 ) , 024516 - 024519 .
huiqian luo , meng wang , chenglin zhang , xingye lu , louis - pierre regnault , rui zhang , shiliang li , jiangping hu , and pengcheng dai , spin excitation anisotropy as a probe of orbital ordering in the paramagnetic tetragonal phase of superconducting bafe@xmath74ni@xmath75as@xmath8 , phys .
( 2013 ) , 107006 - 107011 .
xingye lu , j. t. park , rui zhang , huiqian luo , andriy h. nevidomskyy , qimiao si , pengcheng dai , nematic spin correlations in the tetragonal state of uniaxial - strained bafe@xmath16ni@xmath10as@xmath8 , science 345 ( 2014 ) , 657 - 660 .
|
through detailed electronic structure simulations we show that the electronic orbital ordering ( between d@xmath0 and d@xmath1 bands ) takes place due to local breaking of in - plane symmetry that generates two non - equivalent @xmath2 , @xmath3 directions in 122 family of fe - based superconductors .
orbital ordering is strongly anisotropic and the temperature dependence of the corner zone orbital order maps to that of the orthorhombicity parameter .
orbital anisotropy results in two distinct spin density wave nesting wave vectors and causes inter - orbital charge and spin fluctuations .
temperature dependence of the orbital order is proportional to the nematic order and it sets in at a temperature where magnetic fluctuation starts building .
magnetic fluctuations in the orthorhombic phase is characterized through evolution of stoner factor which reproduces experimental findings very accurately .
orbital ordering becomes strongly spin dependent in presence of magnetic interaction .
occupation probabilities of all the fe - d - orbitals exhibit temperature dependence indicating their possible contribution in orbital fluctuation .
this need to be contrasted with the usual definition of nematic order parameter ( n@xmath4-n@xmath5 ) .
relationship among orbital fluctuations , magnetic fluctuations and nematicity are established .
fe - based superconductors , orbital fluctuation , nematicity , spin density wave
|
in the preprint @xcite , oliinychenko , bugaev and sorin have considered the role of conservation laws , the values of hard core radii along with the effects of the lorentz contraction of hadron eigen volumes in discussing the weaknesses of thermal models which are utilized in describing the hadron multiplicities measured in the central nucleus - nucleus collisions .
regardless the unawareness of earlier literature , the authors concluded that none of the criteria for the chemical freeze - out is robust . in doing this
, they entirely disregarded the experimental results in baryo - chemical potentials @xmath1 and their corresponding temperatures @xmath2 .
a systematic analysis of the four criteria describing the chemical freeze - out is introduced in @xcite .
furthermore , a comparison between these four criteria is elaborated in @xcite .
starting from phenomenological observations at sis energy , it was found that the averaged energy per averaged particle @xmath3gev @xcite , where boltzmann approximations are applied in calculating @xmath4 , this constant ratio is assumed to describe the whole @xmath5 diagram . for completeness , we mention that the authors assumed that the pions and rho - mesons get dominant , at high @xmath2 and small @xmath1 .
the second criterion assumes that total baryon number density @xmath6@xmath7 @xcite . in framework of percolation theory , the authors of ref . @xcite
have suggested a third criterion . as shown in fig .
2 of @xcite , the last two criteria seem to give almost identical results .
all of them are stemming from phenomenological observation . a fourth criterion based on lattice qcd simulations
was introduced in ref .
accordingly , the entropy normalized to cubic temperature is assumed to remain constant over the whole range of baryo - chemical potentials , which is related to the nucleus - nucleus center - of - mass energies @xmath8 @xcite .
an extensive comparison between constant @xmath4 and constant @xmath9 is given in @xcite .
the thermodynamic quantities deriving the chemical freeze - out . in framework of hadron resonance gas
are deduced @xcite .
explicit expressions for @xmath10 at vanishing and finite temperature are introduced @xcite .
the motivation of suggesting constant normalized entropy is the comparison to the lattice qcd simulations with two and three flavors .
we simply found the @xmath11 for two flavors and @xmath12 for three flavors .
furthermore , we confront the hadron resonance gas results to the experimental estimation for the freeze - out parameters , @xmath2 and @xmath1 .
the hadron resonances treated as a free gas @xcite are conjectured to add to the thermodynamic pressure in the hadronic phase ( below @xmath13 ) .
this statement is valid for free as well as strong interactions between the resonances themselves .
it has been shown that the thermodynamics of strongly interacting system can also be approximated to an ideal gas composed of hadron resonances with masses @xmath14gev @xcite .
such a mass cut - off is implemented to avoid the hagedorn singularity @xcite .
therefore , the confined phase of qcd , the hadronic phase , is modelled as a non - interacting gas of resonances .
the grand canonical partition function reads z(t , v ) & = & , where @xmath15 is the hamiltonian of the system and @xmath2 is the temperature .
the hamiltonian is given by the sum of the kinetic energies of relativistic fermi and bose particles .
the main motivation of using this hamiltonian is that it contains all relevant degrees of freedom of confined and strongly interacting matter .
obviously , it can be characterized by various - but a complete - set of microscopic states and therefore the physical properties of the quantum systems turn to be accessible in approximation of non - correlated _ free _ hadron resonances .
each of them is conjectured to add to the overall thermodynamic pressure of the _ strongly _ interacting hadronic matter .
it includes implicitly the interactions that result in resonance formation .
in addition , it has been shown that this model can submit a quite satisfactory description of particle production in heavy - ion collisions @xcite . with the above assumptions
the dynamics the partition function can be calculated exactly and be expressed as a sum over _ single - particle partition _
functions @xmath16 of all hadrons and their resonances .
[ eq : lnz1 ] z(t , _
i , v)&=&_i z^1_i(t , v)=_i_0^ k^2 dk ( 1 ) , where @xmath17 is the @xmath18th particle dispersion relation , @xmath19 is spin - isospin degeneracy factor and @xmath20 stands for bosons and fermions , respectively .
the switching between hadron and quark chemistry is given by the relations between the _ hadronic _ chemical potentials and the quark constituents ; @xmath21 , where @xmath22(@xmath23 ) being baryon ( strange ) quantum number . the chemical potential assigned to the light
quarks is @xmath24 and the one assigned to strange quark reads @xmath25 .
the strangeness chemical potential @xmath26 is calculated as a function of @xmath2 and @xmath27 under the assumption that the overall strange quantum number has to remain conserved in heavy - ion collisions @xcite .
the hrg calculations assume quantum statistics and an overall strangeness conservation . with this regard
, the strangeness chemical potential @xmath26 is calculated at each value of @xmath2 and @xmath1 assuring that the number of strange particles should be the same as that of the anti - strange particles .
it is worthwhile to mention that no statistical fitting has been applied in determining all thermodynamic quantities , including entropy and number density derived from eq .
( [ eq : lnz1 ] ) .
from the entropy and equilibrium , the gibbs condition simply leads to [ eq : thrml ] & = & ( + -_b ) , the rhs is positive as long as @xmath28 , where the thermodynamic quantities , @xmath29 , @xmath30 and @xmath31 are supposed to be calculates at the @xmath5 diagram of the chemical freeze - out .
[ fig : tmu1 ] shows the experimental estimation for the freeze - out parameters @xmath2 and @xmath1 .
it is obvious that increasing @xmath1 leads to decreasing @xmath2 and therefore all values of the thermodynamic quantities decrease as well .
_ @xcite suggested an empirical @xmath5 relation [ eq : tmu ] t & = & a - b _ b^2 - c _
b^4 , where @xmath32 , @xmath33 and @xmath34 are fitting parameters . in light of this discussion ,
the value given to @xmath10 ca nt remain unchanged with increasing @xmath1 .
left panel of fig .
[ fig : sn1 ] presents the values of the three criteria @xmath4 , @xmath35 and @xmath10 calculated in hrg , section [ sec : hrg ] , at @xmath12 .
it is obvious that all four criteria seem to remain constant , especially at high @xmath8 . at low energies ,
the value assigned to @xmath10 @xcite is larger than the actual one , the value resulted from order conditions . the reason is illustrated in the right panel . at different values for @xmath1 , the thermal evolution of @xmath10 is presented .
it is obvious that @xmath10 never reaches @xmath0 at @xmath36mev .
it is essential to bear in mind that the value @xmath0 has almost no physical interpretation .
it is just an _ ad hoc _ value .
this makes it inapplicable at alternating gradient synchrotron ( ags ) and schwer - ionen - synchrotron ( sis ) energies .
almost same kind of restriction would be valid for @xmath37 . according to eq .
( [ eq : thrml ] ) , & = & t + _ b - . the physics of constant @xmath9 has been discussed in ref .
it combines the three thermodynamic quantities , @xmath38 , @xmath39 and @xmath40 & = & + - _ b . at chemical equilibrium ,
the particle production at freeze - out is conjuncted to fully fulfil the laws of thermodynamics , as eq .
( [ eq : thrml ] ) .
the hadronic abundances observed in the final state of heavy - ion collisions are settled when @xmath9 drops to @xmath41 i.e. , the degrees of freedom drop to @xmath42 .
meanwhile the changing in the particle number with the changing in the collision energy is given by @xmath1 , the energy that produces no additional work , i.e. the stage of vanishing free energy , gives the entropy at the chemical equilibrium . at the chemical freeze - out
, the equilibrium entropy represents the amount of energy that ca nt be used to produce additional work . in this context , the entropy is defined as the degree of sharing and spreading the energy inside the system that is in chemical equilibrium @xcite .
once again , related literature on lattice qcd simulations is not cited in @xcite . for example , borsanyi _ et al . _
@xcite studied the trajectories of constant @xmath10 , where @xmath43 and @xmath44 , on the phase diagram and thermodynamic observables along these isentropic lines .
this was not the only work devoted to such line of constant physics @xcite . in stefan - boltzmann limit
, the ratio @xmath10 is assumed to remain unchanged with increasing @xmath1 ( appendix a of @xcite ) . in doing this , lowest order in perturbation theory is assumed , where strangeness chemical potential @xmath26 likely vanishes . for @xmath45 , a limiting behavior for the isentropic lines on the phase diagram
is obtained .
the ratio @xmath10 has been measured at various @xmath8 @xcite .
it is concluded that in limits of low temperatures , increasing the chemical potential results in an overestimation for the ratio @xmath10 even beyond the applicability region of the taylor - expansion method , which is applied in lattice qcd simulations at finite chemical potential .
two remarks are now in order .
first , the values of @xmath10 seem to depend on the chemical potential @xmath1 or @xmath8 .
this is confirmed in different experiments @xcite and lattice gauge theory @xcite .
second , the ratio @xmath10 as calculated in the lattice qcd simulations @xcite is suggested to characterized the qcd phase diagram @xcite .
the qcd phase diagram is likely differs from the freeze - out diagram @xcite , especially at large chemical potential @xmath1 or small @xmath8 so that at fixed @xmath1 the critical temperature differs from the freeze - out temperature .
in fig . [ fig : tmu1 ] , the freeze - out parameters , @xmath2 and @xmath1 , measured in various heavy - ion collisions experiments are compared with the three criteria , @xmath46gev ( dashed line ) , @xmath47 ( dotted line ) and @xmath12 ( solid line ) .
the experimental data are taken from @xcite and the reference therein .
the quality of each criterion is apparent .
all conditions are almost equivalent at very high energy
. the ability of the condition @xmath46gev at very low energies are not as much as that of @xmath12 . as discussed in section [ sec : phys ] , @xmath47 seems to fail to reproduce the freeze - out parameters at @xmath36mev . to illustrate the reason for this observation , the thermal evolution of @xmath10 at very high chemical potential calculated in hrg is presented in the right panel of fig .
[ fig : sn1 ] .
details on hrg are elaborated in section [ sec : hrg ] .
it is obvious that the value assigned to @xmath10 would be achieved at @xmath36mev . in other words , it is obvios that the behavior of @xmath10 is non - monotonic .
the left panel of fig .
[ fig : sn1 ] presents the energy scan for the three criteria , @xmath4 , @xmath35 and @xmath10 calculated in hrg , section [ sec : hrg ] at @xmath12 .
the calculations in hrg are performed as follows . starting with a certain @xmath1
, the temperature is increased very slowly . at this value of @xmath1 and at each increase in @xmath2 ,
the strangeness chemical potential @xmath26 is determined to assure strangeness conservation .
having the three values of @xmath1 , @xmath2 and @xmath26 , then all thermodynamic quantities are calculated .
when the ratio @xmath9 reaches the value @xmath41 , then the three quantities @xmath4 , @xmath35 and @xmath10 are registered .
this procedure is repeated over all values of @xmath1 .
we find that @xmath12 assures @xmath47 and @xmath46gev at small @xmath1 ( large @xmath8 ) . at large @xmath1 ( small @xmath8 ) , the values of @xmath10 gets smaller values , so that the applicability of @xmath47 is limited to @xmath48mev . in conclusion ,
the robustness of @xmath47 is very much limited in comparison to the four criteria : percolation @xcite , baryon number @xcite , energy per particle @xcite and normalized entropy @xcite .
that @xmath9 is accompanied with constant @xmath10 has been introduced in ref . @xcite . that authors of @xcite argue that @xmath47 is novel likely reflects an ignorance of related literature .
the four criteria @xcite are based on physical observation either phenomenological and/or theoretical .
the authors of @xcite are suggesting an _ ad hoc _
value for the ratio @xmath10 .
it is inapplicable at ags and sis energies .
its relation to @xmath9 is apparently overseen . the same is valid for the comparison with other criteria ( some of them are ignored , completely ) and ignorance of the experimental measurements .
ad hoc _ value assigned to @xmath10 is obviously much robuster than any other criterion .
unawareness of literature and underestimating or even ignoring previous work are violation of rules of the scientific research .
ardill , m. creutz , k.j.m .
moriarty and s. samuel , lines of constant physics for su(3 ) lattice gauge theory in four - dimensions , bnl-35081 ( 1984 ) ; s. ejiri , f. karsch , e. laermann , and c. schmidt , phys .
d * 73 * , 054506 ( 2006 ) ; c. bernard , _ et al .
d * 77 * , 014503 ( 2008 ) ; c. detar , _ et al .
_ , phys . rev .
d * 81 * , 114504 ( 2010 ) .
|
oliinychenko , bugaev and sorin [ arxiv:1204.0103 [ hep - ph ] ] considered the role of conservation laws in discussing possible weaknesses of thermal models which are utilized in describing the hadron multiplicities measured in central nucleus - nucleus collisions . they argued to analyse the criteria for chemical freeze - out and to conclude that none of them were robust .
based on this , they suggested a new chemical freeze - out criterion .
they assigned to the entropy per hadron the _ ad hoc _ value @xmath0 and supposed to remain unchanged over the whole range of the baryo - chemical potentials . due to unawareness of recent literature
, the constant entropy per hadron has been discussed in ref .
[ fizika b18 ( 2009 ) 141 - 150 , europhys.lett .
75 ( 2006 ) 420 ] .
furthermore , it has been shown that the constant entropy per hadron is equivalent to constant entropy normalized to cubic temperature , an earlier criterion for the chemical freeze - out introduced in ref .
[ europhys.lett .
75 ( 2006 ) 420 , nucl.phys.a764 ( 2006 ) 387 - 392 ] . in this comment , we list out the ignored literature , compare between the entropy - number density ratio and two criteria of averaged energy per averaged particle number and constant entropy per cubic temperature .
all these criteria are confronted to the experimental results .
the physics of constant entropy per number density is elaborated .
it is concluded that this ratio ca nt remain constant , especially at large chemical potential related to ags and sis energies . .
|
there is a resurgent interest in studying the properties of polyelectrolytes in multivalent salt solutions because such system reveals many fascinating phenomena .
one vital example is the dna condensation , in which dna , a negatively - charged polyelectrolyte , undergoes a dramatic condensation from an extended structure to a compact , highly - ordered structure while multivalent salt is added to the solution @xcite .
the condensation induced by multivalent salt is not a privilege reserved only for dna but a common feature of polyelectrolytes @xcite .
usually trivalent salts or charged molecules of higher valence are demanded to induce the condensation @xcite .
experiments have shown that the morphology of the condensate depends strongly on the chain bending rigidity which balances the attractive collapsing force at a solvent quality .
flexible polymers generally collapse to disordered globules , and stiff ones collapse to ordered structures such as toroid or folded chain @xcite .
the influence of chain stiffness on chain morphology has been investigated by simulations , including the condensation of polyelectrolyte induced by multivalent salt @xcite or by increasing coulomb strength parameter @xcite and the collapse of neutral polymer driven by temperature decrease @xcite .
the results confirmed that chain rigidity plays a decisive role in determination of the condensed structure .
a phenomenon which is less well - known appears when an excess of multivalent salt is presented in the solution : the salt - induced condensation redissolves into the solution and the system returns to a homogeneous phase @xcite .
this phenomenon is called `` reentrant condensation '' @xcite and has been discovered since seventy - five years ago in the study of the coacervation of colloids by oppositely charged ions @xcite .
it occurs for flexible chains , for example , polystyrene sulphonate @xcite , and also for stiff chains , for example , dna @xcite .
the persistence length in the latter example is two order of magnitude larger than in the previous .
as the condensation takes place , the bare chain charge is almost neutralized by the condensed counterions dissociated from the multivalent salt @xcite .
these counterions not only screens out the long - range coulomb repulsion but also creates a short - range attraction due to correlated fluctuation , resulting in the condensation @xcite . on the other hand , for the redissolution of polyelectrolytes at high salt concentration ,
no comprehensive explanation is well established .
recently , two theories were proposed to explain the phenomenon but verification is still ongoing .
the first theory suggested that the redissolution is a consequence of charge inversion occurring when polyelectrolyte binds so many counterions that its net charge alters sign @xcite , whereas the second one argued that it has an origin of the screening of the short - range attraction @xcite . the second theory ,
furthermore , predicted that bjerrum association plays an important role on the organization of polyelectrolyte - ion complexes @xcite , and therefore , charge reversal is not always happened .
the folding transition of a polymer between an elongated state and a compact state is called `` coil - globule transition '' @xcite .
theorists predicted that for semiflexible chains , such transition is first order , using mean - field theory @xcite .
the prediction has been confirmed experimentally @xcite . despite continuous for the whole ensemble ,
the coil - globule transition is discrete at the level of single chain for polymers with sufficient stiffness @xcite . in this case , the transition is similar to the disordered - ordered transition between a gas - like and a crystal - like state . on the other hand , for single polymers with small stiffness ,
the transition is continuous , resembling to the disordered - disordered transition between a gas - like and a liquid - like state .
this topic has been recently studied in simulations @xcite .
the results showed that the discrete transition can be induced by adding condensing agent @xcite or by decreasing temperature @xcite , while chains are stiff .
however , the discussions were mainly focused on the chain condensation occurring in low temperature region or in low salt regime . since an excess of multivalent salt
can lead condensed chains reentering into the solution , it becomes very relevant to know whether the reentering transition behaves in a discrete manner or not . to our knowledge
, there is no simulation nor experiment , hitherto , discussing the conformational transition on the chain reentrance occurring at high salt concentrations . to get insight of it , we investigate in this paper the behavior of single polyelectrolytes with addition of multivalent salt by means of computer simulation
the salt covers a broad range of concentration so that single chains show , in turn , the condensation and the reentrance transitions in a microscopic way .
our study , therefore , offers a good opportunity to investigate thoroughly the coil - globule transition happened at low and at high salt concentrations .
there have been many simulations devoting to the study of the behavior of polyelectrolytes in salt - free solutions @xcite . only recently
, the study with addition of salt becomes numerically feasible because of the progress of computing power .
however , the salt valence studied in the literatures was usually monovalent or divalent @xcite .
for the study with high salt valence , the focus was mainly on the condensation of polyelectrolyte and the salt concentration was not high enough to investigate the chain reentrance into a solution @xcite .
one indirect way to discuss dna condensation and redissolution is to measure the effective interaction between two immobile rigid polyelectrolytes in the environment of multivalent salt @xcite .
recently , progress has been made in a series of study @xcite : by carefully choosing the simulation parameters , we were able to investigate salt - induced condensation and redissolution thoroughly in a direct way .
we found that the dimension of flexible polyelectrolytes undergoes subsequently two continuous transitions upon addition of multivalent salt , the collapsing transition and the swelling transition , which are respectively the microscopic representation of the condensation and the redissolution of polyelectrolyte .
we demonstrated that the excluded volume of ion plays an imperative role on the chain properties and should be inevitably incorporated in a theoretical analysis @xcite .
the study of the potential of mean force showed that like - charge attraction between chains happens only when salt concentration is intermediate around _ the equivalence point _ ( see sec .
[ sec : rg ] for definition ) and the size ratio of ion to monomer lies within a window around unity @xcite .
moreover , it was shown that only salt valence greater than two can induce apparently the two structural transitions @xcite , which is consistent with experimental observations @xcite .
nonetheless , the effect of chain stiffness was not discussed in these studies . since
semiflexible polyelectrolytes exhibit several intriguing phenomena , not being seen in flexible polymers , we focus in this work how chain stiffness influences chain morphology .
toroid structure is presumed to be the ground state of a compacted dna @xcite .
however , folded - rod structure and a number of other morphology , such as racquet , are also often observed in experiments @xcite .
the long lifetime of some of these structures indicates that dna has complicated energy landscape .
numerical studies showed that the energy levels of toroid and rodlike structure are comparable @xcite .
these two structures are surrounded by many metastable states and hindered by high energy barriers .
debate continues until now in deciding which structure is the real ground state of a condensed dna and under what condition it applies .
one of the important factors which affects the chain morphology is the chain length .
it was observed that longer chains display higher appearance percentage for toroids than for rods @xcite .
also , the degree of ion condensation on a chain was found to increase with chain length @xcite .
recently , chain length effect on the state diagram of a single chain has been discussed @xcite .
limited to today s computational power , time period able to be studied by molecular simulations is very short .
it is , thus , difficult to investigate properly the evolution of chain morphology , because of the long lifetime of some of the chain structures .
therefore , many authors found that the formation of a toroidal or a rod - like structure pertains to the initial chain configuration used in simulations @xcite . in this article , we study the morphology of a single polyelectrolyte with fixed chain length .
for the chain stiffness in which the morphological evolution is slow , several simulations are performed , starting with different initial configurations , to make sampling more complete . by this way
, we calculate static properties .
the rest of the article is organized as follows . the simulation model and the method used in this study
are described in sec .
[ sec : model_method ] .
the results and discussions are given in sec .
[ sec : result_discussion ] .
the topics include the radius of gyration ( sec .
[ sec : rg ] ) , the degree of chain swelling ( sec .
[ sec : crg ] ) , the end - to - end distance ( sec . [
sec : re ] ) , the probability density distribution ( sec .
[ sec : pdd ] ) , and the dynamics of chain conformation ( sec .
[ sec : x - t ] ) .
simulation snapshots are presented in sec .
[ sec : snapshot ] .
we summarize the conformation of a single polyelectrolyte in a state diagram in sec .
[ sec : state_diagram ] and give our conclusions in sec .
[ sec : conclusion ] .
our system consists of a single polyelectrolyte modeled by a negatively charged bead - spring chain and spherical counterions dissociated from the chain .
there are 48 beads ( monomers ) on the chain .
each of the beads carries a @xmath0 charge and each of the counterions carries a @xmath1 charge .
tetravalent salt is added into the system .
it is dissociated into tetravalent cations ( counterions ) and monovalent anions ( coions ) , modeled by charged spheres .
solvent molecules are not incorporated explicitly in the simulations .
we suppose them forming a medium of constant dielectric constant @xmath2 .
the collision between particles and solvent molecules is considered , and the effect is described by langevin equation . the system is placed in a cubic box and periodic boundary condition is applied .
four types of interaction are involved in the simulations .
the first one is the excluded - volume interaction , which is applied for all the particles , including monomers , counterions and coions .
it is modeled by a purely repulsive lennard - jones ( lj ) potential @xmath3 + \varepsilon_{lj } & \mbox{for } r \leq 2^{1/6}\sigma\\ 0 & \mbox{for } r > 2^{1/6}\sigma \end{array}\right.\ ] ] where @xmath4 is the interaction strength and @xmath5 is the diameter of a particle .
the second interaction is the coulomb interaction @xmath6 where @xmath7 and @xmath8 are respectively the valences of particle @xmath9 and @xmath10 , @xmath11 is the permittivity in vacuum , @xmath2 is the dielectric constant of the medium
. it can be re - expressed as @xmath12 where @xmath13 is the boltzmann constant , @xmath14 is the temperature , and @xmath15 is the bjerrum length , which denotes the distance between two unit charges at which the coulomb energy is equal to the thermal energy @xmath16 . the third interaction is the bond connectivity interaction .
two adjacent monomers are connected by a _ spring _ , modeled by a finitely extensible nonlinear elastic potential @xmath17 where @xmath18 is the length of a bond , @xmath19 is the maximum bond extention , and @xmath20 is the spring constant .
the forth interaction is a harmonic angle potential , which gives chain stiffness , @xmath21 where @xmath22 is the angle between two consecutive bonds , @xmath23 is the equilibrium angle , and @xmath24 is the force constant .
we perform langevin dynamics simulations in this study .
the equation of motion of a particle @xmath9 is described by the stochastic differential equation @xmath25 where @xmath26 is the mass of particle @xmath9 , @xmath27 denotes the friction coefficient and @xmath28 is a random force modeling the collisions by solvent molecules .
@xmath29 has zero mean over time and satisfies the fluctuation - dissipation theorem : @xmath30 where @xmath31 and @xmath32 are the kronecker delta and the dirac delta function , respectively .
an advantage of using langevin dynamics is that the temperature control is naturally incorporated . in this work
, we assume that all the particles have identical mass @xmath33 , diameter @xmath5 , and lj interaction strength @xmath4 .
we set @xmath34 and control temperature at @xmath35 .
coulomb interaction is computed using the technique of ewald sum .
@xmath20 is chosen to be @xmath36 and @xmath19 to be @xmath37 . under this setup , the average bond length is @xmath38 @xcite .
we study the effect of chain stiffness by altering the spring constant @xmath24 from @xmath39 to @xmath40 .
the equilibrium angle @xmath23 is set to be @xmath41 .
the size of the simulation box is @xmath42 and the concentration of tetravalent salt ranges from @xmath43 to @xmath44 .
all the particles are subjected to a friction force proportional to the particle velocity .
we set the damping constant @xmath45 to @xmath46 to mimic roughly an aqueous environment where @xmath47 is the time unit .
simulation time step @xmath48 is chosen equal to @xmath49 .
the initial chain configuration begins with an extended structure .
an equilibrium phase takes @xmath50 to @xmath51 time steps and a production run @xmath52 to @xmath53 time steps , depending on chain stiffness and salt concentration .
we found that in the region of intermediate salt concentration , the transition between different chain structures is slow . in order to sample more completely different structures ,
five to ten independent runs are performed starting with different initial configurations .
the simulations were run using lammps package . for the case of the highest salt concentration ,
the system contains 7776 charged particles .
the size of a polyelectrolyte can be characterized by the radius of gyration , @xmath54 .
it is calculated by taking the square root of the formula , @xmath55 where @xmath56 is the number of monomers on the chain , @xmath57 is the position vector of monomer @xmath9 , and @xmath58 is the center of mass of the chain .
we studied the effects of salt concentration and chain stiffness on the root - mean - square radius of gyration , @xmath59 .
the results are shown in fig.[fig : rg_line_vcs ] .
we observed that the variation of @xmath59 behaves in three distinct manners .
the first behavior applies to the chain with small @xmath24 , such as @xmath60 . @xmath59
firstly decreases with increasing @xmath61 , the concentration of the tetravalent salt , until the total charge of the tetravalent cations is equal to the negative charge carried on the chain backbone .
we call this particular salt concentration _ the equivalence point_. in this work , the equivalence point occurs at @xmath62 because the monomer concentration is fixed at @xmath63 .
@xmath59 then increases slowly while @xmath61 surpasses the equivalence point .
therefore , the chain undergoes two structural transitions upon addition of tetravalent salt : a shrinking transition and a followed swelling transition . the increasing rate of @xmath59 in the chain swelling region is apparently lower than the decreasing rate in the chain shrinking region . as a result , the curve of @xmath59 looks similar to a tilted ` l ' in the semilog plot . notice that at high salt concentration , the chain is not swollen back to its original size in the absence of salt .
these phenomena have been recently studied in detail by molecular dynamics simulations @xcite .
the second behavior applies to the chain with intermediate @xmath24 , for instance , @xmath64 . at the beginning of addition of salt ,
the chain size decreases gently .
the gentle decrease turns to become a sharp decrease once @xmath61 is increased beyond some critical value @xmath65 , and quickly , @xmath59 reaches a small constant value .
if @xmath61 is further increased over a second critical value @xmath66 , an abrupt increase in @xmath59 takes place .
the equivalence point @xmath67 is situated in the region between @xmath68 and @xmath69 . in the high - salt region
@xmath70 , the chain reattains roughly its original size before the condensation occurred .
these two sudden changes suggest that the chain morphology is altered , from an extended structure to a condensed structure , and vice versa , in a discontinuous way .
the chain possibly commits a first - order conformational transition @xcite . in the semilog plot
, the shape of the @xmath59 curve looks like the capital letter ` u ' .
a relevant phenomenon has been observed in the experiments of dna precipitation by polyamines where dna percentage presented in the supernatant shows a similar u - shaped variation as a function of spermidine concentration @xcite .
suppose that the occurrence of the chain precipitation is directly related to the chain size in the solution .
this variation will then reflect how the dimension of a semiflexible dna varies with salt concentration .
our results are in agreement with the experimental observation .
the third behavior is observed for the chain with large @xmath24 , for example , @xmath71 .
the influence of the adding salts upon the chain size is not significant .
@xmath59 stays essentially at a constant value , and in the figure , the curve is roughly a straight line . based upon the variation of chain size against @xmath61 , a polyelectrolyte of fixed chain length can be classified into three categories according to the chain stiffness : * _ flexible chain _ , characterized by a tilted - l - shaped @xmath59 curve ; * _ semiflexible chain _ , characterized by a u - shaped @xmath59 curve ; * _ rigid chain _ , characterized by an almost straight @xmath59 curve .
one way to justify if a chain belongs to the category _ rigid chain _ or not is to compare the bare persistence length with the chain length . if the bare persistence length is larger than the chain length , the chain is considered to be _
rigid_. in our study , the bare persistence length @xmath72 is related to the force constant @xmath24 of the harmonic angle potential by @xmath73 where @xmath74 is the average bond length . therefore , @xmath72 becomes larger than the chain length while @xmath75 .
this simple criterion works reasonably well because the chain shows a rigid rod structure at all @xmath61 for @xmath76 and @xmath40 but not for the cases with smaller @xmath24 , as having been shown in fig.[fig : rg_line_vcs ] . in the simulations
, we observed that , no matter how stiff the chain is , the polyelectrolyte is in its most stretched state while no salt is added into the solution .
we found that at a given salt concentration , a stiffer chain displays a larger @xmath59 , and hence , occupies a larger space .
the results obtained here also demonstrate that the chain stiffness plays an imperative role in the determination of the type of the coil - globule transition to be continuous or discrete , both in the condensation and in the reexpansion of a polyelectrolyte . in order to understand the degree of swelling of a polyelectrolyte in salt solutions ,
we calculate the size of the uncharged counterpart of the polyelectrolyte in a salt - free solution and use it as the reference of comparison
. the root - mean - square radius of gyration of the uncharged chain , @xmath77 , is equal to @xmath78 , @xmath79 , and @xmath80 for the cases @xmath81 , @xmath82 and @xmath83 , respectively .
we define the degree of chain swelling by @xmath84 and present the results in fig .
[ fig : beta_line_vcs ] . we observed that in the low - salt region , @xmath85 is positive .
it means that the charged chain occupies a larger space than the uncharged chain , apparently resulting from the electrostatic repulsion between monomers on the chain .
the value of @xmath85 for the flexible chain ( @xmath86 ) is much larger than the ones for the semiflexible chain ( @xmath87 ) and for the rigid chain ( @xmath88 ) .
therefore , the major contribution to the dimension of a semiflexible or a rigid polyelectrolyte comes from the intrinsic chain stiffness , whereas the size of a flexible chain is mainly determined by the coulomb repulsion on the chain backbone . in the mid - salt region around the equivalence point , the degree of swelling is negative for the flexible chain and for the semiflexible chain .
the chains are , hence , in collapsed states . in this region
, electrostatics dominates the chain stiffness
. the maximum shrinkage , in reference of the size of the neutral chain , can be as large as @xmath89 .
for the rigid chain , the collapsing transition does not happen , and thus , @xmath85 is close to zero . in the high - salt region
, we saw that the value of @xmath85 approaches to zero for the semiflexible chain .
the polyelectrolyte hence occupies a space volume similar to its uncharged counterpart . for the flexible chain , @xmath85
is increased with @xmath61 but still takes a finite negative value . in our previous study @xcite , flexible polyelectrolytes in the region of much higher salt concentration have been investigated , and the results showed that the chain attains a size similar to the neutral polymers .
therefore , @xmath85 is expected to tend toward zero if @xmath61 was increased to a much higher value in the current study . in fig .
[ fig : beta_line_vka ] , we present the variation of @xmath85 as a function of @xmath24 at different @xmath61 . at small @xmath61
, @xmath85 is positive and monotonically decreases to zero with increasing @xmath90 .
opposite trend of behavior is observed at large @xmath61 , for instance , @xmath91 , where @xmath85 is negative and monotonically increases to zero . in the mid - salt region , @xmath85 displays a v - shaped curve and the maximum shrinkage is happened around @xmath92 .
thus , by tuning @xmath90 and varying @xmath61 , one is able to control the degree of chain swelling to design desired functions in real applications .
the results obtained here tell us that in a salt - free solution , the larger the value of @xmath90 , the less significant the size difference between a charged chain and its uncharged counterpart will be . since at large @xmath61
a chain occupies a space similar to the one of a neutral polymer , a semiflexible chain will reattain roughly its original size at zero @xmath61 by addition of many salts into the solution , but a flexible chain will not reattain it . conventionally , the end - to - end distance @xmath93 is used to characterize the size of a polymer . in fig .
[ fig : re_line_vcs ] we present how the root - mean - square end - to - end distance , @xmath94 , varies with salt concentration for different chain stiffness , where @xmath95 and @xmath96 are the position vectors of the two ends of a chain .
the behavior of @xmath93 resembles the behavior of @xmath54 : a flexible chain displays a tilted l - shaped @xmath93 curve , a semiflexible one displays a u - shaped @xmath93 curve , and a rigid one displays a straight @xmath93 curve .
nonetheless , the value of @xmath93 does not precisely reflect the _ real _ size of a chain . for example , @xmath93 for a semiflexible chain becomes smaller than that for a flexible chain in the mid - salt region , although a semiflexible chain has a larger size ( cf .
[ fig : rg_line_vcs ] ) .
this inconsistency results from the formation of ordered structures which gives a small @xmath93 , such as hairpin or ring - like structure .
therefore , the quantity @xmath93 is not always suitable to characterize the dimension of a chain in a salt solution , particularly when the chain stiffness is intermediate . the mean - square end - to - end distance of a chain has been derived theoretically under the framework of the worm - like chain ( wlc ) model and reads as @xmath97 where @xmath98 is the contour length and @xmath99 is the persistence length @xcite . for a polyelectrolyte chain , @xmath99 is the sum of two contributions : the first one is the bare persistence length @xmath72 , coming from the intrinsic chain stiffness , and the second one is the electrostatic persistence length @xmath100 , due from the electrostatic interaction .
odijk @xcite and , independently , skolnick and fixman @xcite ( osf ) have deduced a formula for the electrostatic persistence length in the limit of stiff chain and low ionic strength .
this formula reads as @xmath101 where @xmath102 is the manning parameter , defined as the ratio of @xmath103 over the effective charge distance on the chain , and @xmath104 is the inverse debye length . in order to make a comparison of the simulation results with the theory
, we adopted the manning - oosawa ( mo ) condensation theory @xcite to estimate @xmath102 . in our case , @xmath102 is renormalized to a value @xmath105 , due to the condensation of tetravalent counterions , while @xmath106 . for small salt concentration @xmath107
, all the tetravalent counterions are expected to condense on the chain , but the number of these ions is not sufficient to renormalize @xmath102 to @xmath105 . at this moment , @xmath102 decreases linearly from 1 to @xmath105 with @xmath61 , or equivalently , @xmath108 . plugging @xmath102 and @xmath109 into eq .
( [ formula : le ] ) , we obtained @xmath110 , and then , added it to @xmath73 to get the persistence length @xmath99 . finally , using eq .
( [ formula : re2 ] ) , we calculated @xmath111 and the results for different chain stiffness are plotted in dashed curves in fig .
[ fig : re_line_vcs ] .
we observed that the wlc model , combining with the osf and the mo theories , gives a qualitative prediction for @xmath111 at low salt concentration for semiflexible and stiff chains .
surprisingly , it also gives a not - bad description of @xmath111 for the flexible chain with zero @xmath24 and even captures the decreasing behavior up to @xmath112 , although the osf and the mo theories are not expected to work well neither for flexible chains nor for a solution with multivalent salt .
the theoretical value is found to be smaller than the value obtained from the simulations .
this deviation can be attributed to the effect of the excluded volume interaction , which is neglected in the derivation of the wlc model and should give a positive contribution to the chain size if it is taken into account . in the mid - salt regime , the theory fails to predict @xmath111 for semiflexible chains . in this region ,
the chain is condensed by tetravalent counterions .
the small chain size suggests that the persistence length is significantly reduced , and as a consequence , the electrostatic persistence length takes a negative value .
the negative electrostatic persistence length has been experimentally observed @xcite . at high salt concentration ,
the wlc model again describes well the end - to - end distance except for the case with zero @xmath90 ; at this moment , @xmath110 is almost zero and the persistence length is dominated by the bare persistence length . before processing a more quantitative analysis , we present in this section snapshots of our simulations to give readers a vital illustration how the chain looks like in different regions of salt concentration and chain stiffness .
[ fig : flexiblechain ] shows snapshots of a flexible chain at low , middle , and high salt concentrations .
the morphology of the chain is elongated at low salt concentration , and displays large - angle turns near the places where the counterions condense , as shown in fig .
[ fig : flexiblechain](a ) . in the mid - salt region around the equivalence point ,
the chain collapses and exhibits a randomly - arranged compact structure ( see fig .
[ fig : flexiblechain](b ) ) . at high salt concentration ( fig .
[ fig : flexiblechain](c ) ) , the chain swells and acquires a zigzagged , extended structure , but not as elongated as at low salt concentration .
[ fig : semiflexiblechain ] shows snapshots of a semiflexible chain at three salt concentrations .
the shape of the chain is smoother than that of the flexible chain . at low salt concentration ( fig .
[ fig : semiflexiblechain](a ) ) , the chain is elongated and not zigzagged . at middle salt concentration ,
the chain is condensed , and noticeably , forms ordered structure , in contrast to the disordered globule formed by a flexible chain .
typical ordered structures are shown in fig .
[ fig : semiflexiblechain](b ) .
they are , from left to right , hairpin , racquet , and toroid .
these typical structures have been observed in the experiments of dna condensation @xcite . unlike the case for flexible chains , a condensed counterion can hardly induce a large - angle turn on the chain backbone .
if it is the case , the turn happens to be the folded end of a hairpin structure and counterions are condensed between the two branches of the hairpin .
the whole structure looks similar to peas in a pod . at high salt concentration
, the chain displays again an elongated structure ( see fig . [
fig : semiflexiblechain](c ) ) . fig .
[ fig : stiffchain ] shows the snapshot of a rigid chain .
the chain exhibits a stretched structure no matter how many salts are added into the solution . the chain is stiff enough to resist from bending or twisting . from these snapshots
, we found that a compact chain structure is formed when condensed tetravalent counterions can overcome the chain stiffness to bridge between monomers and have enough number to tightly bind the whole chain structure .
we observed that , regardless of the chain stiffness , the number of tetravalent counterions condensed on a chain increases with @xmath61 .
when @xmath61 is smaller than the equivalence point @xmath67 , almost all the tetravalent counterions are condensed on the chain .
only when @xmath113 , non - condensed tetravalent counterions can appear in the bulk solution .
moreover , the number of condensed tetravalent counterions surpasses the number needed to neutralize the bare chain charge when @xmath113 , and as a consequence , the chain is _ locally overcharged _ . for flexible chains , there are mainly two effects to drive chain reentrance at high salt concentration .
the first one is the coulomb repulsion between condensed tetravalent counterions .
since the chain is surcharged by the condensed tetravalent counterions , the net repulsion between these ions manifests its effect in the swelling of the chain size .
the second one is the presence of the non - condensed tetravalent counterions in the bulk solution .
these non - condensed counterions interact with the chain , and can dynamically replace previously condensed counterions and become newly condensed counterions . during this process
, the chain morphology oscillates from a loose globule to an expanded structure and vice versa .
increasing salt concentration increases the concentration of non - condensed tetravalent counterions in the bulk , which makes easier the replacement of tetravalent counterions to take place .
the chain hence more frequently stays in an expanded state and the averaged chain size increases . for semiflexible chains ,
there is , in addition , a third effect taking part in the reentrant transition the chain stiffness .
the chain stiffness always gives an effect against the formation of a compact structure .
together with the instability induced by the presence of non - condensed tetravalent counterions , a chain can dramatically leave a condensed state to an elongated structure at high salt concentration . in the previous sections ,
the conformation of a polyelectrolyte was studied by averaged quantities , such as the root - mean - square @xmath54 and the root - mean - square @xmath93 .
in fact , averaged quantities can not reflect precisely the chain shape since chain morphology varies with time during simulations . in order to understand the details of chain morphology
, we investigate in this section the probability density distribution ( pdd ) of conformational quantities .
two quantities are studied : the first one is the asphericity @xmath114 and the second is the radius of gyration @xmath54 .
the asphericity @xmath114 measures the deformation of chain morphology from a spherical geometry and is defined by @xcite @xmath115 @xmath116 , @xmath117 , and @xmath118 are the three eigenvalues of the gyration tensor of the chain calculated by @xmath119 where the subfixes @xmath85 and @xmath120 denote the three cartesian components @xmath121 , @xmath122 and @xmath123 . the value of @xmath114 ranges between 0 and 1 .
it is equal to 0 for a perfect sphere , 0.25 for a perfect ring , and 1 for a straight line . for a coil chain ,
@xmath124 is 0.431 obtained from simulation @xcite . in fig .
[ fig : histod_ka0 ] , parts ( a ) and ( b ) , we present pdd of asphericity , @xmath125 , and pdd of radius of gyration , @xmath126 , for the flexible chain with null @xmath24 .
we observed that @xmath125 shows a peak near @xmath127 in the salt - free solution , indicating that the chain favors an extended structure .
the peak gradually moves toward a small @xmath114 as @xmath128 is increased , and the width of the peak becomes broader .
the width then decreases while @xmath128 approaches to the equivalence point @xmath67 ( case vi in the figure ) . at @xmath67
, @xmath125 shows the most pronounced peak at @xmath129 , and therefore , the favored morphology is a compact sphere - like structure .
if @xmath128 is further increased , the peak becomes broaden and the position of the peak shifts not much toward a larger value . in this salt region
, the chain behaves like a random coil and alternates between a globule and an extended structure . for @xmath126 ,
similar behavior was observed .
the peak of @xmath126 moves left to a small @xmath54 and then moves slightly right as @xmath61 is increased . also , the most pronounced peak appears at the turning point @xmath112 .
[ fig : histod_ka8](a ) and ( b ) shows , respectively , @xmath125 and @xmath126 for the semiflexible chain with @xmath130 . in the low - salt region
, @xmath125 displays a single peak at @xmath131 , and therefore , the chain favors to exhibit a rod - like structure .
if the salt concentration is increased to some intermediate value , a second peak appears at @xmath132 .
it indicates the formation of a ring - like ( toroid ) structure .
the second peaks first increases in its height with @xmath128 and then decreases .
it disappears while @xmath61 surpasses some critical value , and the chain reattains a rod - like structure .
more information can be obtained if we study @xmath126 at the same moment .
for example , at @xmath133 ( case iii in the figure ) , @xmath126 shows a doubly peaked distribution and the second peak is roughly situated at the middle position between the first peak and @xmath134 .
we know from @xmath125 that the chain exhibits a rod - like structure at this salt concentration .
the second peak in @xmath126 , hence , corresponds to hairpin structure because hairpin is a two - folded rod and its radius of gyration is equal to half of that of the unfolded rod .
consequently , the chain morphology alternates between unfolded rod and hairpin structure . at higher salt concentration ,
the first peak in @xmath126 disappears and a new peak appears at @xmath135 .
this new peak is attributed to toroidal structure because it corresponds to the peak of @xmath125 at @xmath132 , which is the typical value for a ring structure .
therefore , in this salt region coexist two condensed structures , hairpin and toroid .
we noted that probability density for @xmath114 or @xmath54 is nearly zero between the two peaks when @xmath61 is not large , which indirectly indicates that the structural transition between hairpin and toroid rarely happens during a run .
if @xmath61 is large , for instance , @xmath136 , a finite value appears between the two peaks , indicating that the two structures can frequently transit between each other .
we mention that these results were obtained by doing statistics combined from several independent runs starting with different initial configurations to circumvent the problem of slow transition . at @xmath137 ( case x in the figure ) ,
@xmath126 shows three peaks , indicating a coexistence between the two condensed and the extended - chain structures .
one thing is worth to be noticed : in this middle salt region , the peak in @xmath126 , which corresponds to a toroidal structure , shifts toward a large value as @xmath61 is increased .
it suggests that the toroidal size grows up with salt concentration .
this phenomenon is in agreement with the observation of conwell _ et al .
_ @xcite . in the experiment
, they found that monovalent or divalent salt causes an increase in diameter of a condensed dna toroid .
finally , if @xmath61 goes even higher , @xmath126 displays only one peak at @xmath138 , and hence , the favored shape is an unfolded chain .
we emphasize that the pdds for the semiflexible chain shows different behavior , compared with the ones for the flexible chain . in the middle salt region , they are multiply peaked and the peaks appear or disappear in a sudden way . moreover , each peak corresponds to an ordered structure such as toroid or hairpin . on the other hand , the pdds for the flexible chain display only single peaks and the peak position moves continuously in the space with salt concentration . as a consequence ,
the coil - globule transition for the semiflexible chain occurs in a discrete manner , whereas in a continuous way for the flexible chain .
both the discrete and the continuous transition have been demonstrated in simulations by other authors @xcite .
however , the subjects were focused on the chain collapsing from a coil state to a globule state , happened at low salt concentration or at low temperature .
our study presented here goes further and gives a thorough description of how the pdds varies with salt concentration , not only in the low salt region but also in the high salt region . the sudden appearance and disappearance of peaks
show that the discrete nature of transition takes place also at high salt concentration when a semiflexible chain redissolves from an ordered globule state to a coil state . to our knowledge , this is the first time such observation at high @xmath61 is reported .
we present in fig .
[ fig : histod_ka100](a ) and fig .
[ fig : histod_ka100](b ) @xmath125 and @xmath126 , respectively , for a rigid chain with @xmath139 .
@xmath125 shows a sharp peak near @xmath140 and @xmath126 a sharp peak at @xmath54 around @xmath141 for all the salt concentration investigated .
the chain exhibits a fully extended structure .
the results can be verified by estimating the chain length from the radius of gyration .
it is known that the length @xmath142 of a rod is related to the radius of gyration by @xmath143 .
consequently , @xmath144 yields @xmath142 equal to @xmath145 , which is in good agreement with the length of a fully extended chain since our chain has 47 bonds and the average bond length is @xmath38 .
we remark that the results presented here were obtained from long simulations .
we have calculated the correlation time for the quantities @xmath114 and @xmath54 at each salt concentration .
the results shows that our simulations contains hundreds to thousands independent data for the cases of flexible chain and rigid chain .
therefore , we are confident that fig . [
fig : histod_ka0 ] and fig .
[ fig : histod_ka100 ] report correct equilibrium populations . for the case of semiflexible chain , particularly in the mid - salt region ,
the system shows rich morphologies and the transition between different morphologies is slow .
we have preformed several independent runs to sample different trajectories in order to improve the statistics .
the analysis of the correlation time shows that at least , thirty independent samples were cumulated in the simulations .
the results reported in fig . [ fig : histod_ka8 ] , hence , may not be very accurate .
however , the pdds show consistent trend of variation against @xmath61 , and also in fig . [
fig : rg_line_vcs ] , no remarkably large fluctuation is appeared in the @xmath59 curve .
therefore , these results still give an appreciable description of the equilibrium populations .
we have shown that a polyelectrolyte can be condensed by multivalent salt . particularly , when the chain is semiflexible , the condensed chain displays an ordered structure , such as hairpin or toroid . in order to get insight of the phenomena
, we take the semiflexible chain of @xmath146 as an example , and study in this section how the conformational quantities , @xmath114 , @xmath54 and @xmath93 , evolve with time .
we focus at first the conformational dynamics at @xmath147 . in fig .
[ fig : histod_ka8 ] , we have demonstrated that the pdds of the semiflexible chain are doubly peaked at this salt concentration .
the chain stays in a hairpin state or a toroid state and the transition between the two states is not frequent . for this case ,
several independent runs were performed starting with different initial configurations . in order to get insight of the dynamics ,
we show firstly in fig .
[ fig : yt_plot_s8c4 ] the time evolution of @xmath114 , @xmath54 and @xmath93 for one of the independent runs , in which the chain exhibits hairpin structure .
the asphericity @xmath114 fluctuates steadily around @xmath148 as shown in fig .
[ fig : yt_plot_s8c4](a ) . at the same moment
, the radius of gyration fluctuates around @xmath149 and @xmath93 fluctuates around a small value ( see fig .
[ fig : yt_plot_s8c4](b ) )
. therefore , the chain has its two ends fluctuating near the open end of the hairpin structure .
readers can see from the figures that this structure is lasted for a duration of at least @xmath150 simulation steps in this run
. please be aware that @xmath150 simulation steps is a long time period in a typical molecular simulation . in our case , it is approximately equal to a duration of @xmath151 , provided that @xmath5 is @xmath152 , @xmath14 is 300k and @xmath33 is 200da .
we then show in fig .
[ fig : yt_plot_s8c5 ] the dynamics of these quantities when the chain exhibits toroidal structure .
the asphericity @xmath114 fluctuates around @xmath153 except at four places , fluctuating around @xmath154 ( cf .
[ fig : yt_plot_s8c5](a ) ) .
at these four places , the chain displays a twisted ring structure , similar to the shape of the digit ` 8 ' .
since the two ends of a toroidal chain is generally not very close to each other , @xmath93 fluctuates around a large value ( see fig . [
fig : yt_plot_s8c5](b ) ) , compared to the value for hairpin structure ( fig .
[ fig : yt_plot_s8c4](b ) ) . resembling the previous run , the toroid structure is continued for nearly @xmath150 simulation steps .
the morphological transition between toroid and hairpin is , hence , not frequent at this salt concentration .
the transition between different ordered structures becomes observable in simulations if the salt concentration is high .
for example , at @xmath136 ( shown in fig .
[ fig : yt_plot_s96](a ) ) , the asphericity fluctuates around two values , at @xmath155 and at @xmath156 , which indicates the transition between ring - like and rod - like structures . moreover , fig .
[ fig : yt_plot_s96](b ) shows that the radius of gyration associated to the rod - like structure is larger than that associated to the ring - like structure , but the end - to - end distance is smaller .
it enables us to identify this rod - like structure to be a two - folded chain ( _ i.e. _ hairpin ) , which was also confirmed from the simulation snapshots .
the second example is chosen at @xmath157 ( see fig . [
fig : yt_plot_s192 ] ) .
the asphericity shows similar fluctuation around the two values .
however , the value , @xmath158 , which corresponds to rod - like structure , is now associated to two behaviors of @xmath54 , fluctuating at @xmath159 and @xmath160 . combining the information obtained from @xmath93
, we assert that the first fluctuation of @xmath54 corresponds to a two - folded chain structure and the second one corresponds to an unfolded - chain structure .
therefore , the chain transits between three states , which are unfolded chain state , hairpin state , and toroid state . recently , dynamic exchange between toroid and folded - chain structures in a salt - free solution of semiflexible polyelectrolyte has been numerically demonstrated by tuning the coulomb strength parameter @xcite .
the work showed a back - and - forth flipping between the two structures , similar to our observations .
our study , moreover , showed that transitions between different chain states can be enhanced by addition of multivalent salt beyond the equivalence point @xmath67 .
the mechanism is understood as follows .
below @xmath67 , most of the multivalent counterions are condensed on the chain .
roughly while @xmath61 is larger than @xmath67 , multivalent counterions start to exist in the bulk solution .
therefore , increasing salt concentration increases the number of the multivalent counterions surrounding but not condensed on the chain . due to the presence of these surrounding counterions , the chance for a chain to escape from its current condensed state
is increased .
the surrounding multivalent counterions strongly interact with the chain , and can induce local change of the chain morphology by the replacing process : some of the condensed ions on the chain are released into the solution and replaced with the surrounding ions . occasionally , the chain forms a newly condensed structure .
the conformation of our single polyelectrolytes , classified by the three categories : _ flexible _ chain , _ semiflexible _ chain , and _ rigid _ chain , is summarized in a schematic state diagram shown in fig .
[ fig : state_diagram ] .
flexible chain exhibits an extended structure in a salt - free solution , and continuously collapse to a compact , spherical structure as @xmath61 is increased to the equivalence point . further increasing @xmath61 gradually leads chain discondensation . at very high salt concentration
, the chain behaves similar to a random coil .
semiflexible chain shows different conformations as multivalent salt is added into the solution . at very low salt concentration , the chain exhibits an elongated structure . if @xmath61 is increased beyond the first critical value @xmath65 , the chain bends over upon itself and forms two - folded rod structure ( hairpin ) .
this structure is not so stable that the chain morphology alters between a hairpin and an extended structure . further increasing @xmath128
firstly stabilizes the hairpin structure , and eventually , leads to the appearance of toroidal structure . at this moment ,
hairpin and toroid structures coexist in the solution .
we found that the rate of transition between hairpin and toroid increases with the salt concentration . therefore , the transition becomes observable during limited time period of a molecular simulation while @xmath128 is large enough .
if @xmath128 is not large enough , the two structures can be observed only in independent runs , since our model system contains solely one chain .
the formation of these ordered structures has been proposed to be a consequence of a nucleation - growth process derived from the thermal fluctuations in chain morphology @xcite . at higher salt concentration
, the chain can even transit between three morphological states : the unfolded - rod state , the hairpin state , and the toroid state . finally ,
if @xmath61 surpasses the second critical value @xmath66 , the chain can no longer be collapsed , and consequently , reattains a unfolded , rod - like structure .
rigid chain , just as its name denotes , is rigid , no matter how many salts are added in the solution , and always shows a rodlike , fully stretched structure in simulations .
the state diagram presented here is in accordance with the results obtained by other authors , although the employed models are different .
for example , ou and muthukumar investigated the conformation of single polyelectrolytes in a salt - free solution and the chains were collapsed by increasing coulomb strength parameter @xmath161 @xcite .
they observed that a semiflexible chain stays , in turn , in a coil state , a folded - rod state , and a toroid state , as @xmath161 is increased ( cf .
fig . 6 of ref .
the result is consistent with what we observed in the region where chains are collapsed .
for the second example , we mention the works done by ivanov and his coworkers @xcite . in their works ,
the effects of chain stiffness and chain length on the conformation of neutral polymers were studied .
they found that finite chain length can make unsharpened the coil - to - toroid transition driven by temperature reduction , and several intermediate states , such as folded - rod and disk - like structures , were observed in the transition region .
the proposed state diagram is in analogy of our results in the low - salt region .
we remark that chain length is an important factor in determination of the chain morphology .
experiments have shown that short dna molecules are more easily condensed to folded rods than toroids @xcite .
this phenomenon gives an explanation of why the hairpin structure appears frequently in our state diagram .
it is because the modeled chain is not long in our study .
nonetheless , the schematic state diagram obtained here has gone beyond the works done by other authors .
we investigated the salt - induced conformation of polyelectrolytes thoroughly in a _ direct _
way since this topic has attracted much attention in scientific community @xcite , and explored the situations from a salt - free condition to a very high salt concentration so that chain condensation and redissolution are both occurred .
the findings give deep insight of the phenomena of dna condensation at molecular level and are relevant for a broad range of salt - induced complexation phenomena .
we have performed langevin dynamics simulations to study single polyelectrolytes in tetravalent salt solutions under the framework of a coarse - grained model .
the effect of salt concentration and the role of chain stiffness on the properties of chain conformation have been discussed . in order to understand thoroughly the reentrant condensation ,
a broad range of salt concentration has been investigated so that the single polyelectrolytes display , in turn , collapsing transition and swelling transition upon addition of salt .
we have observed that chain morphology crucially depends on chain stiffness . at a fixed chain length ,
the radius of gyration @xmath162 of a polyelectrolyte varies in three different ways with the salt concentration @xmath61 . based upon these variations
, we can classify polyelectrolytes into three categories .
the first category is _ flexible _ chain for which @xmath59 shows a tilted l - shaped curve as a function of @xmath61 in the semilog plot .
the second one is _
semiflexible _ chain which is characterized by an u - shaped @xmath59 curve . and
the third one is _ rigid _
chain for which @xmath59 is roughly a constant , no matter how many salts are added in the solution .
the end - to - end distance shows the similar variation as the radius of gyration .
however , its value does not always reflect the real chain size , particularly when the chain is condensed to an orderly packed structure such as hairpin .
moreover , we have found that the wlc model with persistence length obtained by the osf theory qualitatively predicts the end - to - end distance at low and at high salt concentrations for semiflexible and rigid chains .
surprisingly , these theories also give a not - bad description of the decreasing behavior of @xmath111 against @xmath128 for flexible chain up to @xmath163 , although they are not expected to work well neither for flexible chains nor for a solution with multivalent salt .
we have found that in a salt - free or vary low - salt solution , the degree of swelling of a polyelectrolyte decreases with increasing the chain stiffness , in reference of the size of the uncharged counterpart of the chain . on the other hand , in a high - salt region
, it tends toward zero , regardless of the chain stiffness . as a consequence , at high @xmath61 ,
a semiflexible chain reattains a dimension close to the chain size in the absence of salt , but not does a flexible one .
we have discussed the reasons for the occurrence of chain reentrance or swelling at high salt concentration . for flexible chains
, two effects can possibly drive this phenomenon : ( 1 ) the net coulomb repulsion between condensed multivalent counterions which locally overcharges the chain , ( 2 ) the instability derived from the presence of non - condensed multivalent counterions in a bulk solution . for semiflexible chains , the chain stiffness plays , in addition , the imperative role on reexpanding the chain .
the chain stiffness has been witnessed to influence both the local and the global structures of a polyelectrolyte .
a flexible chain shows a zigzagged local structure in the presence of multivalent salt and the condensed structure is disordered .
in contrast , a semiflexible chain locally shows smooth morphology .
large - angle turn on the chain backbone can only be observed when the chain is condensed to a hairpin structure .
moreover , the global structure of a condensed chain is orderly packed . hairpin and
toroid are the two favored structures .
the chain stiffness affects also the nature of the structural transition . by studying the probability density distributions of asphericity @xmath114 and of @xmath54
, we have demonstrated that a semiflexible chain undergoes a coil - globule transition in a discrete manner between an extended structure and a condensed , ordered structure , whereas a flexible chain undergoes a continuous transition .
this discrete nature is happened not only in the low - salt region when chain condensation occurs , but also in the high - salt region when chain reenters into the solution .
we have studied the dynamics of the conformational quantities of a semiflexible chain .
upon addition of salt beyond the first critical concentration @xmath68 , the chain is folded to a hairpin structure .
however , it is a dynamic equilibrium process and the chain morphology alters between a hairpin and a unfolded - chain structure . further increasing @xmath61
firstly stabilizes the hairpin structure and then triggers the birth of toroid structure .
transition between hairpin and toroid becomes more frequent as @xmath128 getting higher .
when @xmath61 approaches the second critical concentration @xmath66 , these two structures become unstable and the chain can stay again in a unfolded state , coexisting with the two condensed structures . at the end
, we summarized the conformation of individual polyelectrolytes in a state diagram .
although the modeled chain is not very long , our results agree with experiments and other numerical works in many aspects in the chain condensation region .
the work presented here goes further to investigate in detail the microstructure and the dynamics of a chain in the chain redissolution region , which gives us a thorough vision of the behavior of polyelectrolytes in salt solutions .
since chain length is also a relevant factor to affect the morphology of a chain , it will be very interesting to study the effect of finite chain length .
that will be the topic of our future work .
this material is based upon work supported by the national science council , the republic of china , under the contract no . nsc 95 - 2112-m-007 - 025-my2 .
a large part of the simulations was run using the resources of the national center for high - performance computing under the project `` taiwan knowledge innovation national grid '' .
the authors express their gratitude to the members and the staffs of the council and the center . 75 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , * * , ( ) . , , , * * , ( ) . , * * , ( ) .
, in _ _ , edited by ( , , ) , pp . . , , , , , , * * , ( ) . , , , ,
* * , ( ) . , , , * * , ( ) . ,
* * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , * * , ( ) . , ed . , _ _ , vol .
( , , ) . , , , * * , ( ) . , * * , ( ) . ,
* * , ( ) .
, , , * * , ( ) . , _ _ ( , , ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . ,
* * , ( ) .
, , , * * , ( ) . , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , , * * , ( ) . , , , * * , ( ) . , , , *
* , ( ) .
, , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , , ,
, , * * , ( ) . , , , * * , ( ) . , , , * * , ( ) . ,
* * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , * * , ( ) . , , , * * , ( ) . , _ _ ( , ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , eds . , _ _ , vol .
( , , ) . , ed . ,
_ _ ( , ) .
* @xmath59 as a function of @xmath128 for the chains of different stiffness @xmath24 .
* @xmath164 as a function of @xmath128 for the chains of different stiffness @xmath90 . *
@xmath165 as a function of @xmath90 at different salt concentration @xmath128 .
* @xmath166 as a function of @xmath128 for the chains of different @xmath24 .
the dashed curves denote the prediction of the wlc model combined with the osf and mo theories for different values of @xmath24 , indicated on the left - hand side of the corresponding curve .
* snapshots of simulation for a flexible chain ( @xmath167 ) at ( a ) @xmath168 , ( b ) @xmath169 , and ( c ) @xmath170 .
the gray bead - spring chain represents the polyelectrolyte , the black spheres represent the tetravalent counterions , and the white spheres represent the monovalent counterions .
coions are not shown for the reason of clarity .
* snapshots of simulation for a semiflexible chain ( @xmath146 ) at ( a ) @xmath168 , ( b ) @xmath171 , and ( c ) @xmath170 .
the gray bead - spring chain represents the polyelectrolyte , the black spheres represent the tetravalent counterions , and the white spheres represent the monovalent counterions .
coions are not shown for the reason of clarity . *
snapshots of simulation for a rigid chain ( @xmath172 ) at @xmath169 .
the gray bead - spring chain represents the polyelectrolyte , the black spheres represent the tetravalent counterions , and the white spheres represent the monovalent counterions . coions are not shown for the reason of clarity . * \(a ) @xmath125 , and ( b ) @xmath126 of the flexible chain with zero stiffness for the cases of different @xmath128 .
each case is numbered in roman number along ` case'-axis .
the value in the parenthesis following the roman number denotes the salt concentration in the unit @xmath173 . *
\(a ) @xmath125 , and ( b ) @xmath126 of the semiflexible chain with @xmath174 for the cases of different @xmath128 .
each case is numbered in roman number along ` case'-axis .
the value in the parenthesis following the roman number denotes the salt concentration in the unit @xmath173 . *
\(a ) @xmath125 , and ( b ) @xmath126 of a rigid chain with @xmath139 for the cases of different @xmath128 .
each case is numbered in roman number along ` case'-axis .
the value in the parenthesis following the roman number denotes the salt concentration in the unit @xmath173 .
* time evolution of ( a ) @xmath114 , and ( b ) @xmath54 and @xmath93 , at @xmath175 where the chain exhibits a hairpin structure .
* time evolution of ( a ) @xmath114 , and ( b ) @xmath54 and @xmath93 , at @xmath175 where the chain shows a toroidal structure .
* time evolution of ( a ) @xmath114 , and ( b ) @xmath54 and @xmath93 , at @xmath176 where the chain alters between a hairpin and a toroid structure .
* time evolution of ( a ) @xmath114 , and ( b ) @xmath54 and @xmath93 , at @xmath177 where the chain morphology alters between a hairpin , a toroid , and an unfolded - chain structure . * schematic state diagram for our single polyelectrolytes .
a double - headed arrow appeared between two states denotes that the structural transition can take place in a typical simulation run at that condition .
= 1 + ( a ) + + ( b ) + + ( c ) + + ( a ) + + ( b ) + + ( c ) +
|
conformation of single polyelectrolytes in tetravalent salt solutions is investigated under the framework of a coarse - grained model , using langevin dynamics simulations .
the chain size , studied by the radius of gyration , shows three different variational behaviors with salt concentration , depending on the chain stiffness . according to the size variations , polyelectrolytes of fixed chain length
are classified into three categories : ( 1 ) flexible chain , for which the variation shows a curve similar to a tilted letter l ; ( 2 ) semiflexible chain , whose curve looks resemble of the letter u ; ( 3 ) rigid chain , for which the curve is a straight line . the worm - like chain model with persistence length predicted by the odijk - skolnick - fixman theory
is found to be able to qualitatively describe the end - to - end distance at low salt concentration , not only for semiflexible and rigid chains but also for flexible chain . in a low - salt region
, a flexible polyelectrolyte extends more significantly than a semiflexible chain , in reference of the size of their uncharged counterparts , and in a high - salt region , regardless of chain stiffness , a chain attains a dimension comparable to that of its neutral polymer .
the chain stiffness influences both the local and the global chain structures .
a flexible chain exhibits a zigzagged local structure in the presence of salt ions and the condensed structure is a disordered , random globule .
a semiflexible chain is locally smooth , and the condensed structure is orderly packed , taking a form such as hairpin or toroid .
moreover , the chain stiffness can also affect the nature of the coil - globule transition .
the transition is occurred in a discrete manner for semiflexible chain , whereas in a continuous way for flexible chain .
this discrete feature is happened not only at low salt concentration when a semiflexible chain is collapsed , but also at high salt concentration when the collapsed chain is reexpanded .
at the end , the effects of chain stiffness and salt concentration on the conformation of single polyelectrolytes are summarized in a schematic state diagram .
|
an ionization process is removal of one or more electrons from a quantum system , such as atom , molecule , quantum heterostructure etc .
, in any manners .
if the ionization process is induced by a photon , this process is called photoionization .
the photoionization of atomic systems provides the opportunity to investigate the dynamic interplay among many body electron - electron correlations and relativistic effects.@xcite it is the simplest process giving detailed information on the atomic and molecular structure.@xcite numerous experimental and theoretical studies on photoionization cross section of different atoms have been published by many authors.@xcite photoionization cross sections are of great significance due to its numerous applications in space research , astrophysics , radiation protection , laser designing , controlled thermonuclear research and different types of laboratory plasma.@xcite on the other hand , because quantum dot ( qd ) heterostructures exhibit atomic properties such as energy levels , density of states etc .
, they are called as artificial atoms.@xcite the investigation of photoionization cross section of a hydrogenic impurity in a qd is very important especially for different optoelectronic device applications such as quantum dot infrared photodetectors .
although there are no experimental studies , the photoionization cross section of a hydrogenic impurity in low - dimensional quantum structures has been studied theoretically by many authors.@xcite the cross section strongly depends on the impurity binding energy and its wave function.@xcite the photoionization cross section is still studied extensively under different physical parameters ( i.e electric field , magnetic field , hydrostatic pressure ) in low - dimensional quantum nano structures .
recently , ahin@xcite has investigated the photoionization cross section of both @xmath0 and @xmath1 impurities in a qd for an on - center impurity as a function of the dot radii and the normalized photon energies .
he demonstrated that the cross section drastically depends on the dot size , the photon energy and the number of electrons in the qd .
barseghyan et al.@xcite studied the behavior of the binding energy and the photoionization cross - section of a donor impurity in a cylindrical - shape gaas / algaas quantum dot , under the effects of hydrostatic pressure and applied electric and magnetic fields in - growth direction .
they showed that when the hydrostatic pressure is increased , the photoionization cross section grows .
ham and spector@xcite analyzed the dependence of the photoionization cross section on the energy and polarization of the photons in a spherical qd as a function of dot radius and impurity locations for both infinite and finite potential barriers .
they found that the cross section is independent of the polarization of the photons for an on - center impurity , while it depends on the polarization of the photon field for an off - center one .
sajeev and moiseyev@xcite investigated the autoionization and photoionization in two - electron spherical quantum dots .
baldea and koppel@xcite showed that the photoionization is a valuable method , which can be used along with transport measurements , to study single electron transistors .
most recently , lin and ho@xcite have studied the photoionization cross section of a hydrogen impurity in spherical quantum dots which are modelled by finite oscillator and gaussian potentials .
all of these studies have been reported for a single core / shell qd .
the production of the spherical multi - layered quantum dot ( mlqd ) has become possible using wet chemical techniques.@xcite the electronic properties of multilayered spherical quantum dots under various physical effects have been investigated by some authors.@xcite these studies are related to electronic energy level and impurity binding energy calculations .
it is seen from these studies that the energy levels and impurity binding energies are very sensitive to the layer thickness . until now , to the best of our knowledge , although the electronic structure has been studied by many authors , the photoionization cross section properties have not been investigated in a mlqd .
the investigation of photoionization cross section in mlqd may give rise to an opportunity for producing a new generation of quantum devices such as laser source , photodetectors , single electron transistor etc .
for example , a single electron transistor can work as depending on the photoionization cross section.@xcite the controllability of the shell thicknesses may provide different advantages in the production of these kind of devices working based on the photoionization cross section .
the main aim of this work is to calculate the photoionization cross section of a hydrogenic impurity in a spherical multi - layered semiconductor quantum dot , and to investigate the dependence of the photoionization on the layer thickness . for this purpose , the ground and first excited energy levels and corresponding wave functions
are calculated by solving the schrdinger equation with shooting method for an electron confined in a mlqd .
the variations of the impurity binding energies and the photoionization cross section are examined as a function of shell thicknesses and photon energies .
the rest of this paper is organized as follows : in the next section , the model and calculations are presented .
the calculation results and discussion of their probable reasons are given in section iii .
a brief conclusion is presented in the last section .
in this study , a spherically symmetric core / shell / well / shell quantum dot structure is considered . as seen from fig .
[ fig:1 ] , gaas core is a sphere of radius @xmath2 .
the well region is that another gaas spherical shell of inner radius @xmath3 and outer radius @xmath4 is concentric with the core region .
the gap between the core and well regions is filled with algaas as a shell ( barrier ) zone which isolates the core and well regions and also allows electron to tunnel between these regions .
the shell thickness is @xmath5 and the well region thickness is @xmath6 . all structure is embedded in an algaas matrix material . in the effective mass approximation , for a spherically symmetric quantum dot
, the single - particle schrdinger equation is given as @xmath7r_{n,\ell } \left ( r \right ) = \varepsilon _ { n,\ell } r_{n,\ell } \left ( r \right),\ ] ] where @xmath8 is reduced planck constant , @xmath9 is the position - dependent electron s effective mass , @xmath10 is charge of the impurity , @xmath11 is the position dependent dielectric constant , @xmath12 is the position of the impurity , @xmath13 is the angular momentum quantum number , @xmath14 is the finite confining potential , @xmath15 is the electron energy eigenvalue , and @xmath16 is the radial wave functions of the electron .
it should be noted that , in the case without the impurity , the charge of the impurity is @xmath17 and in the existence of the impurity , it is @xmath18 . in the calculations ,
we have taken into consideration an on - center impurity and hence the @xmath19 .
the mathematical expression of the confining potential is @xmath20 schematic representation and potential profile of a core / shell / well / shell qd.,width=326 ] in order to determine the single particle energy levels and corresponding wave functions , eq.([eq1 ] ) is full numerically solved by using the shooting method .
as is well known , this technique converts an eigenvalue problem to an initial value problem . for this purpose
, hamiltonian operator is discretized on a uniform radial mesh in 1d using the finite differences , then eq.([eq1 ] ) can be reduced to an initial value equation . here , @xmath21 width between two mesh points is chosen as 0.005 .
all details of this method can be found in ref .. the photoionization process is an optical transition that takes place from the impurity ground state as the initial state to the conduction subbands ( continuum ) , which starts above the confining potential of the dot .
the sufficient energy is required in order for the transition to occur@xcite .
the photoionization cross section can be defined as the ionization probability of the electrons from the bound state under an external optical excitation .
it is strongly dependent on the confinement potential and the photon energy .
the excitation energy depends on the photoionization cross section associated with an impurity and , as in the bulk case , it can be calculated in frame of the fermi s golden rule and the dipole approximation as following@xcite @xmath22\frac{{4\pi ^2 } } { 3}\beta _ { fs } \hbar
\omega \sum\limits_f { \left| { \left\langle { \psi _ i \left| \textbf{r } \right|\psi _ f } \right\rangle } \right| } ^2 \delta \left ( { e_f - e_i - \hbar \omega } \right),\ ] ] where @xmath23 is the refractive index of the semiconductor , @xmath24 is the dielectric constant of the medium , @xmath25/ @xmath26 is the fine structure constant , and @xmath27 is the photon energy .
@xmath28 is the ratio of the effective electric field @xmath29 of the incoming photon and average field @xmath30 in the medium@xcite .
@xmath31 is the matrix element between the initial and final states of the dipole moment of the impurity . @xmath32 and @xmath33 are the wave functions of the final and initial states , and @xmath34 and @xmath35 are the corresponding energy eigenvalues of these states , respectively . these eigenvalues and eigenvectors are determined by the shooting method . in a spherical qd , the selection rules ( @xmath36 ) determine the final state of the electron after the impurity photoionization .
the 1@xmath37 impurity state is taken as ground level for @xmath38 and the 1@xmath39 level of one electron system without the impurity is taken as the final state for @xmath38 as similar with previous studies.@xcite in order to calculate the numerical values of the photoionization cross - section given by eq.([eq4 ] ) , @xmath28 is taken as approximately unity because the calculation of this quantity is very difficult and it has no effect on the photoionization cross - section shape@xcite .
also , the initial and final state wave functions ( @xmath33 and @xmath32 ) are determined by the multiplication of the radial wave function with the spherical harmonics i.e. @xmath40 . here , @xmath41 is the spherical harmonics .
in addition , the @xmath42-function in eq.([eq4 ] ) is replaced by a narrow lorentzian by means of , @xmath43}}.\ ] ] here , @xmath44 is the hydrogenic impurity linewidth and taken as 0.1@xmath45 .
the atomic units have been used throughout the calculations , where @xmath46 .
effective bohr radius is @xmath47 and effective rydberg energy is @xmath48 mev . the confinement potential has been taken as @xmath49 mev .
the material parameters have been taken as @xmath50 , @xmath51 , @xmath52 , @xmath53 .
also the effective masses of electrons inside gaas and algaas are @xmath54 and @xmath55 , and the dielectric constants are @xmath56 and @xmath57 , respectively .
the position - dependent effective mass and the dielectric constant may be defined as follows@xcite @xmath58 the binding energy of a neutral donor impurity ( @xmath0 ) in a qd is given as @xcite @xmath59 here , @xmath60 is ground state energy of the single electron qd without the impurity and @xmath61 is that of the neutral donor impurity .
impurity binding energy of a multi - layered qd as a function of core radius , @xmath62 , for constant @xmath63 and @xmath64.,width=288 ] figure [ fig:2 ] shows the variation of the binding energy of the @xmath0 impurity as a function of the core radius of the mlqd , @xmath62 . here , the shell thickness is @xmath65 and the well width is @xmath66 . as seen from the figure , in small core radius , the binding energy increases rapidly with increasing @xmath62 .
it reaches a maximum value at about @xmath67 and decreases smoothly with further increasing of the @xmath62 and goes to a constant value .
this behavior is rather similar to the binding energy of a neutral hydrogenic donor impurity in a single qd.@xcite ( color online ) impurity binding energy of a core / shell / well / shell qd as a function of shell thickness for constant well width and two different core radii.,width=288 ] figure [ fig:3 ] shows the binding energy of the impurity as a function of the shell thickness for @xmath66 , and two different core radii , @xmath68 and @xmath69 . as seen from the figure , the behaviors of the binding energies are completely different from the each other . when the core radius is @xmath70 , the binding energy decreases with increasing shell thickness . on the other hand , when the core radius is @xmath67 , the binding energy increases firstly and remains at a constant value for almost @xmath71 .
these changes are explained as follows : in the case of small core radius , the electron is confined in the well region for both cases without ( @xmath17 ) and with ( @xmath18 ) the impurity because the core region is rather small .
however , in the presence of the impurity , the electron feels the effect of the impurity at smaller shell thickness and hence the energy level is pulled down because of the attractive coulomb potential of the impurity . as a result of this , the binding energy becomes larger at small shell thickness .
in contrast , while the shell thickness increases , the attractive effect of the impurity on the electron decreases so that , the electron is less affected from the impurity .
these behaviors are clearly seen from the left panel of fig.[fig:4 ] .
accordingly , the binding energy decreases with increasing @xmath63 . by further increase in the @xmath63 ,
the electron is confined completely in the well region for both @xmath17 and @xmath18 cases , and it is insubstantially affected from the impurity . in the case of large core radius , while the electron is confined in both the core and well regions for @xmath17 case , it is almost confined in the core region for @xmath18 case . as a result , the energy difference is relatively small .
when the shell thickness , @xmath63 , is larger , the electron is completely confined in the core region for both @xmath17 and @xmath18 cases and the effect of the well region vanishes as seen from the right panel of fig.[fig:4 ] .
hence , the attractive coulomb effect of the impurity becomes more significant which results in the larger binding energy as seen in fig .
[ fig:3 ] .
( color online ) ground state energies and probability densities at same @xmath66 for @xmath17 and @xmath18 cases and different layer thickness .
the other layer parameters are given on the figure.,width=326 ] ( color online ) variation of the impurity binding energy as a function of the well widths , @xmath64 , for @xmath72 , and @xmath73 and @xmath74.,width=288 ] we expect that the impurity binding energy will be affected drastically from the width of well region . in fig .
[ fig:5 ] , the impurity binding energies are plotted as a function of the well width , @xmath64 , for two different shell thickness at @xmath72 .
when we look at the figure , we see similar treatments in the binding energies for both shell thicknesses .
nevertheless , we observe that the variation of the binding energy is smoother for @xmath75 as compared with case of @xmath65 .
the physical reasons of these treatments can be explained as follows : for the first value of the shell thickness , in spite of the fact that the electron is confined inside the core region for both @xmath17 and @xmath18 cases , the well region influences the energy levels even for small values of the well width .
moreover , the energy difference between @xmath17 and @xmath18 cases is larger , so that the binding energy increases .
when the well region becomes much larger , the electron tunnels easily to the well region and the energy levels approaches to each other .
this situation is shown in left panel of fig.[fig:6 ] . on the other hand ,
when the shell thickness is @xmath65 , the tunneling is not so easy and the electron is confined almost completely in the core region in comparison with the @xmath75 case as seen in right panel of fig.[fig:6 ] . as opposed to the @xmath75 case , when the well region is much larger , the electron tunnels completely to the well region and as depending on this , the decreasing in the binding energy becomes sharper with increasing shell thickness as observed in fig .
[ fig:5 ] .
( color online ) ground state energies and probability densities at same @xmath72 for @xmath17 and @xmath18 cases and different layer thickness.,width=326 ] the photoionization cross section of a donor impurity is an important phenomena especially for device applications of quantum heterostructures . if the photon energy is equal to the energy difference between the initial and the final states ( i.e. resonant energy , @xmath76 ) , all terms of eq.[eq3 ] , except that photon energy and dipole matrix element , contribute as a constant to the photoionization cross section . in this resonant photon energy case ,
the lorentz function becomes also another constant as @xmath77 .
on the other hand , the energy difference between the initial and the final states and the overlapping of their wave functions ( hence the dipole matrix element ) play an effective role on the shape of the photoionization cross section.@xcite the photoionization cross section of the hydrogenic donor impurity vs. core radius at constant @xmath63 and @xmath64.,width=288 ] fig .
[ fig:7 ] shows the variation of the peak values of the photoionization cross section with the core radius , @xmath62 , for constant shell and well widths , i.e. @xmath65 and @xmath66 .
these peak values correspond to the resonant photon energies .
as seen from the figure , while the photoionization has very large value at @xmath78 , it decreases too sharp with increasing @xmath62 and reaches the its smallest value at @xmath72 .
when the core radius is @xmath79 , it increases instantly and remains stable with further increasing of the @xmath62 .
the physical reason of this treatment is as follows : at very small @xmath62 values , the wave functions of both ground and excited states are localized in the well region as similar to fig .
[ fig:4 ] and hence the overlapping of the wave functions becomes larger . as a result of this
, the photoionization cross section is also larger.in this situation , the photoionization takes place in the well region and this may be important for a photodetector application which is sensitive to higher photon energies ( blue shift ) relatively , because the energy levels become larger , in smaller core radii.the increase in the core radius causes a decrease in the overlapping of the wave functions because the impurity wave function is confined in the core region while the excited states wave function is still confined in the well region .
therefore , the photoionization cross section has small values .
with further increasing of the core radius , both ground and excited states wave functions are localized in the core region and hence the overlapping of the wave functions becomes large and as depending on this , the cross section is also to be maximum . in this case , the photoionization occurs in the core region .
this situation may be important to produce a photodetector which is sensitive to lower energies ( red shift ) .
( color online ) the variation of the photoionization cross section of the impurity depending on shell thickness for two different core radii and constant @xmath64.,width=288 ] figure [ fig:8 ] shows the variation of the photoionization cross section as a function of the shell thickness , @xmath63 , for two different values of the core radius , @xmath62 , and the same well thickness , @xmath66 .
when we look at the figure , we see that the cross section increases with increasing shell thickness for @xmath70 . on the other hand , for @xmath80 , in contrast with the previous case , the cross section exhibits completely different behavior and it decreases rapidly with increasing @xmath63 .
similar treatments are also observed for @xmath81 values .
when @xmath70 , the wave functions of ground and excited states are confined in the well region and the finding probability in the core region vanishes almost for large values of the shell thickness .
therefore , the overlap between the wave functions becomes much larger in the well region as similar to the left panel of fig .
[ fig:4 ] .
this is the reason of an increase in the photoionization cross section which happens in the well region .
when @xmath80 , ground state wave function is confined in the core region because of the attractive coulomb potential of the impurity . at small values of @xmath63 , the tunneling of the excited state wave function to the core region is much more . in this case , the overlapping of the wave functions becomes larger and hence the photoionization cross section becomes greater .
when the barrier thickness increases , the overlapping of the wave functions decreases because of the localizing of excited state wave function in the well region and therefore the cross section becomes less .
we conclude that the sensitivity of a device can be also tuned with the shell thickness of mlqd to a narrower or larger region .
( color online ) the variation of the photoionization cross section of the impurity as a function of well width for two different shell thickness and constant core radius.,width=288 ] the variation of the photoionization cross section with the well width , @xmath64 , is seen in fig .
[ fig:9 ] for @xmath72 and two different shell thickness , @xmath73 and @xmath74 .
as seen from the figure , in small well widths , the photoionization cross section remains a bit stable and then it decreases rapidly with increasing well widths and reaches the minimum values at about @xmath82 . the minimum value of the photoionization for @xmath65 becomes smaller than that for @xmath75 .
the cross section increases again for both cases with further increasing of the @xmath64 .
we can explain this behavior as follows : in small @xmath64 values , namely narrow well region case , the wave functions of ground and excited states are localized in the core region and therefore the overlapping of them becomes greater .
the photoionization occurs in the core region .
the excited state wave function starts to tunnel through the well region with increasing well width and it is completely localized in this region when the well width is about @xmath82 as seen from fig .
[ fig:10 ] . in this case , the overlapping of the wave functions is minimum and hence , the photoionization cross section is very small .
this decreasing is much more for @xmath65 , because the tunneling of excited state wave function to the core region is smaller as shown at bottom panel of fig .
[ fig:10 ] . by further increase of the well width
, the ground state wave function also begins to tunnel to the well region and as a result of this , the overlapping of the wave functions increases again .
when @xmath83 , both of the wave functions are localized in the well region and the cross section is to be higher .
the photoionization cross section occurs in the well region anymore .
( color online ) localization and energy cases for ground state of z=1 ( black lines ) and excited state of z=0 ( red lines ) .
the left panel for @xmath75 and the right panel for @xmath65.,width=288 ] the investigation of the photoionization cross section as depending on the photon energy is another important process .
figure [ fig:11 ] shows the variation of the photoionization cross section of the neutral donor impurity as a function of the normalized photon energy @xmath84 for different core radii , @xmath62 , and shell thickness , @xmath63 .
as seen from the figure , the maximum photoionization corresponds to the threshold photon energy for all @xmath62 and @xmath63 values and it decreases with increasing photon energy .
these general trends are quite similar to those in single qds.@xcite also , the cross sections go towards lower values with increasing @xmath62 and decreasing @xmath63 .
when @xmath62 is small , both ground and excited state wave functions are confined in the well region which results in a strong overlap between the wave functions and therefore in a larger photoionization cross section .
the increase in the core radius , which corresponds to the slimming of the shell wall , leads to a possibility of tunneling of ground state wave function through the core region due to the attractive coulomb potential of the impurity .
so this possibility reduces the value of dipole matrix element and also as depending on this , the cross section becomes smaller with increasing normalized photon energy .
( color online ) photoionization cross section of the hydrogenic donor impurity as a function of normalized photon energy for different @xmath62 , and @xmath63 and constant @xmath64.,width=288 ] ( color online ) photoionization cross section of the hydrogenic donor impurity as a function of normalized photon energy for different @xmath64 and constant @xmath62 , and @xmath63.,width=288 ] in fig .
[ fig:12 ] , variation of the photoionization cross section with normalized photon energy is shown for different well widths at @xmath67 and @xmath85 . as seen from the figure
, the maximum photoionization cross section corresponds to the threshold energy and decreases with increasing photon energies for all well widths .
when the well width increases , the photoionization cross section increases and this increase is more obvious for @xmath86 .
this is originated from the overlapping of the wave functions in the well region .
that is , the photoionization happens in the well region . as seen from figs .
[ fig:11 ] and [ fig:12 ] , the maximum absorption occurs at the photoionization threshold and the photoionization cross section decreases monotonically from the maximum value with increasing photon energy like the true hydrogenic model .
this transition requires higher photon energies than those in quantum wells and quantum wires.@xcite the reason for this behavior is that the electron is completely confined in all spatial dimensions in qd structures .
this interesting result , which seems to be the characteristic and the signature of the qds , is not observed in quantum well and quantum wire structures.@xcite
in this study , a detailed examination of the electronic structure and photoionization cross section of a donor impurity have been performed for a multi - layered spherical qd . in calculations ,
we have used shooting method with finite difference techniques and effective mass approximation for determining energy eigenvalues and their wave functions .
first , we have calculated the donor binding energies as a function of the layer thickness .
we have found that the donor binding energy strongly depends on the layer thickness .
then , we have computed photoionization cross section of the donor impurity as a function of layer thickness and normalized photon energies .
we have showed that the photoionization cross section can be controlled with adjusting the core radius or layer widths .
these properties can be useful for producing some device applications such as quantum dot infrared photodetectors . although there is no experimental or theoretical reports in the literature to compare our results , general trend of our results are in a good agreement with previous studies of single quantum dots .
we believe that this study will be rather useful and will contribute to the understanding of the photoionization cross section of mlqds .
we hope that this study will also stimulate both experimental and theoretical investigations of the photoionization cross section of mlqds .
00 h. kjeldsen , _ absolute photoionization cross sections : measurements and applications _ , dissertation , department of physics and astronomy , university of aarhus , denmark ( 2006 ) .
m. rafiq , _ photoionization cross section and oscillator strength measurements of the excited states of sodium and magnesium _ ,
ph.d thesis , department of physics , quaid - i - azam university , pakistan ( 2008 ) .
m. saleem , n. amin , s. hussain , m. rafiq , s. mahmood , and m. a. baig , eur .
j. d * 38 * , 377 ( 2006 ) .
h. kjeldsen , j. phys .
b : at . mol .
. phys . * 39 * , r325 ( 2006 ) . v. schmidt , reports on progress in physics * 55 * , 1483 ( 1992 ) .
m. ahin , phys .
b * 77 * , 045317 ( 2008 ) . m.g .
barseghyan , a.a .
kirakosyan , and c.a .
duque , eur .
j. b * 72 * , 521 ( 2009 ) .
h. ham and h. n. spector , j. appl . phys . * 93 * , 3900 ( 2003 ) . m. el - said and m. tomak , sol .
commun . * 82 * , 721 ( 1992 ) . m. i. el - kawni and m. tomak , surf .
* 260 * , 319 ( 1992 ) .
k. f. ilaiwi and m. el - said , phys .
( b ) * 187 * , 93 ( 1995 ) .
a. sali , h. satori , m. fliyou , and h. loumrhari , phys .
( b ) * 232 * , 209 ( 2002 ) . h. ham and c. j. lee , j. appl
96 * , 335 ( 2004 ) .
j. d. correa , o. cepeda - giraldo , n. porras - montenegro , and c. a. duque , phys .
( b ) * 241 * , 3311 ( 2004 ) .
y. sajeev and n. moiseyev , phys .
b * 78 * , 075316 ( 2008 ) .
i. baldea and h. koppel , phys .
b * 79 * , 165317 ( 2009 ) . c. y. lin and y. k. ho , phys . rev .
a * 84 * , 023407 ( 2011 ) .
d. dorfs and a. eychmller , nano letters , * 1 * , 663 ( 2001 ) .
a. mews , a. eychmller , m. giersig , d. schooss , and h. weller , j. chem . phys . * 82 * , 552 ( 1994 ) . f. k. boz , s. aktas , a. bilekkaya , and s. e. okan , appl . surf .
science * 255 * , 6561 ( 2009 ) .
f. k. boz , s. aktas , a. bilekkaya , and s. e. okan , appl .
surf . science * 256 * , 3832 ( 2010 ) .
s. aktas and f. k. boz , physica e * 40 * , 753 ( 2008 ) .
p. harrison , 2009 , _ quantum wells , wires and dots : theoretical and computational physics _ ,
3rd edition , john wiley , chippenham , great britain .
m. lax , proceedings of the 1954 atlantic city conference on photoconductivity , ( wiley , new york 1956 ) .
b. k. ridley , _ quantum process in semiconductors _
( clarendon press , oxford , 1982 ) .
r. buczko and f. bassani , phys .
b * 54 * , 2667 ( 1996 ) .
n. porras - montenegro and s. t. perez - merchancano , phys .
b * 46 * , 9780 ( 1992 ) .
a. sali , m. fliyou , h. satori , and h. loumrhari , j. phys .
solids * 64 * , 31 ( 2003 ) .
|
in this study , we have investigated the photoionization cross section of an on - center hydrogenic impurity in a multi - layered spherical quantum dot .
the electronic energy levels and their wave functions have been determined fully numerically by shooting method .
also , we have calculated the binding energy of the impurity by using these energy values .
the photoionization cross section has also been computed as a function of the layer thickness and normalized photon energies .
we have discussed in detail the possible physical reasons behind the changes in the binding energies and photoionization cross section . it is observed that both the binding energies and the photoionization cross sections depend strongly on the layer thickness and photon energies .
|
at high radio frequencies ( @xmath10ghz ) the bright ( @xmath11mjy ) extragalactic radio population is dominated by blazars ( bl lacs and flat - spectrum radio quasars , fsrqs ) , interpreted as sources whose relativistic jets are directed very close to the line - of - sight ( see , e.g. , blandford & knigl 1979 ) .
the jets are collimated by intense magnetic fields .
the blazar radio emission is synchrotron radiation from a series of blobs along the jet .
the observed flux densities are blue - shifted and greatly boosted by relativistic effects . in this scenario ,
the flat radio spectra are interpreted as the combination of components self - absorbed up to different frequencies .
the spectra steepen when the emission becomes optically thin .
young and compact objects are characterised by spectra peaking at high frequencies while older sources are more extended and have spectra steepening at lower frequencies .
occasionally the spectra may be dominated by a single flaring blob .
measurements of polarimetric properties of radio - loud active galactic nuclei ( agn ) at centimetric and millimetric wavelengths carry information on magnetic fields and on the plasma in the inner , unresolved regions of their relativistic jets . even if synchrotron radiation can be up to @xmath12 polarized , the polarization degree of extragalactic sources is rarely observed to be higher than @xmath13 ; the median values are @xmath14 at @xmath3ghz ( massardi et al .
this is the result of several depolarisation effects .
the measured polarised flux is an average of the emission from sub - structures with different orientations of the magnetic field .
further depolarisation may be due to differential faraday rotation that may be amplified by its frequency dependence within the bandwidth .
as a consequence , polarimetric observations generally require high sensitivities and a detailed knowledge of instrumental properties to perform accurate calibration ( see massardi et al .
2016a ) , a delicate procedure at high frequencies .
for these reasons polarimetric surveys of large source samples at @xmath15ghz have become possible only recently thanks to the advent of large bandwidth instruments ( see galluzzi et al .
2016 for an up - to - date summary of the available observations ) . to set the stage we give here an overview of the main high frequency polarisation surveys in the table [ tab : polcom ] , reporting the detection rates and median polarisation fractions found at the frequency @xmath16ghz .
[ tab : polcom ] lccccl & & & & & + & & & & & + & & & & & + & & & & & + + & & & mjy . ] & & + & & & & & + & & & & & + & & & & & + & & & considering the sub - sample of point - like objects . ] & & + & & & ghz observations . ] & & + these studies did not detect any statistically significant relationship between polarisation fraction and total flux density ( but all covered a limited flux density range ) or with frequency . on the contrary , various studies of samples mostly selected at @xmath17ghz and covering broader flux density ranges , reported an increase of the polarisation degree with decreasing flux density and with increasing frequency ( mesa et al .
2002 ; tucci et al . 2004
; taylor et al .
2007 ; grant et al . 2010 ; subrahmanyan et al .
these results were however challenged by hales et al .
( 2014 ) . on the whole ,
our understanding of the high radio frequency polarisation properties of extragalactic sources is still limited .
this has motivated the new high - sensitivity ( @xmath0mjy , a factor @xmath18 better than the at20 g sensitivity ) and nearly simultaneous , multi - frequency polarimetric observations over a wide spectral range ( @xmath1ghz ) whose results are reported here .
we observed with the atca the _
planck_-atca coeval observations ( paco ) `` faint sample '' ( massardi et al .
2016b ) , a complete sample of at20 g sources with @xmath19mjy .
the instrumental capabilities were exploited to reconstruct the polarimetric spectral profile to unprecedented detail .
high frequency observations are easily affected by spectral , detection and variability - related biases . the present work , owing to the completeness of the sample , the high sensitivity and the close - in - time multi - frequency characterisation , allows us to minimise all these effects .
thus , we provide a new statistical assessment of the polarimetric properties of high frequency extragalactic radio sources , analysing and describing their spectral and polarisation behaviour with frequency .
moreover , exploiting also the at20 g and paco observations in total intensity , we investigate the source variability for @xmath8 and @xmath20yr time lags at several frequencies up to @xmath21ghz .
multi - frequency and multi - epoch polarimetric observations of radio sources are useful cosmological tools for several reasons .
radio sources are the most relevant foreground contaminants of the cosmic microwave background ( cmb ) polarisation maps ( on scales of up to @xmath22arcmin ) in the relatively clean @xmath23ghz frequency range ( massardi et al .
a proper characterisation of the radio source contribution to the power spectrum in polarisation is essential for a precise assessment of the lensing b - mode signal ( hanson et al .
2013 , polarbear : ade et al .
this , in turn , is essential for the detection and the characterisation of the power spectrum of the primordial b - modes associated with the stochastic background of gravitational waves , the most ambitious goal of current and future cmb projects ( prism collaboration 2014 ) . because of the broad variety of polarized emission spectra , extrapolations from low frequencies ( @xmath24ghz ) are inadequate to model the radio source contribution in cmb polarisation maps ( huffenberger et al .
. at the same time , radio sources are the privileged calibrators both of the polarized intensity and of the polarisation angle for cmb experiments .
the systematic errors due to inaccuracies in the calibration of the polarisation angle are becoming the limiting factor for cmb polarization experiments ( e.g. kaufman et al .
unfortunately , the number of compact , bright , highly polarised and stable enough extragalactic sources , suitable for accurately calibrating cmb polarization maps are rare .
one of the purposes of our polarisation survey was to identify good candidates .
the paper is organised as follows . in section [ obs ]
we briefly present the paco project , its main results and the polarimetric observations of the selected sub - sample held in september 2014 with atca . in section
[ sec : maths ] we describe the data reduction . in section [ dataanaly ]
we discuss the data analysis and the spectral behaviours in total intensity and polarisation . in section [ variability ]
we address the variability properties for our sample .
finally , in section [ discusseconcl ] we draw our conclusions .
the paco project yielded observations of @xmath25 at20 g sources in @xmath26 epochs between july 2009 and august 2010 in the @xmath27ghz frequency range with atca , nearly simultaneously ( within @xmath28 days ) with the _ planck _ observations ( massardi et al . 2016b ) .
the main goal of the project was to characterize , in combination with _
data , the total intensity spectra and their variability over a wide frequency range ( at least from @xmath29ghz to @xmath30ghz , but up to @xmath31ghz for some sources ) .
a multi - frequency observing run was dedicated to polarimetry in september 2014 with atca .
the sample was extracted from the `` faint '' paco sample ( @xmath32mjy ) selecting sources in the southern ecliptic pole region ( ecliptic latitude @xmath33 ) , where the satellite scanning strategy implies a better sensitivity .
the @xmath34 extended sources in the sample ( i.e. sources larger than the paco resolution of @xmath35arcsec ) were left out as the techniques to extract flux densities used in the paco project underestimate their flux densities . the final complete sample comprises @xmath36 compact sources with @xmath32mjy in the at20 g catalogue . the spectral setup consisted in @xmath37 sets of @xmath38ghz cabb ( compact array broadband backend ) bands centred at @xmath39ghz , @xmath40ghz and @xmath41ghz , similar to those for the paco program .
a total of @xmath42hr were allocated in three slots of @xmath43hr each ( including overheads and calibration ) dedicated to each of the three bands over a period of least three days .
each object was observed for @xmath44min in two @xmath45min scans at different hour angles for each frequency .
weather conditions were very good during all the campaign .
the array configuration was h214 , a hybrid ( antenna are displayed also along the n - s direction ) compact array with a nominal spatial resolution ranging from @xmath46 to @xmath29arcsec in the @xmath1ghz interval if we consider only the @xmath34 most packed antennas .
the resolution goes down to @xmath47arcsec considering also the longest baselines with the sixth antenna , but with a lower sensitivity . adopting a hybrid configuration results in a more homogeneous @xmath48-plane coverage of the largest spatial scales in less time .
this results in a better imaging with respect to a non - hybrid solution , at least for our sample of mostly point - like objects .
the data reduction was done using the miriad software package ( sault , teuben & wright 1995 ) treating each frequency separately , as indicated in the atca user s guide . while loading the data , we corrected for the time - dependent instrumental xy - phase variation , exploiting the known signal injected from a noise diode mounted in one of the feeds of each antenna .
the reference flux density calibrator ( at all frequencies ) was the source pks1934 - 638 , a bright ( point - like ) gigahertz peaked - spectrum ( gps ) radio galaxy , stable , unpolarised ( at least below @xmath49ghz ) , whose model ( see reynolds 1994 ; sault 2003 ; partridge et al .
2016 ) is encoded into miriad itself .
it is the only known source with all these characteristics in the southern sky .
once the calibration tables were derived , all solutions were ingested in the code for flux density extraction .
to better characterize the source spectra , we decided to split each @xmath50ghz - wide frequency band in sub - bands .
each of them was calibrated separately . for total intensity
we split each band into @xmath51mhz - wide sub - bands , as done in the past analyses for the paco project . for polarised flux densities we split bands in only @xmath52 sub - bands to limit the @xmath53 degradation in sensitivity .
flux densities were estimated via the miriad task uvflux .
our sources are known to exhibit linear polarisation ( up to @xmath13 , massardi et al .
2008 , 2013 ) , defined by the @xmath54 and @xmath55 stokes parameters .
observations of the circular polarisation of extragalactic radio sources have demonstrated that it is very low , generally below @xmath56 , at least one order of magnitude lower than the linear polarisation ( rayner et al . 2000 ) , hence generally undetectable for our sources , given the @xmath57mjy sensitivity .
hence , the rms @xmath58 ( of the retrieved @xmath59 stokes parameter ) is used as a noise estimator of the total intensity flux density @xmath60 . the polarised emission , @xmath61 , is given by the simple relation : @xmath62 where the @xmath63 term removes the noise bias on @xmath61 under the assumption of equal noise values , @xmath64 , for the @xmath54 and @xmath55 parameters ( e.g. wardle & kronberg 1974 ) .
the polarisation angle @xmath65 and fraction @xmath66 ( usually in terms of a percentage ) are : @xmath67 assuming gaussian noise , the error scales as @xmath68 , @xmath69 being the number of correlations at a given @xmath70 .
a suitable estimate of the total intensity error is the sum in quadrature of @xmath58 with a systematic term , mainly accounting for the calibration uncertainty .
based on the past experience with paco observations and on a comparison between flux densities obtained from different calibrators , the calibration error amounts to @xmath14 of the @xmath60 flux density , giving for @xmath71 , the global error on i : @xmath72 the error , @xmath73 , on the polarised flux density , @xmath61 , can be derived from the eq .
( [ equ : polfluden ] ) .
a conservative @xmath13 error in the polarisation calibration is adopted in this case .
then , we have : @xmath74 @xmath75 being the rms errors on the stokes parameters @xmath54 and @xmath55 , respectively .
this estimate of the calibration error is consistent with the differences between flux densities obtained using two different calibrators for reducing our data . from error propagation
, the global error on the polarisation fraction is : @xmath76 and that on @xmath65 is : @xmath77 where , again , the calibration error on the polarisation angle , @xmath78 , is added in quadrature . under the hypothesis that the calibration error affects equally the @xmath54 and @xmath55 parameters ( @xmath79 ) , @xmath80 gives @xmath81 .
finally : @xmath82 implying @xmath83 . this estimation turns out to be consistent with differences in the polarisation angle obtained by using different calibrators . to check our assumption of target compactness at all frequencies and the flux density extraction we created maps for each stokes parameter by means of the standard miriad procedure
. we adopted the natural weighting , the standard for point sources , to ensure the lowest noise level .
we used multi - frequency synthesis imaging and a standard hgbom algorithm .
the detection threshold was set at @xmath5 in both @xmath60 and @xmath61 .
the @xmath84 was derived from the @xmath60 image region with no emission .
we use the task uvflux for flux density measurements ( except for a few cases described below ) .
this task is designed for point - like sources and works directly on visibilities , minimizing artifacts and phase - instabilities which may affect more image - based measurements . since our sample
excludes extended sources ( at least up to @xmath3ghz ) , we expect that flux densities provided by uvflux and by imstat ( the miriad task for flux density estimation from imaging ) show differences no greater than @xmath85 in total intensity ( due to residual phase instabilities in the images ) .
moreover , we assumed a source pointing accurate enough to keep objects in the phase center of the @xmath48-plane .
the latter hypothesis is supported by the fact that the paco catalogue is drawn from the at20 g one , for which all positions are known with an accuracy @xmath86arcsec ( murphy et al .
2010 ) .
afterwards , we compared the imstat image peak ( in mjy / beam ) and the uvflux flux density estimation ( for all the objects for which the uv - coverage allows imaging , i.e. @xmath87 of the cases ) to reveal , in both i and p , whether an extended or displaced ( from the phase centre ) component is present .
the median discrepancies , @xmath88 and @xmath89 , are @xmath90 and @xmath91 , respectively . in total intensity
the relative discrepancy is @xmath92 for @xmath93 of the images .
these large discrepancies are registered for frequencies @xmath94ghz ; they reach maximum values between @xmath95 and @xmath96 for @xmath52 objects ( at20gj080633 - 711217 and at20gj080649 - 610131 , respectively ) at @xmath21ghz . in all the cases the image peak values are higher than the uvflux estimation .
a visual inspection of the images confirms that the sources are barely resolved . in polarisation
the fractional discrepancy is @xmath97 in @xmath98 of the available images and @xmath99 in @xmath100 of the cases . in particular , there is one object , at20gj040848 - 750720 , for which the excess of flux density ( with respect to uvflux ) measured in correspondence of the peak in the image reaches @xmath101 at @xmath102ghz and @xmath103 at @xmath21ghz .
we show images for this source ( see fig . [
fig : j0408 ] ) at all the @xmath34 frequencies at which they could be obtained in total intensity .
polarised emissions are instead displayed by contour levels .
it can be seen that the object is marginally resolved at the lower frequencies . at the higher frequencies a second component appears and the polarised emission mainly comes from it . according to morganti et al .
( 1999 ) at20gj0408 - 7507 is a bright frii radio galaxy at @xmath104 , dominated by two bright lobes .
both lobes have high depolarization , slightly higher in the eastern one . given the good matching between flux densities from imaging and from visibilities for point - like objects ( both in total intensity and in polarisation ) , we decided to integrate over a suitable region on @xmath60 images to recover the total intensity flux densities in case of slightly resolved objects . in case of the flux discrepancies in polarisation ,
since images reveal an emerging point - like component displaced from the phase center , we estimated the flux density by considering the peak of the @xmath61 image . the complete catalogue ( flux densities and fitting parameters , polarisation fractions and angles ) for the observations discussed in the present paper is available online , as supplementary material . ,
@xmath105 , @xmath106 , @xmath107 and @xmath21ghz .
the two - lobe structure is resolved at the higher frequencies .
the eastern lobe appears to be strongly depolarised.,title="fig : " ] , @xmath105 , @xmath106 , @xmath107 and @xmath21ghz .
the two - lobe structure is resolved at the higher frequencies .
the eastern lobe appears to be strongly depolarised.,title="fig : " ] , @xmath105 , @xmath106 , @xmath107 and @xmath21ghz .
the two - lobe structure is resolved at the higher frequencies .
the eastern lobe appears to be strongly depolarised.,title="fig : " ] , @xmath105 , @xmath106 , @xmath107 and @xmath21ghz .
the two - lobe structure is resolved at the higher frequencies .
the eastern lobe appears to be strongly depolarised.,title="fig : " ] , @xmath105 , @xmath106 , @xmath107 and @xmath21ghz .
the two - lobe structure is resolved at the higher frequencies .
the eastern lobe appears to be strongly depolarised.,title="fig : " ] , @xmath105 , @xmath106 , @xmath107 and @xmath21ghz .
the two - lobe structure is resolved at the higher frequencies .
the eastern lobe appears to be strongly depolarised.,title="fig : " ]
we obtained @xmath108 detections in polarisation , with a median error of @xmath109mjy , for @xmath110 of our sources .
the detection rate is nearly uniform across the observed frequencies ( @xmath111 sources detected at @xmath112ghz and @xmath113 detected at @xmath21ghz ) . to check whether our data are affected by intra - band depolarisation at any frequency
we have subdivided each @xmath50ghz - wide band into two @xmath114ghz - wide sub - bands and compared the polarisation degrees measured in the sub - bands . as illustrated by the examples shown in fig .
[ fig : ppoldiag ] no systematic difference was found .
ghz - wide sub - bands of the @xmath50ghz - wide bands at @xmath112ghz ( upper panel ) and at @xmath102ghz ( lower panel ) .
red asterisks , blue @xmath115 signs and green diamonds refer to steep spectrum , flat spectrum and peaked spectrum sources , respectively ( the spectral classification is described in sect . [ sec:4.2 ] ) .
the bisector is shown as a dashed line.,title="fig : " ] ghz - wide sub - bands of the @xmath50ghz - wide bands at @xmath112ghz ( upper panel ) and at @xmath102ghz ( lower panel ) .
red asterisks , blue @xmath115 signs and green diamonds refer to steep spectrum , flat spectrum and peaked spectrum sources , respectively ( the spectral classification is described in sect . [ sec:4.2 ] ) .
the bisector is shown as a dashed line.,title="fig : " ] to properly fit source spectra we start considering a double power law represented by the expression @xmath116 or by a concave version of it ( needed in two cases ) @xmath117 where @xmath118 , @xmath119 , @xmath120 and @xmath121 are free parameters .
this function properly fits the total intensity data in @xmath122 of the cases , confirming what found in previous works ( e.g. , massardi et al .
2016b ) . in two cases
( sources at20gj0546 - 6415 and at20gj0719 - 6218 ) the double power law provided a poor fit and we resorted to a triple power law model which requires 3 additional parameters , i.e. @xmath123 , @xmath124 and @xmath125 : @xmath126 as for fitting procedures in polarisation , we need detections at no less than @xmath127 frequencies ( over a maximum of @xmath128 ) in case of a double power law and at no less than @xmath129 frequencies for a triple power law . if a point source was not detected in one ( or both ) of the split frequency ranges of a band , we used the corresponding non - split detection , when available . given the small fraction ( @xmath130 ) of non - detections we did not consider the upper limits in doing the spectral fits . in only three cases
( at20gj054641 - 641522 , at20gj062524 - 602030 and at20gj075714 - 735308 ) we do not have detections at enough frequencies to get a proper fit . about @xmath131 of our source spectra
could be fitted with either a double power law or by a concave version of it ( needed in @xmath127 cases ) .
the median reduced @xmath132 are @xmath133 and @xmath134 for i and p , respectively .
this seems to confirm that our approach in error estimates is indeed conservative .
figure [ fig : spettri1 ] shows the data , the fitting curves and , when available , the previous paco best epoch ( 2009 - 2010 ) observations in total intensity and the at20 g best epoch ( 2004 - 2008 ) observations in total intensity and in polarisation .
.distribution of sources per spectral type in total intensity and in polarisation .
the row `` na '' refers to the three objects classified in total intensity but missing a spectral fit in polarisation .
the last row reports the total for a given spectral class in total intensity , while the last column does the same in polarisation .
[ cols="<,^,^,^,^,^,^ " , ]
we have presented high sensitivity ( @xmath135mjy ) polarimetric observations of a complete sample of @xmath36 compact extragalactic radio sources drawn from the `` faint '' paco sample .
thanks to the high sensitivity we achieved a high ( @xmath136 ) @xmath108 detection rate in polarisation .
the continuum spectra of over @xmath6 of sources are well fitted by double power laws , both in total intensity and in polarisation . however
, substantial variations of the polarisation degree with frequency were found , implying different spectral shapes in polarisation compared to total intensity . within the spectral range covered by our observations ( @xmath1ghz )
most sources can be classified either as steep - spectrum or as peaked - spectrum , both in total intensity and in polarisation , although less than half of sources have the same classification in the two cases . between @xmath137 and @xmath21ghz
all spectral indices are steep , so that the dichotomy between flat and steep spectra , associated to compact and extended sources , well established at low frequencies , no longer holds at these frequencies .
no significant trend of the fractional polarisation with either flux density or frequency was found , although we caution that the limited ranges covered by our sample and the variety of spectral behaviours may hamper the detection of weak trends . however , steep - spectrum sources show higher polarisation fractions than peaked - spectrum ones at all the observed frequencies .
we found evidence of faraday rotation for @xmath52 sources .
the derived values of the rotation measures are @xmath138 and @xmath139 , respectively .
the mean variability index in total intensity of steep - spectrum sources increases with frequency for a @xmath8 year lag , while no significant trend shows up for peaked - spectrum sources and for the @xmath9 year lag . in polarisation ,
the variability index , that could be computed only for the @xmath9 year lag , is substantially higher than in total intensity and has no significant frequency dependence .
we thank the anonymous referee for useful comments . we acknowledge financial support by the italian _
ministero dellistruzione , universit e ricerca _ through the grant _ progetti premiali 2012-ialma _ ( cup c52i13000140001 ) . partial support by asi / inaf agreement 2014 - 024-r.1 for the _ planck _ lfi activity of phase e2 and by asi through the contract i-022 - 11 - 0 lspe is acknowledged .
we thank the staff at the australia telescope compact array site , narrabri ( nsw ) , for the valuable support they provide in running the telescope and in data reduction .
the australia telescope compact array is part of the australia telescope which is funded by the commonwealth of australia for operation as a national facility managed by csiro .
ab acknowledges support from the european research council under the ec fp7 grant number 280127 .
vc acknowledges dustpedia , a collaborative focused research project supported by the european union under the seventh framework programme ( 2007 - 2013 ) call ( proposal no .
the participating institutions are : cardiff university , uk ; national observatory of athens , greece ; ghent university , belgium ; universit paris sud , france ; national institute for astrophysics , italy and cea ( paris ) , france . lt and lb acknowledges partial financial support from the spanish ministry of economy and competitiveness ( mineco ) , under project aya-2015 - 65887-p .
agudo i. , thum c. , wiesemeyer h. , krichbaum t. p. , 2010
, apjs , 189 , 1 agudo i. , thum c. , gmez j. l. , wiesemeyer h. , 2014 , a&a , 566 , a59 arkin h. , colton r. r. , 1970 , statistical methods , barnes & noble , 5th ed .
battye r. a. , browne i. w. a. , peel m. w. , jackson n. j. , dickinson c. , 2011 , mnras , 413 , 132 blandford r. d. , knigl a. , 1979 , apj , 232 , 34 di serego alighieri s. , 2015 , int
d , 24 , 1530016 galluzzi v. , massardi m. , 2016 , int .
j. of mod .
d , 25 , 1640005 .
grant j. k. , taylor a. r. , stil j. m. , landecker t. l. , kothes r. , ransom r. r. , scott d. , 2010 , apj , 714 , 1689 hales c. a. , norris r. p. , gaensler b. m. , middelberg e. , 2014 , mnras , 440 , 3113 hanson d. et al . , 2013 , prl , 111 , 141301 huffenberger k. m. , et al . , 2015 , apj , 806 , 112 impey c. d. , neugebauer g. , 1988 , aj , 95 , 307 lavalley m. , isobe t. , feigelson e. , 1992 , in worrall d. m. , biemesderfer c. , barnes j. , eds , asp conf .
25 , astronomical data analysis software and systems i. astron .
, san francisco , p. 245 massardi m. , et al .
, 2008 , mnras , 384 , 775 massardi m. , et al . , 2011a , mnras , 412 , 318 massardi m. , bonaldi a. , bonavera l. , lpez - caniego m. , de zotti g. , ekers r. d. , 2011b , mnras , 415 , 1597 massardi m. et al . , 2013 , mnras , 436 , 2915 massardi m. , galluzzi v. , paladino r. , burigana c. , 2016a , int .
j. of mod .
d , 25 , 1640009 .
massardi m. , bonaldi , a. , bonavera l. , de zotti , g. , lpez - caniego , m. , galluzzi , v. , 2016b , mnras , 455(3 ) , 3249 mesa d. , baccigalupi c. , de zotti g. , gregorini l. , mack k .- h .
, vigotti m. , klein u. , 2002 , a&a , 396 , 463 morganti r. , oosterloo t. , tadhunter c. n. , aiudi r. , jones p. , villar - martin m. , 1999 , astron .
140 , 355 - 372 murphy t. et al . , 2010 , mnras , 402
, 2403 partridge b. , lpez - caniego m. , perley r. a. , stevens j. , butler b. j. , rocha g. , walter b. , zacchei a. , 2016 , apj , 821 , 61 polarbear : ade p. a. r. et al .
, 2014 , phys .
113 , 021301 polarbear : ade p. a. r. et al .
, 2015 , phys .
d 92 , 123509 prism collaboration 2014 , jcap , 2 , 006 rayner d. p. , norris r. p. , sault r. j. , 2000 , mnras , 319 , 484 reynolds j. e. , 1994 , atnf memo , at/39.3/040 ricci r. , prandoni i. , gruppioni c. , sault r. j. , de zotti g. , 2004 , a&a , 415 , 549 sadler e. m. et al .
, 2006 , mnras , 371 , 898 sajina a. , partridge b. , evans t. , stefl s. , vechik n. , myers s. , dicker s. , korngut p. , 2011 ,
apj , 732 , 45 sault r. j. , teuben p. j. , wright m. c. h. , 1995 , in r. a. shaw , h. e. payne , & j. j. e. hayes ed .
, astronomical data analysis software and systems iv vol .
77 of astronomical society of the pacific conference series , a retrospective view of miriad .
pp 433 sault r. j. , atnf memo , at/39.3/124 simmons j. f. l. , stewart b. g. , 1985 , a&a , 142 , 100 subrahmanyan r. , ekers r. d. , saripalli l. , sadler e. m. , 2010 , mnras , 402 , 2792 taylor a. r. et al . , 2007 , apj , 666 , 201 tucci m. , martnez - gonzlez , e. , toffolatti l. , gonzlez - nuevo j. , de zotti , g. , 2004 , mnras , 349 , 1267 tucci m. , toffolatti l. , 2012 , adast , 2012 , 624987 wardle j. f. c. , kronberg p. p. , 1974 ,
apj , 194 , 249
|
we present high sensitivity polarimetric observations ( @xmath0mjy ) in 6 bands covering the @xmath1ghz range of a complete sample of 53 compact extragalactic radio sources brighter than @xmath2mjy at @xmath3ghz .
the observations , carried out with the australia telescope compact array ( atca ) , achieved a @xmath4 detection rate ( at @xmath5 ) . within this frequency range the spectra of about @xmath6 of sources
are well fitted by double power laws , both in total intensity and in polarisation , but the spectral shapes are generally different in the two cases .
most sources were classified as either steep or peaked spectrum but less than @xmath7 have the same classification in total and in polarised intensity .
no significant trends of the polarisation degree with flux density or with frequency were found .
the mean variability index in total intensity of steep - spectrum sources increases with frequency for a @xmath8 year lag , while no significant trend shows up for the other sources and for the 8 year lag . in polarisation ,
the variability index , that could be computed only for the @xmath9 year lag , is substantially higher than in total intensity and has no significant frequency dependence .
[ firstpage ] galaxies : active radio continuum : galaxies polarisation .
|
it is currently widely accepted that inflationary cosmological scenarios offer most promising explanations for the initial conditions for structure formation in our universe .
almost all inflationary scenarios predict that the primordial density perturbation spectrum deviates slightly from a power law and is dominated by scalar density fluctuations .
typically , the scalar density perturbation spectrum is parameterized as @xmath5 where @xmath6 is a scale dependent function which is usually called the running ( scalar ) spectral index ( running refers to the change in the spectral index as a function of wavenumber @xmath7 ) . combining observations of cmb , galaxy surveys , and ly@xmath8 forest , there have been claims for evidence of strongly running spectral index ( see e.g. @xcite ) , but at the moment , combined data set favors no spectral index running ( e.g. @xcite ) . nonetheless , a significant running of the spectral index is still a debatable possibility that will be settled by future experiments . according to the cmb and 2dfgrs galaxy survey data @xcite , @xmath9 furthermore ,
as pointed out by @xcite , within the context of single field inflationary models , there is some indication that the spectral index quantity @xmath10 runs from positive values ( blue ) on long length scales to negative values ( red ) on short length scales ( positive to negative within about 5 e - folds ) .
it is well known @xcite that one robust check of slow roll inflationary scenario is what is usually referred to as the single field self - consistency condition @xmath11 where @xmath12 is the spectral index of tensor perturbation power spectrum parameterized as @xmath13 and @xmath14 is the slow roll parameter characterizing the tensor to scalar power spectrum ratio . in @xcite , it was pointed out that if running occurs from blue to red , then there may be another approximate consistency condition that we may observationally aim at checking regarding inflationary scenarios .
namely , within the single field slow roll scenario , there should be an approximate coincidence between the length scale @xmath15 at which @xmath16 and the length scale @xmath17 at which the tensor to scalar ratio reaches a minimum , i.e. @xmath18 if true , this kind of new consistency condition would be important because currently there is only a very limited number of observational consistency checks that could support the picture of inflationary origin of density perturbations .
the degree to which @xmath19 depends on a ) how fast @xmath20 runs and b ) the slow roll parameter @xmath14 ( this is true at least in the leading slow roll order approximation ) .
hence , it is not clear how compelling the expectation of this coincidence is . for example , does eq .
( [ eq : maincoincidence ] ) occur only for 1 out of 10,000 slow roll models of inflation with running spectral index or does it occur for 9000 out of 10,000 slow roll models of inflation ?
furthermore , because of the strongly running behavior of the spectral index , it was not clear that the leading order presented in @xcite was robust .
hence , in this paper , we quantify the degree to which one would find the approximate consistency condition a compelling consistency check of single field inflationary scenario .
furthermore , we check that the qualitative expectation based on first order slow roll expansion is also valid at second order slow roll expansion . in the spirit of @xcite , we examine a grid of slow roll models and ask how the distribution of @xmath21 changes as a function of increasing the requirement of running spectral index . expressed in terms of difference in the e - folds of inflation ( expressed as @xmath22 ) , we find @xmath23 has a mean of about @xmath24 with a @xmath25 width of @xmath26 when a strong running condition is imposed as when both @xmath27 and @xmath28 are satisfied . hence ,
slow roll inflationary scenarios predict that there should be a coincidence of @xmath15 and @xmath17 within a factor of two if the spectrum runs from blue to red , and the mismatch is such that @xmath29 . furthermore ,
since the @xmath25 width of the distribution without strong running condition is about @xmath30 , we find quantitative evidence that if the spectral index runs strongly , there is an increase in the coincidence as expected .
there are two main caveats to our results . as discussed in section [ sec : discconc ] , because the b - mode polarization sensitive to tensor perturbations is expected to peak on relatively large angular scales where it is measurable with current forecasts @xcite while the minimum of tensor to scalar ratio lies on shorter angular scales , measurements testing the proposed approximate consistency condition may be difficult .
if strong running is relevant to cosmology , ingenuity of scientists in the future will hopefully overcome this obstacle .
the second is that the grid of models that we choose are sampled with equal weight for the initial conditions of the slow roll equation .
although this is commonly found in the literature @xcite and does not seem to be an unreasonable way to sample the possible set of models , this kind of probability measure does not have any justification from first principles .
a similar ambiguity of measure plagues most `` landscape '' or anthropic principle arguments ( e.g. @xcite ) .
this paper is organized as follows . in the next section ,
we analytically compute the coincidence in terms of slow roll parameters . following that we present numerical results .
measurement prospects are then discussed in section [ sec : discconc ] . in the appendix , we collect some useful slow roll formulas used in the derivations in the paper . throughout this paper
, we use the convention @xmath31 .
let us see why there should be an approximate consistency condition as described in the introduction . in @xcite
, it was shown that within the context of a single real scalar field @xmath32 slow roll inflationary model with a potential @xmath33 , one can write @xmath34 where @xmath35 is the usual inflationary slow roll parameter in terms of the potential , the upper ( lower ) sign is for @xmath36 ( @xmath37 ) . since the tensor to scalar ratio is given by @xmath38 , the vanishing of the right hand side of eq .
( [ eq : oldwork ] ) corresponds to the scale at which the tensor to scalar power reaches an extremum .
now , the observation that was made was that if the spectrum runs from blue to red , then @xmath39 will be positive initially , but as @xmath40 becomes negative to the point of canceling the @xmath41 , @xmath42 will vanish .
as long as @xmath43 is small and/or @xmath40 runs strongly negative , the scale at which @xmath40 crosses zero will coincide with the extremum of @xmath43 which in turn corresponds to the extremum of the tensor to scalar ratio . in this section
we will extend the previous derivation to second order in slow roll and confirm the qualitative validity of the previous results .
this is reassuring given that the validity of leading order results was questionable in light of strongly running behavior .
to accomplish this task , we will work with hubble flow slow roll expansion techniques ( see , e.g. @xcite ) .
the advantage of the hubble flow approach is that it is simpler to go to higher orders in derivative expansion . using the conventions of @xcite
, we use the definitions for the slow roll parameters in terms of the hubble function @xmath44 as follows : @xmath45\cr { } ^\ell\lambda_{\rm h } & \equiv & \left({m_p^2 \over 4 \pi}\right)^\ell { \left(h'\right)^{\ell-1 } \over h^\ell } { d^{(\ell+1 ) } h \over d\phi^{(\ell + 1)}}.\end{aligned}\ ] ] here , the hubble function is defined by the hamilton - jacobi equations @xmath46 which has the advantage of having a readily solvable model of power law inflation . as is well known ,
hubble flow slow roll expansion that we summarize below is a derivative expansion in @xmath47 about the power law inflationary model .
we used @xmath48 as the measure of time during inflation , the number of e - folds before the end of inflation , which increases as one goes _
backward _ in time : @xmath49 with the sign convention @xmath50 these equations implies a relationship of @xmath14 to @xmath51 and @xmath48 : @xmath52 the evolution of the higher order parameters during inflation is determined by a set of `` flow '' equations @xcite , @xmath53 \left({}^\ell\lambda_{\rm h}\right ) + { } ^{\ell+1}\lambda_{\rm h}.\label{eqfullflowequations}\end{aligned}\ ] ] as one can see , in general any slow roll parameter could be expressed in terms of one single slow roll and its derivatives .
some of these relationships are explicitly given in the appendix .
note that @xmath14 has the interpretation through the fact that tensor to scalar power spectrum is given by @xmath54 where @xmath55 and @xmath56 are the usual power spectrum for tensor and scalar gauge invariant metric perturbations .
let us now solve for the difference between the length scale at which @xmath14 reaches an extremum ( @xmath57 ) and the length scale at which @xmath40 crosses 0 ( @xmath58 ) . instead of giving the length scale in terms of the inverse wave vector , we will express it terms of the difference in the horizon exit e - folds @xmath23 . to translate that into ratio of lengths ,
one simply has @xmath59 where @xmath60 is the expansion rate when @xmath17 left the horizon and @xmath61 is the expansion rate when @xmath15 left the horizon .
since @xmath62 is of order of the slow roll parameter during the early stages of inflation , one can disregard this term when interpreting @xmath63 as long as it is larger than about @xmath64 . defining @xmath65 as the point where the extremum of @xmath14 occurs ,
i.e. @xmath66 , we can linearize @xmath40 around it to solve approximately for @xmath23 . using the formulas given in the appendix
, we can write @xmath67 @xmath68 @xmath69\dot{\epsilon } + \ddot{\epsilon}(3-c)-8\epsilon^3(\epsilon+1)\}.\ ] ] solving @xmath70 for @xmath23 , we obtain @xmath71 the constant @xmath72 , where @xmath73 is euler s constant . as it can be seen from this formula @xmath23 clusters around the negative value @xmath74 and it can be positive or negative .
we can also observe that stronger running ( associated with a higher slope for @xmath0 ) drifts @xmath23 toward more negative values , and that this effect is modulated by the value of @xmath14 at @xmath65 .
, @xmath56 ( scalar power spectrum ) , and @xmath40 as functions of the number of e - folds @xmath48 , it can be seen that the extrema of @xmath14 and @xmath56 are closed to the point where @xmath40 is zero .
the dot dashed curve is @xmath75 , the solid curve is @xmath76 , and the dotted curve is @xmath40 .
the horizontal axis is oriented such that inflation ends at @xmath77 ( i.e. time flows right to left ) . ]
we will see that numerical results agree well with the analytical approximation . before going onto numerical analysis , we would like to note that beyond eq .
( [ eq : oldwork ] ) , there may be many more `` approximate consistency '' conditions associated with strong running since we are merely labeling various features of the slow roll equations .
for example , as it can be inferred already from the first order formula @xmath78 the power spectrum will have an extremum roughly coinciding with the extrememum of @xmath14 . to check this numerically
, we can examine an example of an inflationary model ( as explained in the next section ) . as it can be seen in fig .
[ fig : p - e - n - plot ] , the scalar power spectrum has an extremum very closed to the point where @xmath40 is zero .
we could proceed similarly to the way we have done for eq .
( [ eq : maincoincidence ] ) , comparing analytical approximation to numerical data for this `` approximate consistency '' condition , which in this case would not depend on the strength of the running of the spectral index , but could be another test for more general inflation models .
in this section , we study our proposed approximate consistency condition question numerically .
one scientific question we would like to address is , `` what is the expected coincidence of the value @xmath15 satisfying @xmath79 with the value @xmath17 satisfying @xmath80 ? ''
the second question is whether a stronger negative running of the spectral index makes the expected @xmath81 significantly smaller . to this end
, we examine a particular grid ( to be specified below ) of slow roll inflationary models . in this grid of models
, we make a set of `` cuts '' which select out subsets of the models with scalar spectral index running from blue to red . in this way
, we can make a histogram of models as a function @xmath23 ( defined in eq .
( [ eq : deltan ] ) ) .
the peak and the width of the histogram quantifies the expected coincidence of @xmath15 with @xmath17 , answering our first question . we then make the cuts more stringent such that only those slow roll models with more strongly running spectral indices are included in the histogram .
the degree to which the width of the histogram changes as we make the required running stronger answers our second question .
let us now explain the details which closely follows the methods of @xcite . assigning a set of initial conditions to eqs .
( [ eqfullflowequations ] ) and integrating them specifies a flow of the slow roll function .
each of the initial conditions corresponds to a particular choice of inflaton potential and the inflaton initial conditions .
we choose initial values for the parameters at random from the following ranges , assuming a uniform probability distribution : @xmath82\cr \epsilon & = & \left[6\ 10^{-76},0.6\right]\cr \sigma & = & \left[-0.5,0.5\right]\cr { } ^2\lambda_{\rm h } & = & \left[-0.05,0.05\right]\cr { } ^3\lambda_{\rm h } & = & \left[-0.025,0.025\right],\cr & \cdots&\cr { } ^{m+1}\lambda_{\rm h } & = & 0.\label{eqinitialconditions}\end{aligned}\ ] ] we truncated the expansion to order 5 by setting @xmath83 .
we calculate the values of the tensor / scalar ratio @xmath84 , the spectral index @xmath85 , and the `` running '' of the spectral index @xmath86 according to : @xcite @xmath87,\label{eqrsecondorder}\ ] ] @xmath88 note that the variable @xmath84 here is defined with a different normalization than in @xcite where @xmath89 to leading slow - roll order .
a comoving scale @xmath7 crossed the horizon a number of e - folds @xmath90 before the end of inflation : @xcite @xmath91 here @xmath92 is the potential at horizon exit , @xmath93 is the potential at the end of inflation , and @xmath94 is the energy density after reheating
. note that since slow roll inflation evolves toward decreasing potential , we can write @xmath95 where @xmath96 is a function of @xmath7 that varies as to keep @xmath93 independent of @xmath7 . furthermore , since the reheating energy density must be smaller than @xmath93 , we can write @xmath97 where the proportionality constant satisfies @xmath98 . using eq .
( [ eq : comovingnk ] ) , derivatives with respect to wavenumber @xmath7 can be expressed in terms of derivatives with respect to @xmath48 as @xcite @xmath99 hence , the running of the spectral index can be expressed to third order in slow roll parameters as @xmath100\end{aligned}\ ] ] we will now use superscripts a and b to denote quantities evaluated at or corresponding to the length scale @xmath101 or @xmath102 . the algorithm for generating the histogram of fig .
1 can then be described as follows : 1 .
a set of random initial conditions was generated for the value of the slow roll parameters at @xmath103 .
the following constraints are imposed : @xmath104 @xmath105 @xmath106 where @xmath107 is fixed to 0.3 and @xmath108 varies from 0 to 0.225 for different cuts with a 0.075 increment . if these constraints are respected , proceed to step 2 otherwise go back to step 1 .
2 . integrate the flow equation ( using lsoda from odepack @xcite ) to @xmath109 and then check the following constraints : @xmath110 where @xmath111 is fixed to -0.3 and @xmath112 varies from 0 to -0.225 for different cuts with a 0.075 increment . if these are respected go to step 3 otherwise go to step 1 .
evolve forward in time ( @xmath113 ) until inflation ends ( @xmath114).if the total number of e - folds @xmath48 from the beginning to the end of inflation is in the range [ 40,70 ] add this model to the ensemble of acceptable models .
if not go back to step 1 .
repeat step 1 through 3 until the desired number of acceptable models have been found .
once the set of acceptable models has been generated we solve numerically the following two equations : @xmath115 @xmath116 it is important to observe that the presence of the square root of @xmath14 in the definition of the differential operator @xmath117 ( eq . ( [ eq : ddn ] ) ) causes numerical solutions to eq .
( [ eq : epsprime ] ) to be difficult for small values of @xmath14 , since small values of the derivative may be associated to the smallness of @xmath14 more than to the actual presence of an extremum . in other words , even though the derivatives are taken with respect to @xmath48 , for inflaton potentials with very small @xmath14 , the derivatives become very small , requiring high numerical precision . for this reason and to discard inflection points , extrema are identified numerically using the second derivative information as well . before making the histogram , we must further check that the models in the histogram are reasonable from an inflationary phenomenology point of view .
this is done in lieu of executing precision fits to all available data since it has been well demonstrated by previous analyses ( see for example @xcite ) that the current cmb data are not extremely constraining in terms of the details of the slow roll inflationary models .
hence , our histogram should not be sensitive to the lack of precision in our cuts .
on the other hand , given that the combined fits including sdss data as carried out by @xcite prefer small running , the strongly running cases of the current analysis may have been statistically disfavored had precision fits to combined data been made . given that most of the numerically generated models do not have strongly running spectral index for the weakest cut , the results of the weakest cut should be robust with respect to imposing additional fit constraints , while the results of the stronger cuts should be interpreted with appropriate caution .
let us now write explicitly the constraint equations . since the wmap analysis @xcite gives @xmath118
we use the approximate formula for the power spectrum @xmath119 to impose a constraint on @xmath14 and @xmath92 .
the requirements of having sufficient energy density at the end of inflation to reheat the universe to a temperature consistent with big bang nucleosynthesis gives @xmath120 we also know from observations that @xmath121 and that eq .
( [ eq : powerobserve ] ) is true . substituting @xmath92 obtained from eqs .
( [ eq : comovingnk ] ) , ( [ eq : vevkrel ] ) , and ( [ eq : verhorel ] ) into eq .
( [ eq : approxpzeta ] ) and solving for @xmath14 we obtain @xmath122 where @xmath123 and the upper bound comes from eq .
( [ eq : rcond ] ) and @xmath124 ( again note that the normalization used in this paper is given by eq .
( [ eqrsecondorder ] ) , and specifically is not @xmath125 ) . consider how eq .
( [ eq : epsyrel ] ) gives a constraint .
note that for any given flow trajectory , because of eqs .
( [ eq : powerobserve ] ) and ( [ eq : approxpzeta ] ) , @xmath126 is fixed .
furthermore , the flow trajectory itself determines @xmath127 ( the value of the potential at the end of inflation ) .
hence , the only adjustable parameter is @xmath128 which determines the reheating temperature . for every flow trajectory with
@xmath129 fixed , eq .
( [ eq : epsyrel ] ) then provides constraints on the initial conditions of inflation .
for example , consider a flow trajectory having @xmath130 .
( [ eq : epsyrel ] ) gives @xmath131)\sim 10^{-16 } \label{eq : fkconstraint}\ ] ] this is a significant constraint since this says that the energy density at the end of inflation is at least @xmath132 smaller than the the energy density at @xmath133 e - foldings before the end of inflation .
the physical reason for eq .
( [ eq : yconst ] ) is that upper and lower bounds on the potential ( coming from tensor perturbations and reheating temperature , respectively ) directly translates into upper and lower bounds @xmath14 because of eqs .
( [ eq : powerobserve ] ) and ( [ eq : approxpzeta ] ) . the physical origin of eq .
( [ eq : epsyrel ] ) is that any mismatch in the inflationary stretching of the wavelength and physical @xmath134 must be compensated by postinflationary expansion which is related to the potential energy at the end of inflation which in turn is related to @xmath14 through eqs . ( [ eq : powerobserve ] ) , ( [ eq : approxpzeta ] ) , and ( [ eq : vevkrel ] ) . in practice , eq .
( [ eq : fkconstraint ] ) does not play a significant role in the numerical exploration since the number of e - foldings is generically small when there is significant running .
.1 in our final histogram can be seen in fig .
[ figpk ] and its distributional characterization can be seen in table [ tbl : mainresult ] . to have a measure of how @xmath23 clusters around zero we followed this procedure : a ) we have divided @xmath23 in the range @xmath135 $ ] into 25 intervals of width 0.25 each containing @xmath136 models and then calculated for each of these intervals the average @xmath137 for the models in that interval and the standard deviation @xmath138
; b)we calculated @xmath139 , a weighted second order moment around 0 of the 25 averages of @xmath137 for the intervals defined above , using the inverse of the corresponding variance @xmath140 as weight .
we also similarly calculated the weighted standard deviation @xmath141 .
the total number of slow roll models we considered to construct the histogram was around @xmath142 , and the number of models fulfilling the cuts is shown in the @xmath143 column of tbl .
[ tbl : mainresult ] .
.basic distributional characterization of fig .
[ figpk ] .
@xmath144 characterizes the deviation from 0 .
@xmath145 is the fraction of the models within 2@xmath146 of @xmath147 .
@xmath143 is the number of models analyzed for the given range of the @xmath0 . [
cols=">,>,>,>,>,>,>,>",options="header " , ] notice in table [ tbl : mainresult ] that both @xmath148 and @xmath149 regardless of the strength of running .
the column labeled `` @xmath150 '' quantifies ( at least with respect to our choice of model sampling ) the extent to which we should expect a coincidence between @xmath15 and @xmath17 : i.e. we should expect with roughly `` 95% confidence '' that the length scale for which the tensor to scalar ratio reaches a minimum will coincide with the length scale for which the @xmath3 up to a factor of about 2 and @xmath29 ( i.e. wavelength at which the tensor to scalar ratio reaches a minimum is longer than the wavelength at which @xmath40 goes through a zero ) .
as far as the effect of different strengths of running is concerned , one sees in table [ tbl : mainresult ] that stronger running corresponds to a larger magnitude of @xmath151 and a more negative value of @xmath152 .
although `` statistically '' marginal , the smaller @xmath146 means that strong running seems to narrow the distribution of @xmath23 .
in this paper , we have studied the robustness of the approximate condition suggested by @xcite .
we have quantified distributionally the extent to which one should consider the approximate consistency condition to be a prediction of single field slow roll inflation .
we find that if the spectrum runs from blue to red , we should look for an approximate coincidence between the wave vector @xmath15 satisfying @xmath79 and the wave vector @xmath17 satisfying @xmath153 , up to a factor of about 2 . according to a wmap analysis @xcite , this should occur at @xmath154 .
given that the consistency condition is only approximate , even if one finds that observations contradict the consistency condition ( for example , if observations deduce @xmath155 ) it will be very difficult to make any rigorous conclusions about the models of inflation .
furthermore , given that we have sampled each of the initial condition ranges with equal weight , it is not clear what the distribution that we computed has to do with the real world .
( a similar argument can be made about much of the current literature that use similar sampling @xcite . ) nonetheless , if this set of initial sampling turns out to be a good approximation to the real inflationary model distributions coming from a more fundamental theory , tests coming from this kind of approximate consistency condition may be useful . to turn this around
, there may be a way to classify the vacua of more fundamental theories according to whether they satisfy roughly equal probability sampling of the slow roll initial condition or whether they prefer a very specific distribution which makes the approximate consistency condition more ( or less ) stringent . before concluding ,
let us briefly consider the measurement prospects for the approximate consistency condition . according to current ideas of anticipated measurements @xcite
, b - mode polarization observations are necessary to reliably infer the tensor perturbation amplitudes .
as is well known , b - mode polarization has the advantage over the e - mode polarization in that at the last scattering surface , the b - mode polarization can not be generated by scalar perturbations alone because to leading order , the scalar perturbations do not generate nonzero @xmath156 amplitudes in the temperature anisotropy while the tensor modes can .
hence , in principle , assuming no large contamination from vector perturbations , b - mode polarization measurement offers a direct way to infer tensor perturbation amplitudes .
the difficulty with measuring the b - mode polarization spectrum coming from gravity waves is that because its amplitude is proportional to the local quadrupole anisotropy induced by the time derivative @xmath157 ( where @xmath158 is the amplitude of the gravitational wave ) localized at the last scattering surface ( due to the requirement of thomson scattering ) , its amplitude is suppressed at least by order @xmath159 on long wavelengths with respect to the temperature anisotropy ( where @xmath160 and @xmath161 are the hubble radius and the scale factor at recombination , respectively ) .
furthermore , since gravitational wave mode amplitudes inside the horizon are diluted by the expansion of the spatial volume , the transfer function should behave approximately as @xmath162 where @xmath163 is the scale factor today and @xmath164 is the redshift to the last scattering surface .
( here , we have considered only the modes that entered the horizon after matter domination , i.e. @xmath165 . ) hence , the b - polarization temperature perturbation amplitude can be written as @xmath166 where @xmath167 is the power spectrum of the tensor perturbations .
( for discussions of analytic treatments of cmb polarization , see e.g. @xcite . )
this peaks at @xmath168 , and even there , the amplitude is suppressed by about @xmath169 relative to the temperature anisotropies if the tensor perturbation amplitude is about the same as scalar perturbations amplitudes .
to make the situation slightly worse , @xmath15 is expected to be at around @xmath170 .
( note that to map the wave vectors @xmath171 to the multipole moment number @xmath172 , one can use the approximate formula @xmath173 . )
this means that to actually measure @xmath21 , one must go to short scales ( @xmath174 ) while the b - mode polarization measurements prefer large length scales ( @xmath175 ; for more accurate plots of examples , see e.g. @xcite ) .
hence , experimental confirmation of the approximate consistency conditions will be a challenge .
there is one more obstacle that makes it difficult for tensor perturbation amplitudes to be extracted from b - mode polarization measurements .
this is due to the fact that e - modes can be converted into b - modes through gravitational lensing @xcite .
hence , obtaining a tensor spectrum to check the approximate consistency condition requires success in extracting the tensor perturbation contribution even at relatively short length scales , around where the lensing contribution most likely dominates over the gravitational wave signal .
this contamination from gravitational lensing can be subtracted out , at least in principle , if the lens distribution can be accurately deduced . despite the difficulties , we are optimistic that the ingenuity of researchers in the future will allow a test of the approximate consistency conditions considered here .
we thank niayesh afshordi and scott dodelson for useful discussions and helpful , detailed comments on the manuscript .
here we report some formulas which can be used to express slow roll parameters in terms of @xmath176 and its derivatives respect to the field @xmath32 or the operator @xmath117 . denoting @xmath177 and @xmath178 : @xmath179 99 a. vallinotto , e. j. copeland , e. w. kolb , a. r. liddle and d. a. steer , `` inflationary potentials yielding constant scalar perturbation spectral indices , '' phys .
d * 69 * , 103519 ( 2004 ) [ arxiv : astro - ph/0311005 ] .
h. v. peiris _ et al .
_ , astrophys .
j. suppl .
* 148 * , 213 ( 2003 ) [ arxiv : astro - ph/0302225 ] .
u. seljak _ et al .
_ , phys .
d * 71 * , 103515 ( 2005 ) [ arxiv : astro - ph/0407372 ] . c. l. bennett _ et al .
_ , astrophys .
j. suppl .
* 148 * , 1 ( 2003 ) [ arxiv : astro - ph/0302207 ] .
w. j. percival _ et al . _
[ the 2dfgrs collaboration ] , mon . not .
327 * , 1297 ( 2001 ) [ arxiv : astro - ph/0105252 ]
. j. e. lidsey , a. r. liddle , e. w. kolb , e. j. copeland , t. barreiro and m. abney , rev .
* 69 * , 373 ( 1997 ) [ arxiv : astro - ph/9508078 ] .
d. j. h. chung , g. shiu and m. trodden , phys .
d * 68 * , 063501 ( 2003 ) [ arxiv : astro - ph/0305193 ] .
w. h. kinney , e. w. kolb , a. melchiorri and a. riotto , phys .
d * 69 * , 103516 ( 2004 ) [ arxiv : hep - ph/0305130 ] .
s. dodelson , w. h. kinney and e. w. kolb , inflation phys .
d * 56 * , 3207 ( 1997 ) [ arxiv : astro - ph/9702166 ] .
l. a. boyle , p. j. steinhardt and n. turok , arxiv : astro - ph/0507455 .
t. banks , m. dine and e. gorbatov , jhep * 0408 * , 058 ( 2004 ) [ arxiv : hep - th/0309170 ] .
a. c. hindmarsh , `` odepack , a systematized collection of ode solvers , '' in scientific computing , r. s. stepleman et al .
( eds . ) , north - holland , amsterdam , 1983 ( vol . 1 of imacs transactions on scientific computation ) , pp .
55 - 64 [ http://www.netlib.org/odepack/ ] .
m. kamionkowski , a. kosowsky and a. stebbins , phys .
d * 55 * , 7368 ( 1997 ) [ arxiv : astro - ph/9611125 ] .
m. kamionkowski , a. kosowsky and a. stebbins , phys .
lett . *
78 * , 2058 ( 1997 ) [ arxiv : astro - ph/9609132 ] .
m. zaldarriaga and u. seljak , phys .
d * 55 * , 1830 ( 1997 ) [ arxiv : astro - ph/9609170 ] .
u. seljak , arxiv : astro - ph/9608131 .
w. hu and m. j. white , phys .
d * 56 * , 596 ( 1997 ) [ arxiv : astro - ph/9702170 ] .
j. r. pritchard and m. kamionkowski , waves : an annals phys . *
318 * , 2 ( 2005 ) [ arxiv : astro - ph/0412581 ] .
w. hu , u. seljak , m. j. white and m. zaldarriaga , phys .
d * 57 * , 3290 ( 1998 ) [ arxiv : astro - ph/9709066 ] .
notes from princeton university course phy 564 course ( spring term 2004 ) taught by uros seljak ( can be accessed from http://wwwphy.princeton.edu/@xmath180uros/phy564.html ) .
|
density perturbations generated from inflation almost always have a spectral index @xmath0 which runs ( varies with the wavelength ) .
we explore a running spectral index scenario in which the scalar spectral index runs from blue ( @xmath1 ) on large length scales to red ( @xmath2 ) on short length scales . specifically , we look for a correlation between the length scale at which @xmath3 and the length scale at which tensor to scalar ratio @xmath4 reaches a minimum for single field slow roll inflationary models . by computing the distribution of length scale ratios ,
we conclude that there indeed is a new approximate consistency condition that is characteristic of running spectral index scenarios that run from blue to red .
specifically , with strong running , we expect 96% of the slow roll models to have the two length scales to be within a factor of 2 , with the length scale at which the tensor to scalar ratio reaching a minimum longer than the wavelength at which @xmath3 .
|
the sunyaev - zeldovich ( sz ) effect has long been recognized as a powerful tool for probing the physics of the intra - cluster medium , large - scale structure formation and the dark energy equation of state @xcite .
simple but robust analytical arguments @xcite and hydrodynamical simulations of clusters @xcite have indicated that the integrated sz flux ( the integral of the temperature decrement across the surface area of a cluster ) should correlate tightly with cluster mass .
combined with a greater sensitivity than optical or x - ray surveys to high redshift objects , this makes sz - selected cluster samples well suited for measuring the evolution of the cluster mass function over a wide range in redshift .
experiments such as the atacama cosmology telescope @xcite and the south pole telescope @xcite are currently surveying the microwave sky to develop large catalogs of galaxy clusters that are uniformly selected by the sz flux .
however , in order to use cluster samples for this purpose , it is necessary to have a good understanding of both the selection function of the survey @xcite and of the mapping between the measured integrated sz flux and cluster mass ( and how this evolves with redshift ) . besides the slope and normalization of the flux - mass ( y - m ) relation
, one must also gain some measure of the scatter around the mean relation in order to be able to differentiate between different dark energy cosmologies @xcite .
a detailed characterization of the y - m relation may require x - ray follow - up of a representative sub - sample of sz - selected clusters @xcite .
complete optical coverage of surveyed fields is necessary for obtaining cluster redshifts and to help determine the selection function .
thus , although there is much power in this method , much work is required before it can be applied to its full potential .
the sz effect can also be detected as a secondary anisotropy in the cmb temperature anisotropy power spectrum , appearing as ` excess power ' ( over the predicted primary anisotropy signal ) on angular scales below several arc - minutes . in principle , measuring the sz angular power spectrum is an easier task than serendipitous cluster detection as one is searching for temperature fluctuations ( on a given angular scale ) to which clusters over a wide mass and redshift range contribute significantly , without the need to resolve individual clusters at high signal - to - noise .
predictions for the sz power spectrum amplitude @xmath6 ( henceforth @xmath0 ) can be made using the halo model ( e.g. * ? ? ?
* ) and estimates of the radial pressure profile of intra - cluster gas .
assuming that the cluster gas resides in hydrostatic equilibrium in the potential well of the host dark matter halo ( with an additional steepening of the gas density profile in the outer regions ) , @xcite demonstrated that the ensemble averaged power spectrum amplitude @xmath7 has an extremely sensitive dependence on @xmath1 , where @xmath8 .
although there remains a certain ambiguity in the exact amplitude and shape of the predicted signal due to uncertainties in cluster gas physics ( and understanding of the contribution of sub - mm and synchrotron sources to the power spectrum provides an added complication ) , the sz angular power spectrum represents a robust observable with which competitive constraints on @xmath1 can be obtained . rather than attempting to resolve individual clusters , small - scale cmb experiments have focused on detecting sz power at arc - minute scales and measuring the amplitude of @xmath0 and thus @xmath1 .
bima surveyed 0.2 square degrees on the sky measuring a fluctuation amplitude of @xmath9 , from which they inferred a value of @xmath10 @xcite and cbi recently measured an excess of @xmath11 above @xmath12 @xcite .
clearly there is some tension between the value of @xmath1 inferred from these experiments and the current value of @xmath13 favored by wmap and x - ray cluster number counts @xcite .
however , we note that recent results from sza @xcite and quad @xcite measurements of the sz power spectrum infer a value of @xmath1 that is more consistent with those derived from other methods .
one possible explanation for this discrepancy is that the icm gas models used to calculate the sz power spectrum do not include or correctly capture the relevant physics that determine the shape and amplitude of pressure profiles , resulting in a large modeling uncertainty in the amplitude of @xmath0 .
analytic arguments @xcite and hydrodynamical simulations @xcite suggest that the inclusion of more detailed physics ( radiative cooling , star formation and energy feedback ) only moderately influence the amplitude of the power spectrum .
this is partly because these effects mostly influence the temperature and density of gas in cluster cores , whereas it is the gas at larger radii that contributes most to @xmath0 @xcite .
however , to date there has not yet been a systematic comparison between simulated and analytic power spectra , partly due to the limited total area of sz maps that can be produced from the hydrodynamical simulations , and so the actual magnitude of the theoretical uncertainty in @xmath0 remains uncertain .
a second explanation relates to the fact that the probability density function for the field to field distribution of @xmath0 , p(@xmath0 ) , is inherently non - gaussian , with a long tail towards high values of @xmath0 .
the significant non - gaussian contribution is due to poisson fluctuations in the sampling of the halo mass function for a given survey field .
hence , although the average power @xmath7 measured over many maps should reproduce the predicted halo model spectrum ( assuming the cluster gas physics is correctly modeled ) , the power measured in any single map may be several times greater than the ensemble - averaged value , if that field happens to contain several very massive clusters . these objects , while not contributing strongly to the mean sz power spectrum drive the significant non - gaussian contribution to p(@xmath0 ) .
@xcite derived an analytic expression for p(@xmath0 ) using the halo model approach and demonstrated that the degree of non - gaussianity is most significant at large angular scales , where massive clusters contribute most of their signal .
until now , it has not been possible to generate a large enough sample of independent simulated sz maps in order to test the analytic expressions .
the main purpose of this paper is to demonstrate , using both halo model calculations and a large sample of simulated sz sky maps generated from a ` lightcone ' simulation , that the non - gaussianities in p(@xmath0 ) are indeed driven by shot noise due to the number of very high mass clusters in a given field , and that one can reduce this poisson contribution to the total variance in @xmath0 by masking these objects from sky maps .
we compare the power spectra measured from our simulated maps with the halo model analytic predictions for @xmath7 and the non - gaussian contribution to p(@xmath0 ) . in section [ sec : halomodel ] we describe the halo approach to calculating the sz angular power spectrum and trispectrum , and demonstrate how masking clusters will increase the sensitivity of the measured @xmath0 to @xmath1 . in section [ sec : methods ] we compare the analytic and simulated results for @xmath0 , p(@xmath0 ) and @xmath14 , and demonstrate in practice the utility of masking the brightest clusters . in section [ sec : foregrounds ]
we discuss the impact of astrophysical foregrounds and instrument noise on our results . henceforth , in this paper cluster mass is measured within the spherical region @xmath15 , defined as the region enclosing a mean overdensity of @xmath16 .
@xmath17 is calculated using the fitting formula provided by @xcite .
the fiducial assumed cosmology has parameters @xmath18 , @xmath19 , @xmath20 , @xmath21 and @xmath22 .
the thermal sz effect is a distortion of the cmb caused by inverse compton scattering of cmb photons ( at temperature @xmath23 ) off electrons ( at @xmath24 ) in the high temperature plasma within galaxy clusters .
to first order , the temperature change at frequency @xmath25 of the cmb is given by @xmath26 , where @xmath27 , @xmath28 , and @xmath29 is the normal compton parameter @xmath30 where the integral is along the line of sight .
there is a host of higher order effects that can be important at the @xmath31 level , including bulk motions and relativistic effects @xcite . unless stated otherwise we assume henceforth that @xmath32 , which is appropriate in the rayleigh - jeans limit .
we can also calculate the sz power spectrum , assuming that clusters are poisson - distributed , by simply summing up the squared fourier - space sz profiles , @xmath33 of all clusters : @xmath34 where v(z ) is the comoving volume per steradian and @xmath35 is the number density of objects of mass @xmath36 at redshift @xmath37 . for the latter we use the fitting function of @xcite , while for the sz profiles as a function of mass and redshift we use either a simple @xmath38-model ( where @xmath39 ) with self - similar scaling for the central @xmath40 @xcite or the hydrostatic model of @xcite ( henceforth referred to as the ks model ) .
we have extensively experimented with different profiles , with most prescriptions showing differences from each other at the level of tens of percent in the power for consistent assumptions about gas fractions and cosmological parameters . the dominant contribution to
the non - gaussian error comes from the shot noise in galaxy clusters .
this poisson contribution to the variance at a single value of @xmath41 comes from the one - halo contribution to the trispectrum @xcite . for closely separated bins in @xmath41 space ,
the variance in the sz power spectrum is given by @xmath42 \label{eqn : variance}\ ] ] where the trispectrum due to shot noise is @xmath43 henceforth , we will refer to the second term within the brackets in eqn .
[ eqn : variance ] , as the ` non - gaussian ' variance ( or trispectrum ) , @xmath44
. we will refer to the first term as the ` gaussian ' variance , @xmath45 .
we note that the ratio of these two quantities is independent of the observed map area @xmath46 .
figure [ fig : cl_mz ] demonstrates how clusters at a given mass and redshift contribute to @xmath0 and @xmath47 at @xmath48 and for @xmath49 ( assuming a @xmath38-model for the gas pressure profiles ) . the black and blue contours represent equal contribution to @xmath0 and @xmath47 , respectively .
clusters in the redshift range @xmath50 and of mass @xmath51 contribute most to the power spectrum , whereas the trispectrum is driven by clusters at approximately the same redshift but with more than twice the mass . since the sz flux scales as @xmath52 , the trispectrum is clearly skewed toward much larger masses than the power spectrum .
this leads to a significant amount of non - gaussianity in the statistics of the sz power spectrum being driven by the most massive objects .
it is clear from figure [ fig : cl_mz ] that although the most massive clusters do not strongly contribute to the mean power spectrum at @xmath48 , they dominate the trispectrum .
these clusters are a significant source of noise in any measurement of @xmath0 .
therefore , if one simply clips the most massive objects from the map , the field - to - field variance of @xmath0 will be reduced and the statistics will become more gaussian .
the three lines intersecting the contours in figure [ fig : cl_mz ] are lines of constant integrated flux ( y ) , each separated by a factor of 2 .
they represent thresholds above which one could mask clusters so that the impact of the trispectrum ( or non - gaussian variance ) is greatly suppressed , while not removing a significant amount of signal from a power spectrum measurement .
this then allows for a more precise measurement of the ` true ' or ensemble - averaged sz power spectrum amplitude , @xmath7 , and thus on @xmath1 .
figure [ fig : dcldsig8 ] demonstrates the improvement in sensitivity to @xmath1 when clusters above a given threshold are masked out of the analysis .
we define our threshold in terms of a cumulative surface density on the sky , such that all bright clusters down to surface density @xmath53 ( deg@xmath5 ) are masked out from the map .
we adopt a surface density threshold rather than an integrated flux threshold so that the number of clusters masked is relatively insensitive to @xmath1 and modeling of cluster gas physics ( see discussion below ) . in practice
we have chosen three values of @xmath54 : 0.01 , 0.1 and 0.9 deg@xmath5 .
spt has already demonstrated that clusters can be detected at @xmath55 significance down to the 0.1 deg@xmath5 threshold @xcite , and both act and spt are predicted to detect clusters down to the 0.9 deg@xmath5 threshold .
therefore , it should be possible for both these experiments to identify and mask or remove clusters down to the thresholds that we examine here .
the top panel in figure [ fig : dcldsig8 ] shows the sz angular power spectrum for no masking ( black ) and masking down to 0.01 , 0.1 and 0.9 clusters per square degree ( blue , green and red lines ) .
the solid and dashed lines show the results for @xmath49 and @xmath56 , respectively . as expected , masking the brightest clusters primarily removes power at larger scales ( @xmath57 ) .
removing the brightest 0.01 clusters per square degree has little impact above @xmath58 , however masking down to 0.1 ( 0.9 ) deg@xmath5 reduces the power spectrum at this scale by 33% ( 66% ) .
it is clear that the change in power due to masking is greater than that due to the 0.03 change in @xmath1 , so the masking must be included in the theory prediction for the signal .
although the second panel indicates that an unmasked map has the greatest raw sensitivity to @xmath1 , the third panel demonstrates that the masked maps have much lower total variance , and are therefore more precise . the reduced variance more than compensates for the slight reduction in raw sensitivity , such that the ability to use the sz power spectrum as a cosmological tool is enhanced by using maps with the brightest sources masked
this does not include the extra information encoded in these individually - detected clusters and is based solely on the information in the power spectrum .
the bottom panel shows the improvement in true sensitivity ( @xmath59 ) having cut out bright clusters , relative to the uncut case . at @xmath58 , masking 0.1 clusters / deg@xmath2 results in a factor of 2.6 improvement in the cosmological sensitivity of the sz power spectrum . at @xmath60
there is still a 50% improvement .
we note that given full knowledge of the number density of clusters and the thermal properties of the icm over a wide range in mass and redshift , one could construct an optimized scheme for measuring the power spectrum amplitude by de - weighting the signal from high mass clusters .
however , as one would already know the cluster mass function which is more sensitivity to cosmological parameters the power spectrum would then contain no further cosmological information . in the absence of such an optimal estimator ,
a simple solution is to apply zero weight ( i.e. mask ) the highest mass clusters from power spectrum measurements ( to which they mostly contribute just noise ) while utilizing them to sample the cluster mass function , to which they contribute significant statistical information .
we now investigate the difference between applying cuts defined by a surface density or integrated sz flux threshold . in figure
[ fig : densitycut_fluxcut ] we plot the fractional error in the amplitude of the sz power spectrum as a function of angular multipole number ( assuming full sky coverage ) .
the black solid and dotted lines show the unmasked results for @xmath49 and @xmath61 respectively .
the solid blue , green and red curves show the results having applied our normal surface density cuts ( @xmath53 = 0.01 , 0.1 and 0.9 clusters deg@xmath5 ) for @xmath49 .
the dotted lines show the fractional error for @xmath62 having applied exactly the same cut in flux .
the dashed lines show the results for @xmath62 having now used the surface density rather than flux thresholds .
the plot demonstrates that using cuts in surface density produces a fractional error that is approximately independent of @xmath1 . on the other hand , adopting a flux threshold means that the number of clusters masked and thus @xmath63 is highly sensitive to @xmath1 .
larger values of @xmath1 result in greater numbers of clusters with flux greater than a given value .
this will ultimately make the statistical error on measurements of @xmath0 more difficult to determine .
furthermore , the effectiveness of applying flux cuts is also dependent on the accuracy with which one is able to measure cluster fluxes observationally , and the mass and redshift dependence of the intrinsic scatter in the m - y relation .
using surface density thresholds neatly circumvents these issues .
to date , it has proven difficult to compare the statistics of the sz power spectrum as predicted by analytic calculations and numerical simulations as the typical box size of hydrodynamical simulations limits the number of independent maps that can be created , while also poorly sampling the extreme high - mass end of the halo mass function . here
we combine an n - body ` lightcone ' simulation with a semi - analytic model for cluster gas to generate a large number of maps and thus enable such a comparison .
we also use our large sample of simulated maps to demonstrate the utility of masking clusters to improve the precision of measurements of the sz power spectrum . to generate the simulated sz maps , we begin with the output of a large ( @xmath64 particles ) cosmological dark matter simulation .
cosmological parameters were chosen to be consistent with those measured from the 3rd - year wmap data combined with large - scale structure observations @xcite .
the simulated volume is a periodic cube of size @xmath65mpc ; the particle mass @xmath66 and the cubic spline softening length @xmath67kpc .
this simulation was carried out with exactly the same methods used for the simulation described in @xcite , @xcite and @xcite , so for a description of the computational details we refer the reader to these papers .
the very large box size of this simulation makes it ideal for generating a large area of simulated sky .
the matter distribution in a light cone extending to @xmath68 was saved in 348 time slices .
dark matter halos in this light cone were identified with the friends - of - friends algorithm using a comoving linking length parameter @xmath69 . in total
, the lightcone covers a single octant on the sky ( @xmath70 deg@xmath2 ) , containing over two million clusters with @xmath71 . for the purposes of the analysis here we retain only clusters with @xmath72 .
the cluster gas distribution in each halo was calculated using the semi - analytic model described in @xcite and @xcite . in brief , a 3d mesh ( with cell side - length @xmath73 kpc ) is placed around each cluster , with the gas pressure and density determined in each mesh cell assuming hydrostatic equilibrium and a polytropic equation of state ( with adiabatic index @xmath74 , * ? ? ?
it is also assumed that the gas in the densest cluster regions has cooled and condensed , forming stars . at z = 0 ,
the stellar / gas mass ratio is set to 0.1 @xcite . to compute the star / gas ratio at @xmath75
, the star formation rate was assumed to follow a delayed exponential model ( eqn . 1 of * ? ? ?
* ) , with decay time @xmath76 gyr .
as discussed in detail in @xcite , the most important free parameter in this model is the energy input into the cluster gas via non - thermal feedback processes , such as agn outflows and supernovae .
this is set through the parameter @xmath77 , such that the feedback energy is @xmath78 , where @xmath79 is the stellar mass in the cluster .
it was demonstrated that setting @xmath80 provides good agreement between the @xmath81 and @xmath82 scaling relations obtained from our simulation plus gas model cluster sample and those from x - ray observations @xcite . from the output of our icm model ,
we produce a 2-d ` postage - stamp ' @xmath29 image for each cluster by summing up the electron pressure @xmath83 in each mesh cell along one direction ( for example , see fig . 7 of * ? ? ?
we then construct a library of images for all clusters in the lightcone octant with @xmath84 .
note that each postage - stamp is produced having rotated the 3d mesh around the cluster so that one face lies parallel to the plane of the lightcone ` sky ' .
sz cluster sky maps are thus produced by projecting down the lightcone , summing up the contribution of all the clusters along the line of sight .
in practice , we produce individual sz sky maps of size 5x5 degrees .
we set the image resolution ( pixel size ) to 0.25 arcminutes . in order to allow robust comparison with the analytic results
, we also create a second set of simulated maps in which we replace the gas model cluster ` postage - stamp ' images with a @xmath38-model profile with @xmath85 and normalized in the same way as for the analytic calculation described in the previous section .
we henceforth refer to the maps produced with the @xcite gas prescription as the bo gas model sample .
in total we produce 96 25 deg@xmath2 maps ( 2400 deg@xmath2 in total of simulated sky ) for each gas profile . in figure [
fig : power_allfields ] we plot the sz angular power spectrum measured for each of the 25 deg@xmath2 simulated maps ( grey lines ) in our bo gas model sample .
the solid black line represents the mean power spectrum measured for the bo gas model maps ; the red solid line the halo model predictions for the ks gas model .
the black error bars denote the @xmath86 statistical errors around the mean value @xmath7 for the bo map sample .
the blue error - bars denote the standard deviation expected from gaussian cosmic variance alone ( @xmath87 , the first term in eqn .
[ eqn : variance ] ) .
the simulated power spectra are band - averaged in bins of @xmath88 and are apodized with a 2d gaussian window function with unity variance .
the analytic power spectra have been convolved with a gaussian of fwhm 0.25 arcminutes in order to approximately account for the pixelisation of the simulated maps .
we also note that the analytic power spectra are calculated omitting the contribution of clusters with @xmath89 .
this is because the simulated maps are only 100% complete down to this mass , and so we have omitted lower mass clusters from the simulated sky - maps .
we note that the power spectra measured for the bo gas model simulations agree with the ks power spectra at very small angular scales ( @xmath90 ) but have significantly more power at larger scales .
this additional power is most likely due to the impact of the feedback parameter in the bo model .
the additional energy added to the gas has the effect of heating and ` puffing out ' the gas , reducing the slope of the pressure profiles in the outer regions and increasing the large - scale signal .
we also note that the ks model has a relatively steep sz profile at larger radii , reducing the signal at larger scales .
the full impact of non - gravitational heating sources on the icm gas temperature and density profiles is currently poorly understood ; the ks presciption is not astrophysically well - motivated and there is no reason to expect the model to match either the bo gas model or real clusters at high redshift .
hence there remains a large theoretical uncertainty in @xmath0 which must be accounted for when attempting to constrain @xmath1 via the sz power spectrum .
the exact range of uncertainty in @xmath0 between different models and simulations has not been quantified and we leave this issue to future work . we have confirmed that the power spectra measured from our sample of @xmath38-model maps matches the equivalent predictions from the halo model approach . as demonstrated by the error
bars in figure [ fig : power_allfields ] , the total field - to - field variation in the power spectrum amplitude is several times greater than the gaussian error , with the factor decreasing with increasing @xmath41 .
this is due to poisson noise associated with the discrete sampling of the halo mass function in each field . comparing the individual simulated power - spectra to the error - bars ,
it is also clear that the distribution of @xmath0 is highly non - gaussian , especially for @xmath91 .
the power spectra for individual maps ( grey lines ) indicate that the non - gaussian variance is being driven by fields with @xmath92 ( where @xmath7 represents the mean ) , corresponding to fields that contain at least one very massive cluster . in figure
[ fig : tll_allmods ] we plot the ratio of the non - gaussian and gaussian contributions to the total standard deviation in the statistical distribution of @xmath0 , @xmath93 ( the second term divided by first term in eqn .
[ eqn : variance ] ) , as a function of angular multipole number .
the total variance of p(@xmath0 ) is therefore just the sum of @xmath94 and @xmath95 , although we reiterate p(@xmath0 ) is a strongly non - gaussian function at larger scales and thus the total variance ( and mean ) does not completely define this distribution .
@xmath96 is measured from the simulated maps by calculating the variance in @xmath0 for each set of maps , multiplying this by @xmath97 and deducting the gaussian variance , calculated using @xmath98 .
the heavy lines represent the results for our simulated @xmath38-model ( solid ) and bo gas model ( dashed ) maps .
the thin lines are the equivalent results for the @xmath38-model and ks halo model calculations , respectively . comparing thick and thin solid black lines , we find that the non - gaussian error measured for our set of simulated @xmath38-model maps is approximately one third greater than that predicted by the trispectrum . as we find a good agreement between the simulated and predicted power spectra , this discrepency must be due to effects that increase the width of p(@xmath0 ) without influencing strongly the ensemble - averaged mean .
one such effect is the non - linear bias , or small scale halo clustering , which is not accounted for in the trispectrum . as noted in section [ sec : halomodel ] and @xcite , the two halo ( and higher - order ) contribution to the trispectrum which accounts for the linear bias is insignificant , including the two - halo term increases @xmath99 by only a few percent .
we have found that by remeasuring @xmath99 for both sets of simulated maps having first randomised the halo positions reduces @xmath100 such that the simulated @xmath38-model line becomes into better agreement with that of the trispectrum result , although remains above the predicted curve at all @xmath41 .
the thick and thin dashed lines in figure [ fig : tll_allmods ] show the ratio of the non - gaussian to gaussian variance for bo gas model simulations and the analytic ks model .
the simulations demonstrate a shallower dependence on @xmath41 than the halo model calculation . given that @xmath101 , the flatness of the simulated curve is likely due to the larger mean power simulated spectrum in the simulated maps compared to the ks model predictions .
we note again however that there is no reason why an agreement is expected .
indeed , as the simulation results include large variations in the underlying dark matter halo structure ( including morphology and substructure ) we would expect @xmath93 to be greater , especially at small angular scales .
in section [ sec : halomodel ] we showed that the primary contribution to the sz power spectrum comes from clusters occupying a different region of the cluster mass - redshift plane to those that determine the trispectrum .
clusters at @xmath102 and of mass @xmath103 contribute most strongly to the power spectrum , whereas clusters at @xmath104 are the foremost contributors to the trispectrum .
therefore , if one wishes to measure accurately the mean sz power spectrum , very high mass clusters constitute a source of noise .
one can mask or remove the brightest clusters detected in a sz map down to some threshold density in order to suppress the trispectrum contribution to the total variance in @xmath0 without removing a large component of the signal from the power spectrum .
we now demonstrate this in practice using our sample of simulated maps .
to evalulate the sensitivity of the sz power spectrum to @xmath1 , we use a second lightcone simulation to create an additional sample of mock sz maps .
the cosmological parameters were chosen to be consistent with those measured from the 5th - year wmap data @xcite , therefore @xmath105 for this simulation ( compared to @xmath106 as for our wmap3 run ) .
the simulated volume is a periodic cube of size @xmath107mpc ; the particle mass @xmath108 and the cubic spline softening length @xmath109kpc .
we have applied the @xcite gas prescription to the halos in the wmap5 lightcone , and created a second set of mock sz sky maps using the procedure described in the previous section .
we refer to the @xmath49 maps as the wmap3 simulation , and the @xmath110 maps as the wmap5 simulation .
clusters are masked from the simulated maps in the following manner .
for each of the 96 maps in the wmap3 & 5 samples , there is an associated catalogue listing all clusters in map .
we combine all 96 catalogues and rank - order clusters by their integrated sz flux @xmath111 .
starting with the brightest cluster , we then fit and remove an isothermal @xmath38-model ( with @xmath38 fixed at 1 ) from the map at the location of the cluster .
this is then repeated until all clusters down to the desired surface density threshold @xmath53 have been masked .
this process can be performed in practice on real sz survey maps by ranking detected clusters by their detection significance , and removing each in turn down to the required threshold surface density ( and is somewhat analogous to the application of an optimal matched - filter , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the upper panel of figure [ fig : cl_tll_masked ] demonstrates the mean power spectrum measured with no clusters masked ( black line ) , and having masked clusters down to 0.01 , 0.1 and 0.9 deg@xmath5 ( blue , green and red lines ) .
the solid lines show the results for our wmap3 simulation , the dashed lines for our wmap5 lightcone simulation . for minimal masking , power is removed only on large angular scales , shifting to smaller scales as increasing numbers of clusters are masked . removing 0.01 clusters per square degree
barely changes the average power spectrum , whilst masking 0.1 clusters per square degree reduces power by less than a third at @xmath58 , and negligibly at @xmath112 .
masking 0.9 clusters deg@xmath5 removes power over a wide range of scales , reducing the mean band - averaged power by a factor of 2.5 at @xmath58 and by 27% at @xmath113 .
as expected , the most massive clusters contribute most of their power at large angular scales , and that at scales of @xmath114 low mass clusters contribute the majority of the signal .
these results are similar to those shown in figure [ fig : dcldsig8 ] . in the lower panel of figure [ fig : cl_tll_masked ]
we plot the sensitivity of the sz power spectrum to the difference in @xmath1 between the two simulation samples ( @xmath115 ) including cluster masking , divided by the sensitivity for the unmasked case .
we define sensitivity as being the difference in @xmath7 between the two simulations divided by the standard deviation in @xmath0 , @xmath116 measured in the @xmath49 sample of maps .
the line colours are the same as in the upper panel .
it is immediately evident that whilst removing down to 0.01 clusters per square degree has little impact on the mean power spectrum , there is a clear gain in sensitivity below @xmath3 . masking just the 10 brightest clusters per 100
deg@xmath2 improves the sensitivity by a factor of two at @xmath3 and 1.5 at @xmath4 .
masking the 90 brightest clusters in the same area enables similar improvements at @xmath117 and @xmath118 .
the improvement in sensitivity measured for our simulated maps agrees well with the analytic predictions from the halo model approach ( see the lowest panel in figure [ fig : dcldsig8 ] ) .
in addition , as one masks out the brightest clusters the pdf of @xmath0 becomes significantly more gaussian .
this is demonstrated in figure [ fig : pcl ] , in which we plot the distribution of @xmath119 at l = 1000 , 2500 , 6000 and 10,000 for our wmap3 sample .
the unshaded , dashed histogram represents p(@xmath120 ) for the unmasked case , the shaded histogram represents p(@xmath119 ) having masked bright clusters down to 0.9 clusters per square degree .
for the unmasked case , it is clear that the level of non - gaussianity is significantly greater at large scales , where massive clusters contribute most of their power .
it is also immediately apparent that the masked distribution is significantly more gaussian than the unmasked case .
this can be quantified by the reduction in skewness between the masked and masked distributions , which is 1.0 , 0.5 , 0.2 and 0.1 at @xmath41 = 1000 , 2500 , 6000 and 10,000 .
we have demonstrated the utility of masking clusters from sz sky maps to reduce the non - gaussian variance in the amplitude of the sz power spectrum and thus improve the precision with which one can measure the ensemble - averaged @xmath0 .
however , there still remains a question of the overall gain in sensitivity once one accounts for astrophysical foregrounds , such as the primary cmb temperature anisotropies and the contribution from radio and sub - mm point source populations , and instrument noise . while masking bright clusters reduces the field - to - field variance in @xmath0 , the amplitude of the ensemble - averaged @xmath0 is also reduced . masking increasing numbers of clusters thus reduces the amount of sz power relative to that of the foregrounds .
the variance in the foreground signal is proportional to @xmath121 , therefore as the sz signal is reduced by masking , the total fractional error on the sz amplitude due to the error in the foreground signals increases .
the relevant question is , how does the total intrinsic error ( gaussian plus non - gaussian ) on the sz signal @xmath122 compare to the error ( or uncertainty ) on the the primary cmb and point source power spectra amplitudes within a given field ?
if @xmath99 dominates the error in the total power measured on the sky , @xmath123 , where @xmath124 then one can simply remove the foreground components from the measured power spectrum , and clipping the bright clusters will improve the precision on the measured @xmath125 .
on the other hand , if the error due to cosmic variance on @xmath126 or @xmath127 is comparable to @xmath128 then reducing the amplitude of the sz signal by masking clusters may degrade the measured sz power spectrum amplitude . for mm - wavelength experiments ,
the variance on the total power spectrum due to all sources can be written as the sum in quadrature of the gaussian sample variance term ( including cmb , point sources , gaussian sz error and instrument noise ) , the non - gaussian sz error and a theory error , @xmath129 where we have included an error on the predicted sz power spectrum , @xmath130 , due to the uncertainties in modeling cluster gas physics .
using eqn .
[ eqn : variance ] this can be rewritten as @xmath131 + t^2 \ ; , \label{eqn : total_variance}\ ] ] where @xmath132 and @xmath133 is the ratio of the masked to unmasked @xmath134 s for a given mask surface density threshold @xmath53 , and @xmath135 is the fractional error on @xmath134 due to the theoretical uncertainties in cluster gas physics .
@xmath136 is defined as the ratio of foreground to sz power , @xmath137 where @xmath138 is the fourier domain profile of the telescope beam .
we henceforth assume a 10% theoretical uncertainty on @xmath0 , and @xmath139 instrument noise in an arcminute pixel for a single - frequency survey at 90ghz of 25 deg@xmath2 of sky with a gaussian beam of fwhm 1.67 arcminutes .
the values of @xmath140 are measured from our wmap3 simulation sz map sample , with @xmath141 .
the point source power spectrum is calculated assuming radio and sub - mm source populations and spectral index distributions as given in @xcite and @xcite .
the upper panel of figure [ fig : frac_var_foregrounds ] shows the total fractional error on @xmath134 as a function of angular multipole number .
the solid black line gives the total fractional error ( as given in eqn .
[ eqn : total_variance ] ) for no masking ( @xmath142 ) .
the black dot - dashed , dashed and dotted lines show the contribution of the sz trispectrum , the total error on the combined foreground signals ( including instrument noise ) , and the theory uncertainty on @xmath0 , respectively .
the red lines show the contributions of the cmb plus gaussian sz variance ( dashed ) and that of the point sources and instrument noise ( dotted ) .
it is evident that in the range @xmath143 the non - gaussian variance in the sz power spectrum amplitude is the dominant source of error on @xmath134 , with cosmic variance in the cmb power dominating at larger scales , and that of the point sources and instrument noise dominating at smaller scales .
it is clear that in this range , @xmath134 can be more accurately measured if one can reduce the non - gaussian variance @xmath99 by masking clusters .
the lower panel of figure [ fig : frac_var_foregrounds ] demonstrates the reduction in the fractional error on @xmath134 having masked bright clusters down to the surface density threshold @xmath54 ( given in the legend ) , compared to the unmasked case . for @xmath144 , the gaussian variance on the cmb amplitude dominates and so masking down to 0.1 clusters deg@xmath5 ( dashed line )
merely reduces the amplitude of the sz signal with no associated reduction in the error ( @xmath116 ) .
thus the total fractional error increases relative to the unmasked case .
however , masking clusters down to 0.25 per square degree ( dotted ) gives a 20% reduction in the fractional error on @xmath134 compared to the unmasked case in the range @xmath145 .
we have found that 0.9 clusters deg@xmath5 ( dash - dotted ) gives the best results ; more aggressive masking reduces the amplitude of the sz power to the point at which the error on the foregrounds becomes significant . above @xmath146 , the variance due to point sources and instrument noise dominates and so masking clusters simply diminishes the amplitude of the sz signal without an associated reduction in the variance .
as at @xmath147 , this results in a decreased sensitivity to @xmath148 at these scales .
the red dotted line shows the reduction in the fractional error ( for @xmath149 ) if one can perfectly remove the primary cmb contribution using multi - frequency observations .
overall , our results show that , after accounting for the various foregrounds , for a simulated single frequency survey with @xmath139 instrumental noise a 20% improvement in the measurement of the amplitude of the sz angular power spectrum can be obtained if one masks out the brightest clusters from the map down to a surface density threshold of at least 0.25 clusters degree@xmath5 .
this corresponds to a reduction of @xmath150 on the error on @xmath1 , although we note the additional gain in the gaussianized statistics of p(@xmath134 ) .
we have demonstrated that the sz power spectrum has a significant amount of variance contributed by rare massive clusters .
while these objects do not contribute heavily to the ensemble - average sz power spectrum , they do add significant non - gaussian noise to measurements of the power spectrum amplitude and thus degrade cosmological utility .
furthermore , it is difficult to use numerical simulations to characterize the statistics of the sz power spectrum because one must have sufficiently large volumes to be able to adequately sample the rarest objects in the universe . we have also shown that the magnitude and scale dependence of the non - gaussian variance is sensitive to the modelling of the intra - cluster gas .
a simple solution is to mask out the largest objects in the map before estimating the power spectrum .
we have demonstrated using both halo model calculations and a large sample of simulated sz sky maps that fairly conservative masking can reduce the variance and gaussianize the statistics significantly , increasing the precision of the measured @xmath0 and thus the sensitivity to cosmological parameters .
for example , masking the 10 brightest clusters in a 100 deg@xmath2 sz map will improve the precision on @xmath0 by a factor of two at @xmath151 and 1.5 at @xmath4 . masking the 90 brightest clusters in the same region
should improve the accuracy by the same factors at @xmath4 and @xmath118 .
this is in agreement with work related to information content in the matter power spectrum in the trans - linear regime , where it has been found that a few massive objects are responsible for much of the variance @xcite .
@xcite have also demonstrated that knowledge of the number of massive clusters within a given survey field can help improve the precision of measurements of the weak lensing shear power spectrum . choosing which objects to mask is non - trivial ; we found that using a fixed sky density produced a well - defined and well - behaved estimate that can easily be applied to real maps .
this skirts the issue ( at least partially ) of finding an observable that is well - correlated with mass , and can be easily applied to any survey ( for example , weak lensing ) .
we have also shown that the non - gaussian variance in the sz power spectrum remains a significant source of error in the range @xmath152 , even in the presence of the astrophysical foregrounds provided by the primary cmb temperature anisotropies and unresolved point source populations .
we have demonstrated that even single frequency experiments with no ability to spectrally subtract the cmb and point sources , can improve the accuracy on @xmath134 by more than 20% in the range @xmath145 by masking the brightest clusters down to between 0.25 and 0.9 clusters per square degree .
we emphasize that one is not losing cosmological information by masking bright clusters since our procedure is a way to discard the signal from massive clusters whose information we do not know how to analyze but which still contributes to the variance .
halo model calculations of the sz power spectrum demonstrate that the ensemble - averaged power spectrum ( @xmath153 ) to which high mass clusters do not contribute significantly has a sensitive dependence on @xmath1 @xcite . if one had prior knowledge of the number density and gas properties of clusters over a wide mass and redshift range it would be possible to construct an optimal power spectrum estimator in which the signal from high mass clusters would be down - weighted . however , in this case one would already know the cluster mass function which is more sensitive to cosmological parameters and thus power spectrum measurements would reveal no additional information . in the absence of complete knowledge of the mass function , one can simply apply zero weight to the brightest clusters as they effectively contribute only noise to the power spectrum . on the other hand ,
masked clusters carry much statistical weight when used to measure the cluster mass function .
there are still theory challenges to predicting the detailed shape and amplitude of the sz power spectrum , but we have demonstrated that the field - to - field variance can be reduced to allow precise tests of the physics and astrophysics involved .
this work supported by the natural sciences and engineering research council ( canada ) through the discovery grant awards to gph .
gph would also like to acknowledge support from the canadian institute for advanced research and the canada research chairs program .
this research was facilitated in part by allocations of time on the cosmos supercomputer at damtp in cambridge , a uk - ccc facility supported by hefce and pparc .
computer simulations were supported by the national science foundation through teragrid resources provided by pittsburgh supercomputing center and the national center for supercomputing applications under grant ast070015 .
we would like to thank paul bode for providing the n - body simulations , and niayesh afshordi , keith vanderlinde , paul bode and nicholas hall for useful discussions .
|
using both halo model calculations and a large sample of simulated sz maps , we demonstrate that high - mass clusters add significant non - gaussian variance to measurements of the sz power spectrum amplitude .
the difficulty in correctly accounting theoretically for the contribution of these objects to the uncertainty in @xmath0 leads to a reduced sensitivity to @xmath1 .
we show that a simple solution is to mask out the brightest clusters in the map before measuring the power spectrum .
we demonstrate that fairly conservative masking can reduce the variance and gaussianize the statistics significantly , thus increasing the sensitivity to cosmological parameters . choosing which objects to mask is non - trivial ; we found that using a fixed sky density produced a well - defined and well - behaved estimate that can easily be applied to real maps .
for example , masking the 10 ( 90 ) brightest clusters in a 100 deg@xmath2 sz map will improve the sensitivity to @xmath0 by a factor of two at @xmath3 ( 2000 ) and 1.5 at @xmath4 ( 4000 ) .
we show that even in the presence of astrophysical foregrounds ( primary cmb and point sources ) and instrument noise , one can increase the precision on measurements of @xmath1 by masking up to 0.9 clusters deg@xmath5 .
|
many real - world networks show properties of small - world networks as their neighborhood connectivity , generally denoted by the clustering coefficient , is higher than in comparable random networks@xcite .
the local clustering coefficient for an individual node @xmath2 with @xmath3 neighbors and @xmath4 edges between its neighbors is @xmath5 this formula is basically not defined if the number of neighbors @xmath3 becomes zero or one as the denominator becomes zero @xcite .
these cases are usually treated as @xmath6 although some authors also set these values to one @xcite . in the current scheme
, these values would be part of the global calculation @xmath7 in addition , we tested an alternative and more widely used definition of the clustering coefficient @xcite in which @xmath8 this might lead to biased assessments of neighborhood clustering in the sense that values that are not defined ( division by zero ) should not be included in the averaging .
thus , instead of using @xmath9 as the number of evaluated nodes for the global @xmath10 , a new number @xmath11 indicating all nodes with defined local clustering should be used for a global measure @xmath12 .
we show that using such an adjusted measure for the clustering coefficient has several implications for network analysis and can help to identify the contribution of leafs and isolated nodes on average clustering . on a conceptual level ,
the adjusted value @xmath12 is more intuitive as the clustering coefficient is commonly called a measure of neighborhood connectivity : if 30% of the local coefficients are zeros from cases where no neighbors exist , how can the classical definitions still give information about neighborhood ? cases of leafs and isolated nodes are more likely in sparse networks where the edge density @xmath13 , the number of existing divided by the number of possible connections ( @xmath14 for a network with @xmath9 nodes and @xmath15 directed edges or arcs ) , is low .
therefore , the classical definition is a mixed measure of neighborhood clustering and sparseness ( edge density ) or more precisely the frequency of leafs and isolated nodes .
a general problem of network measures , such as the clustering coefficient , is whether sampling or perturbations change the values of these measures .
network measures are frequently used for the classification of different networks @xcite or of topological changes ( addition or deletion of nodes or edges ) within the same network .
incomplete sampling only observing a sub - network of a larger network can lead to the wrong classification of a network as being a scale - free network @xcite .
this occurred , for example , for comparing the partial and complete protein - protein interaction networks @xcite and the router and underlying communication network @xcite .
in addition to sampling , false scale - free classifications can also arise due to statistical errors @xcite .
whereas previous studies investigated the effect of sampling on the degree distribution , a recent study @xcite looked at the sensitivity to sampling and network perturbation for a range of measures : the clustering coefficient , as well as the hierarchical clustering coefficient , the hierarchical degree , and the divergence ratio were found to be least sensitive to perturbations of the topology .
therefore , classifications using the clustering coefficient ( e.g. small - world classification @xcite ) are less affected by the sampling problem . however , as we show here , the definition of the clustering coefficient can have a considerable effect on network classification .
we tested the effect of different definitions for the clustering coefficient on several real - world networks .
all but one network , the german highway system , were small - world networks .
the _ caenorhabditis elegans _ neuronal network consisted of individual neurons as nodes and existing synaptic connections as edges @xcite .
the metabolic networks of _ c. elegans _ , _ saccharomyces cerevisiae _ , and 41 other organisms included metabolic substrates as nodes and reactions as edges @xcite .
the protein - protein interaction network of _ s. cerevisiae _ ( yeast ) included proteins as well as interactions as discovered by the yeast two - hybrid method ( http://dip.doe-mbi.ucla.edu , dataset from 2 dec 2007 ) .
the german highway ( autobahn ) system consisted of location nodes ( that is , highway exits ) and road links between them ( autobahn - informations - system , ais , from http://www.bast.de ) @xcite .
only the gross level of highways were included in the analysis , discarding smaller and local roads ( bundesstrassen and landstrassen ) .
for the power grid , nodes represent generators , transformers and substations , and edges represent high - voltage transmission lines between them @xcite . for the world - wide - web ,
individual pages are the nodes and links between them the edges @xcite .
information about the size of the networks as well as a reference to the source of the datasets is included in table [ tab.2 ] . for comparisons
, we also generated random networks with the same number of nodes and edges as the original networks described above .
in such erds - rnyi random networks @xcite , the probability @xmath16 that an individual connection between two nodes is established equals the edge density @xmath13 of the desired network . [ cols="<,>,<,<,<",options="header " , ] many real - world networks show features of small - world networks @xcite . in these networks ,
the characteristic path length @xmath17 remains comparable with random benchmark networks whereas the average connectivity between neighbors ( clustering coefficient ) @xmath18 of a node is much higher than for random networks , that means @xmath19 but @xmath20 @xcite .
one way to assess the extent of small - world features is calculating the small - worldness @xmath21 ( note that a comparison of small - worldness is only meaningful for similar edge densities as the edge density influences the possible increase in the clustering coefficient ) .
how do these small - world features , in particular the clustering coefficient component of the small - worldness @xmath22 , change with the definition of the clustering coefficient ?
using the measure @xmath12 would lead to higher small - worldness @xmath22the ratio between the clustering coefficient in the original and a comparable random network if numerator @xmath18 for the original networks increases at least as much as the denominator @xmath23 for the random benchmark networks .
we tested the increase of changing from definition @xmath10 to @xmath12 which can be calculated by the ratio @xmath0 ( cf .
equation [ eq.4 ] ) : a larger value of @xmath0 results in a larger increase of the clustering coefficient .
therefore , the small - worldness would increase as long as @xmath0 is larger for the original rather than the random benchmark network .
we tested the increase for the 43 metabolic networks by generating 50 random networks for each metabolic network and using the maximum value of @xmath0 out of the random networks .
for all 43 metabolic networks , @xmath0 was larger for the original network than for random benchmark networks ( fig .
[ fig.4]a ) .
we also generated artificial small - world networks with 100 nodes and a variable edge density ranging from the minimum ( 0.5% ) to the maximum ( 2% ) value of the metabolic networks .
for each edge density , 20 small - world networks were generated and for each such small - world network , 50 comparable random networks were analyzed .
in contrast to the previous results of real - world networks , random networks show a higher ratio of leafs and isolated nodes than the generated small - world networks ( fig .
[ fig.4]b ) .
the reason is that the small - world networks were generated starting from a lattice model followed by random rewiring of the network @xcite . despite the rewiring ,
the strong neighborhood connectivity of the lattice model remains and prevents the occurrence of leafs and isolated nodes . to remove the effect of the lattice network being the starting point of rewiring
, we developed a small - world network generator with _ _ inverse rewiring _ _ : the model starts with a random network and rewires edges so that the connectedness but also the number of isolated nodes increases . for a given network with @xmath15 edges ,
@xmath24 rewiring steps were performed . at each step ,
an existing edge is chosen and deleted .
thereafter , another existing edge is chosen and the starting node of that edge is connected with a randomly chosen node that has not before been connected to that node .
each step elongates an existing chain of nodes by adding an edge , potentially leading to the formation of triangles , whereas the removed edge is either the internal or terminal part of a chain , leading towards a leaf node . for this model , in accordance with the results from the real - world metabolic networks , @xmath0 for the generated small - world network was below the value for random networks ( fig .
[ fig.4]c ) .
we have seen in the previous section that the small - worldness @xmath22 of a network increases , or at least stays the same , when the new measure @xmath12 is used compared to the classical measure @xmath10 .
looking at both measures @xmath10 and @xmath25 could it be the case that networks that were previously classified as random would be classified as small - world with the new measure @xmath12
? this would be the case if the ratio @xmath26 is lower or close to 1 for the classical measures but much higher than 1 for @xmath12 .
again we tested artificial small - world networks with 100 nodes and a variable edge density using standard and inverse rewiring as described above . for each edge density ,
200 networks were generated and 100 benchmark random networks were evaluated for each generated network . the definition of a change in classification from random to small - world was a clustering coefficient ratio @xmath27 for @xmath10 or @xmath25 and a ratio @xmath28 when using the measure @xmath12 . for standard rewiring ( fig .
[ fig.classification]a ) , the fraction of changed cases was zero except for a small range of edge densities where the classification changed in up to 2.5% of the cases when shifting from @xmath25 to @xmath12 . for inverse rewiring ( fig .
[ fig.classification]b ) , however , classification changed in around 10% of the cases ( up to 15% for some edge densities ) when shifting from @xmath25 to @xmath12 in the edge density range of 0.6%-1.5%
. for shifting from @xmath10 to @xmath12 , classification only changed up to 1% of the cases ; a fluctuation which might be due to the small sample size .
in addition , a shift also occurred between the classical measures @xmath10 and @xmath25 : whereas changing from @xmath10 to @xmath25 affected few cases , a shift from @xmath25 to @xmath10 affected up to 3% of the cases for standard rewiring and up to 14% for inverse rewiring . therefore , changes in classification are possible for all clustering measures ; in particular when using inverse rewiring .
( x markers ) or @xmath25 ( o markers ) to @xmath12 .
( a ) small - world networks generated by rewiring starting with a lattice model ( inset ) ( b ) small - world networks generated by condensation starting with a random model ( inset ) . ]
we have shown that current definitions underestimate neighborhood clustering in sparse networks with many isolated or leaf nodes .
in addition , the outcome of comparisons of the extent of small - world features between different networks critically depended on the applied definition of the clustering coefficient .
furthermore , networks formerly classified as random can be classified as small - world when isolated or leaf nodes are excluded from the calculation of the average clustering coefficient .
this can also happen when switching from @xmath25 to @xmath10 .
could the clustering coefficient definitions impact the analysis of small - world networks ? there are three consequences of this study .
first , small - world networks regarding previous measures @xmath10 and @xmath25 will still be detected as small - world using @xmath12 as this value will be higher than the previous values .
consequently , the small - worldness @xmath22the ratio of the clustering coefficient in the original and random benchmark networks divided by the unchanged ratio of the characteristic path lengths in original and random networks will be higher .
second , networks which are currently not classified as small - world networks may be regarded as small - world due to the increase in clustering coefficient .
this case will occur when the path length is comparable to that of random networks but the clustering coefficient , concerning previous definitions @xmath10 and @xmath25 , is not significantly higher than that of random networks .
third , comparison of networks could lead to opposite conclusions using the new measure . in conclusion
, the novel measure @xmath12 gives a clearer view of neighborhood connectivity and is more independent of the sparseness of edge density . a problem of the proposed measure @xmath12is that the percentage @xmath0 of nodes that are excluded from analysis could be considerably high ( tab . [ tab.2 ] ) .
the percentage of excluded nodes could be as high as 14% for metabolic networks and as high as 59% for man - made networks ( power grid ) .
note that the value for the protein interaction network in yeast is also high at 56% as edge density is low and isolated nodes are not part of the largest connected cluster observed here .
in general , however , exclusion from the average affected less than 10% for most of the networks .
in addition , using a subset of defined nodes is comparable to the procedure for calculating shortest paths or the characteristic path length where unreachable paths with otherwise infinite distance are not included in calculating the average path length
. an alternative solution would be to describe the clustering coefficient using inverse neighborhood clustering .
for the shortest paths , for example , the inverse measure of efficiency @xcite where unreachable paths contribute @xmath29 to the local efficiency circumvents the need for excluding unreachable paths .
similarly , the ( neighborhood ) disconnectedness @xmath1 could be defined as : @xmath30 @xmath31 here , nodes which are leafs or isolated will contribute a zero value to the average @xmath1 as one of the degrees will be zero for these nodes .
@xmath1 will be high when neighbors are not connected and low ( @xmath321 ) for high connectivity between neighbors .
the correlations between disconnectedness and measures of the clustering coefficient are shown in figure [ fig.5 ] . and measures for the clustering coefficient in 43 metabolic networks .
correlations are @xmath33 between @xmath10 and @xmath1 ( a ) , @xmath34 between @xmath25 and @xmath1 ( b ) , and @xmath35 between @xmath12 and @xmath1 ( c ) . ]
including the percentage @xmath0 in publications could help to understand the validity of the applied definition of the clustering coefficient regardless of whether it is the watts - strogatz definition @xmath10 , the newman - strogatz - watts definition @xmath25 , or the alternative definition @xmath12 presented here .
in addition , this information is critical for the classification as small - world networks .
we therefore suggest that information about the applied definition and the number of leafs and isolated nodes , the ratio @xmath0 , should be included in addition to the value of the average clustering coefficient .
|
many networks exhibit the small - world property of the neighborhood connectivity being higher than in comparable random networks .
however , the standard measure of local neighborhood clustering is typically not defined if a node has one or no neighbor . in such cases , local clustering
has traditionally been set to zero and this value influenced the global clustering coefficient .
such a procedure leads to under - estimation of the neighborhood clustering in sparse networks .
we propose to include @xmath0 as the proportion of leafs and isolated nodes to estimate the contribution of these cases and provide a formula for estimating a clustering coefficient excluding these cases from the watts & strogatz @xcite definition of the clustering coefficient . excluding leafs and isolated nodes leads to values which are up to 140% higher than the traditional values for the observed networks indicating that neighborhood connectivity is normally underestimated .
we find that the definition of the clustering coefficient has a major effect when comparing different networks . for metabolic networks of 43 organisms ,
relations changed for 58% of the comparisons
when a different definition was applied .
we also show that the definition influences small - world features and that the classification can change from non - small - world to small - world network .
we discuss the use of an alternative measure , disconnectedness @xmath1 , which is less influenced by leafs and isolated nodes .
|
time dilation is a fundamental property of an expanding universe . in fact the increase of timescale by a factor of @xmath0 represents a basic link between redshift and time which is essentially related to the definition of time and is independant of cosmological model parameters . as a consequence , time dilation has generally been assumed to be a property of the universe even though it has proved hard to measure directly .
recently there has been new interest in time dilation as a result of experiments where its effect is large and must be taken into account .
there have been a number of claims by groups working on gamma ray bursters @xcite that time dilation is seen in the stretching of peak - to - peak timescales .
this has then been used to support the argument that the bursts are at cosmological distances .
it is not clear however that the argument can be inverted to provide convincing evidence for the existence of time dilation .
a more direct observation of time dilation has come from the measurement of the decay time of distant supernova light curves and spectra @xcite . here
one can make a good case that the rest frame timescale is known , and hence directly detect any time dilation effect at high redshift .
the results so far published are very convincing , and strongly imply that time dilation has been observed .
another situation where one would expect to see a time dilation effect is in the light curves of quasars , which are at cosmological distances and vary on a timescale of years . a number of groups have looked for time dilation in quasar light curves @xcite , but so far it seems fair to say that no convincing detection has been made , although one might argue that the sample sizes and data analysis procedures were not adequate to detect the effect of time dilation if it were present .
in order to measure time dilation in quasar light curves it is necessary to find a way of characterising the timescale of variation .
the most popular parameterisation to date has been the structure function , and several groups have measured it for samples of light curves @xcite , as well as predicting its shape for various models of quasar variability @xcite . the autocorrelation function , which is closely related to the structure function , has also been used @xcite .
the main drawback of these functions is that the points are not independent of each other , which causes difficulties with error analysis , and makes them very hard to interpret .
fourier power spectrum analysis has not been used much in the analysis of quasar light curves , probably because it requires a long run of evenly spaced data to be effective . however ,
given such a dataset it provides some significant advantages over other methods , perhaps the most important of which is the relative ease with which it may be interpreted . in this paper
we apply it to a large sample of quasars which have been homogeneously monitored every year for 24 years @xcite .
the survey is based on a long series of uk 1.2 m schmidt plates of the eso / serc field 287 centred on 21h 38 m , -45@xmath1 .
the plates were taken over a variety of timescales from hours to years , and in various passbands including @xmath2 , @xmath3 and @xmath4 .
of particular relevance to this paper is a regular yearly monitoring of the field in the @xmath5 passband ( kodak iiia - j emulsion with a schott gg395 filter ) from 1977 till 2000 . for most years
four plates were obtained , but in a few cases it was only possible to obtain one . a similar series of 18 yearly measures was also obtained in @xmath3 , from 1983 till 2000 .
the plates were measured by the cosmos or supercosmos machines at the university of edinburgh to give a catalogue of photometric measures for some 200,000 objects in the central 19 square degrees of the field .
the photometric error on an observation from a single plate is about 0.08 mag .
when 4 plates were available the mean magnitude was used giving a photometric error of about 0.04 mag . more details of the reduction procedure and error analysis are given by @xcite and references therein .
the quasars in the field were found by a variety of techniques , including ultra - violet excess , variability , blue drop - out and objective prism .
altogether some 600 quasars have now been identified , with confirming redshifts in the range @xmath6 .
there are sufficient numbers that the quasars can be binned in both redshift and luminosity to avoid the well - known degeneracy between these two parameters .
all the quasars used in this study fluctuated significantly in brightness over the 24 year monitoring period , with an amplitude of mode 0.6 mag and a tail extending to 2 mag . in order to compare the spectrum of variations of subsamples of quasars from the survey ,
a fourier power spectrum was calculated for each light curve .
the quasars were then binned in redshift and luminosity , in such a way that each bin contained approximately 100 objects , with a total of 407 quasars used for the analysis .
for the study of time dilation we first make the assumption that the light variations are intrinsic to the quasars , and so the light curves are subject to the effects of time dilation .
thus for each light curve we rescale the time interval by a factor of @xmath7 where @xmath8 is the redshift of the quasar , and re - sample the power spectra on a uniform scale .
this should remove the effects of time dilation , with the result that there should be no trend of timescale with redshift .
this has the effect of shifting the contributions of all quasars to higher frequencies .
it is a big effect , especially for high redshift quasars , in several cases resulting in low frequency bins being completely emptied .
[ fig1 ] shows the power spectra of samples of quasar light curves , binned according to redshift and luminosity .
each point represents the average of value of all the contributions to that frequency interval .
the top two panels show results for two luminosity bins , with power spectra for low and high redshift quasars being represented by filled and open circles respectively .
if the timescales of the quasar light curves were subject to time dilation one would expect no displacement between the two curves .
in fact , a @xmath9 test shows that for both luminosity bins the power spectra are not coincident at the 99% level .
the two power spectra are separated by about 0.15 in the log in both luminosity bins .
this is close to the offset produced by allowing for a @xmath0 scale change , on the basis of the mean redshifts of the bins .
there is also some indication that the power spectra have moved horizontally rather than vertically from the morphology of the distributions .
the bottom two panels show similar data for two redshift bins . in this case
low and high luminosity quasars are represented by closed and open circles respectively .
the mean redshift for the high and low luminosity data in each plot only differs by about 0.03 in the log , and so the removal of the @xmath0 factor should make little difference .
in fact the two luminosity bins are well separated , implying more power or shorter timescales for low luminosity objects .
( 240,240)(-120,0 ) we now make no assumptions about the nature of the quasar variability , but carry out the power spectrum analysis in the observer s reference frame .
[ fig2 ] shows power spectra of quasar light curves as for fig .
[ fig1 ] but this time with no correction for time dilation .
for the two luminosity bins in the top two panels it will be seen that all spectra show well - defined linear ( power law ) relations . on the basis of the mean redshift for each of the bins , the effects of time
dilation should result in a horizontal offset of 0.15 between the power spectra in each of the two top panels .
in fact , in each case the high and low redshift data are superimposed , showing no change of timescale with redshift .
this is confirmed by a @xmath9 test which shows both pairs of power spectra compatible at the 30% level .
the two redshift bins in the bottom panels again show well defined power laws for the power spectra , and in this case it is clear that low luminosity quasars have more power on shorter timescales .
this effect can also be seen by comparing the slopes of the power spectra for low and high luminosity quasars in the top two panels , and confirms that the power spectra are consistently measuring changes in timescale .
the implication from figs . [ fig1 ] and [ fig2 ] that quasars do not suffer the effects of time
dilation is hard to avoid .
there are however two possible ways out .
if the timescale of quasar variation were a function of wavelength in the sense that timescales were shorter in bluer passbands then this might possibly exactly offset the effect of time dilation .
however , this can be directly tested @xcite .
[ fig3 ] shows power spectra for a sample of about 200 quasar light curves in blue and red passbands . if there were a correlation of timescale with wavelength of the kind described above , then the two power spectra in fig .
[ fig3 ] should be separated by 0.18 on the log scale , corresponding to the effective wavelengths of 436 nm and 665 nm for the blue and red passbands .
in fact , the data for the blue light curves appear to be systematically offset from the red by 0.06 in the log , but a @xmath9 test shows no significant difference at the 10% level .
the only other alternative is that the timescale of quasar variation decreases by a factor @xmath0 towards high redshift by some as yet unspecified physical process , to exactly cancel out the time dilation effect .
such a ` cosmic conspiracy ' has no independent motivation , and would require a considerable degree of fine tuning .
in fact the shape of the plots in fig . [ fig1 ] would appear to make such fine tuning an unrealistic possibility .
there would appear to be three possible explanations for the lack of a time dilation effect in quasar light curves , all of which conflict with broad consensus in the astronomical community .
firstly , time dilation might not in fact be a property of the universe , which would effectively mean that the universe was not expanding .
apart from the overwhelming support for the big bang theory , the direct measurements of time dilation quoted above strongly argue against this .
the second possibilty is that quasars are not at cosmological distances .
this is an argument which was hotly disputed in the 1970s , with an emerging consensus favouring cosmological distances .
this has subsequently been strongly confirmed by studies of quasar host galaxies at high redshift .
the third possibility is that the observed variations are not intrinsic to the quasars but caused by some intervening process at lower redshift , such as gravitational microlensing .
although this idea has been strongly argued @xcite , there is an opposing view that variations in quasars are dominated by instabilities in the central accretion disc .
the reality of this mode of variability in active galactic nuclei is supported by detailed observations of seyfert galaxies @xcite and gravitationally lensed quasars @xcite , where the presence of intrinsic variations can not be in doubt .
the debate centres on whether this mechanism is responsible for the long timescale large amplitude variations which dominate the power spectra discussed in this paper .
taking the various arguments outlined above at face value , and accepting the case against microlensing , there does not appear to be a satisfactory explanation for the absence of a time dilation effect in quasar power spectra .
the arguments resting on an expanding universe and cosmological distances for quasars seem beyond challenge .
the argument against microlensing is not so secure .
apart from the statistical evidence from quasar light curves @xcite , microlensing has been unambiguously shown to take place in gravitationally lensed quasar systems @xcite , and dominates at long timescales . if this were a general phenomenon in quasars at cosmological distances then the apparent absence of a time dilation effect in quasar light curves would be explained .
|
the timescale of quasar variability is widely expected to show the effects of time dilation . in this paper
we analyse the fourier power spectra of a large sample of quasar light curves to look for such an effect .
we find that the timescale of quasar variation does not increase with redshift as required by time dilation .
possible explanations of this result all conflict with widely held consensus in the scientific community .
|
galactic cosmic rays ( gcrs ) are high energy particles that pass through the interstellar space in our milky way galaxy .
the most abundant components of gcrs are protons and @xmath0 particles .
these energetic particles can interact with the matters both in the interstellar space and interplanetary space .
the former can be used to investigate the origin and evolution of gcr itself@xcite , and the latter provide us an useful tool to investigate the cosmic ray exposure history of extraterrestrial bodies such as the planets and the meteorites@xcite . among these bodies
the moon is an unique one in our solar system . due to the atmosphere
free and old surface , the moon has recorded rich information about the solar system evolution and variations of the galactic cosmic rays ( gcrs ) in the past several billion years .
gcrs can bombard directly the solid surface of the moon .
the interactions of gcrs ( primary particles ) and the matters of the lunar surface can produce many protons , neutrons and mesons ( secondary particles ) and some nuclei that can not be synthesized by thermonuclear reactions .
these nuclei are called cosmogenic nuclei .
the materials returned by apollo and luna missions has provided good opportunities to study the cosmogenic nuclei with very high accuracy .
it can be expected that the lunar samples returned by chinese lunar exploration program in the near future will promote enormously the development of lunar and planetary science including the investigation of cosmogenic nuclei . to provide reliable method to interpret the measured data in the near future , it is time to investigate this subject beforehand from theoretical points of view . in principle ,
if both the concentration and production rate of a stable cosmogenic nuclei are known , one can calculate the cosmic - ray exposure ( cre ) age of that material .
the concentration of nuclei can be measured accurately in laboratory by accelerator mass spectrometer ( ams ) , but the calculation of the absolute production rate is much more difficult in general speaking .
fortunately , there is an important exception .
if the cre age is much longer than the half - life of a given radioactive cosmogenic nucleus , its production rate equals to the activity that can be measured directly ( see below ) . therefore , most of the cre ages are determined by using the radioactive - stable nuclei pairs , for instance @xmath1be-@xmath3ne , @xmath2al-@xmath3ne , @xmath4cl-@xmath4ar@xcite .
the cre ages of lunar rocks on the rim of lunar craters are particularly important to date the impact events that excavated these rocks to gcr irradiation@xcite .
these ages can be used to calibrate the relative ages of lunar surface determined by the morphology craters@xcite . in this work ,
we develop a monte carlo model based on geant4 software package to calculate the production rate of cosmogenic nuclei . before applying this model to extensive studies
, it is needed to test the validity of it . to this end
, we choose the apollo 15 long drill core as research object .
it is because its chemical composition , location , shielding depth of samples are known clearly .
in addition , apollo 15 drill core is the least disturbed sample among the three deep drill cores , apollo 15 , 16 and 17 .
the production rates of long - lived nuclei @xmath1be and @xmath2al are calculated and compared with experimental data .
this paper is oganized as follows : in section 2 we describe the theoretical model , and in section 3 we discuss the numerical results , a summary and outlook is given in section 4 .
the shape and size of meteoroids ( i.e. , the precursor of the meteorites ) are difficult to determine directly from that of meteorites due to the unknown degree of ablation and fragmentation . for simplicity ,
the meteoroids are usually regarded as a sphere with radius @xmath5 and uniform chemical composition . under this assumption , the general formula of the production rate of cosmogenic nuclei @xmath6 at the depth @xmath7 under the surface of a meteoroid
is shown as follows@xcite : @xmath8 where @xmath9 is the concentration of element @xmath10 in the meteoroid , @xmath11 is the differential flux of particle @xmath12 ( in this work we only consider protons and neutrons ) , @xmath13 is the excitation function of the involved nuclear reaction for the production of nuclide @xmath6 from element @xmath10 induced by particle @xmath12 at incident energy @xmath14 . before the meteoroids impacting the earth s atmosphere , they are irradiated by galactic cosmic rays . in this work , we take the most abundant proton and @xmath0 particles in gcrs as primary particles , which bombard the surface of meteoroids isotropically .
the fluxes of proton and @xmath0 particle are taken as follows@xcite : @xmath15 @xmath16 in these two formulas @xmath14 denotes the energy per nucleon in mev , @xmath17 is the solar modulation parameter in mv , and @xmath18 mev is the rest mass of proton . in eq.(2 ) @xmath19(@xmath20 s mev)@xmath21 , @xmath22 , @xmath23 . in eq.(3 ) @xmath24(@xmath20 s mev)@xmath21 , @xmath25 , @xmath26 .
these formulas are implemented into the geant4 code to sample the primary protons and @xmath0 particles .
when the high energy protons and @xmath0 particles bombard the meteoroid , they will interact with the atoms and nuclei therein .
the dominant processes are ionizing energy loss and spallation reactions . by ionizing energy loss
the energy of primary particles will decrease continuously . and by the spallation reactions
many secondary particles , such as protons , neutrons and mesons , are produced .
the secondary particles , especially neutrons will induce lower - energy ( @xmath27 mev ) reactions , and the very low - energy neutrons will account for neutron capture reactions . as shown in eq.(1 ) , in order to calculate the production rate one must search for a method to obtain the differential fluxes of proton and neutron in the irradiated body .
the monte carlo simulation is a good choice for these calculations . among various monte carlo
based software , the geant4@xcite has been used more and more widely in nuclear physics and space science@xcite . in our recent work@xcite
, the neutron fluxes escaped from the lunar surface has been investigated using the geant4 software . in the present work
we will also use this software to calculate the proton and neutron fluxes beneath the meteoroid surface . in order to obtain the differential fluxes of active particles ,
the energies of primary and secondary proton and neutron at different depths are recorded and saved as root files .
then we obtain the numerical differential fluxes of different particles by analyzing the root file .
however , due to the statistical nature of the monte carlo method , there are statistical fluctuations in the fluxes . to calculate the production rate of cosmogenic nuclei by eq.(1 )
, it is necessary to obtain the analytical expressions of proton and neutron fluxes by the best - fit procedure .
once the analytical expressions for fluxes are obtained , one can calculate the production rate of interesting nuclei by combining these fluxes formulas with the nuclear reaction cross sections .
the cross sections of proton induced reactions can be taken from experimental data .
however , the cross sections for high energy neutron induced reactions are obtained in indirect ways , for instance the thick target irradiation experiment .
it is because the neutron is a neutral particle that can not be accelerated to any energy . in this paper
we will use the evaluated proton cross sections by nishiizumi et al.@xcite and neutron cross sections by reedy@xcite . up to now , the physical model to calculate the production rate of cosmogenic nuclei has been established .
after apply this model to extensive studies , it is needed to test the validation of this model . to this end , we choose the lunar drill core as research object .
it is because its chemical composition , location , shielding depth of samples are known clearly .
in addition , apollo 15 drill core is the least disturbed sample among the three deep drill cores , apollo 15 , 16 and 17@xcite . in this case
the radius of moon can be regarded as infinite , and the surface can be seen as a plane . in simulations
we establish a sufficient large box filled with the average chemical composition of apollo 15 drill core@xcite , the gcrs incident on the upper surface of it . in this work ,
the box is taken as 100 m long @xmath28 100 width @xmath28 20 m thickness .
this box is filled with uniform lunar soils .
the chemical composition is : o ( 43.0% wt ) , si ( 22.1% wt ) , al ( 7.67% wt ) , ca ( 7.52 % wt ) , mg ( 5.89 % wt ) , fe ( 11.57 % wt ) , ti ( 1.12 % wt ) , na ( 0.332 % wt ) , and k ( 0.21 % wt ) . here
, % wt denotes the weight percentage .
the gcrs , sampled according to eqs.(2 ) and ( 3 ) , incident on the upper surface of this box isotropically .
considering both the statistical accuracy and time cost , 5 million particles are sampled .
once the production rate is calculated one can derive the cosmic ray exposure ( cre ) age by using the relations as follows : for a stable cosmogenic nuclide @xmath29 and for a radioactive cosmogenic nuclide with the decay constant @xmath30 @xmath31 where @xmath32 is the cre age and @xmath33 ( @xmath5 ) and @xmath34 ( @xmath35 ) are the concentration and production rate of stable ( radioactive ) nuclide .
the basic unit of the cosmogenic nuclei concentration is the number of atoms per gram sample ( atoms/(g sample ) ) . in literatures ,
the concentrations of radioactive nuclei are often reported as radioactive activity , in unit of disintegrations per minute per kilogram ( dpm kg@xmath21 ) .
the relationship between the concentration and the activity is @xmath36 from this equation , one can see that if the cre age is much longer than the half - life of this nuclide , the radioactive equilibrium will be reached and the activity is approximately equal to the production rate . for this case
the measured quantity can be compared directly with the calculated ones . in this paper
, we will choose the nuclei whose half - lives are much longer than the time by which the lunar core are returned from the moon and much shorter than the cre age .
@xmath1be ( @xmath37 ) , @xmath2al ( @xmath38 ) are good candidates .
according to the theoretical model described in sec.2 , we write the geant4 code to simulate the transportation and interaction of gcrs in the lunar surface . in simulations ,
the neutrons and protons at different depths are recorded and saved as root files , from which we obtain the differential fluxes for many energy bins .
then the differential fluxes for neutron and proton at different energy ranges are fitted , respectively .
after careful analysis we found that the proton and neutron fluxes can be expressed quite well by the following formulas : @xmath39 @xmath40 in these formulas the energy @xmath14 is in mev , and the @xmath41 s are depth and chemical dependent parameters .
these formulas are much simpler than that used by arnold@xcite and reedy@xcite .
the parameters are determined by the best - fit procedure using the root software . considering the limit of space we will not give the numerical results of theses parameters .
the proton and neutron fluxes calculated using eqs.(7 ) and ( 8) at depths @xmath42 g/@xmath20 are shown in fig.1 . here
the depth @xmath7 is the geometrical depth times the density , the unit is g/@xmath20 . in this figure
the fluxes are not normalized to particles/(@xmath20 s mev ) since we are interested to the relative magnitude of proton and neutron fluxes . from this figure
, one can see that at high energy range ( @xmath43 mev ) the proton flux is larger than that of neutron , but at low energy range ( @xmath27 mev ) the neutron flux is larger than that of proton .
it is because the protons beneath the lunar surface include the primary and secondary ones , but the neutrons are purely secondary particles .
the secondary protons will suffer ionizing loss and stop quickly .
g/@xmath20.,width=377 ] as for the cross sections for proton and neutron induced reactions , the most newly evaluated data are used .
the proton cross sections are taken from@xcite , and the neutron cross sections are taken from@xcite .
the numerical data for the cross sections involved are shown in figs.2 - 4 .
be on target elements o , mg , al , si , fe .
these data are taken from@xcite.,width=377 ] the excitation function for proton induced reactions to produce @xmath2al on target elements mg , al , si , ca , ti , fe . these data are taken from@xcite.,width=377 ] be on target elements o , si , and @xmath2al on elements al , si . these data are taken from@xcite.,width=377 ] from these figures one can see clearly that the excitation functions for these reactions are not smooth .
it is because the spallation reaction is a very complex process , the spallation cross sections are very difficult to calculate .
therefore , these functions are obtained by interpolating the experimental data .
unfortunately , there are only a few data available , even for neutron induced reactions .
be in apollo 15 drill core .
the experimental data are taken from @xcite .
, width=377 ] al in apollo 15 drill core .
the experimental data are taken from @xcite .
, width=377 ] finally , the production rates of cosmogenic nuclei @xmath1be and @xmath2al are calculated by integrating the depth - dependent fluxes of active particles with the excitation functions of nuclear reactions involved . by using this method , the production rate of @xmath1be and @xmath2al in apollo 15 drill core are calculated . in calculations ,
the production rates result from proton induced reactions and neutron induced reactions are considered . as for proton induced reactions the target elements we considered are o , mg , al ,
si , fe for @xmath1be , al , si , ca , ti , fe for @xmath2al .
as for neutron induced reactions the cross sections are very scattered .
the target elements we considered are o and si for @xmath1be , al and si for @xmath2al . in order to compare with experimental data , we introduce two normalization factors for the production rates for neutron and proton induced reactions : @xmath44 where @xmath35 is the total production rate of radioactive nuclei , @xmath45 and @xmath46 are the contributions due to proton and neutron induced reactions , respectively , calculated using the theoretical model shown in sec.2 .
the main reasons of introducing these factors are as follows : ( 1 ) the excitation functions are obtained by interpolating a small number of cross section data , particularly the neutron cross section data are obtained by indirect method ; ( 2 ) the secondary proton and neutron yields calculated by geant4 depend on the theoretical model of spallation reactions ; ( 3 ) the spectrum of primary cosmic rays are assumed to be the current mean value averaged during the solar activity cycle .
that is to say the uncertainties on the primary gcr fluxes , secondary particle fluxes and excitation functions in the theoretical model are normalized by two parameters @xmath45 and @xmath46 .
these two parameters can be obtained by the best - fit procedure .
the theoretical and experimental results are shown in figs.5 - 6 . in these figures
the dotted line and dashed line represent the contributions from proton and neutron induced reactions , respectively , and the solid lines denote the total production rate .
the experimental data ( denoted by circles ) for @xmath1be are taken from @xcite , and the data for @xmath2al are taken from@xcite . from these two figures
one can see that the theoretical production rate of @xmath1be and @xmath2al agree well with the measured data .
in addition , the production rate from neutron induced reactions are larger than that of proton induced reactions . in other words , the neutron reactions dominate the production of cosmogenic nuclei , hence the high - energy neutron reaction cross sections play crucial roles .
in summary , a physical model based on the monte carlo method is proposed to calculate the production rate of cosmogenic nuclei .
a geant4 code has been developed to simulate the transport and interactions of the primary and secondary particles in the extra - terrestrial body . as a test of this model
we calculate the production rate of long - lived nuclei @xmath1be , @xmath2al in the apollo 15 drill core .
the information of neutrons and protons at different depths are recorded and saved as root file .
then the neutron and proton differential fluxes are obtained by using root software . by combining the differential fluxes and the newly evaluated spallation cross sections
the production rates of @xmath1be , @xmath2al are calculated and compared with experimental data .
the results show that the theoretical production rates agree quite well with the measured data for the apollo 15 long drill core .
it means that our model is suitable to investigate the production rate of cosmogenic nuclei in extra - terrestrial body .
it is reasonable to believe that this model can be applied to the investigation of production rate of cosmogenic nuclei in meteorites and earth s atmosphere .
furthermore , according to the chinese lunar exploration program , lunar samples will be returned 10 years later or so .
when lunar samples are obtained the cosmic ray exposure history of the landing sites is an important subject of research .
for instance , the cosmic ray exposure age of the crater rims can be used to date the impact events and to calibrate the geological age of that region defined by the counting rate of impact craters . to reach this goal ,
it is needed to improve this model by using better physical model of high energy nuclear reactions and more accurate cross sections of proton and neutron induced reactions . as for the former
, we can expect that with the improvement of the nuclear reaction model , for instance li@xmath47ge intranuclear cascade ( incl ) model@xcite , the proton and neutron spectrum can be calculated more and more accurate . in the next , we will study the influences of different nuclear reaction models on the proton and neutron fluxes . as for the latter , the neutron cross sections measured by using the quasi - monoenergetic neuron beams produced by @xmath48li(p , n)@xmath48be reaction within the european hindas ( high- and intermediate - energy data for accelerator - driven systems ) project@xcite will give us unprecedented opportunity to study the production rate of cosmogenic nuclei .
|
a physical model of geant4-based simulation of galactic cosmic ray ( gcr ) particles interaction with the lunar surface matter has been developed to investigate the production rate of cosmogenic nuclei . in this model
the gcrs , mainly very high energy protons and @xmath0 particles , bombard the surface of the moon and produce many secondary particles such as protons and neutrons .
the energies of proton and neutron at different depths are recorded and saved into root files , and the analytical expressions for the differential proton and neutron fluxes are obtained through the best - fit procedure under the root software . to test the validity of this model
, we calculate the production rates of long - lived nuclei @xmath1be and @xmath2al in the apollo 15 long drill core by combining the above differential fluxes and the newly evaluated spallation reaction cross sections .
numerical results show that the theoretical production rates agree quite well with the measured data .
it means that this model works well .
therefore , it can be expected that this model can be used to investigate the cosmogenic nuclei in lunar samples returned by chinese lunar exploration program and can be extended to study other objects , such as the meteorites and the earth s atmosphere .
|
the mechanism of electroweak symmetry breaking ( ewsb ) remains the most pressing puzzle in elementary particle physics .
experimentally , this question will be addressed at the large hadron collider ( lhc ) . theoretically , several interesting possibilities have been proposed . in this article , we will concentrate on the `` little higgs '' proposal @xcite . in this approach ,
the higgs emerges as a pseudo - nambu - goldstone boson , whose properties are constrained by global symmetries .
these global symmetries are not exact , and their breaking allows the higgs to participate in non - derivative ( i.e. gauge and yukawa ) interactions .
at the same time , there is enough global symmetry left to ensure that the higgs mass term vanishes at tree level , and is only logarithmically sensitive to the unknown short distance ( ultraviolet , or uv ) physics at the one - loop level .
the usual quadratic sensitivity of the higgs mass parameter on the uv physics first appears at two loops , and the incalculable uv effects remain subleading as long as the cutoff of the theory ( the scale at which it becomes strongly coupled ) is at or below about 10 tev . with this requirement ,
the higgs mass terms are dominated by the one - loop contribution from the top loops , which has the appropriate sign to trigger the electroweak symmetry breaking , and produces the higgs vacuum expectation value ( vev ) of the right order of magnitude .
thus , little higgs theories provide an attractive explanation of ewsb .
the originally proposed implementations of the little higgs approach suffered from severe constraints from precision electroweak measurements @xcite , which could only be satisfied by finely tuning the model parameters .
the most serious constraints resulted from the tree - level corrections to precision electroweak observables due to the exchanges of additional heavy gauge bosons present in the theories , as well as from the small but non - vanishing vev of an additional weak - triplet scalar field .
motivated by these constraints , several new implementations of the little higgs were proposed @xcite .
particularly interesting is the approach of refs .
@xcite , which introduces a discrete symmetry , dubbed `` t parity '' in analogy to r parity in the minimal supersymmetric standard model ( mssm ) .
t parity explicitly forbids any tree - level contribution from the heavy gauge bosons to the observables involving only standard model ( sm ) particles as external states .
it also forbids the interactions that induced the triplet vev . as a result , in t parity symmetric little higgs models , corrections to precision electroweak observables
are generated exclusively at loop level .
this implies that the constraints are generically weaker than in the tree - level case , and fine tuning can be avoided @xcite .
the main goal of this paper is to investigate the electroweak precision constraints on the models with t parity in more detail .
we will concentrate on the t parity symmetric version of the littlest higgs ( lh ) model , based on an @xmath2 global symmetry breaking pattern @xcite .
some phenomenological aspects of this model have been analyzed in ref .
the model possesses an attractive dark matter candidate , the t - odd partner of the hypercharge gauge boson , which has the correct relic abundance in certain regions of the parameter space .
it also leads to an interesting set of signatures at the lhc ; in particular , an excess of events with large missing transverse energy is expected . in this paper
, we will compute the corrections to the properties of @xmath3 bosons induced by the new particles present in the lh model , and perform a global fit to precision electroweak observables .
we will show that a consistent fit can be obtained in a large region of the model parameter space , so that no significant fine tuning is required .
we will also demonstrate that the lh model allows for consistent fits with values of the higgs mass as large as 800 gev , far in excess of the upper bound obtained within the standard model .
finally , we will show that there exists a non - vanishing overlap between the region allowed by precision electroweak fits and the region where the lh model provides all of the observed dark matter .
the rest of the paper is organized as follows .
after briefly reviewing the @xmath2 littlest higgs model with t parity in section [ model ] , we will present the calculation of the corrections to precision electroweak observables in section [ electroweak ] . in section
[ fits ] , we present the constraints on the parameter space of the model resulting from a global fit to precision electroweak observables .
section [ conc ] contains our conclusions .
a discussion of some aspects of the lh model in the renormalizable @xmath4 gauge , which we find useful in our calculations , is presented in the appendix [ app ] .
in this section , we will review the lh model with t parity @xcite , emphasizing the features that will be important for the analysis of this paper .
the littlest higgs model @xcite embeds the electroweak sector of the standard model in an @xmath2 non - linear sigma model ( nl@xmath5 m ) . a global @xmath6 symmetry
is broken to @xmath7 by the vev of an @xmath6 symmetric tensor @xmath8 of the form _ 0 = ( ccccc 0 & 0 & 0 & 1 & 0 + 0 & 0 & 0 & 0 & 1 + 0 & 0 & 1 & 0 & 0 + 1 & 0 & 0 & 0 & 0 + 0 & 1 & 0 & 0 & 0 + ) .
the low energy dynamics of the nl@xmath5 m is described in terms of the field @xmath9 where @xmath10 is the `` pion matrix '' containing the goldstone degrees of freedom , and @xmath11 tev is the nl@xmath5 m symmetry breaking scale , or `` pion decay constant '' .
an @xmath12 ^ 2 $ ] subgroup of the global @xmath6 symmetry is gauged .
the gauged generators have the form @xmath13 the kinetic term for the @xmath8 field can be written as _ kin = d_(d^ ) ^ , where @xmath14\,,\ ] ] with @xmath15 . here ,
@xmath16 and @xmath17 are the @xmath18 and @xmath19 gauge fields , respectively , and @xmath20 and @xmath21 are the corresponding coupling constants .
the vev @xmath22 breaks the extended gauge group @xmath23 ^ 2 $ ] down to the diagonal subgroup , which is identified with the standard model electroweak group @xmath24 .
the fourteen pions of the @xmath2 breaking decompose into representations of the electroweak gauge group as follows : @xmath25 we will denote the fields in the above four representations as @xmath26 , @xmath27 , @xmath28 and @xmath29 , respectively .
the field @xmath28 has the appropriate quantum numbers to be identified with the sm higgs ; after ewsb , it can be decomposed as @xmath30 , where @xmath31 gev is the ewsb scale and @xmath32 is the physical higgs field .
explicitly , the pion matrix in terms of these fields has the form = ( ccccc -^0/2- / & -^+/ & -i ^+/ & -i ^++ & -i + -^-/ & ^0/2- / & & -i & + i ^-/ & ( v+h - i ^0)/2 & & -i ^+/ & ( v+h+i ^0)/2 + i ^ & i & i ^-/ & -^0/2 - / & - ^-/ + i & & & - ^+/ & ^0/2- / ) , where the superscripts indicate the electric charge .
the fields @xmath26 and @xmath27 are eaten and @xmath27 mix with the field @xmath29 at order @xmath33 , and it is a linear combination of these that is eaten .
see appendix [ app ] for details .
] when the extended gauge group is broken down to @xmath34 , whereas the @xmath35 fields are absorbed by the standard model @xmath3 bosons after ewsb .
the fields @xmath32 and @xmath29 remain in the spectrum . including the ewsb effects , the vev of the @xmath8 field has the form = ( ccccc 0 & 0 & 0 & 1 & 0 + 0 & -(1-c_v ) & s_v & 0 & ( 1+c_v ) + 0 & s_v & c_v & 0 & s_v + 1 & 0 & 0 & 0 & 0 + 0 & ( 1+c_v ) & s_v & 0 & -(1-c_v ) + ) , where s_v= , c_v=. these formulas will prove very useful for analyzing the spectrum of the model .
the gauge generators are embedded in the @xmath6 is such a way that any given generator commutes with an @xmath36 subgroup of the @xmath6 .
this implies that if one pair of gauge couplings ( @xmath37 or @xmath38 ) is set to zero , the higgs field @xmath28 would be an exact goldstone boson and , therefore , exactly massless .
thus , any diagram renormalizing the higgs mass vanishes unless it involves at least two of the gauge couplings . at one loop , all diagrams satisfying this property are only logarithmically divergent :
the `` collective '' symmetry breaking mechanism protects the higgs mass from quadratic divergences .
the first quadratic divergence appears at two loop level . the original littlest higgs model described above turned out to be significantly constrained by precision electroweak observables @xcite .
t parity , a discrete @xmath39 symmetry , was introduced by cheng and low @xcite to avoid this difficulty , and it also provides a potential weak scale dark matter candidate . in the gauge sector ,
t parity is an automorphism of the gauge groups which exchanges the @xmath40_1 $ ] and @xmath40_2 $ ] gauge fields @xcite .
the lagrangian in eq .
is invariant under this transformation provided that @xmath41 and @xmath42 . in this case , the gauge boson mass eigenstates ( before ewsb ) have the simple form , @xmath43 , @xmath44 , where @xmath45 and @xmath46 are the standard model gauge bosons and are t - even , whereas @xmath47 and @xmath48 are the additional , heavy , t - odd states .
( typically , @xmath48 is the lightest t - odd state , and plays the role of dark matter @xcite . ) after ewsb , the t - even neutral states @xmath49 and @xmath46 mix to produce the sm @xmath50 and the photon .
since they do not mix with the heavy t - odd states , the weinberg angle is given by the sm relation , @xmath51 , where @xmath52 and @xmath53 are the sm gauge couplings , and @xmath54 at tree level .
as will be shown below , all the sm fermions are also t - even , implying that the @xmath47 and @xmath48 states generate no corrections to precision electroweak observables at tree level .
the transformation properties of the gauge fields under t parity and the structure of the lagrangian imply that t parity acts on the pion matrix as follows : t : - , where @xmath55 .
this transformation law ensures that the complex @xmath56 triplet @xmath29 is odd under t parity , while the higgs doublet @xmath28 is even .
the trilinear coupling of the form @xmath57 is therefore forbidden , and no triplet vev is generated . eliminating this source of tree - level custodial @xmath58 violation
further relaxes the precision electroweak constraints on the model . in the original lh model ,
the fermion sector of the standard model remained unchanged with the exception of the third generation of quarks , where the top yukawa coupling had to be modified to avoid the large quadratically divergent contribution to the higgs mass from top loops . in the model with t parity , however , the standard model fermion doublet spectrum needs to be doubled to avoid compositeness constraints @xcite . for each lepton / quark doublet , two fermion doublets @xmath59 and @xmath60
are introduced .
( the quantum numbers refer to representations under the @xmath61 gauge symmetry . )
these can be embedded in incomplete representations @xmath62 of the global @xmath6 symmetry .
an additional set of fermions forming an @xmath7 multiplet @xmath63 , transforming nonlinearly under the full @xmath6 , is introduced to give mass to the extra fermions ; the field content can be expressed as follows : @xmath64 these fields transform under the @xmath6 as follows : _
1 v^ * _ 1 , _ 2 v _ 2 , ^c u^c , where @xmath65 is the nonlinear transformation matrix defined in refs . @xcite .
the action of t parity on the multiplets takes _
1-_0 _ 2 , ^c -^c .
these assignments allow a term in the lagrangian of the form @xmath66 where @xmath67 .
this term gives a dirac mass @xmath68 to the t - odd linear combination of @xmath69 and @xmath70 , @xmath71 , together with @xmath72 ; the t - even linear combination , @xmath73 , remains massless and is identified with the standard model lepton or quark doublet .
to give dirac masses to the remaining t - odd states @xmath74 and @xmath75 , additional fermions with opposite gauge quantum numbers can be introduced @xcite . to complete the discussion of the fermion sector
, we introduce the usual sm set of the @xmath56-singlet leptons and quarks , which are t - even and can participate in the sm yukawa interactions with @xmath76 .
the yukawa interactions induce a one - loop quadratic divergence in the higgs mass ; however , the effect is numerically small except for the third generation of quarks .
the yukawa couplings of the third generation must be modified to incorporate the collective symmetry breaking pattern ; this is discussed in the next subsection . in order to avoid large one - loop quadratic divergences from the top sector , the @xmath77 and @xmath78 multiplets for the third generation must be completed to representations of the @xmath79 ( `` upper - left corner '' ) and @xmath80 ( `` lower - right corner '' ) subgroups of @xmath6 .
these multiplets are @xmath81 they obey the same transformation laws under t parity and the @xmath6 symmetry as do @xmath77 and @xmath78 , see eqs . and .
the quark doublets are embedded such that @xmath82 in addition to the sm right - handed top quark field @xmath83 , which is assumed to be t - even , the model contains two @xmath56-singlet fermions @xmath84 and @xmath85 of hypercharge 2/3 , which transform under t parity as @xmath86 the top yukawa couplings arise from the lagrangian of the form _ t & = & _ 1 f _ ijk _
xy u_r + & & + _ 2 f ( |u_l1 u_r1 + |u_l2 u_r2)+ h.c . where @xmath87 is the image of the @xmath8 field under t parity , see eq .
, and the indices @xmath88 run from 1 to 3 whereas @xmath89 .
the t parity eigenstates are given by q_= ( q_1 q_2 ) , u_l = ( u_l1 u_l2 ) , u_r = ( u_r1 u_r2 ) . in terms of these eigenstates , eq . has the form ^t_m = _ 1 f u_r + _ 2 f ( |u_l+ u_r+ + |u_l- u_r- ) + h.c . where we have used eq . .
the t - odd states @xmath90 and @xmath91 combine to form a dirac fermion @xmath92 , with mass @xmath93 .
the remaining t - odd states @xmath94 receive a dirac mass from the interaction in eq . , and are assumed to be decoupled .
the mass terms for the t - even states are diagonalized by defining t_l & = & u_l+ - u_l+ , t_l+ = u_l+ + u_l+,t_r & = & u_r - u_r+ , t_r+ = u_r + u_r+ , where @xmath95 is identified with the sm top and @xmath96 is its t - even heavy partner .
the mixing angles are given by & = & ^-1 , & = & ^-1 .
to leading order in the @xmath97 expansion , = , = .
the masses of the two t - even dirac fermions are given by m^2_t , t_+ = f^2(1 ) , where = ( _ 2 ^ 2+(s_v^2+(1+c_v)^2 ) ) .
to leading order in @xmath97 , m_t= , = f. it is interesting to note that the t - odd states do not participate in the cancellation of quadratic divergences in the top sector : the cancellation only involves loops of @xmath95 and @xmath96 , and the details are identical to the lh model without t parity @xcite . using the above equations , it is straightforward to obtain the feynman rules for the top sector of the lh model ; we list the rules relevant for the calculations in this paper in table [ tab : feynman ] .
the introduction of t parity automatically eliminates the tree level electroweak precision constraints that plagued the original littlest higgs model : since the external states in all experimentally tested processes are t - even , no t - odd state can contribute to such processes at tree level .
the only non - sm t - even state in our model , the heavy top @xmath96 , can only contribute at tree level to observables involving the sm top quark to other observables are suppressed by small off - diagonal ckm matrix elements .
] , such as its couplings to @xmath98 and @xmath50 bosons @xcite . at present , however , these couplings have not been measured experimentally , so no contraints arise at the tree level . at one - loop level , however , precision electroweak observables receive contributions from the @xmath96 as well as the t - odd particles .
it is these contributions that determine the allowed parameter space of the littlest higgs model with t parity .
we will evaluate the leading corrections in this section , and use them to perform a global fit to precision electroweak observables in section [ fits ] . in the sm , one - loop contributions to precision electroweak observables from the top sector , enhanced by powers of the top yukawa coupling @xmath99 ,
dominate over contributions from the gauge and scalar sectors .
we expect that the same hierarchy of effects will hold in the littlest higgs model , and our main focus will be on analyzing the effects of the top sector
. however , we will also include the custodial - symmetry violating contributions from the gauge sector and the t - odd partners of light fermions , which become important in certain regions of the parameter space .
in addition , we will show explicitly that the contributions from the complex scalar triplet , which were shown to be potentially important in the original lh model @xcite , completely decouple in the t parity symmetric case due to the absence of the triplet vev . before proceeding with the calculations ,
let us make the following comment .
the nl@xmath5 m underlying the lh model is a non - renormalizable effective theory , valid up to a cutoff scale @xmath100 .
every operator consistent with the symmetries of the low - energy theory will be generated at the cutoff scale @xmath101 , and will contribute to the precision electroweak observables .
however , we do not include such operators in the fit .
( the only exception we make is to include the leading operator contributing to the @xmath1 parameter , eq .
, since this parameter plays the most important role in constraining the model . )
this is justified by the following considerations .
first , while the contribution of the tev - scale states that we will compute and the operator contributions that we will ignore are naively of the same order , @xmath102 , the former are logarithmically enhanced by a factor of @xmath103 .
second , while a cancellation between the corrections computed below and the operator contribution is in principle possible if the uv physics produces an operator with a large coefficient , any significant change in the fits due to such a cancellation would represent a fine - tuning between the effects generated at two different energy scales , @xmath0 and @xmath101 .
the largest corrections to precision electroweak observables in the lh model are induced by the one - loop diagrams involving the t - even @xmath96 quark shown in fig .
[ fig : stu ] .
these oblique corrections can be described in terms of the peskin - takeuchi @xmath104 , @xmath1 , and @xmath65 parameters @xcite .
the calculation of these parameters is straightforward if somewhat tedious ; the result is s & = & , t & = & , u & = & - , where @xmath105 , @xmath106 is the sine of the left - handed @xmath107 mixing angle given in eq . ,
and @xmath108 is the sine of the weinberg angle . in the limit when @xmath109 , these formulas simplify considerably and we obtain s & = & ( ) ^2 , + t & = & ( ) ^2 , + u & = & ( ) ^2 . the leading - order result for
the @xmath1 parameter is in agreement with the previous analyses of lh models with and without t parity @xcite .
in contrast to @xmath96 , the t - odd top partner @xmath92 does not contribute to @xmath110 or @xmath65 since it is an @xmath56 singlet which does not mix with the sm top .
however , @xmath92 loops do affect precision electroweak observables at the level of @xmath111 corrections which are not captured by the formalism of peskin and takeuchi .
the corrections to the two most precisely measured observables , @xmath112 and @xmath113 ( the value of the weak mixing angle implied from @xmath50 decay asymmetries ) , are given by s_*^2-s_0 ^ 2=- , s_w^2-s_*^2=0 , where @xmath114 is the reference value of the weak mixing angle inferred from s_0 ^ 2 ( 1-s_0 ^ 2 ) = .
these corrections are very small , and we do not include them in the global fit performed in section [ fits ] . it is clear from eqs . that the @xmath1 parameter induced by the @xmath96 loops is about 20 times larger than the @xmath104 and the @xmath65 for the same model parameters , and therefore the constraints on the model are largely driven by the @xmath1 parameter .
this parameter also receives a contribution from the gauge sector of the model .
while in general subdominant , this correction becomes important when the @xmath107 mixing is suppressed ( namely when the ratio @xmath115 is small ) , and we will include it in the fit .
this contribution arises from the custodial @xmath58-violating tree level mass splitting of the t - odd heavy @xmath116 and @xmath117 gauge bosons . neglecting effects of order @xmath118 ,
the mass splitting is given by m^2 m^2(w^3_h)-m^2(w^_h ) = ( 1-c_v)^2 g^2 . at one loop , this effect induces a contribution to the @xmath1 parameter @xcite : . ] t_w_h = - m^2 .
note that the result is divergent , and depends on the uv cutoff of the theory @xmath101 .
this should not be surprising since the theory we re dealing with is non - renormalizable : indeed , the mass splitting in eq
. comes from a dimension-6 operator of the form @xmath119 , which appears when the @xmath8 fields in eq .
are expanded to order @xmath120 .
the uv divergence signals the presence of a `` counterterm '' operator of the form @xcite _ c = _ c f^2_i , a tr , where @xmath121 is an order - one coefficient whose exact value depends on the details of the uv physics . ( the normalization of eq .
is fixed by naive dimensional analysis @xcite . )
including the counterterm , the full contribution of the gauge sector to the @xmath1 parameter has the form t= - ( _ c+ ) = - ( _ c+ ) .
where we have assumed @xmath122 and used @xmath123 . as expected , this contribution is parametrically subdominant to the correction from the top sector , eq . ,
in the limit @xmath124 .
furthermore , in agreement with the discussion at the beginning of this section , the effect of the operator induced at the cutoff scale is subdominant , by a factor of @xmath125 , compared to the calculable contribution in eq . . in section
[ fits ] , we will also be interested in the effects of varying the higgs mass .
to leading order , the higgs contribution to the oblique parameters is given by @xcite s & = & , + t & = & - , + u & = & 0 , where @xmath126 is the `` reference '' value of the higgs mass used to obtain the sm predictions for precision electroweak observables bosons will receive corrections of order @xmath97 , which have been neglected in eq . .
this will not affect any of the conclusions of our analysis in section [ fits ] . ] .
interestingly , the negative contribution to the @xmath1 parameter from a heavy higgs can be partially cancelled by the positive contribution from the @xmath96 . as we will show below
, this cancellation allows for a consistent fit to precision electroweak observables with the higgs mass well above the upper bound obtained in the sm @xcite .
finally , the lh model contains an additional t - odd @xmath56-triplet scalar field @xmath29 , with the mass m_^2~(1 tev)^2 .
after ewsb , a mass splitting of order @xmath127 between various components of the triplet is generated , for example , by operators of the form @xmath128 . neglecting this mass splitting ,
the triplet contributions to @xmath104 , @xmath1 and @xmath65 parameters vanish ; keeping the terms of order @xmath129 , its contributions to @xmath113 and @xmath130 are given by s_*^2-s_0 ^ 2=- , s_w^2-s_*^2=. if the mass splitting is taken into account , non - zero contributions to the peskin - takeuchi parameters are induced ; however , these effects are of order @xmath131 , and are thus subleading to the corrections given in eq . .
we conclude that the effects of the triplet @xmath29 on the precision electroweak observables in the lh model with t parity decouple with growing @xmath132 , and are negligible for @xmath132 in its natural range , around 1 tev .
we will not include these effects in the global fit of section [ fits ] .
.,width=384 ] to implement t parity in the fermion sector of the lh model , it is necessary to introduce a t - odd fermion partner for each lepton / quark doublet of the sm ( see section [ light ] ) .
these particles are vector - like , and their effects on precision electroweak observables must decouple in the limit when their mass is taken to infinity . however , box diagrams involving the exchanges of goldstone bosons @xmath27 and @xmath26 , see fig .
[ baksho ] , generate four - fermion operators whose coefficients _ increase _ if the mass of the t - odd fermions is increased while @xmath0 is kept fixed . this non - decoupling is easy to understand qualitatively : to increase the mass of the t - odd fermions , it is necessary to increase the yukawa coupling @xmath133 in eq . , which in turn makes the four interaction vertices in the box diagrams stronger .
assuming that the couplings @xmath133 are flavor - diagonal and flavor - independent , the generated operators have the form _ 4-f= - |_l ^_l |^_l _ ^_l , where @xmath134 and @xmath135 are ( distinct ) sm fermions , and we ignore the corrections of order @xmath136 .
the experimental bounds on four - fermi interactions involving sm fields provide an _ upper _ bound on the t - odd fermion masses ; the strongest constraint comes from the @xmath137 operator , whose coefficient is required to be smaller than @xmath138 @xcite .
this yields m_tev 4.8 f^2_tev , where @xmath139 and @xmath140 are the values of the t - odd fermion masses and the symmetry breaking scale , respectively , in tev .
in addition , a lower bound on the masses of the t - odd fermions can be obtained from non - observation of these particles at the tevatron ; in analogy with squarks of the mssm , we expect the bound to be in the neighborhood of 250300 gev .
the contribution of each t - odd doublet to the @xmath1 parameter is given by t_t - odd= - ( ) ^2 , where we omit terms of order @xmath141 and higher .
note that , for a fixed value of @xmath0 , this contribution _ increases _ with increasing t - odd fermion mass ; eq
. implies that |t_t - odd|0.05 , independent of @xmath0 .
( note that this bound relies on the assumption that the @xmath133 couplings are flavor - independent . ) nevertheless , the t - odd fermions can have a noticeable effect on the precision electroweak fits due to a large number ( twelve ) of doublets in the sm ; this will be illustrated in the next section . with the assumptions of flavor - diagonal yukawa couplings @xmath133 , the non - oblique corrections induced by the t - odd fermions at one loop are flavor - universal , and can be absorbed in the @xmath104 parameter .
vertex renormalization at one - loop in the @xmath4 gauge.,width=480 ] in the sm , the largest non - oblique correction is the renormalization of the @xmath143 vertex by top quark loops .
this effect is non - decoupling in the sense that it is proportional to the square of the top mass .
this non - decoupling is most easily seen if the calculation is performed in the t hooft - feynman gauge @xcite . in this gauge ,
the non - decoupling part of the vertex correction comes purely from the diagrams involving the exchange of a goldstone boson @xmath144 , since its couplings to the top and bottom quarks are enhanced by the top yukawa @xmath99 .
these diagrams are shown in fig .
[ zbbsmloops ] .
the diagrams involving the exchange of the gauge bosons are subdominant in the large-@xmath145 limit , and neglecting their contribution only induces an error of order @xmath146% in the vertex correction calculation .
we have calculated the one - loop correction to the @xmath147 vertex in the lh model with t parity .
we used the t hooft - feynman gauge ( for a brief discussion of the @xmath4 gauges in the lh model , see appendix [ app ] ) .
as in the sm case , the diagrams involving @xmath144 exchanges dominate in the large-@xmath145 limit , and we neglected all other contributions .
( while a more precise calculation could be done , the effort would not be justified as the @xmath147 correction turns out to have only a small effect on precision electroweak fits . )
these diagrams are of three kinds .
first , the same diagrams as in the sm appear , but with the top coupling to the @xmath50 modified according to table [ tab : feynman ] .
second , all the diagrams in fig .
[ zbbsmloops ] also appear with the top replaced by the t - even heavy top partner @xmath96 .
third , the `` mixing '' diagrams shown in fig . [ zbbbsmloops ] appear as a result of the mixing between @xmath95 and @xmath96 . these diagrams can be easily calculated using the couplings given in table [ tab : feynman ] and in eq . of appendix [ app ] . to leading order in the limit @xmath148 ,
the result is g_l^b|b= , where @xmath149 is the correction received by the @xmath150 vertex in the lh model _ in addition _ to the usual sm one - loop correction .
it is interesting to note that this leading order contribution comes entirely from the mixing diagrams in fig .
[ zbbbsmloops ] .
the correction to the @xmath151 vertex is negligible since it is not enhanced by the top yukawa coupling .
note also that the correction in eq . does not have the correct sign to alleviate the well - known deviation of the measured value of the forward - backward asymmetry in @xmath142 decays from the sm prediction inferred from the other precision electroweak observables @xcite .
vertex renormalization at one loop in the lh model with t parity.,width=384 ] a calculation of @xmath147 in a general theory containing an extra heavy quark that mixes with @xmath95 has been carried out in ref .
accounting for the fact that @xmath152 is a vector isosinglet and including the appropriate mixing specific to the lh model , eq . agrees with the results of this analysis in the limit @xmath148 .
the results of ref .
@xcite are more general , valid for arbitrary values of @xmath153 and @xmath145 .
however , we find that using these expressions instead of eq
. does not lead to noticeable changes in the global fits performed in section [ fits ] . with the assumption of flavor - diagonal and flavor - independent yukawa couplings
@xmath133 made in section [ stuff ] , the one - loop vertex corrections due to loops of t - odd fermions are flavor - universal , and can therefore be absorbed in the redefinitions of gauge couplings .
they will not induce an observable shift in @xmath154 couplings .
and the symmetry breaking scale @xmath0 . the contribution of the t - odd fermions to the t parameter is neglected . from lightest to darkest ,
the contours correspond to the 95 , 99 , and 99.9 confidence level exclusion . ]
-0.3 cm , see eq . , and @xmath0 .
from lightest to darkest , the contours correspond to the 95 , 99 , and 99.9 confidence level exclusion . ]
-0.3 cm to obtain constraints on the parameter space of the lh model with t parity , we have performed a global fit to precision electroweak observables , including the lh contributions evaluated in the previous section .
the lh contributions are parametrized by two dimensionless numbers , @xmath155 and @xmath121 , and the symmetry breaking scale @xmath0 . in the fit ,
we have used the values of the 21 @xmath50 pole and low - energy observables listed in ref .
@xcite ; the equations expressing the shifts in these observables in terms of the oblique parameters and @xmath156 are given in ref .
we take the top mass to be 176.9 gev @xcite , and do not include the uncertainty associated with the top mass . in each constraint plot
, we draw the @xmath157 , @xmath158 , and @xmath159 confidence level contours in the context of a @xmath160 analysis with two degrees of freedom . ] . in the first part of the analysis , we have fixed the higgs mass at its reference value , @xmath161 gev . in fig .
[ fr ] , we plot the constraints in the @xmath162 plane , assuming @xmath163 . in fig .
[ deltac ] , we fix @xmath164 and plot the constraints in the @xmath165 plane , neglecting the t - odd fermion contribution . it is clear that a large part of the parameter space is consistent with precision electroweak constraints , including regions where the symmetry breaking scale @xmath0 is as low as 500 gev .
( in these regions , a partial cancellation between the top and gauge sector contributions to the @xmath1 parameter takes place . ) in some cases , we have even obtained consistent fits for values of @xmath0 as low as 350 gev .
however , since our analysis neglects all higher - derivative operators generated at the scale @xmath166 , which can contribute significantly to precision electroweak observables when @xmath167 tev , we estimate that the fits can not be trusted for @xmath168 gev , and do not show that part of the parameter space in the plots .
and the symmetry breaking scale @xmath0 .
from lightest to darkest , the contours correspond to the 95 , 99 , and 99.9 confidence level exclusion .
contours of constant values of the fine - tuning parameter @xmath169 are also shown ; the solid and dashed lines correspond to @xmath170 and @xmath171 , respectively . ]
-0.3 cm as has been shown in section [ oblique ] , top sector loops in the lh model provide a sizable , positive contribution to the @xmath1 parameter .
this raises an interesting possibility : since the contribution of a heavy sm higgs to the @xmath1 parameter is negative , it is possible that these two effects partially cancel parameter is generated at tree level ; however , this requires a rather high value of @xmath0 , of order @xmath172 tev @xcite .
a similar cancellation of the heavy higgs and new physics contributions to @xmath1 also occurs in top seesaw models @xcite ; see ref .
we are grateful to bogdan dobrescu for bringing this paper to our attention .
] , and a consistent fit is obtained for @xmath173 far in excess of the usual sm upper bound , currently about 250 gev . this possibility is illustrated in fig .
[ hmass ] , where we fix @xmath174 , @xmath163 , and plot the constraints in the @xmath175 plane .
remarkably , values of @xmath173 as high as 800 gev are allowed at 95% confidence level .
( note that the approximation made in eq . , where the corrections of order @xmath97 in the higgs contribution to the oblique parameters have been neglected , is justified in the region of interest , since @xmath0 is still of order 1 tev .
) thus , the lh model provides an explicit , well - motivated example of a theory in which the sm upper bound on the higgs mass is avoided .
moreover , from the point of view of fine tuning in the higgs potential , the high values of @xmath173 are more natural in the context of this model @xcite .
for example , let us use the ratio of the one - loop top contribution to @xmath176 to the full @xmath176 , f = , as a quantitative measure of the fine tuning .
( larger values of @xmath169 correspond to higher degree of fine tuning . ) plotting the contours of constant @xmath169 indicates that , in the region of the parameter space consistent with precision electroweak constraints , the degree of fine tuning increases with the decreasing higgs mass . large values of @xmath173 are clearly preferred from the point of view of naturalness in the higgs potential . plane . in the narrow bands between the pairs of dashed lines ,
the ltp relic density is within 2@xmath5 of the central value provided by the wmap collaboration @xcite . for the detailed analysis of the ltp relic density ,
see ref .
@xcite . ]
-0.3 cm if t parity is an exact symmetry ( including the theory completing the description above the scale @xmath101 ) , the lightest t - odd particle ( ltp ) is stable .
generically , the ltp is the t - odd partner of the hypercharge gauge boson , which is electrically neutral and can play the role of wimp dark matter .
the ltp relic density has been computed in ref .
@xcite , and a region in the parameter space where the ltp can account for all of the observed dark matter has been identified . in fig .
[ darkmattertoo ] , the 2 sigma contours on the dark matter relic density are superposed over a plot of the precision electroweak constraints where @xmath174 , @xmath163 , and @xmath173 and @xmath0 are allowed to vary .
there is a region of the allowed parameter space in which the ltp can account for all of the dark matter .
and the symmetry breaking scale @xmath0 .
the contribution of the t - odd fermions to the t parameter is included assuming that it has the maximal size consistent with the constraint from four - fermi interactions , eq . . from lightest to darkest ,
the contours correspond to the 95 , 99 , and 99.9 confidence level exclusion . ]
-0.3 cm in figs .
[ fr][darkmattertoo ] , the contribution of the t - odd fermions to the @xmath1 parameter is neglected .
this approximation is justified as long as the t - odd fermions are sufficiently light : for example , for t - odd fermion mass of 300 gev , their total contribution to the @xmath1 parameter is very small , and does not have any noticeable effect on the fits . on the other hand
, heavier t - odd fermions can have a substantial effect .
this is illustrated in fig .
[ fr1 ] , where the t - odd fermion contribution has been assumed to have the maximal size consistent with the constraint from four - fermi interactions , eq . .
( this corresponds to the t - odd fermion masses saturating the upper bound in eq . . )
while the constraints in this case are more severe , consistent fits can still be obtained for values of @xmath0 below 1 tev .
in this paper we have calculated the dominant corrections to the precision electroweak observables at the one - loop level in the littlest higgs model with t parity @xcite .
we performed a global fit to the precision electroweak observables and found that a large part of the model parameter space is consistent with data . in particular , a consistent fit can be obtained for values of the nl@xmath5 m symmetry breaking scale @xmath0 as low as 500 gev .
furthermore , we found that the lh model can fit the data for values of the higgs mass far in excess of the sm upper bound , due to the possibility of a partial cancellation between the contributions to the @xmath1 parameter from higgs loops and new physics . combining our results with those of ref .
@xcite , we found that there are regions of parameter space allowed by precision electroweak constraints where the lightest t - odd particle can account for all of the observed dark matter .
we have argued that the corrections to low energy observables in the lh model are dominated by the top sector , and our analysis was primarily focused on those contributions . it would be interesting to perform a more detailed analysis of the effects from the gauge and scalar sectors
; however , we do not expect these effects to substantially modify our conclusions .
the analysis of the t - odd fermion sector in this paper relied on rather restrictive simplifying assumptions : in particular , the yukawa couplings in the t - odd sector were assumed to be flavor - diagonal and flavor - independent
. a possible non - trivial flavor structure of their couplings could have interesting experimental consequences .
moreover , these fermions should be sufficiently light to be pair - produced at the lhc , or even at the tevatron .
it is therefore important to analyze that sector of the model in more detail .
these additional fermions are not required for naturalness , but are needed to implement the sm fermions in a way consistent with t parity and compositeness bounds . while their effects on the precision electroweak observables are expected to be small , a possible non - trivial flavor structure of their couplings could have interesting experimental consequences . moreover , these fermions could be sufficiently light to be pair - produced at the lhc , and may affect the relevant phenomenology
. it would be interesting to analyze that sector of the model in more detail .
in conclusion we find that the littlest higgs model with t parity is only weakly constrained by precision electroweak data , and provides a viable alternative for physics at the tev scale .
apart from being theoretically attractive , the model has several features that are of interest for planning future experiments .
two examples are the possibility of a relatively heavy higgs , as discussed in this paper , and the similarity of many of the collider signatures of this model to the benchmark susy signatures , which will inevitably complicate the lhc analysis @xcite .
we hope that our analysis , which explicitly demonstrates the viability of the lh model , will open the door for further detailed studies of its collider phenomenology .
we would like to thank bogdan dobrescu and ian low for useful discussions related to this work .
we would also like to acknowledge helpful correspondence with thomas gregoire .
this work is supported by the national science foundation under grant phy-0355005 .
while the higher - order corrections for observable quantities in gauge theories must be gauge independent , an appropriate gauge choice can greatly reduce the complexity of a loop calculation , and make the underlying physics more transparent .
this is especially important in the case of spontaneously broken gauge symmetries .
while many issues in these theories are most easily understood in the unitary gauge , this gauge is ill - suited for loop calculations , leading to complicated intermediate expressions , and , in some cases , ambiguous answers boson contribution to the anomalous magnetic moment of the muon @xcite . ] .
experience with the sm radiative correction calculations indicates that it is best to use the renormalizable , or @xmath4 , gauges ; a special case of @xmath177 , the t hooft - feynman gauge , is especially useful . in this paper , we have used this gauge to calculate the @xmath147 coupling shift , see section [ zbbar ] .
since only unitary - gauge feynman rules have appeared in the literature so far for the lh models @xcite , we will briefly discuss the @xmath4 gauges for the lh model with t parity in this appendix .
we will focus on the charged gauge boson sector ; the analysis of the neutral sector is similar .
even though the calculations in the paper do not require it , in this appendix we will keep all correction of order @xmath33 , since an interesting effect of @xmath178 mixing first appears at that order .
the charged gauge boson mass matrix follows from eq . ; to order @xmath179 ( where @xmath180 ) it has the form @xmath181 diagonalizing this matrix results in the mass eigenstates @xmath182 in the lh model with t parity , the gauge couplings are set equal , @xmath183 and the mixing angle is given by @xmath184 .
the charged gauge boson mass eigenvalues are then @xmath185 , \ \ \ \ m_l^2 = \frac{g^2 v^2}{4 } \left [ 1 - \frac{1}{6 } \eps^2 + \cdots \right].\ ] ] spontaneous breaking of the gauge symmetries leads to the kinetic mixing between the gauge bosons and the goldstone boson fields in eq . .
in the mass eigenbasis , the mixing terms have the form @xmath186 { \nonumber \\ } & & + m_l w_l^- \pd_\mu \pi^+ \left ( 1 - \frac{1}{12 } \eps^2 \right)+ \mathrm{h.c.}\end{aligned}\ ] ] at order @xmath179 , the goldstone boson fields in eq . are not canonically normalized . to canonically normalize the goldstone fields ,
we perform the following rescaling : @xmath187 after this redefinition , there are still kinetic mixing terms involving the @xmath188 fields and the complex triplet , @xmath29 : @xmath189 these terms are diagonalized with the redefinition @xmath190 in terms of these new canonically normalized fields , the gauge boson - goldstone mixing terms are given by _
w & = & m_h w_h^- + & & + m_l w_l^- _ ^+ + a final rotation which leaves the kinetic terms diagonal , @xmath191 identifies @xmath192 as the combination of goldstones eaten by the heavy gauge bosons , and @xmath193 as the uneaten combination . a similar , but algebraically more involved
, analysis carries through for the massive neutral gauge bosons . following the usual logic of renormalizable gauges ,
we add a gauge - fixing term which , after integration by parts , cancels the mixing terms : @xmath194 the mass eigenstates in the eaten goldstone sector are @xmath144 , with mass @xmath195 , and @xmath192 , with mass @xmath196 .
( we have used the t hooft feynman gauge , @xmath197 , in section [ zbbar ] . ) note that the @xmath35 fields do not mix with @xmath188 and @xmath29 at any order in @xmath97 , since such mixing is forbidden by t parity .
the situation would be considerably more involved in the case of the littlest higgs without t parity . given the exact identification of the @xmath144 fields in eq .
with the lighter mass eigenstate in the @xmath4 gauges , it is straightforward to obtain the @xmath198 and @xmath199 vertices required in the calculation of section [ zbbar ] . expanding the @xmath8 matrices in eq . to linear order in @xmath10 , and using eqs .
, to transform to the mass eigenbasis for the top sector , we obtain -i_1 |b_l+ ^- ( t_r + t_r+ ) + h.c . = -i_t |b_l+ ^- ( t_r + t_r+ ) + h.c . where @xmath99 is the sm yukawa coupling , and @xmath200 is identified with the sm @xmath201 quark .
note that the couplings involving the @xmath200 and any one of the t - odd goldstone bosons , @xmath29 , @xmath27 or @xmath26 , vanish due to the structure of the lagrangian and the fact that the field @xmath83 is t - even .
n. arkani - hamed , a. g. cohen , e. katz and a. e. nelson , jhep * 0207 * , 034 ( 2002 ) [ arxiv : hep - ph/0206021 ] .
for a recent review and a comprehensive collection of references , see m. schmaltz and d. tucker - smith , arxiv : hep - ph/0502182 . c. csaki , j. hubisz , g. d. kribs , p. meade and j. terning , phys .
d * 67 * , 115002 ( 2003 ) [ arxiv : hep - ph/0211124 ] ; phys .
d * 68 * , 035009 ( 2003 ) [ arxiv : hep - ph/0303236 ] .
j. l. hewett , f. j. petriello and t. g. rizzo , jhep * 0310 * , 062 ( 2003 ) [ arxiv : hep - ph/0211218 ] .
m. perelstein , m. e. peskin and a. pierce , phys .
d * 69 * , 075002 ( 2004 ) [ arxiv : hep - ph/0310039 ] .
m. c. chen and s. dawson , phys .
d * 70 * , 015003 ( 2004 ) [ arxiv : hep - ph/0311032 ] .
d. e. kaplan and m. schmaltz , jhep * 0310 * , 039 ( 2003 ) [ arxiv : hep - ph/0302049 ] ; m. schmaltz , jhep * 0408 * , 056 ( 2004 ) [ arxiv : hep - ph/0407143 ] .
s. chang and j. g. wacker , phys .
d * 69 * , 035002 ( 2004 ) [ arxiv : hep - ph/0303001 ] .
e. katz , a. e. nelson and d. g. e. walker , arxiv : hep - ph/0504252 . h. c. cheng and i. low , jhep * 0309 * , 051 ( 2003 ) [ arxiv : hep - ph/0308199 ] . h. c. cheng and i. low , jhep * 0408 * , 061 ( 2004 ) [ arxiv : hep - ph/0405243 ]
. i. low , jhep * 0410 * , 067 ( 2004 ) [ arxiv : hep - ph/0409025 ] .
j. hubisz and p. meade , phys .
d * 71 * , 035016 ( 2005 ) [ arxiv : hep - ph/0411264 ] . c. f. berger , m. perelstein and f. petriello , arxiv : hep - ph/0512053 . m. e. peskin and t. takeuchi , phys .
d * 46 * , 381 ( 1992 ) .
a. manohar and h. georgi , nucl .
b * 234 * , 189 ( 1984 ) ; h. georgi and l. randall , nucl .
b * 276 * , 241 ( 1986 ) .
s. eidelman _ et al . _
[ particle data group ] , phys .
b * 592 * , 1 ( 2004 ) .
j. bernabeu , a. pich and a. santamaria , phys .
b * 200 * , 569 ( 1988 ) .
m. s. chanowitz , phys .
* 87 * , 231802 ( 2001 ) [ arxiv : hep - ph/0104024 ] .
p. bamert , c. p. burgess , j. m. cline , d. london and e. nardi , phys .
d * 54 * , 4275 ( 1996 ) [ arxiv : hep - ph/9602438 ] .
c. p. burgess , s. godfrey , h. konig , d. london and i. maksymyk , phys .
d * 49 * , 6115 ( 1994 ) [ arxiv : hep - ph/9312291 ] .
g. marandella , c. schappacher and a. strumia , nucl .
b * 715 * , 173 ( 2005 ) [ arxiv : hep - ph/0502095 ] ; arxiv : hep - ph/0502096 . b. a. dobrescu and c. t. hill , phys .
lett . * 81 * , 2634 ( 1998 ) [ arxiv : hep - ph/9712319 ] ; r. s. chivukula , b. a. dobrescu , h. georgi and c. t. hill , phys .
d * 59 * , 075003 ( 1999 ) [ arxiv : hep - ph/9809470 ] ; h. collins , a. k. grant and h. georgi , phys .
d * 61 * , 055002 ( 2000 ) [ arxiv : hep - ph/9908330 ] .
r. s. chivukula , c. hoelbling and n. j. evans , phys .
lett . * 85 * , 511 ( 2000 ) [ arxiv : hep - ph/0002022 ] .
j. a. casas , j. r. espinosa and i. hidalgo , jhep * 0503 * , 038 ( 2005 ) [ arxiv : hep - ph/0502066 ] .
d. n. spergel _ et al .
_ , astrophys .
j. suppl .
* 148 * , 175 ( 2003 ) [ astro - ph/0302209 ] .
k. fujikawa , b. w. lee and a. i. sanda , phys .
d * 6 * , 2923 ( 1972 ) . g. burdman , m. perelstein and a. pierce , phys . rev
. lett . * 90 * , 241802 ( 2003 ) [ erratum - ibid . * 92 * , 049903 ( 2004 ) ] [ arxiv : hep - ph/0212228 ] .
t. han , h. e. logan , b. mcelrath and l. t. wang , phys .
d * 67 * , 095004 ( 2003 ) [ arxiv : hep - ph/0301040 ] .
|
we compute the leading corrections to the properties of w and z bosons induced at the one - loop level in the su(5)/so(5 ) littlest higgs model with t parity , and perform a global fit to precision electroweak data to determine the constraints on the model parameters .
we find that a large part of the model parameter space is consistent with data .
values of the symmetry breaking scale @xmath0 as low as 500 gev are allowed , indicating that no significant fine tuning in the higgs potential is required .
we identify a region within the allowed parameter space in which the lightest t - odd particle , the partner of the hypercharge gauge boson , has the correct relic abundance to play the role of dark matter .
in addition , we find that a consistent fit to data can be obtained for large values of the higgs mass , up to 800 gev , due to the possibility of a partial cancellation between the contributions to the @xmath1 parameter from higgs loops and new physics .
hep - ph/0506042 + .5 cm * electroweak precision constraintson the littlest higgs model .3 cm with t parity * .2 cm * jay hubisz , patrick meade , andrew noble , maxim perelstein * + _ institute for high energy phenomenology , + f.r .
newman laboratory of elementary particle physics , + cornell university , ithaca , ny 14853 , usa _
+ 0.3truecm
|
the authors thank professors i. a. sanda and a. soni for useful discussions .
m. a. acknowledges support from the science and technology agency of japan .
e. k. acknowledges support from the japanese society for the promotion of science .
3.0 cm * figure 1 * : non - spectator contribution to @xmath99 decays .
+ 0.5 cm * figure 2 * : the current - current ( tree ) contribution to @xmath42 decay .
+ 0.5 cm * figure 3 * : the triangle quark loop diagram for gluon - gluon-@xmath4 vertex .
+ 0.5 cm * figure 4 * : branching ratio versus cp asymmetry as @xmath63 varies from @xmath100 to @xmath101 for three sets of @xmath102 values .
|
we calculate the branching ratio and direct cp asymmetry in nonleptonic two body @xmath0 decays @xmath1 . it is shown that the tree diagram and gluon fusion mechanism via penguin diagram have comparable contributions to these decays which , as a result , could provide an interesting venue for investigating cp violation .
our estimate shows that the direct cp asymmetry in the above decays could be as large as @xmath2 which along with a branching ratio @xmath3 should be accessible to experiment in the near future .
the discovery by the cleo collaboration of a larger than expected branching ratio for fast @xmath4 production in hadronic @xmath0 decays@xcite has led to extensive theoretical work on investigating the underlying mechanism .
one explanation which is based on the gluon - gluon-@xmath4 anomalous coupling , has been proposed by atwood and soni@xcite . in this mechanism , @xmath4 is produced by the fragmentation of the virtual gluon of the qcd @xmath5 penguin .
on the other hand , in order to formulate the inclusive @xmath6 and exclusive @xmath7 decays under the same mechanism , we proposed a nonspectator gluon fusion process using anomaly driven @xmath8 vertex@xcite .
the exclusive branching ratio @xmath9 obtained this way is in good agreement with experimantal data . in this paper , we focus on a different hadronic decay mode of @xmath0 mesons which may receive a significant contribution from the above mechanism .
indeed , if the anomalous @xmath8 vertex in conjunction with the qcd penguin is the underlying process for the fast @xmath4 production in @xmath0 meson decays , one could expect that the same mechanism also be an important part of the two body @xmath1 decay modes . the difference between these decay modes and @xmath7
decay is in that , unlike the latter one , the former decays receive a comparable contribution from the tree diagrams . as a result
, one could expect that the above decay modes provide a suitable avenue to investigate the direct cp asymmetry in charged @xmath0 meson decays . in this work
, we estimate the branching ratio and direct cp asymmetry in @xmath1 decays and point out a possible large cp asymmetry which should be accessible to experiment in the near future .
we start by repeating the derivation of the nonspectator contribution to @xmath7 matrix element from reference @xcite . using the dominant chromo - electric component of the qcd penguin , we obtain the following effective hamiltonian corresponding to figure 1 .
@xmath10 where @xmath11\;\ ; .\ ] ] the coefficient function @xmath12 is defined as @xmath13 where @xmath14 with @xmath15 being the internal quark mass@xcite .
@xmath16 is the form factor parametrizing the @xmath8 vertex @xmath17 using the decay mode @xmath18 , @xmath19 is estimated to be approximately 1.8 gev@xmath20@xcite . a re - arrangement of eq . ( 1 ) via fierz transformation @xmath21\;\ ; , \end{aligned}\ ] ] and using the definition of the decay constants for the @xmath0 and @xmath22 mesons in the context of factorization
, we can express the matrix element for @xmath7 decay as the following : @xmath23 leading to the exclusive decay rate @xmath24 where @xmath25 and @xmath26 are the three momentum of the k meson @xmath27}^{\frac{1}{2}}\;\ ; , \ ] ] and the energy transfer by the gluon emitted from the light quark in the b meson rest frame , respectively . inserting @xmath28 gev@xmath29 , @xmath30 gev and @xmath31 gev in eq .
( 7 ) results in the following exclusive branching ratio : @xmath32 which is in good agreement with the experimental data@xcite . on the other hand
, we observe that if @xmath8 anomalous coupling along with qcd @xmath5 penguin ( fig .
1 ) is indeed the dominant underlying mechanism for @xmath33 decay , then one could expect that a similar process , with @xmath34 quark replacing the strange quark , to contribute significantly to @xmath35 decay mode .
however , in this case , unlike @xmath7 decay , the tree and penguin amplitudes receive a ckm suppression factor of the same order and therefore there is a good possibility for existence of a large direct cp asymmetry in @xmath1 decay modes . the nonspectator contribution to @xmath36
can be obtained from eq .
( 6 ) by replacing @xmath37 with @xmath38 and @xmath39 with @xmath40 , where @xmath41+v_{ub}v_{ud}^*[e(x_u)-e(x_c)]\right ) \;\ ; .\ ] ] thus , including the expression for the tree amplitude ( fig .
2 ) which are derived by using the factorization assumption , the total matrix element for @xmath42 can be written as : @xmath43 @xmath44 and @xmath45 are the combinations of the wilson coefficients of the current - current operators and @xmath46 and @xmath47 , are defined as follows : @xmath48 @xmath49f_1^{b\to p}(q^2 ) \\ & + & \frac{m_b^2-m_p^2}{q^2}q^\mu f_0^{b\to p}(q^2 ) \;\ ; , \end{aligned}\ ] ] where @xmath50 .
we estimate @xmath51 gev by using the two - angle formalism for the @xmath52 mixing@xcite . as for the form factors , we use the numerical results @xmath53 and @xmath54 obtained in the bsw model@xcite . using the wolfenstein parametrization@xcite and the unitarity triangle convention for the phases of the quark mixing matrix elements@xcite @xmath55 where @xmath56 and @xmath57 are well determined@xcite .
the parameters @xmath58 and @xmath59 are correlated and their range have been determined from ckm unitarity fits@xcite . in order to show the dependence of the branching ratio and cp asymmetry on these parameters , we present our results for central values of @xmath60 and @xmath61 and three different sets of @xmath62 values from the constrained range . we note that different weak ( cp odd ) phases enter nonspectator ( penguin ) and tree amplitudes . also , an overall strong ( cp even ) phase @xmath63 for the nonspectator amplitude
is assumed which in combination with the weak phases leads to cp asymmerty in @xmath1 decays .
in fact , one possible source of this strong phase could be the form factor which parametrizes the @xmath8 vertex .
for example , in the sigma model formulation of @xmath64 where this form factor is derived from a quark triangle loop ( fig .
3 ) , an absorptive part is generated for @xmath65 ( @xmath66 is the mass of the quarks participating in the triangle loop i.e. , @xmath67 in the @xmath4 case)@xcite .
for @xmath68gev@xmath29 the magnitude of this strong phase can be around @xmath69 to @xmath70 depending on the input value for @xmath66 .
inserting @xmath71 ( @xmath72gev@xmath20 as mentioned before ) in eq . ( 11 ) and using the same input values for various parameters as in @xmath7 case , we obtain the following branching ratio for @xmath42 @xmath73\;\ ; .\end{aligned}\ ] ] the first term in eq .
( 15 ) is due to current - current operators ( tree ) while the second term is the contribution of the nonspectator gluon fusion process ( penguin ) and the last entry is the cross term .
the branching ratio for the cp conjugate process @xmath74 is obtained from the eq .
( 15 ) by changing the sign of the weak angles @xmath75 and @xmath76 .
consequently , the cp asymmetry which is defined as @xmath77 is proportional to @xmath78 and @xmath79 . in figure 4
, we illustrate the variation of @xmath80 and @xmath81 when @xmath63 takes on values in the range @xmath82 for three different sets of @xmath58 and @xmath59 .
it is quite interesting that a large asymmetry is possible due to the comparable nonspectator and tree contributions to this process .
for example , for @xmath83 and @xmath84 ( i.e. @xmath85 ) , @xmath86 could be as large as @xmath2 and @xmath3 if @xmath87 .
a larger branching ratio @xmath88 is possible for @xmath89 and @xmath90 ( i.e. @xmath91 ) however , the asymmetry is somewhat smaller around @xmath92 . on the other hand , for the preferred values @xmath93 and @xmath94 ( i.e. @xmath95 ) , the resulting branching ratio and asymmetry are @xmath96 for @xmath97 , which is the case if we assume that the absorptive part of the gluon - gluon-@xmath4 form factor @xmath16 is responsible for the strong phase . in conclusion
, we emphasize that a comparison between our results and other theoretical estimates@xcite indicates that @xmath98 and @xmath81 could receive a significant enhancement due to the inclusion of the nonspectator gluon fusion mechanism .
these decay modes could be accessible to experiment in the near future and therefore , provide a clean testing ground for cp violation mechanism in the standard model .
|
type ia supernovae ( sne ia ) are currently the most mature cosmological distance indicator which led to the discovery of the acceleration of the universe ( riess et a. 1998 ; perlmutter et al .
their use as distance indicators relies on the well calibrated light curve characteristics , namely a phenomenological relation between the peak luminosity and the light curve width ( `` phillips relation '' or `` stretching factor '' ; phillips 1993 ; perlmutter et al . 1997 ; phillips et al .
1999 ) . by clarifying natures of progenitor system(s ) of sne ia ( livio 2000 : hillebrandt & niemeyer 2000 ; nomoto et al . 2003 )
, one expects to obtain deep knowledge on the origin of the light curve relation for the better luminosity calibration , as well as new application of sne ia to a range of cosmological study . a progenitor of normal sne ia ( branch et al .
1993 ; li et al .
2001 ) is believed to be a white dwarf ( wd ) having nearly the chandrasekhar - mass .
recently , a few sne ia have been suggested to have originated from a wd whose mass exceeds the chandrasekhar - mass of a non - rotating wd , raising a possibility that not all sne ia are from a single class of progenitor system .
the suggestion mainly relies on their exceptionally large luminosity , which requires more than @xmath1 of @xmath0ni synthesized and ejected during the explosion .
the thermonuclear explosion converts the initial wd compositions partly to @xmath0ni .
@xmath2-rays from radioactive decay @xmath0ni @xmath3 @xmath0co @xmath3 @xmath0fe , through compton scatterings and thermalization , power the optical emission of sne ia .
thus , the peak luminosity is closely related to the mass of @xmath0ni initially synthesized ( hereafter @xmath4 ) . historically , the first observationally - based suggestion for the super - chandrasekhar wd explosion was made for sn ia 1991 t , as is summarized by lira et al .
( 1998 ) and fisher et al .
however , the distance to this sn has been actively debated , and the smaller value is now favored ( e.g. , saha et al . 2001 ) . as a result ,
@xmath4 required to explain the peak bolometric luminosity has been reduced down ; for the reddening @xmath5 and the distance modulus @xmath6 as adopted by stritzinger et al .
( 2006 ) , the peak @xmath7-band magnitude is @xmath8 .
stritzinger et al .
( 2006 ) derived the mass of @xmath0ni , @xmath9 , based on the peak bolometric luminosity @xmath10 erg s@xmath11 .
this amount of @xmath0ni could be marginally explained even by a chandrasekhar - mass model , although one may need an extreme , purely detonation - driven explosion ( khokhlov , mller , & hflich 1993 ) .
the progress had to wait until a discovery of over - luminous sne ia , which require that @xmath12 .
the first candidate , sn 2003fg ( snls-03d3bb ) , was reported by howell et al .
the sn reached @xmath13 mag at peak and @xmath14 erg s@xmath11 . from this , howell et al .
( 2006 ) estimated that @xmath15 , using an approximate relation between the peak luminosity and @xmath4 .
since there should be other elements required both theoretically ( i.e. , all the well studied chandrasekhar - mass models do not produce more than @xmath1 of @xmath0ni ; khokhlov et al .
1993 ; iwamoto et al .
1999 ; rpke et al .
2007 and references therein ) and observationally ( i.e. , the maximum - light spectrum of sn 2003fg was dominated by intermediate mass elements ) , howell et al .
( 2006 ) argued that sn 2003fg should come from a super - chandrasekhar - mass wd .
two other over - luminous sne ia , thus candidates of the super - chandrasekhar - mass wd explosions , have been reported : sne ia 2006gz ( hicken et al .
2007 ) and 2007if ( yuan et al . 2007 ) . for sn 2006gz ,
hicken et al . (
2007 ) adopted @xmath16 and @xmath17 ( but also see 2 ) , deriving @xmath18 at peak and @xmath19 erg s@xmath11 , and thus @xmath20(@xmath0ni ) @xmath21 .
in addition to the large luminosity , they are unique also in their spectra .
temporal sequence of the optical spectra of sn 2006gz is presented by hicken et al .
they identified c ii @xmath224745 , 5490 , 6580 , and 7324 , in the spectra taken @xmath23 days before the maximum brightness , with the equivalent width reaching @xmath24 and at the velocity of @xmath25 km s@xmath11 .
this is the strongest evidence for unburned carbon of any sne ia observed so far ( sn 1990n by leibundgut et al .
1991 ; sn 2006d by thomas et al .
adding to this , a si ii 6355 absorption velocity was unusually low well before the maximum ( @xmath26 km s@xmath11 at @xmath27 days before the maximum ) . without rapid decline observed in typical sne ia ( benetti et al .
2005 ) , the si ii velocity at the maximum brightness settled down to @xmath28 km s@xmath11 , a value typical of normal sne ia . based on this unusually slow decline in the si ii velocity , hicken et al .
( 2007 ) speculated that this might be a result of a dense unburned c+o envelope overlying a si - rich region ; the dense c+o envelope may be expected in the merger of two wds leading to the formation of the super - chandrasekhar - mass wd progenitor .
howell et al .
( 2006 ) also reported the probable detection of a c ii feature in the spectrum of sn 2003fg around the maximum brightness ; if pre - maximum spectra had been available , the c ii feature may have been in the spectra early on .
these spectroscopic features suggest that the strong c ii and the slow evolution of si ii velocity may be a distinct feature of over - luminous sne ia . however , the spectral sequence is only available for sn 2006gz , and it has not been clarified to what extent these features are common in over - luminous sne ia .
furthermore , it has not been clarified yet how these features , e.g. , the dense c+o envelope , are related to the progenitor system ( e.g. , see maeda et al .
2008b for a caution of interpreting the c+o - rich region as an outcome of merging two wds ) .
summarizing , the main argument for the super - chandrasekhar model is the large peak luminosity and the mass of @xmath0ni , with spectroscopic features only indicative .
detailed study on observed characteristics of super - chandrasekhar - mass models is missing . clearly , the luminosity is not the only quantity that is dependent on the underlying models .
the suggestion by howell et al .
( 2006 ) , about the super - chandrasekhar - model for sn 2003fg , is actually based on another observed character , i.e. , the velocity of the expanding sn materials as deduced from its spectra .
sn 2003fg showed exceptionally small velocity around the maximum brightness ( @xmath29 km s@xmath11 ) compare to that of normal sne ia ( @xmath30 km s@xmath11 ) . at the first look
, this seems fully consistent with the expectation , since the super - chandrasekhar - mass wd provides the large binding energy , and thus the small kinetic energy per mass as compared to an explosion of a chandrasekhar - mass wd .
jeffery , branch , and baron ( 2006 ) confirmed this statement on the relation between the wd mass and the velocity - scale .
although the argument may sound satisfactory , it is not a whole story .
studying emission processes should add further constraints on the underlying models .
there are two over - luminous sne ia whose observed characteristics are available in the literature , i.e. , 2003fg and 2006gz .
extensive observational data set for the moderately over - luminous sn 1991 t is also available . by examining these observations ( 2 ) , we clarify that they have quite different properties .
for example , sn 2006gz showed the velocity similar to normal sne ia - then , the argument for the super - chandrasekhar - mass model for sn 2006gz may be flawed .
the aim of this paper is to examine if the observed characteristics can be explained by super - chandrasekhar - mass models , by means of radiation transfer calculations .
we especially focus on sne 2003fg and 2006gz . in 3 , we describe sn ia models and summarize a method for the radiation transfer calculations .
results based on an extensive set of model calculations are shown in 4 .
discussion is given in 5 , where we examine uncertainties involved in our calculations .
we also discuss the nature of these peculiar over - luminous sne ia .
the emphasis is placed on our finding that not every observed characteristics can be interpreted within the context of the simple ( spherical ) super - chandrasekhar - mass wd models , and on our suggestion that the progenitor should largely deviate from spherical symmetry to account for the observed features ( at least for sn 2003fg ) .
implications on sn 1991 t are also discussed .
the paper is closed in 6 with concluding remarks .
in this section , we summarize observed features relevant to this work , for sne 1991 t , 2003fg , and 2006gz .
howell et al .
( 2006 ) presented their observational data of sn ia 2003fg .
the multi - band light curves are fitted well by @xmath31-corrected template light curves with the stretching factor ( perlmutter et al .
1997 , 1999 ) of @xmath32 .
the corresponding @xmath33 ( the magnitude change in the first 15 days past @xmath34 maximum : phillips 1993 ) is the following : @xmath35 , using equation ( 5 ) of perlmutter et al .
( 1997 ) for sn 2003fg in their figure 4 .
although this is different from what we derived here , the expected @xmath36 is @xmath37 days , which is well within the errors of our estimate ] .
we then convert @xmath33 to @xmath36 , the time since the maximum luminosity to half the maximum luminosity as measured in a bolometric light curve ( contardo , leibundgut , & vacca 2000 ) .
this is possible thanks to a correlation between these two quantities .
note that @xmath36 in this paper is a _
post - maximum _ quantity , not the _ rise time _ frequently used in the related field . for the conversion
, we fit an observationally derived set of ( @xmath33 , @xmath36 ) for nearby 9 sne ia presented in contardo et al .
( 2000 ) , assuming a functional form of @xmath38 .
we obtain @xmath39 and @xmath40 . using this relation , @xmath41 days for sn 2003fg .
the derivation of @xmath36 in sn ia 2006gz is straightforward .
the _ ubvri _
bolometric curve is presented by hicken et al .
( 2007 ) , and it directly gives @xmath42 days .
the light curve width is very different between the two sne ia .
they are very different in characteristic velocity , too .
sn 2003fg showed a si ii absorption velocity ( @xmath43 ) of @xmath44 km s@xmath11 around maximum brightness ( howell et al .
the same value for sn 2006gz is in the range @xmath45 km s@xmath11 ( hicken et al .
2007 ) .
typically , normal sne ia have @xmath36 in the range of @xmath46 days ( e.g. , contardo , et al . 2000 ) and @xmath43 in @xmath47 km s@xmath11 ( e.g. , hachinger , mazzali , & benetti 2006 ) . if we focus on sne ia with @xmath48 , the phillips relation indicates that @xmath49 ( stritzinger et al .
this corresponds to @xmath50 using the relation we derived above .
the examples are sne 2003du and 1990n ( e.g. , stritzinger et al .
2006 ) , both having @xmath51 km s@xmath11 . the time scale , @xmath36 , in sn 2003fg is at about the upper bound of normal sne ia , while @xmath43 is much lower .
@xmath36 in sn 2006gz is well above the range seen in normal sne ia , while @xmath43 is similar to normal cases . as for the bolometric magnitude at the maximum light ( @xmath52 ) , sn
2003fg has been claimed to be over - luminous ; @xmath53 erg s@xmath11 ( howell et al . 2006 ) .
sn 2006gz has also been claimed to be over - luminous ; @xmath54 erg s@xmath11 assuming @xmath16 and @xmath17 ( hicken et al .
however , there is a caveat for @xmath52 in sn 2006gz ; maeda et al .
( 2008b ) pointed out that the peak luminosity of sn 2006gz derived by hicken et al .
( 2007 ) involves large uncertainty , because the host - galaxy extinction has not been well constrained . if the host extinction would be totally negligible , then @xmath55 at peak and @xmath56 erg s@xmath11 , which is relatively bright but not over - luminous . for sn 1991 t ,
the following values are relevant .
@xmath57 given by contardo et al .
( 2000 ) , based on compilation of multi - band light curve .
@xmath51 km s@xmath11 ( hachinger et al . 2006 and references therein ) . in these properties ,
sn 1991 t represents an intermediate case between sne 2003fg and 2006gz , and indeed close to those of normal sne ia .
the peak luminosity with the favored distance modulus is @xmath58 erg s@xmath11 ( stritzinger et al .
2006 ) which is at the upper bound of normal sne ia , although the original suggestion for the super - chandrasekhar - mass model was based on the brighter estimate ( @xmath59 erg s@xmath11 ; fisher et al . 1999 ; contardo et al .
2000 ) . summarizing , the two most probable candidate super - chandrasekhar - mass wd explosions , i.e. , sne 2003fg and 2006gz , have very different observational characteristics , although this fact has not been emphasized in the literature .
we construct sn ia models starting with the density structure of the w7 model ( nomoto , thielemann , & yokoi 1984 ) which reproduces basic observational features of normal sne ia ( branch et al .
1985 ) . assuming the homologous expansion which should be a good approximation for an explosion of a compact progenitor , the density distribution as a function of velocity is uniquely determined from the normalized reference density distribution ( i.e. , w7 ) , by specifying the following model parameters : 1 .
@xmath60 : the mass of the wd .
@xmath61 : the wd central density at the burning ignition .
@xmath62 : the mass fraction of fe - peak elements produced by strong electron captures , i.e. , fe - peak elements excluding @xmath0ni ( e.g. , @xmath63ni , @xmath0fe , @xmath64fe ) .
@xmath65 : the mass fraction of @xmath0ni .
@xmath66 : the mass fraction of partially burned intermediate mass elements like mg , si , and s. the mass fraction of unburned c+o materials ( @xmath67 ) is then simply given by @xmath68 .
the procedure of the model construction follows that presented in jeffery et al .
( 2006 ) ( see also howell et al . 2006 and maeda et al
. 2008b ) . for given @xmath60 and @xmath61
, we compute the binding energy of the wd ( @xmath69 ) by the formulae given by yoon & langer ( 2005 ) , who examined a sequence of structure of a rotating wd . the energy produced by the nuclear burning ( @xmath70 )
is given by @xmath60 , @xmath62 , @xmath65 , and @xmath66 , by a simple relation @xmath71 the kinetic energy is then simply @xmath72 the reference density structure is then scaled in a self - similar manner , i.e. , @xmath73 and @xmath74 where @xmath75 is a velocity of a lagrangian fluid element , and @xmath76 the density there .
now that we have specified the density structure , we are left to specify the distribution of elements .
we examine three extreme cases so that our models should cover the real situation anyway .
1 . stratified ( model sequence `` a '' ) : characteristic burning layers are totally separated and stratified .
the electron capture region ( where the mass fraction of stable fe - peak elements is set to be unity ) is at the centre , surrounded by the @xmath0ni - rich region ( the mass fraction of @xmath0ni is set to be unity ) , then by the partially burned layer ( mass fractions of si and s are set to be 0.7 and 0.3 , respectively ) , and by the unburned c+o layer at the outermost region ( mass fractions of c and o are set to be 0.5 and 0.5 , respectively ) .
2 . mixing in the fe - rich region ( model sequence `` b '' ) : basically the same with the model sequence `` a '' , except that the innermost electron capture region and the @xmath0ni region are assumed to be fully mixed in the composition structure .
3 . fully mixing ( model sequence `` c '' ) : the compositions are fully and homogeneously mixed throughout the ejecta . the models `` a '' and `` b '' are most likely the case at least in normal sne ia , in a viewpoint of observed spectral evolution ( stehle et al . 2005 ; mazzali et al .
there are observational hints favoring the models `` a '' than `` b '' for some sne ia ( hflich et al .
2004 ; motohara et al .
examples of the density and composition structures are shown in figure 1 . ; solid ) and for sup2.3im ( with @xmath77 ; dashed ) .
also shown is the composition structure of sw7im , for the three different mixing prescription.,width=317 ] we have examined an extensive set of models as shown in table 1 .
@xmath60 is varied from @xmath78 to @xmath79 separated by @xmath80 each .
@xmath61 is set to be @xmath81 g @xmath82 in most models , as the model sequence with varying @xmath61 can be self - similar ( only different in @xmath83 ) .
the dependence of the observed behaviors on @xmath61 is examined only for the models with @xmath84 , and can be generalized to different masses in a straightforward way .
@xmath65 is fixed for given @xmath60 , corresponding to @xmath85 for the chandrasekhar - mass models ( @xmath86 ) and @xmath87 for the super chandrasekhar - mass models ( @xmath88 ) .
@xmath62 is varied from 0 ( or 0.1 so that @xmath83 should be positive ) to @xmath89 , corresponding to the complete burning of whole ejecta to the fe - peak elements , separated by 0.1 each .
either @xmath66 or @xmath67 is set to be zero to reduce the number of model calculations .
these two represent the two extreme cases , and thus should cover the real situation .
the total number of models we construct here is then 246 , including models with different @xmath62 and different mixing prescription .
[ cols="<,<,<,<,<,<,<,<,<",options="header " , ] models with the w7 reference density distribution . for the description of the parameters ,
see the main text .
each column represents a set of models : for example , model sw7im is examined with three different mixing prescription ( mixing `` a '' , `` b '' , and `` c '' ; see the main text ) . for each mixing , a set of ( @xmath62 , @xmath90 ) ( 0 , 0.57 ) , ( 0.1 , 0.47 ) , ( 0.2 , 0.37 ) , ( 0.3 , 0.27 ) , ( 0.4 , 0.17 ) , ( 0.5 , 0.07 ) , and ( 0.57 , 0 ) are examined . for each model which is descritized in 172 radial zones
, we performed radiation transfer calculations based on a numerical code presented in iwamoto et al .
the important assumption in our calculations is spherical symmetry .
the transport of @xmath2-rays from radioactive @xmath0ni / co / fe decay is solved under a gray atmosphere approximation with @xmath91 @xmath92 g@xmath11 , which should be a sufficiently accurate approximation for our present purpose ( maeda 2006a ) . with the energy input by @xmath2-rays ,
the optical photon transport is solved again under the gray approximation .
opacity is provided by thomson scatterings and line scatterings ,
i.e. , @xmath93 the ionization is determined by solving the saha equation ( i.e. , in lte ) , then used to determine @xmath94 at each time step and in each radial grid .
the prescription of @xmath95 is more complicated and uncertain , as it is affected by many weak lines whose atomic data are experimentally not well known ( woosley et al .
2007 ) . in this paper
, we follow a simplified , phenomenological prescription based on modeling observations of normal sne ia , and restrict our selves to focus on bolometric light curves .
the main advantage of this treatment is that the bolometric light curve is insensitive to a color evolution which can be quite complicated .
we follow mazzali et al .
( 2001 ) : @xmath96 \ { \rm cm}^{2 } { \rm g}^{-1 } \ .\ ] ] here , @xmath97 denotes a local ( position - dependent ) mass fraction .
the coefficient ( 0.5 ) is added to the original prescription of mazzali et al .
( 2001 ) so that the sw7 model sequence satisfies the observational constraint of normal sne ia with @xmath48 ( 4 ) .
figure 2 shows examples of the synthetic bolometric light curves . shown in the figure are the models with the mixing `` a '' ( stratified ) and @xmath77 .
if the other parameters are similar , models with larger @xmath60 have broader light curve shape around the maximum brightness , i.e. , @xmath36 is larger for larger @xmath60 .
also , the photospheric velocity at the maximum luminosity ( @xmath98 ) is smaller for larger @xmath60 because of smaller @xmath99 ( note that the velocity in sup1.7im is larger than sw7im because of the larger amount of @xmath4 and thus larger @xmath99 ) . ; sw7im ( thick - solid ) , lw7im ( thin - solid ) , sup1.7im ( thick - dashed ) , sup2.0im ( thin - dashed ) , sup2.3im ( thick - dotted ) , and sup2.6im ( thin - dotted ) . shown in parentheses are the photospheric velocity at the maximum bolometric brightness ( @xmath98 km s@xmath11 ) .
, width=317 ] this example shows that the relation between @xmath36 and @xmath98 can provide a strong tool to check the validity of a model as compared with the observations , although this has not been examined for sne 2003fg and 2006gz in the previous studies .
the constraints are the following : ( 1 ) a model should reproduce @xmath36 derived in 2 .
( 2 ) the photospheric velocity at the maximum luminosity in a model ( @xmath98 ) should be _ equal to or less than _ observed @xmath43 , since the si ii absorption must be formed above the photosphere .
figure 3 shows ( @xmath36 , @xmath98 ) for all the models , except for sup2.0im10 and sup2.0co10 , @xmath98 ) of sup2.0im10/sup2.0co10 can be obtained by simply shifting those of sup2.0im / co , along the curve defined by the mixing `` a '' to the bottom - right direction .
this is because larger @xmath61 is equivalent to lower @xmath83 . thus changing @xmath61
can be mimicked by changing @xmath83 ( i.e. , by changing @xmath62 ) . ] .
the sw7im / co model sequence ( with @xmath85 ) approximately satisfies the ( @xmath36 , @xmath43 ) constraint of the normal sne ia with @xmath100 ( e.g. , sne 2003du , 1990n , as the observed range shown in figure 3 ) .
, @xmath98 ) shown for the models .
the meaning of the symbols are the following : sw7im and sw7co ( filled squares ) , lw7im and lw7co ( open squares ) , sup1.7im and sup1.7co ( filled circles ) , sup2.0im and sup2.0co ( open circles ) , sup2.3im and sup2.3co ( filled triangles ) , sup2.6im and sup2.6co ( open triangles ) .
the sw7 models with different @xmath62 are connected by the lines , separately for different mixing prescription ( `` a '' , `` b '' , and `` c '' ) ; solid lines for sw7im and dotted lines for sw7co .
the observed range for normal sne ia with @xmath48 is shown by the shaded region .
the observed characteristics of sne 2003fg , 2006gz , and 1991 t , are also plotted as the shaded regions with observed uncertainty evaluated in 2 . , width=317 ] important discovery in figure 3 is that even a single model does not satisfy the observed characteristics of sn 2003fg . on the other hand , sn
2006gz can be explained by many models ( note that observed @xmath43 can be larger than synthetic @xmath98 ) , and most naturally by super - chandrasekhar models having @xmath101 , which results in @xmath102 for the favored non - mixing case ( `` a '' ) .
we have examined an extensive set of models including extreme cases , so that our failure to reproduce the observed characteristics of sn 2003fg represents a principal difficulty in the super - chandrasekhar model for this sn ia .
we found that the lower boundary in the @xmath103 plot , covered by varying various parameters for given @xmath60 , is represented by a model sequence of `` a '' with only varying @xmath62 . in the mixing cases ( `` b '' and `` c '' ) ,
@xmath0ni is mixed down to the low velocity , and thus the diffusion time - scale becomes long as compared with the stratified case `` a '' .
this results in larger @xmath36 in mixing `` b '' , `` c '' than `` a '' .
we also found that the curves , obtained by varying @xmath62 , almost overlap between the cases `` i m '' ( @xmath104 ) and `` co '' ( @xmath105 ) in the @xmath103 plane , and the ones with @xmath106 ( e.g. , sw7co ) cover the wider range of parameter space .
thus , the model sequence with the mixing `` a '' and @xmath106 form the lower boundary as we vary the value for @xmath62 , in the @xmath103 plane for given @xmath60 and @xmath4 .
if the observationally derived set of ( @xmath36 , @xmath43 ) is above the curve , the model sequence is acceptable ( i.e. , there are observationally acceptable combinations of parameters for given @xmath60 and @xmath4 ) . if it is below the curve , the combination of @xmath60 and @xmath4 should be rejected .
figure 4 shows the lower boundary of @xmath98 as a function of @xmath36 for different @xmath60 ( note that @xmath4 is fixed for given @xmath60 in this paper ) . plane shown for the w7-sequence models with various @xmath60 .
the meaning of the symbols are the following : sw7co ( filled squares and thick solid line ) , lw7co ( open squares and thin solid line ) , sup1.7co ( filled circles and thick dashed line ) , sup2.0co ( open circles and thin dashed line ) , sup2.3co ( filled triangles and thick dotted line ) , sup2.6co ( open triangles and thin dotted line ) . for the observed values ( shaded regions ) ,
see the caption of figure 3 .
, width=317 ] figure 4 shows that the curve is similar for models having the same @xmath4 .
the behavior can be explained as follows to the first approximation .
the time - scale of the light curve evolution around peak is scaled as follows ( arnett 1982 ) .
@xmath107 where @xmath108 is the opacity averaged over the ejecta .
the estimate of the photospheric velocity at the maximum brightness is complicated . for the present demonstration purpose ,
it is enough to assume that this is scaled with the average velocity , i.e. , @xmath109 then , combining these two expressions , we obtain the following relation : @xmath110 the opacity is provided by fe - peak elements , and stable fe - peak elements below the @xmath0ni - rich region does not significantly contribute to the opacity in non - mixing case `` a '' .
thus , @xmath111 therefore , the lower boundary ( i.e. , non - mixing case `` a '' ) in the @xmath103 plane for a given model is basically determined by @xmath4 .
note that the curves representing the lower boundary in figure 4 do not exactly behaves as @xmath112 , because the models with different @xmath62 are not self - similar to one another .
specifically , a model with lower @xmath62 ( thus larger @xmath36 and lower @xmath98 ) has @xmath36 larger than expected by this simple analytic relation , since they have more centrally concentrated @xmath0ni distribution leading to the larger diffusion time - scale .
one may wonder if our results are consistent with observations of normal sne ia and some peculiar sne ia .
first of all , we emphasize that we restrict ourselves to investigate chandrasekhar models with @xmath113 and super - chandrasekhar models with @xmath114 . as such
, our calculations do not directly cover ( a ) detailed variation in normal sne ia , and especially ( b ) sub - luminous sne ia which requires @xmath115 .
however , these issues are consistent with our results , and thus it is safe to use our calculations to model the super - chandrasekhar sne ia candidates : ( a ) a range of @xmath116 seen in normal sne ia can be explained by varying @xmath4 .
for example , if we take @xmath117 , the model sequence moves to the left [ i.e. , to smaller @xmath36 ; equation ( 10 ) ] .
this also reduces the peak luminosity as @xmath0ni is the energy source .
in this way , ( @xmath36 , @xmath43 ) of normal sne ia can basically be explained , simultaneously satisfying the phillips relation ( mazzali et al .
( b ) there are two sne ia showing @xmath98 smaller than sn 2003fg . they are `` sub - luminous '' peculiar sne ia 2002cx ( li
et al . 2003 ; branch et al . 2004 ; jha et al .
2006 ) and 2005hk ( jha et al .
2006 ; phillips et al . 2007 ;
sahu et al .
2008 ) , suggested to be a less energetic explosion with smaller mass of @xmath0ni than normal .
sn 2005hk showed @xmath118 km s@xmath11 ( @xmath119 according to the spectrum modeling ; sahu et al .
2008 ) and @xmath120 days
. some of our models have small @xmath83 ( tab .
1 ) , which results in small @xmath98 and large @xmath36 . for example
, figure 4 shows that for sufficiently small @xmath83 , @xmath121 km s@xmath11 _ and _ @xmath122 days for @xmath101 .
it has been estimated that @xmath123 in these faint sne ia ( phillips et al .
2007 ; sahu et al . 2008 ) , and a model with such a small amount of @xmath0ni should be considered . from equation
( 10 ) , we see that @xmath124 . applying our result to this relation
, we expect that @xmath125 days for @xmath126 and for @xmath127 km s@xmath11 , as is consistent with the observational features of sn 2005hk .
equations ( 9 ) and ( 10 ) are only illustrative , but already provide a solid argument which was confirmed by a set of our model calculations : ( 1 ) for given @xmath60 , if @xmath83 is larger , then @xmath36 is smaller _ and _ @xmath98 is larger . ( 2 ) to reproduce the observationally derived value of @xmath36 , one can thus play with @xmath83 ( in terms of @xmath62 , for example ) .
the model at the same time predicts smaller @xmath98 for smaller @xmath4 .
the combination ( @xmath36 , @xmath43 ) of sn 2006gz is totally consistent with the expectation from @xmath101 . on the other hand , the above relation between @xmath36 and @xmath98 raises a difficulty in interpreting the observational data of sn 2003fg , since ( @xmath36 , @xmath43 ) falls into the range even below the lower boundary of the chandrasekhar model with @xmath85 . as a function of @xmath36 .
see the caption of fig .
3 for the meaning of the symbols , for sn 2006gz , shown here are the range of the luminosity from hicken et al .
( 2007 ) with likely @xmath128 for the host extinction , and the stringent lower limit with @xmath129 ( see maeda et al .
sn ia samples from contardo et al .
( 2000 ) are shown by open stars ( among which sn ia 1991 t is explicitly indicated , but with the luminosity from stritzinger et al .
the observationally - derived relation between the @xmath36 and @xmath52 ( equivalent to the phillips relation ) is shown by a solid line ( excluding bright sn ia 1991 t ) and by a dashed line ( excluding bright sn ia 1991 t and faint sn ia 1991bg ) .
an approximate linear fit to the results of the sup2.0 sequence ( with mixing `` a '' ) is indicated by the solid line .
two lines ( `` x1.1 '' shown by dashed line , `` x0.8 '' shown by dotted line ) roughly indicate what is expected for models with @xmath130 and @xmath131 , respectively .
, width=317 ] the @xmath103 relation is the main point in the present paper .
we also examine the bolometric peak luminosity ( @xmath52 ) for the self - consistency test in this section .
figure 5 shows @xmath52 as a function of @xmath36 for all the models ( except for sup2.0im10/co10 ) .
normal sne ia ( shown in stars in fig .
5 ) can basically explained by the sw7 model sequence , if we allow dispersion in @xmath4 ( i.e. , @xmath132 if other properties are the same ) .
our model value ( @xmath101 ) is slightly smaller than required by the peak luminosity of sne 2003fg .
the best value is @xmath133 , as is consistent with the estimate by howell et al .
( 2006 ) . for sn 2006gz
, the best value is @xmath134 , if we adopt @xmath135 and @xmath136 ( hicken et al .
the stringent lower limit for sn 2006gz is given by setting the host extinction negligible ( 2 ) ; @xmath137 .
with these values , let us go back to the discussion on the @xmath103 relation . if we take @xmath130 for sn 2006gz , the lower boundary in fig .
4 moves slightly to the right , but still marginally consistent with sn 2006gz .
for sn 2003fg , further increasing @xmath4 makes the situation worse : the lower limit of @xmath98 now increases , making the deviation larger .
thus , the conclusion in the previous section does not change .
so far we have seen that properties of sn 2006gz is consistent with super - chandrasekhar models .
then , can we identify a set of model parameters ( @xmath60 and @xmath4 ) relevant to this sn ?
the @xmath138 constraint can be satisfied as long as @xmath139 .
this is only the upper limit corresponding to the condition @xmath140 . on the other hand
, @xmath141 requires @xmath12 for @xmath128 and @xmath137 for @xmath129 .
thus , we conclude that an explosion producing @xmath142 is the most likely explanation for sn 2006gz , with possible range of @xmath143 . directly deriving @xmath60 , rather than @xmath4 ,
turns out to be difficult : @xmath60 is degenerated in the @xmath103 relation . in terms of @xmath141 ,
@xmath60 is again basically degenerated according to the approximate relation @xmath144 .
although there is a diversity resulting from different @xmath60 , the model luminosity differs up to only @xmath27 per cent for the same mixing prescription , between @xmath145 and @xmath79 .
the variation is smaller than the observed error in @xmath52 ( @xmath146 per cent ) , and thus it is difficult to derive @xmath60 from the analysis presented in this paper .
additional test on the models can be provided by the ejecta temperature .
the discussion in this section is only qualitative : deriving the photospheric temperature requires detailed spectrum modeling ( e.g. , hachinger et al .
2008 ) , which is beyond the scope of the present study .
the ratio of equivalent widths of si ii @xmath225972 to si ii @xmath226355 provides a temperature indicator . in normal sne ia , the ratio is smaller for sne with larger peak luminosity , and thus higher temperature ( nugent et al . 1995 ; hachinger et al . 2008 ) .
the ratio in sn 2003fg is similar to that in relatively faint sn ia 1994d ( fig . 3 of howell et al .
2006 ) , and that in sn 2006gz is similar to normal sn ia 2003du ( fig . 1 of hicken et al .
these infer that the temperature of sn 2003fg is lower than normal sne ia with @xmath147 , and that in sn 2006gz is comparable to the normal case .
figure 6 shows the effective temperature at bolometric maximum brightness in our models .
the effective temperature is generally a decreasing function of @xmath36 ; although the photospheric velocity is smaller for larger @xmath36 ( fig .
4 ) , the peak luminosity is smaller and the photospheric radius ( i.e. , the photospheric velocity multiplied by the peak date ) is larger for larger @xmath36 .
the effect of the latter functions is more important , resulting in the dependence in figure 6 . for models resulting in same @xmath36 ,
the temperature is larger for larger @xmath60 ( see fig .
4 : the photospheric velocity is slightly smaller for larger @xmath60 ) .
the temperature of sn 2003fg looks to be lower than normal ( i.e. , sw7 models ) , and @xmath36 is comparable to normal sne ia ( i.e. , @xmath148 days ) .
this requires that the ejecta mass is smaller than in sw7 .
this is again consistent with our conclusion independently derived using the photospheric velocity .
the temperature of sn 2006gz is comparable to normal sne ia , but @xmath36 ( @xmath149 days ) is larger than normal cases .
this indicates that the ejecta mass is larger than in normal sne ia ( fig .
the meaning of the symbols are same as in figure 4 .
models shown here are for the mixing case `` a''.,width=317 ]
one may wonder that uncertainty involved in our calculations might change our conclusion . in this section , we examine dominant sources of the uncertainty in our calculations . here
we examine effects of ( a ) different opacity prescription and ( b ) density distribution . in short ,
these do not alter our conclusion .
we have performed the same calculations for all the models , with different opacity prescription . here
, the opacity takes the following form : @xmath150 @xmath92 g@xmath11 , independent from composition .
the result is shown in figure 7 .
the combination ( @xmath36 , @xmath98 ) of the super - chandrasekhar wd model is again above that of the chandrasekhar model and above ( @xmath36 , @xmath43 ) of sn 2003fg .
the observed values of sn 2006gz are slightly above the lower boundary of the super - chandrasekhar model , and thus this sn is quite consistent with the super - chandrasekhar explosion . in this case , @xmath151 as the opacity is independent from the composition . the same conclusion as in the previous section applies , only if we use @xmath60 as the major function rather than @xmath4 : to explain the observational characteristics of sn 2003fg , we require @xmath60 smaller than in the normal sne ia . and @xmath85 ( filled circles and thick dashed line ) , exponential with @xmath84 and @xmath152 ( open circles and thin dashed line ) , constant density with @xmath86 and @xmath85 ( filled triangles and thick dotted line ) , constant with @xmath84 and @xmath152 ( open triangles and thin dotted line ) . for comparison ,
the w7-sequence is also shown ( sw7co by filled squares and thick solid , sup2.0co by open squares and thin solid ) .
, width=317 ] the models with exponential or constant density distributions are examined for selected model parameters to check the uncertainty .
most of existing explosion models ( including the w7 model ) predict more or less exponential density distribution ( e.g. , woosley et al .
we constructed the exponential density distribution from @xmath153 and @xmath83 according to description in jeffery et al .
( 2006 ) and woosley et al .
the construction of the constant density distribution is trivial .
these two types of density distribution are examined only for the following cases : 1 .
@xmath86 and @xmath154 2 .
@xmath155 g @xmath82 3 .
@xmath156 ( @xmath86 ) or @xmath157 ( @xmath84 ) 4 .
@xmath105 5 . mixing @xmath158 `` a '' for each model
, @xmath62 is varied to form the lower boundary in the @xmath159 plot for given @xmath60 .
the result is shown in figure 8 .
exponential models are similar to the w7 density models in the @xmath103 plane .
the constant density distribution predicts smaller @xmath98 than in the other models , but not drastically change the result .
the constant density distribution is an extreme assumption , probably resulting in the lowest value of @xmath98 for given @xmath36 .
thus , the uncertainty in the density distribution , as long as spherical symmetry is assumed , does not change our conclusion . in this section ( 5.2 ) and
the following two sections ( 5.3 & 5.4 ) , we summarize our results for individual objects , and discuss implications . for sn 2003fg
, we have found that the observational characteristics can not be put into the super - chandrasekhar - mass wd explosion scenario ; most importantly , the observed @xmath103 relation ( 4.1 ) requires that either @xmath4 or @xmath60 ( or both ) should be smaller than even the chandrasekhar mass , contrary to the earlier expectations ( howell et al .
additional support is provided by the ejecta temperature ( 4.3 ) , which also indicates that the ejecta mass ( i.e. , @xmath60 ) is smaller than the chandrasekhar mass . on the other hand , the large peak luminosity requires that @xmath160 ( 4.2 ; howell et al .
2006 ) .
this is apparently a contradiction . to remedy the problem
, we suggest that the ejecta structure is far from spherical .
our radiation transfer calculations in this paper assume spherical symmetry , and for example proportional coefficient in equation ( 8) for @xmath98 should be a function of the viewing angle in the presence of large deviation from spherical symmetry .
if the viewing angle is such that the effective ( isotropic ) mass is small along the line - of - sight , this may effectively look like an explosion with small amount of @xmath0ni and/or @xmath60 , although the large luminosity can probably be provided by @xmath0ni in the whole ejecta , not only along the light - of - sight ( note that the photospheric velocity is more sensitively affected than the peak luminosity in asymmetric sn models ; maeda et al .
2006b ; tanaka et al .
2007 ) .
the important finding here is that the mass should be effectively small as viewed from an observer to satisfy the @xmath103 constraint .
such an explosion can not be a strongly jetted explosion of a spherically symmetric progenitor star : the jet - type explosion should yield smaller @xmath36 for an observer closer to the jet direction ( thanks to the large isotropic @xmath161 ) but at the same time resulting in larger @xmath98 ( for the same reason ) ( see e.g. , maeda et al .
2006b ; but see hillebrandt , sim , & rpke 2007 ) .
alternatively , we suggest that _ the progenitor star _ is highly aspherical . this may actually be consistent with the super - chandrasekhar model , as such a massive wd should rotate rapidly to support an exceeding mass . although an explosion based on such a deformed configuration has not been examined except for the purely detonation model ( steinmetz , mller , & hillebrandt 1992 ) , we believe it is rational to assume that the structure after the explosion preserves the initial configuration to some extent .
the disc - like structure has effectively small isotropic mass if viewed along the axis of the rotational symmetry .
we suggest this is a situation in sn 2003fg .
hillebrandt et al .
( 2007 ) suggested an off - centre explosion model ( a kind of one - sided jet - like explosion model ) within the context of a chandrasekhar - mass wd explosion , as an alternative explanation for sn 2003fg ( see also sim et al .
their argument is based on the viewing angle effect on the light curve features , i.e. , the larger luminosity and smaller diffusion time - scale for an observer closer to the direction of the @xmath0ni - rich blob .
we emphasize the differences between our and their works : ( 1 ) our suggestion on the ejecta asymmetry is based on the combination of the light curve and spectral features , and ( 2 ) the disc - like / oblate geometry , as we favor in this work , is different from their suggestion .
although we have clarified the need for the ejecta asymmetry for sn 2003fg , the progenitor wd mass ( @xmath60 ) is not conclusively constrained by the present study .
this will need the detailed multi - dimensional radiation transfer calculations ; especially , we need to understand the dependence of the luminosity on the geometry and the viewing angle . we have found that observed features of sn 2006gz are consistent with expectations from the super - chandrasekhar - mass wd explosion scenario .
a difficulty in identifying the underlying model for sn 2006gz is the uncertainty of its host s extinction and the luminosity .
although we can not perfectly reject the possibility that this is an explosion of a chandrasekhar wd , we favor that this is indeed a super - chandrasekhar wd explosion ; the @xmath103 relation ( 4.1 ) and the ejecta temperature ( 4.3 ) are consistent with the expectations from a super - chandrasekhar wd explosion . for sn 2006gz
, it is not necessary to introduce the ejecta asymmetry , unlike sn 2003fg .
these two sne may be intrinsically different , or , sn 2006gz may be explained by the configuration similar to that of sn 2003fg , but viewed at a different angle . here , we point out that the difference in the observational features of these two sne are qualitatively consistent with expected difference arising from the same configuration but viewed at different orientations .
first , these two sne ia require similar amount of @xmath0ni as estimated from the peak luminosity , and the main difference is in the spectroscopic feature ( @xmath98 ) and the light curve width ( @xmath36 ) .
the latter two are expected to be more sensitively dependent on the viewing angle than the estimated value of @xmath20(@xmath0ni ) ( maeda et al .
for the disc - like / oblate configuration , we expect larger @xmath98 and larger @xmath36 for the larger inclination .
these two values in sne 2003fg and 2006gz follow this tendency .
this is , however , only a qualitative argument , and discriminating these two possibilities ( the difference in the ejecta shape or the viewing angle ) needs more detailed , multi - dimensional study .
maeda et al .
( 2008b ) reported that sn 2006gz is peculiar also at late phases , @xmath162 days after the explosion .
applying the standard @xmath0ni / co / fe heating scenario , they estimated @xmath4 by a factor of 5 smaller that estimated with the early - phase data .
this contradiction could mean either of the following two possibilities ; ( 1 ) sn 2006gz was not powered by decays of @xmath0ni / co / fe at the early - phases ( which then casts doubt on the super - chandrasekhar - mass wd progenitor ) , or ( 2 ) it was powered by the decay , but for some reason ( e.g. , thermal catastrophe or dust formation ) , the bulk of the emission might be shifted to nir / mid - ir .
based on the result in this paper , we favor the second possibility . for sn 1991 t
, @xmath36 is at the upper boundary of normal sne ia , and @xmath43 is at the lower boundary .
this is marginally explained by the sw7 model sequence with @xmath85 .
thus , this constraint results in @xmath163 . on the other hand , the other constraint from @xmath52 results in @xmath164 ( fig .
5 ) . strictly speaking ,
these two constraints are not mutually consistent , and the same argument for sn 2003fg can apply to sn 1991 t .
however , the deviation between the observations and models is not as large as in sn 2003fg , and sn 1991 t may be marginally consistent with a spherical explosion with @xmath165 within uncertainties involved in our model calculations .
alternatively , sn 1991 t may also be an aspherical , disc - like explosion like our suggestion for sn 2003fg .
however , the intrinsic property ( e.g. , the explosion geometry , the mass of @xmath0ni ) of sn 1991 t is likely different from that of sn 2003fg ( note that there is a possibility that sn 2006gz is intrinsically similar to sn 2003fg ) . the estimated value of @xmath20(@xmath0ni ) is different , and other features ( @xmath98 , @xmath36 ) do not seem to be consistent with the expectation from the disc - like / oblate geometry which we favor for sn 2003fg .
in this paper , we critically examined super - chandrasekhar - mass wd models for ( candidate ) over - luminous sne ia 2003fg , 2006gz , and moderately over - luminous sn ia 1991 t .
our new approach is to use two observed features , @xmath36 and @xmath98 .
this is equivalent to make use of a combined information of the light curve and spectra , and thus is more powerful than previous studies which mainly relied on the peak luminosity .
our conclusions are summarized as follows : 1 .
somewhat negatively , the observations of sn 2003fg are not readily explained by the standard @xmath0ni / co / fe heating scenario .
the combination of relatively small @xmath36 and small @xmath98 requires that either @xmath60 or @xmath4 ( or both ) should be smaller than in normal sne ia .
the observations of sn 2006gz can be naturally accounted for , by the ( spherical ) super - chandrasekhar - mass model with @xmath142 . although the peak luminosity of sn 2006gz is largely uncertain ( 2 )
, we found that the super - chandrasekhar - mass model is consistent with various observational features of sn 2006gz , and thus we favor the interpretation that sn 2006gz was indeed over - luminous at peak as suggested by hicken et al .
3 . observed features of sn 1991 t is marginally explained by a spherical explosion of a wd and @xmath20(@xmath0ni ) @xmath166 .
this may either be a chandrasekhar wd or a super - chandrasekhar explosion .
the failure of fitting sn 2003fg is _ not _ a result of the @xmath167 relation expected for sne ia . in short
, the large amount of @xmath0ni should inevitably yield too large diffusion time - scale .
one can then try a model with large @xmath83 to reduce the diffusion time - scale , but this inevitably leads to the large velocity - thus the contradiction .
therefore , the observed characteristics of sn 2003fg are indeed inconsistent with any parameter set within the standard @xmath0ni / co / fe heating scenario .
we suggest a solution to remedy the problem - the ejecta asymmetry resulting from a disc - like or oblate density structure of a rapidly rotating progenitor wd
. this may indeed be consistent with the super - chandrasekhar wd scenario . within the results obtained by the present calculations ,
however , we can not conclusively derive the progenitor wd mass .
an interesting possibility is that the different observational features of sne 2003fg and 2006gz may be unified into one scheme , i.e. , an explosion of super - chandrasekhar wd viewed at different directions .
thomas et al . ( 2002 ) pointed out that a signature of ejecta asymmetry can be identified in the absorption strength of si ii @xmath226355 .
they concluded that the typical scale of inhomogeneity should be smaller than the size of the photosphere in normal sne ia , in order to account for uniformity of the absorption depth of the si ii . from fig
. 1 of hicken et al .
( 2007 ) , we see that the absorption strength is indeed different between sne 2003fg and 2006gz ; sn 2003fg seems to show the weaker absorption .
this may support the hypothetical configuration mentioned above , i.e. , disc - like ejecta structure with sn 2003fg viewed closer to the polar than sn 2006gz ; in the polar direction , there are a small amount of materials to absorb the light emitted at the photosphere , leading possibly to the weaker absorption strength in sn 2003fg
. our results can be tested by future observations of over - luminous sne ia .
first , we expect a large degree of polarization in sn 2003fg - like over - luminous sne ia , depending on the viewing direction .
if sn 2006gz is an explosion similar to sn 2003fg , but viewed at different orientation , we also expect a large degree of polarization for sn 2006gz - like sne ia .
next , late - time nebular spectra may be used to infer the distribution of @xmath0ni directly to probe the distribution of @xmath0ni at @xmath168 year after the explosion ( maeda et al .
some unidentified emission process at the late - phase might cause this unexpected behavior ( 5.3 ) , and spectroscopy at relatively early epochs is therefore recommended for the study of geometry for these peculiar sne ia . ]
( motohara et al . 2006 ; maeda et al . 2008a ; modjaz et al . 2008 ) . furthermore ,
if we plot a number of over - luminous sne ia on the @xmath103 plot , it will tell in a statistical way how large the deviation from spherical symmetry on average is , and how much the intrinsic diversity ( e.g. , @xmath60 ) in the super - chandrasekhar - mass wd explosions is .
more detailed , multi - dimensional studies are necessary to confirm our suggestions and speculations .
first , multi - dimensional hydrodynamic simulations , based on a rapidly rotating super - chandrasekhar wd , should be useful to check if our preferred geometry , i.e. , the disc - like or oblate distribution of density and @xmath0ni , can indeed be realized .
next , multi - d radiation transport calculations should be useful to see if the @xmath103 constraint can indeed be satisfied .
we postpone these studies to the future , since in this paper we concentrate on examining a large parameter space to clarify the applicability and difficulty in spherically symmetric models ( practically impossible to do in expensive multi - dimensional study ) , and to clarify the need for the non - spherical ejecta ( for sn 2003fg ) .
identifying the geometry should be useful to evaluate possible contamination of sne ia , which do not follow the phillips relation , in the cosmological study .
sn 2006gz is roughly consistent with the phillips relation if we adopt the peak luminosity derived by hicken et al .
( 2007 ) ( 4.2 and 4.3 ; fig .
5 ) , while sn 2003fg is not ( howell et al .
our scenario for the disc - like / oblate ejecta infers that over - luminous sne ia which do not follow the phillips relation ( e.g. , sn 2003fg ) , should also show the peculiar low photospheric velocity .
thus , by performing spectroscopy , the sne ia with large deviation from the phillips relation can be easily identified , and can be removed from samples for the cosmological study .
the authors would like to thank kenichi nomoto and masaomi tanaka for useful discussion , and the anonymous referee for many constructive comments .
this research has been supported by world premier international research center initiative ( wpi initiative ) , mext , japan , and by the grant - in - aid for young scientists of the jsps / mext ( 20840007 ) .
this research has made use of the cfa supernova archive , which is founded in part by the national science foundation through grant ast 0606772 .
nomoto k. , uenishi t. , kobayashi c. , umeda h. , ohkubo t. , hachisu i. , kato m. 2003 , in from twilight to highlight : the physics of supernovae , ed . w. hillebrandt & b. leibundgut ( berlin : springer ) , 115 ( astro - ph/0308138 )
|
a few type ia supernovae ( sne ia ) have been suggested to be an explosion of a super - chandrasekhar - mass white dwarf ( wd ) to account for their large luminosities , requiring a large amount of @xmath0ni .
however , the candidate over - luminous sne ia 2003fg , 2006gz , and ( moderately over - luminous ) sn 1991 t , have very different observational features : the characteristic time - scale and velocity are very different . we examine if and how the diversity can be explained , by 1d _ spherical _ radiation transport calculations covering a wide range of model parameters ( e.g. , wd mass ) . the observations of sn 2006gz are naturally explained by the super - chandrasekhar - mass model .
sn 1991 t represents a marginal case , which may either be a chandrasekhar or a super - chandrasekhar - mass wd explosion . on the other hand ,
the low velocity and short time - scale seen in sn 2003fg indicate that the ejecta mass is smaller than the chandrasekhar - mass , which is an apparent contradiction to the large luminosity .
we suggest that the problem is solved if the progenitor wd , and thus the sn explosion , is aspherical .
this may reflect a rapid rotation of the progenitor star , likely a consequence of the super - chandrasekhar - mass wd progenitor .
the observed differences between sne 2003fg and 2006gz may be attributed to different viewing orientations .
[ firstpage ] white dwarfs radiative transfer supernovae : individual ( sn 2006gz , sn 2003fg , sn 1991 t )
|
when a sunlight photon is absorbed by a light - sensitive molecule ( such as chlorophyll or carotenoid ) in a light - harvesting photosynthetic complex ( lhc ) , the photon energy is stored in the molecule in the form of an exciton , an excited electron state of the molecule .
the exciton then travels very quickly ( some picoseconds ) inside the lhc and reaches the reaction center ( rc ) , where charge separation , and afterwards relatively slow chemical reactions take place @xcite .
both the primary processes of exciton dynamics and charge separation occur in the presence of a protein ( and solvent ) environment at ambient temperature . in the framework of frster s resonance excitation transfer theory @xcite
, the energy transfer is so fast that both fluorescence and recombination ( due to the environment ) can be neglected . when modeling these primary exciton transfer ( et ) processes , the light - sensitive molecule is usually associated with a geometrically localized site , @xmath0 , having excited electron energy @xmath1 @xcite .
the total number of sites , @xmath2 , depends on the photosynthetic system .
for example , @xmath3 in the cp29 lhc which is closely associated with photosystem ii ( psii ) @xcite .
the sites interact via dipole - dipole ( or exchange ) interaction , which is described by matrix elements @xmath4 .
similar to many molecular systems , the simplest unit of this picture is a dimer , which describes the interaction between two ( not necessarily neighboring ) sites .
the dimer is characterized by a donor ( site @xmath0 ) and an acceptor ( site @xmath5 ) .
denote by @xmath6 the difference between the two excited electron energy levels .
the donor and acceptor can be weakly coupled ( @xmath7 ) or they can be strongly coupled ( @xmath8 ) @xcite .
the dimer is embedded in , and interacting with , a protein - solvent environment .
one then introduces the constants of interaction @xmath9 and @xmath10 characterizing the strength of the interaction between the dimer sites and the environment .
the protein - solvent environment is characterized by its correlation function , which depends on the temperature and some parameters which describe the spectral density ( the ` spectral function ' ) of protein - solvent fluctuations at low and high frequencies . in order to describe different environmental effects
, this correlation function can be taken to vary from a quite standard form @xcite to a rather complicated one @xcite .
the quantum dynamics of the dimer is characterized by its @xmath11 time - dependent reduced density matrix , which is obtained from the ` total ' dimer - environment density matrix by averaging over ( tracing out ) the environmental degrees of freedom .
the reduced dimer dynamics is then formally equivalent to that of an effective spin @xmath12 .
the protein - solvent environment is often modeled by a set of linear quantum oscillators ( bosonic degrees of freedom ) which characterize the dynamics of the protein and solvent atoms . in this
spin - boson " model , the interaction between the dimer and the environment is usually reduced to the self - consistent ( adiabatic ) renormalization of the dimer donor " and acceptor " energy levels @xcite .
see also @xcite for different approaches .
this so - called diagonal interaction " @xcite can be generalized , if needed , to a
non - diagonal interaction " @xcite .
the dynamics of the diagonal components of the reduced dimer density matrix ( relative to the energy basis ) is characterized by the et rate .
the dynamics of the non - diagonal components of the reduced density matrix is characterized by the decoherence rate .
it describes the evolution of quantum coherent effects which are currently the focus of intense theoretical and experimental studies @xcite ( see also references therein ) .
even the relatively simple spin - boson model has surprisingly many unresolved and rather subtle theoretical issues .
the first one is related to the structure of the dimer electron energy spectrum .
a real dimer based , for example , on two chlorophyll molecules , has many vibrational degrees of freedom @xcite .
this means that the hilbert space of the electron energy spectrum of a dimer is in actual fact very complicated ( not two - dimensional ! ) , with the two - level effective spin @xmath12 system only being a reasonable approximation to reality .
the second issue is related to the protein - solvent environment .
there are many models which describe this environment , but there is no consensus or clear understanding as to which one is the right one @xcite .
the third issue is related to the dimer - environment interaction .
the problem is that the interaction constants @xmath9 and
@xmath10 are not small ( compared to @xmath4 ) . a standard perturbation theory can therefore not be used .
indeed , the famous marcus formula for the electron transfer rate , @xmath13 , from an initially populated donor has the form , @xmath14 @xcite , where the interaction constant , @xmath15 , appears in the denominators of both the prefactor and in the exponent . in the standard perturbation theory approach ( such as the bloch - redfield theory @xcite ) , the rate is proportional to @xmath16 .
it is therefore clear that the marcus rate can not be derived by using a standard perturbative approach . in this paper , we resolve the third issue mentioned above , i.e. , we develop a theory which allows for strong ( indeed , arbitrary ) values of the interactions .
the donor and the acceptor of the dimer can be localized close to each other ( @xmath17 ) . for a protein - solvent environment
whose correlation length exceeds this distance , it is reasonable to consider the donor and acceptor to be coupled to a single reservoir , a situation which we call the collective reservoir model .
the opposite case , when each dimer site is coupled to its own , independent environment , is called the local reservoirs model .
collective and local interactions were discussed previously in @xcite ( see also references therein ) and in @xcite for stochastic environments .
the advantage of the current work is that our mathematical analysis is rigorous , meaning that any approximation made can be controlled , and conditions of validity of these approximations can be given .
this is especially important because , as mentioned above , standard perturbation approaches can not be used .
the main result of this work is a controlled expansion , for small @xmath4 , of the reduced dimer density matrix , valid for all times @xmath18 and for arbitrary coupling constants @xmath19 and temperatures @xmath20 . from this expansion ,
we derive the rates of the processes of electron ( excitation ) transfer and decoherence .
we show that in certain regimes , those rates coincide with the expressions obtained previously from the usual marcus formula .
we discuss the conditions on the spectral function of the reservoir at low frequencies which lead to long - time quantum coherences .
this issue is of significant interest in recent research on the et dynamics in lhcs , see e.g. @xcite and references therein .
the paper is organized as follows . in section [ descrmodel ]
, we describe the model and give an outline of our main results . in section [ mainresdetsect ] ,
we give the results in more mathematical detail and present the results of our numerical simulations .
section [ sectdynres ] is devoted to the mathematical approach of the dynamical resonance theory .
we summarize our results in a conclusion section . finally , in appendices [ lsosect]-[reservoirssect ] we present the detailed derivation of a few intermediate results .
we consider two types of dimer - reservoir interactions . in one ,
both dimer sites are coupled to the _ same _ , collective , heat bath . in the other one ,
each dimer site is coupled to an _ independent _ , local , heat bath .
the hamiltonian of the * collective reservoir model * is @xmath21 where @xmath22 here , @xmath23 is the frequency of mode @xmath24 and @xmath25 is the ` form factor ' , determining how strongly the mode @xmath24 is coupled to the dimer . while the quantities in are written down for discrete modes indexed by @xmath24 , we consider a reservoir with _ continuous modes _ , where @xmath24 becomes a continuous variable ( e.g. @xmath26 or a finite interval in @xmath27 , or @xmath28 ) .
protein - solvent environments are usually described by an ensemble of atoms , called a molecular reservoir model " , in contrast to an ensemble of modes like phonons ( see also @xcite ) . in sections
[ sectdynres ] and [ reservoirssect ] , we discuss the relation between these two environment models and we show that they both lead to the same results for the exciton transfer dynamics discussed in this paper . in the continuous mode limit , the form factor @xmath29 and the creation and annihilation operators @xmath30 , @xmath31 become functions of the continuous variable @xmath24 , which we denote by @xmath32 and @xmath33 , @xmath34 , respectively . the hamiltonian of the * local reservoirs model * is @xmath35 now we have _ two _ form factors @xmath36 , @xmath37 and _ two _ sets of creation and annihilation operators , @xmath38 and @xmath39 , @xmath40 .
the values @xmath40 label the reservoirs .
all operators of one reservoir commute with all those of the other .
we consider * initial states * of the form @xmath41 for the collective or local models , respectively . here
, @xmath42 is an arbitrary initial density matrix of the dimer and @xmath43 , @xmath44 , @xmath45 are the reservoir equilibrium states at a temperature @xmath46 . the * reduced dimer density matrix * at time @xmath47 is obtained by evolving the whole dimer - reservoir density matrix and then tracing out the reservoirs , @xmath48 here , @xmath49 or @xmath50 , depending on the model considered . the chlorophyll molecules in the dimer have the optical transitional frequencies , with excited energy levels @xmath51 .
the thermal bath usually has much lower excited modes energy , @xmath52 .
our main goal is to describe the dynamics of the reduced dimer density matrix . in theorems [ thm1 ] and [ thm2 ] below
, we give the law of evolution of the populations ( diagonal elements of the density matrix in the energy basis ) and of decoherence ( off - diagonal ) .
these results exhibit a main term in the dynamics and a remainder which is _ controlled for all times @xmath53_. in order to derive the dynamical laws mathematically , we need to impose a regularity condition on the form factors @xmath54 , @xmath55 in , .
it is best expressed as a condition on the * spectral function of the reservoir * , which is defined as @xcite @xmath56 where @xmath57 , @xmath58 , is the fourier transform of the symmetrized correlation function @xmath59 here , @xmath60 is the average of a reservoir observable in the reservoir equilibrium at temperature @xmath61 , @xmath54 is the form factor appearing in , and @xmath62 , @xmath63 the uncoupled hamiltonian of the reservoir(s ) . of course , in the model where we have two reservoirs , there are two corresponding spectral functions @xmath64 , @xmath65 .
we point out that the product @xmath66 appearing in the definition is actually _ independent _ of @xmath67 , expressible purely in terms of the form factor ( c.f . @xcite and also section [ reservoirssect ] ) .
the mathematical regularity condition we impose ( for each reservoir ) is @xmath68 where @xmath69 and where @xmath70 is a bounded function of @xmath71 .
we remark that in principle , a ` minimal condition ' on the low - modes behavior is @xmath72 , but so far our mathematical techniques require the more restrictive values in .
, we believe that the only ` rigid ' requirement is @xmath72 . this same restriction occurs in leggett s work @xcite and is needed in the application of a unitary transformation ( polaron transformation , which is defined only if @xmath72 ) to obtain a hamiltonian in which @xmath73 do not have to be considered to be small parameters , but where @xmath74 is the small perturbation parameter .
an improvement of our current mathematical approach is likely to remove some of the restriction on @xmath75 .
see sections [ regularitysect ] and [ reservoirssect ] for further detail on this point . ]
we denote the population of site ( level ) @xmath76 by @xmath77_{11},\ ] ] where the dimer site basis for @xmath78 ( or , energy basis ) is @xmath79 the initial population is @xmath80 $ ] .
we show in theorem [ thm1 ] that , for arbitrary values of @xmath81 and @xmath82 and sufficiently small values of @xmath74 , and for _ all times @xmath18 _ , @xmath83 the relation is valid for both the local and collective reservoirs models , with different expressions for the relaxation rate @xmath13 .
namely , @xmath13 takes specific values @xmath84 and @xmath85 ( c.f . ) for the collective and local reservoirs models , which we discuss below in section [ ratessection ] .
it satisfies @xmath86 and @xmath87 for @xmath88 .
the remainder term has the bound @xmath89 for some constant @xmath90 .
the final value @xmath91 is the population of the dimer at equilibrium with the reservoir(s ) , @xmath92 where @xmath93 here , @xmath94 are _ renormalizations _ of the dimer donor and acceptor energies ( see section [ renormsect ] ) , which are caused by the interaction with the reservoir(s ) .
their precise expressions are given in and they satisfy @xmath95 .
* discussion * * _ properties of the final population @xmath91 .
_ + if @xmath96 then we say that the reservoirs are _ coupled in a symmetric way _ to the dimer , and we have @xmath97 .
this happens for instance if @xmath98 and , for the local reservoirs case , if additionally @xmath99 .
+ however , the sign of @xmath100 can be positive or negative depending on which reservoir coupling constant is stronger . from and @xmath95
we see readily that if @xmath101 then @xmath102 and if @xmath103 then @xmath104 . for high temperatures we have @xmath105 the small correction to the value @xmath12 is _ negative _ for the symmetrically coupled case ( @xmath106 ) and if level two is coupled to the reservoir(s ) much more strongly than level one .
if the level one is coupled more strongly , then that correction is _ positive_. this shows that if one level is coupled more strongly to the reservoir(s ) than the other , the final population of the more strongly coupled level is increased .
while this effect is small for high temperatures ( see ) , it is large for low temperatures : @xmath107 we conclude that if @xmath108 , then one can entirely populate level one by coupling it very strongly to the reservoir(s ) , or entirely depopulate it by coupling the other level very strongly to the reservoir(s ) . * _ domain of usefulness of expansion .
_ + the expansion is meaningful if @xmath109 if @xmath110 then the right side is just @xmath111 and so the bound is satisfied if in addition @xmath112 , where @xmath90 is the constant in the upper bound of @xmath113 ( see before ) .
hence holds for @xmath114 this is useful if @xmath111 is not very small .
other domains can be found as follows . if @xmath115 then is satisfied for @xmath116 .
similarly , if @xmath117 , then we have @xmath118 , where @xmath119 and @xmath120 are the level 2 populations . now gives @xmath121 , where the remainder is the same as in .
hence , this remainder is small compared to the main term for @xmath122 . at high temperatures , @xmath123 and the remainder is small for times @xmath124 , regardless of the initial condition . * _ a different remainder estimate . _
the dynamical resonance theory of @xcite which we use to prove the expansion is designed to give a remainder that decays as @xmath125 . instead
, one can modify this theory to yield a remainder which is @xmath126 for all times , but may not vanish at @xmath127 .
let @xmath128_{12 } = { \left\langle { \varphi_1 } , { \rho_{{\rm s}}(t)\varphi_2}\right\rangle}\ ] ] be the off - diagonal density matrix of the dimer in the basis .
we show in theorem [ thm2 ] that for arbitrary @xmath129 , for small enough @xmath74 , and _ all times @xmath53 _ , @xmath130_{12 } = { { \rm e}}^{-\gamma t/2}\ , { { \rm e}}^{-{{\rm i}}t(\hat\epsilon+x_{\rm ls})}\ { { \rm e}}^{-\gamma_\infty } \ , [ \rho_{{\rm s}}(0)]_{12 } + o(v ) + r(t),\ ] ] where @xmath13 is the relaxation rate ( the same as in , see for the explicit expression ) and @xmath131 is the lamb shift ( c.f . ) .
relation is valid for both the local and collective reservoirs models ( the @xmath13 having different expressions ) .
the term @xmath126 is independent of time @xmath47 , and @xmath113 satisfies @xmath89 , for some constant @xmath90 .
the constant @xmath132 is given explicitly in .
it describes the large time decoherence of the dimer under the dynamics with @xmath78 .
namely , for @xmath78 the dimer dynamics can be solved exactly ( see section [ subsectb2 ] ) , @xmath133_{12 } = { { \rm e}}^{-{{\rm i}}t\hat\epsilon } \ , { \cal d}(t ) \ , [ \rho_{{\rm s}}(0)]_{12},\qquad ( v=0)\ ] ] where @xmath134 ( which can be expressed in terms of a reservoir correlation function , see section [ decoorigin ] ) satisfies @xmath135 * discussion * * _ two regimes : full and partial phase decoherence ( @xmath78 ) _ + the dynamics for @xmath78 leaves the populations invariant , and the off - diagonal density matrix element of the dimer ( energy basis ) satisfies @xmath136_{12}\big| = \big| [ \rho_{{\rm s}}(0)]_{12}\big| \left\ { \begin{array}{ll } { { \rm e}}^{-\frac1\pi(\lambda_1-\lambda_2)^2q_2(t ) } & \mbox{collective reservoir}\\ { { \rm e}}^{-\frac1\pi\lambda_1 ^ 2q_2^{(1)}(t)}{{\rm e}}^{-\frac1\pi\lambda_2 ^ 2q_2^{(2)}(t ) } & \mbox{local reservoirs } \end{array } \right .
\quad ( v=0),\ ] ] for some functions @xmath137 ( see , ) .
the process described by is called phase decoherence .
we say that _ full _ phase decoherence takes place if @xmath138_{12}\rightarrow 0 $ ] as @xmath139 .
otherwise we call the phase decoherence _ partial_. full phase decoherence happens if and only if @xmath140 and @xmath141 ( collective ) or at least one @xmath142 and @xmath143 ( local ) .
we show in that @xmath144 where @xmath75 is the parameter in ( see also section [ decoorigin ] ) .
+ in the situation of partial phase decoherence ( for @xmath78 ) , i.e. , when @xmath145 , the spectral function of the reservoir , , satisfies @xmath146 as @xmath147 .
this low - mode behavior of @xmath148 is not always satisfied in noisy protein environments . in particular
, it is not true in the presence of the so - called @xmath149 noise @xcite . at the same time , a vanishing spectral density in the limit of low frequencies can be realized in many lhcs , see @xcite ( as well references therein ) .
+ our technical condition , used to derive the evolution of the dimer mathematically rigorously , places us in the regime of partial phase decoherence .
but as mentioned above , we expect that the mathematical analysis can be pushed to cover the range @xmath72 .
it is then reasonable to include a discussion of the relaxation rates in the situation of full phase decoherence ( @xmath150 ) as well , in what follows . * _ domain of usefulness of expansion . _ + analogously to the discussion of , one sees that the remainder in is small if @xmath151_{12 } < \!\ ! < t<\!\ !
< 1/\gamma \propto v^{-2}.\ ] ] we may view @xmath152_{12}$ ] in as a _ shifted initial condition _ , whose absolute value ( but not phase ) is that of the asymptotic dynamics with @xmath78 . under the dynamics with @xmath78
, the absolute value of the off - diagonal matrix element reaches its final value @xmath153_{12}\big|$ ] at the reservoir correlation time independent of @xmath74 ( c.f .
, , lemma [ lemma1 ] ) . for @xmath74 small , the factor @xmath154 in has not yet started to decay at that point in time .
* _ possible improvement of the resonance expansion for small times . _
+ we expect that an expansion holds with @xmath155 replaced by @xmath76 and a remainder @xmath156 for all times .
however , the existing dynamical resonance theory must be modified to reach that result . in its present form ,
it is not accurate to describe decoherence for small times .
indeed , it follows from that @xmath157_{12}\cdot o(1-{{\rm e}}^{-\gamma_\infty})+o(v)$ ] , which may not be small .
we explain how the factor @xmath155 appears naturally within the resonance description of the dynamics , and why it is not present ( rather , @xmath158 ) in the dynamics of the populations .
see section [ explansect ] , equation . * _ relation between decoherence and relaxation rates for arbitrary interaction strength . _
+ it is well known from the bloch - redfield weak coupling theory that if @xmath15 is small and @xmath159 , then the relaxation rate @xmath160 ( decay of diagonal ) and the decoherence rate @xmath161 ( decay of off - diagonal ) satisfy the relation @xmath162 our results and show that the relation is _ valid for arbitrary @xmath15_. the relaxation rates of the collective and local reservoirs models are given by @xmath163 \exp\left [ -\frac{(\lambda_1-\lambda_2)^2}{\pi}q_2(t)\right ] { { \rm d}}t,\\ \gamma_{{\rm loc}}&= v^2 \lim_{r\rightarrow 0_+}\int_0^\infty\!\!{{\rm e}}^{-r t } \cos(\hat\epsilon t)\ , \cos\left [ \frac{\lambda_1 ^ 2}{\pi}q_1^{(1)}(t ) + \frac{\lambda_2 ^ 2}{\pi}q_1^{(2)}(t)\right ] \exp \left[-\frac{\lambda_1 ^ 2}{\pi}q_2^{(1)}(t ) - \frac{\lambda_2 ^ 2}{\pi}q_2^{(2)}(t)\right ] { { \rm d}}t , \end{split } \end{aligned}\ ] ] where @xmath100 is defined in and where ( collective reservoir ) @xmath164 and ( local reservoirs ) @xmath165 the limit @xmath166 appears in because the analysis involves green s functions close to the real axis .
the correctness of the formula depends crucially on the presence of this limit when the function to be integrated does not decay to zero as @xmath125 .
this happens in the case of partial phase decoherence , where @xmath167 for large times . in the contrary case , when the integrand in for @xmath168 is integrable ,
we can leave out the limit in the expression for @xmath84 and @xmath85 ( just set @xmath168 ) .
both situations are reasonable , but they need to be discussed separately .
note that for @xmath169 , we obtain from that @xmath170 ( for @xmath171 ) .
_ remark . _
the expansion is valid for small values of @xmath74 . in particular ( see ) , @xmath172 , where @xmath173 is @xmath174 or @xmath175 ( which is independent of @xmath74 ) .
one can expand @xmath13 , , for small @xmath15 .
to lowest order ( @xmath16 ) , one then recovers the expression predicted by the bloch - redfield theory ( fermi golden rule ) in the weak coupling limit ( small @xmath15 ) .
this has been shown in @xcite , section 5 .
we now give more manageable expressions of @xmath176 than , in both regimes of full and of partial phase decoherence .
these easier ( approximate ) expressions are given in , and , and are derived from in section [ derivationsubsect ] .
let us consider spectral functions of the reservoir of the form with @xmath72 and with an exponential high - mode cutoff @xmath177 , @xmath178 where @xmath179 . in the local reservoirs model , each reservoir
s spectral function @xmath180 is of the form with possibly different @xmath75 and @xmath181 .
set @xmath182 * dimensionalities . * we adopt units in which @xmath183 .
the dimensions are as follows .
dimensionless quantities : @xmath184 , @xmath185 , @xmath186 , @xmath187 dimension energy : @xmath188 , @xmath189 , @xmath181 , @xmath20 , @xmath190 dimension 1/energy : @xmath191 , @xmath192 , @xmath193 ( c.f . )
since the spectral function is dimensionless , the coefficient @xmath194 has the same dimension as @xmath195 , which is @xmath196 and depends on @xmath75 .
@xmath184 is quadratic in the form factor(s ) @xmath54 ( @xmath197 ) ( c.f . , ) and the form factor(s ) are defined only up to multiplication with the coupling constants @xmath19 , see and .
therefore , only the combination @xmath198 ever appears .
the expressions for the relaxation rates do not depend on @xmath194 .
* relaxation rate in the regime of partial phase decoherence , @xmath145 .
* consider first the * collective reservoir * model and the regime @xmath199 the first inequality in ( [ regime1 ] ) , @xmath200 , is called the high temperature regime @xcite and usually covers room temperatures .
the dimer relaxation rate has the ( approximate ) expression @xmath201 where @xmath202 .
for instance , @xmath203 . similarly , for the * local reservoirs * model , the spectral functions of the reservoirs are given by with ( possibly different ) @xmath204 and @xmath205 .
for ease of notation , we take @xmath206 and @xmath207 ( the following bounds can be easily derived also when this approximate symmetry does not hold ) . in the high - temperature regime
@xmath208 the relaxation rate is given by @xmath209 where @xmath210 . * dimer relaxation in the regime of full phase decoherence , @xmath211 .
* consider first the * collective reservoir * model in the parameter region @xmath212 which also corresponds to the high - temperature regime . the dimer relaxation rate has the ( approximate ) expression @xmath213 + \exp\big[-\frac{(\epsilon+\epsilon_{{{\rm col}},2})^2}{2t(\epsilon_{{{\rm col}},1}+\epsilon_{{{\rm col}},2 } ) } \big ] \big\}\ ] ] where two _ reconstruction energies _ are introduced , @xmath214 we call the * generalized marcus formula * ( see the discussion after ) .
as @xmath215 , the second condition in is equivalent to @xmath216 for symmetric coupling , @xmath217 , we have @xmath218 ( for @xmath40 ) and takes the form @xmath219 which coincides with the heuristically derived rate given in @xcite .
typically , the second exponential is much smaller than the first one ( @xmath220 ) and is therefore neglected . the resulting formula
, @xmath221 is the famous * marcus formula * of electron transfer @xcite . hence the name generalized marcus formula for , which extends the original one to the situation where the donor and acceptor may be coupled in an asymmetric way to a common reservoir .
the energies @xmath222 , , play the role of generalized reorganization ( reconstruction ) energies .
similarly , for the * local reservoirs * model , in the high - temperature regime @xmath223 the dimer relaxation rate has the expression , @xmath224 + \exp\left[-\frac{(\epsilon+\epsilon_{{{\rm loc}},2})^2}{2t(\epsilon_{{{\rm loc}},1}+\epsilon_{{{\rm loc}},2})}\right]\right\},\ ] ] where the generalized reorganization ( reconstruction ) energies are @xmath225 * discussion * * for @xmath226 , @xmath84 reduces to the form derived in @xcite .
if in addition @xmath227 , then @xmath228 .
the rate @xmath85 is a symmetric function of the coupling constants @xmath81 and @xmath82 ( each reservoir acts independently in the same way ) , but @xmath84 is not .
indeed , @xmath85 only depends on the absolute values of the @xmath229 , while @xmath84 depends also on the sign of the product @xmath230 . if @xmath231 or @xmath232 , then @xmath228 . * the _ reconstruction energies _ @xmath222 of the collective reservoir model can be positive or negative , depending on the relative values and signs of the coupling constants .
however , in the local reservoirs model , @xmath233 always .
* _ upper bound on @xmath84 in the high - temperature regime .
_ according to and we have @xmath234 _ derivation of . _
the process is governed by two parameters : the ( finite ) asymptotic value @xmath235 and the characteristic time @xmath236 at which @xmath237 approaches the value @xmath238 . ] here , we have used @xmath239 ( other values @xmath145 give the same conclusion ) and that , due to the high frequency cutoff @xmath181 in @xmath148 ( see ) , the integration is essentially over @xmath240 , and therefore @xmath241 provided @xmath242 . ] to analyze @xmath84 , , we split the integration in two domains . for @xmath243 , we replace @xmath244 . from and upon making a change of variables @xmath245 , one immediately finds @xmath246 , where @xmath90 is a dimensionless constant ( depending on @xmath75 ) . ]
@xmath247 it follows that @xmath248\approx 1 $ ] if @xmath249 .
the last inequality is satisfied for ( see ) @xmath250 thus , under the conditions and , the contribution to @xmath84 from the integration over @xmath251 is @xmath252 where we use in the last step @xmath253 , or @xmath254 now we estimate the contribution to for small times @xmath255 . in this integral , we can set @xmath168 to begin with and we expand the integrand in as ( @xmath47 small ) @xmath256 \
\left[1 -\frac{(\lambda_1-\lambda_2)^2}{\pi}t\nu t^2\right]}\nonumber\\ & & \approx 1-t^2\big(\hat\epsilon^2/2 + ( \lambda_1-\lambda_2)^4\nu^2/\pi^2 + ( \lambda_1-\lambda_2)^2 t\nu/\pi\big ) .
\label{est''}\end{aligned}\ ] ] the integral @xmath257 of is then @xmath258 which we approximate by @xmath259 under the condition that @xmath260 , i.e. , that @xmath261 .
the latter inequality is implied by .
combining this with and gives @xmath262 further , again for @xmath263 , we can replace @xmath264 in by @xmath265 .
then @xmath266 and so follows from .
equation is derived in the same way .
_ derivation of .
_ for @xmath150 we have @xmath267 .
the region of integration in is essentially @xmath268 , due to the high energy mode cutoff .
therefore , for @xmath263 , we replace @xmath269 in , so that @xmath270 where the last approximation is valid for times @xmath47 satisfying @xmath271 ( in order that @xmath272 ) . the approximation @xmath273 has also been used in @xcite .
as @xmath274 is increasing , the exponential factor in for @xmath275 is estimated as @xmath276 provided @xmath277 the region @xmath275 can thus be neglected in the integral and and we can also approximate @xmath278 in by its behaviour for small times ( see also @xcite ) , @xmath279 consequently , @xmath280 \ { { \rm e}}^{-\frac1\pi ( \lambda_1-\lambda_2)^2 t\nu t^2}\ { { \rm d}}t.\ ] ] using @xmath281 and @xmath282 , we evaluate explicitly to obtain .
relation is obtained in the same way .
for the collective reservoir model and the local reservoirs model , repectively , set @xmath283 where the @xmath185 , @xmath186 are given in . consider @xmath284 , and @xmath285 , with @xmath78 and @xmath129 arbitrary .
we show in appendix [ subsectb1 ] that the equilibrium state of the interacting dimer - reservoir system at temperature @xmath46 for the model with the collective reservoir is @xmath286 where @xmath287 is the unitary weyl operator . by tracing out the reservoir in
we find @xmath288 in this sense , the system energies @xmath289 are _ renormalized _ by @xmath290 , due to the interaction with the reservoir .
note though , that _ the system is entangled with the reservoir in the equilibrium state _ @xmath291 .
the situation for the model with the two local reservoirs is analogous .
the interacting dimer - reservoirs equilibrium state ( @xmath78 ) is given by @xmath292 by tracing out the reservoir in we find again a dimer state of the form , with @xmath293 given by .
we point out that if @xmath96 ( e.g. if @xmath294 and @xmath295 ) , then @xmath296 differs from @xmath297 ( with @xmath78 ) just by a constant and the reduced dimer equilibrium state is the same for the coupled and the uncoupled system . in section [ sectdynres ] ,
we prove the following result .
[ thm1 ] let @xmath81 , @xmath82 be arbitrary .
there is a @xmath298 such that if @xmath299 , then @xmath300 where @xmath301 is the level 1 population of the dimer density matrix at equilibrium coupled with the reservoir(s ) .
here @xmath293 are the renormalization energies given in , and the relaxation rate @xmath13 is given by , for the model with the collective reservoir and that with the local ones , respectively .
the remainder term @xmath126 is independent of time @xmath47 and @xmath113 satisfies @xmath302 for some constant @xmath303 .
_ remark .
_ the dynamical resonance theory of @xcite which we use to prove the expansion is set up so as to give a remainder that decays as @xmath125 .
instead , one could modify this theory to yield a remainder which is @xmath126 for all times , but may not vanish at @xmath127 .
we do not explore this modification here .
we now discuss the decoherence properties of the dimer .
the models with @xmath78 can be solved exactly , see section [ subsectb2 ] .
namely , @xmath304_{12 } = { { \rm e}}^{-{{\rm i}}t\hat\epsilon } \ , { \cal d}(t ) \ , [ \rho_{{\rm s}}(0)]_{12},\qquad ( v=0)\ ] ] where @xmath100 is given in and @xmath305 } { { \rm e}}^{-\frac1\pi [ \lambda^2_1 q_2^{(1)}(t ) + \lambda^2_2 q_2^{(2)}(t ) ] } \mbox{\quad ( local ) } \end{array } \right.\ ] ] for the collective and the local resevoirs models , respectively , with @xmath306 and @xmath307 given in and .
the following result is proven in section [ subsectb2 ] , see .
[ prop4 ] we have @xmath308 where @xmath309 for the collective and the local reservoirs models , respectively .
_ remark .
_ for spectral functions @xmath184 of the form we have @xmath310 provided @xmath141 ( collective ) and @xmath311 ( local ) , the divergence of the integrals in for @xmath312 stemming from a non - integrable singularity at low modes ( @xmath313 ) .
we now introduce the _ lamb shift _
, @xmath314 \exp\big[-\frac{(\lambda_1-\lambda_2)^2}{\pi}q_2(t)\big]\nonumber { { \rm d}}t \quad \mbox{(collective)}\\ \\ \int_0^\infty { { \rm e}}^{-r t } \sin(\hat\epsilon t ) \cos\big[\frac{\lambda^2_1}{\pi } q_1^{(1)}(t ) + \frac{\lambda^2_2}{\pi } q_1^{(2)}(t)\big ] \exp \big [ -\frac{\lambda_1 ^ 2}{\pi}q_2^{(1)}(t ) -\frac{\lambda_2 ^ 2}{\pi}q_2^{(2)}(t ) \big ] { { \rm d}}t\quad \mbox{(local)}\nonumber \end{array } \right.\end{aligned}\ ] ] in section [ sectdynres ] , we prove the following result . [ thm2 ] let @xmath81 , @xmath82 be arbitrary .
there is a @xmath298 such that if @xmath299 , then @xmath315_{12 } = { { \rm e}}^{-\gamma_\infty } \ , { { \rm e}}^{-\gamma t/2}\ , { { \rm e}}^{-{{\rm i}}t(\hat\epsilon+x_{\rm ls})}\ [ \rho_{{\rm s}}(0)]_{12 } + o(v ) + r(t),\ ] ] where @xmath13 is given in and @xmath316 is the lamb shift .
the term @xmath126 is independent of time @xmath47 and @xmath113 satisfies @xmath317 .
the spectral functions of the collective and local environments are taken as in , with @xmath318 in the case of local reservoirs ( the results obtained below can be generalized in a straightforward way for the case @xmath319 ) . in the numerical simulations
, we have chosen a weakly coupled dimer based on two chlorophylls , @xmath320 , in their excited states .
the following parameters of the dimer were chosen , @xmath321 ( so @xmath322 ) , and the room temperature : @xmath323 ( @xmath324 ) .
it is convenient to define new dimensionless quantities @xmath325 using this notation , the relaxation rates become @xmath326 \cos\left[\frac{\varepsilon^c_1+\varepsilon^c_2}{2}\mathcal q_1(\tau)\right ] \exp\left [ -\frac{\varepsilon^c_1+\varepsilon^c_2}{2}\mathcal q_2(\tau)\right ] { { \rm d}}\tau,\\ \gamma_{{\rm loc}}&= \beta v^2 \int_0^\infty \cos\big[\big(\varepsilon-\frac{\varepsilon^l_1-\varepsilon^l_2}{2}\big ) \tau\big ] \ , \cos\left [ \frac{\varepsilon^l_1}{2}{\mathcal q}_1^{(1)}(\tau ) + \frac{\varepsilon^l_2}{2}{\mathcal q}_1^{(2)}(\tau)\right ] \nonumber \\ & \times \exp \left[-\frac{\varepsilon^l_1}{2}{\mathcal q}_2^{(1)}(\tau ) -\frac{\varepsilon^l_2}{2}{\mathcal q}_2^{(2)}(\tau)\right ] { { \rm d}}\tau , \label{b12b}\end{aligned}\ ] ] where we understand that a limit @xmath166 has to be performed ( as in ) , and where we have set @xmath327 and @xmath328 ( @xmath329 ) .
we have , @xmath330 where @xmath331 performing the integrations in - , we obtain , @xmath332 here , @xmath333 and @xmath334 denotes the hurwitz @xmath335-function @xcite
. one can see that for @xmath145 the function @xmath336 is bounded from above by @xmath337 using the asymptotic properties of the @xmath335-function , we obtain @xmath338 where @xmath339 is the riemann zeta function , @xmath340 .
one can estimate @xmath341 as , @xmath342 similar considerations yield the asymptotic behavior for the function @xmath343 , @xmath344 in fig .
[ r2 ] , the functions , @xmath345 and @xmath346 , are presented for various values of parameters , @xmath75 and @xmath347 . as one can see , in the high - temperature regime , @xmath348 , the conditions and are satisfied for @xmath349 , while for @xmath145 a saturation of @xmath336 at finite values ( @xmath350 ) takes place .
the asymptotic behavior of the function @xmath336 significantly influences the quantum coherence in the system .
( a ) ( b ) ( c ) ( d ) ( a ) ( b ) ( c ) ( d ) ( a ) ( b ) ( c ) ( d ) the relaxation rates in the generalized marcus limit are given in and .
in the new variables , they are @xmath351 + \exp\left[-\frac{(\varepsilon+\varepsilon^{c}_{2})^2}{2(\varepsilon^{c}_{1}+\varepsilon^{c}_{2})}\right]\right\ } , \\ \gamma_{{\rm loc } } & = \beta\left(\frac{v}{2}\right)^2\sqrt{\frac{2\pi}{\varepsilon^{l}_{1}+\varepsilon^{l}_{2}}}\left\ { \exp\left[-\frac{(\varepsilon-\varepsilon^{l}_{1})^2}{2(\varepsilon^{l}_{1}+\varepsilon^{l}_{2})}\right ] + \exp\left[-\frac{(\varepsilon+\varepsilon^{l}_{2})^2}{2(\varepsilon^{l}_{1}+\varepsilon^{l}_{2})}\right]\right\}. \label{mrl1}\end{aligned}\ ] ] in the case @xmath352 , or @xmath353 , they take the standard marcus form ( see also @xcite ) , @xmath354 + \exp\left[-\frac{(\varepsilon+\varepsilon_{c , l})^2}{4\varepsilon_{c , l}}\right]\right\},\end{aligned}\ ] ] for the collective and local reservoirs models , respectively . in figs.[fig1 ] and [ fig3 ] , we compare the et rates obtained by numerical integration of ( [ b12a ] ) and ( [ b12b ] ) , with the generalized marcus formulas given by - , corresponding to the high - temperature limit @xmath355 .
all two - dimensional surfaces shown in figs.[fig1 ] and [ fig3 ] correspond to the exact eqs .
( [ b12a ] ) and ( [ b12b ] ) .
[ fig1](a ) shows the high - temperature regime , @xmath356 . in this case
, the rate @xmath357 , obtained from the generalized marcus formula , yields a surface that practically coincides with the surface defined @xmath357 obtained from the exact eq .
( [ b12a ] ) , for all @xmath358 .
the large peak in fig .
[ fig1](a ) lies inside the parameter region for the guaranteed domain of applicability of the theory , namely , @xmath359 , see .
the values of @xmath84 for small values of @xmath360 lie outside the domain for which we can guarantee that the generalized marcus formula is approximating correctly the rate ( c.f . ) .
so , in the numerical simulations of fig .
[ fig1](a ) we take @xmath361 . the red curves in fig .
[ fig1 ] correspond to the standard marcus et rate , given by eq .
( [ xs ] ) .
as expected , the red curve in fig .
[ fig1](a ) lies on the surface @xmath362 .
the et rates shown by the red curve in fig .
[ fig1](a ) ( marcus high - temperature regime , @xmath356 ) are close to the experimental values reported in @xcite ( @xmath363 ) .
still , the et rates given by the surface but away from the red curve , corresponding to the generalized marcus formula , can significantly exceed the values on the red curve . in fig . [ fig1](b ) and fig .
[ fig1](c ) we take @xmath364 , which does not correspond to the high - temperature regime any longer .
our numerical simulations demonstrate that the generalized marcus formula for @xmath84 , given by eq . , does not work in this case . in particular , the red curves in fig .
[ fig1](b ) and fig .
[ fig1](c ) partly lie outside the two - dimensional surfaces corresponding to the exact formulas .
the difference between the exact results for @xmath84 and the generalized marcus formula is shown in fig .
[ fig1](d ) , for the dependence @xmath365 , where @xmath366 .
as one can see , only the blue curve , corresponding to the exact formula in the high - temperature regime ( @xmath356 ) , practically coincides with the marcus formula . in fig .
[ fig3 ] , similar results ( to those presented in fig . [ fig1 ] ) are shown for @xmath367 .
the two - dimensional surfaces correspond in this case to the exact formulass of eq .
( [ b12b ] ) .
the high - temperature regime is shown in fig .
[ fig3](a ) ( @xmath368 ) . in this case , the generalized marcus formula , given by eq . , practically describes the exact result in the whole region of the reconstruction energies , ( @xmath369 ) .
in particular , the red curve in fig .
[ fig3](a ) ( corresponding to eq .
( [ xs ] ) ) lies on the two - dimensional surface , @xmath367 .
similar to the case of the collective environment , the et rates , @xmath85 , of the generalized marcus formula , can exceed the et rates corresponding to the standard marcus formula given by eq .
( [ xs ] ) .
the results of our numerical simulations of the exact eq .
( [ b12b ] ) ( shown in figs .
[ fig3](b , c , d ) ) demonstrate that outside the high - temperature regime ( @xmath348 ) , the generalized marcus formula , given by eq . , can not be used .
( however , we do have the correct expressions for @xmath84 , @xmath85 , even in this regime , c.f . - . ) for @xmath78 there is no relaxation , only decoherence in the energy basis occurs @xcite .
the saturated value of the non - diagonal reduced density matrix elements is characterized by the factor @xmath155 given in . in the variables we have @xmath370 as one
can see from eq .
( [ gammatau ] ) , the asymptotic behavior of the function @xmath371 significantly effects the quantum coherence in the system .
we plot the function @xmath372 in fig .
[ fig4a ] ( see also fig . [ r2 ] ) .
one can see that in the high - temperature regime , @xmath355 , @xmath341 increases very fast when @xmath373 increases .
hence one expects strong decoherence in this case .
( a ) ( c ) in the regime of partial phase decoherence , @xmath145 , we can use the asymptotic formulas ( [ q0 ] ) to obtain @xmath374 here , @xmath375 for the collective environment model and @xmath376 for the local environments model .
[ fig4 ] shows the decoherence function @xmath377 for both collective and local environments , and for @xmath145 .
quantum coherence in this case ( @xmath145 ) survives for large times ( for all times if @xmath78 ) .
the level of quantum coherence strongly depends on the value of @xmath347 the for collective environment and on @xmath378 and @xmath379 for the local environments .
namely , quantum coherence is strong for @xmath347 , @xmath380 not too small . indeed , for @xmath356 ( blue curve in fig .
[ fig4]b ) , coherence is very small at large times .
the description of a small system ( dimer ) coupled to infinitely extended bose reservoirs , or continuous modes oscillator reservoirs , is phrased mathematically in terms of the language of @xmath381-dynamical systems .
we note that we start off with a reservoir having continuous modes . often in the physics literature
, one first considers a reservoir of finitely many modes ( quantized for instance because confined to a finite volume ) .
then one performs calculations , reaches expressions for , say , the relaxation rates , and finally takes the limit of continuous modes in the end .
the arguments leading to expressions of relaxation rates in this way ( the procedure being called the time - dependent perturbation theory " ) , are often based on taking large - time limits ( @xmath125 ) .
it is not clear if these arguments are really valid since in general , performing first the large time limit and then the continuous modes limit ( or , infinite volume limit ) is not the same as performing the limits in the opposite order ( as taking first @xmath125 at discrete modes or finite volume depends on ` boundary effects ' ) .
thus , for a rigorous analysis , we first perform the infinite volume limit and get results which hold _ for all times @xmath382_. this approach has been fruitful in many situations recently @xcite .
the total system is a @xmath381-dynamical system @xmath383 , where @xmath384 is a von neumann algebra of observables acting on a hilbert space @xmath385 and where @xmath386 is a group of @xmath387automorphisms of @xmath388 .
the `` positive temperature hilbert space '' is given by @xmath389 depending on whether we have one or two reservoirs .
here , @xmath390 is the fock space @xmath391 it differs from the ` usual zero - temperature ' fock space @xmath392 in that the single - particle space at positive temperature is the ` glued ' space @xmath393 @xcite ( @xmath394 is the uniform measure on @xmath395 ) .
@xmath390 carries a representation of the ccr ( canonical commutation relation ) algebra .
the represented weyl operators are given by @xmath396 , where @xmath397 .
here , @xmath398 and @xmath399 denote creation and annihilation operators on @xmath390 , smoothed out with the function @xmath400 belonging to @xmath393 .
it is easy to see that the ccr are satisfied , namely , @xmath401 the vacuum vector @xmath402 represents the infinite - volume equilibrium state of the free bose field , determined by the formula @xmath403 the ccr algebra is represented on as @xmath404 , for @xmath405 such that @xmath406 .
this representation was first derived by araki and woods @xcite .
we denote the von neumann algebra of the represented weyl operators by @xmath407 . the doubled spin hilbert space in allows to represent any ( pure or mixed ) state of the two - level system by a vector , again by the gns construction .
this construction is also known as the _ liouville description _ @xcite and goes as follows .
let @xmath408 be a density matrix on @xmath409 .
when diagonalized it takes the form @xmath410 , to which we associate the vector @xmath411 ( complex conjugation in any fixed basis we will choose the eigenbasis of @xmath297 , ) .
then @xmath412 for all @xmath413 and where @xmath414 is the identity in @xmath409 .
this is the gns representation of the state given by @xmath408 @xcite .
the von neumann algebra of observables of the total system is @xmath415 the modular conjugation @xmath148 is the antilinear operator defined by @xmath416 where @xmath417 is the matrix obtained from @xmath418 by taking entrywise complex conjugation ( matrices are represented in the eigenbasis of @xmath297 ) .
note that by , we have @xmath419 . by the tomita - takesaki theorem @xcite , conjugation by @xmath148 maps the von neumann algebra of observables into its commutant .
that is , @xmath74 and @xmath420 commute for any @xmath421 .
the dynamics of the spin - boson system is given by @xmath422 it is generated by the self - adjoint liouville operator acting on @xmath385 , associated to the hamiltonian . for the single reservoir model , @xmath423 where @xmath424 where @xmath425 is the second quantization of the operator of multiplication by the radial variable @xmath426 , @xmath427 is the represented field operator and @xmath428 similarly , the liouvillian associated to the hamiltonian is @xmath429 with @xmath430 given as in , the @xmath431 as in ( on their individual reservoir spaces ) and @xmath432 it is convenient to transform the liouvillians unitarily by a polaron transformation " @xcite .
[ propunitrans ] * define the unitaries @xmath433}$ ] and @xmath434 , where @xmath435 then @xmath436 where @xmath437 with renormalization energies @xmath438 .
the transformed interaction is @xmath439,\ ] ] where @xmath440
* define the unitaries @xmath441}$ ] and @xmath434 , where @xmath442 then @xmath443 where @xmath444 is given as in with renormalization energies @xmath445 .
the transformed interaction is @xmath446 .
\nonumber\end{aligned}\ ] ] _ remarks . _
* for a function @xmath447 , @xmath28 , we have @xmath448 .
if @xmath169 ( and @xmath99 in the case of local reservoirs ) , then @xmath449 .
sometimes the following expressions for @xmath426 are useful . in the case of the single collective reservoir , @xmath450 . in the case of the two local reservoirs , @xmath451 .
the proof of proposition [ propunitrans ] is simply a calculation , based on the relations @xmath452 and @xmath453 let @xmath454 be the initial state ( represented as a vector in the gns hilbert space ) .
the expectation of a ( represented ) observable @xmath418 ( of the dimer and/or the reservoir(s ) ) at time @xmath47 is @xmath455 since @xmath456 is separating ( see @xcite ) , there exists an operator @xmath457 from the commutant algebra ( commuting with all observables ) , s.t . @xmath458
using relation together with the fact that @xmath457 commutes with @xmath459 and the invariance @xmath460 gives @xmath461 we now use the unitary transformation @xmath462 described in proposition [ propunitrans ] , @xmath463 where @xmath464 is @xmath465 or @xmath466 ( see , ) , depending on the model considered .
has the following resonance expansion , which follows from a general resonance expansion of propagators proven in @xcite .
set , for ease of notation , @xmath467 then there is a @xmath298 s.t . for all @xmath299 , and all @xmath468 , @xmath469 \psi_2}\right\rangle } + r(t ) .
\nonumber\end{aligned}\ ] ] here , @xmath470 is the coupled dimer - reservoir(s ) equilibrium state .
the `` decay rate '' is given by @xmath471 depending on whether we consider the model with the collective reservoir or the local ones , and where @xmath84 , @xmath85 are given by .
the quantity @xmath131 is the `` lamb shift '' .
moreover , the resonance projections are @xmath472 where @xmath473 ( see also ) and @xmath474 ( projection onto the vacua of the reservoirs ) , depending on whether we consider the collective or local reservoirs models .
the decay rates and lamb shifts are ( to lowest order in @xmath74 ) the real and imaginary parts of _ level shift operators _ which describe the perturbative movement of eigenvalues of @xmath464 for small @xmath74 .
we calculate them explicitly in appendix [ lsosect ] .
the remainder @xmath113 in satisfies the following bounds : there are constants @xmath303 and @xmath475 ( independent of @xmath47 and @xmath74 ) , such that for all @xmath468 , latexmath:[\[\label{a30 } _ remark .
_ to prove the expansion one needs some regularity of the form factors @xmath54 , @xmath55 appearing in and .
we explain them in section [ regularitysect ] .
define the operators @xmath477 , @xmath478 , belonging to the algebra of observables of the joint dimer - reservoir(s ) system , as follows .
* for the model with the single collective reservoir , @xmath479 ( recall the definition of @xmath480 . )
* for the model with the two local reservoirs , @xmath481 ( recall the definition of @xmath480 . )
the main result of this section is the following representation for the dynamics .
[ effdynprop ] let @xmath482 be the initial dimer - reservoir(s ) state and let @xmath418 be a dimer - reservoir(s ) observable .
then @xmath483 where @xmath470 is the dimer - reservoir(s ) equilibrium state , @xmath113 satisfies and @xmath484 where @xmath485 or @xmath486 given in and @xmath316 is the lamb shift .
_ proof of theorem [ effdynprop ] . _
we use the expansion .
the following result follows from an easy calculation combining - with the definition of the unitary @xmath462 given in proposition [ propunitrans ] .
[ lem2 ] the projections @xmath487 , @xmath478 , given in - have the form @xmath488 where the vectors @xmath489 are as follows .
* for the model with the collective reservoir , we have @xmath490 * for the model with the two local reservoirs , we have @xmath491 lemma [ lem2 ] shows that @xmath492 by araki s perturbation theory of kms states ( @xcite , theorem 5.4.4 ) , the equilibrium states with @xmath78 and @xmath493 satisfy @xmath494 thus , @xmath495 the first factor on the right side of is linked to the initial condition @xmath454 , the second one to the equilibrium @xmath496 .
[ lem5 ] the operators @xmath477 defined in , satisfy @xmath497 _ proof of proposition [ lem5 ] .
_ the equilibrium states for the collective and local models are @xmath498 the relations , and are then easily seen to hold .
since the operators @xmath477 commute with @xmath457 , we have @xmath499 moreover , @xmath500 combining and with and , the expansion now follows from .
this completes the proof of theorem [ effdynprop ] . *
dynamics of populations and proof of theorem [ thm1 ] . * choosing @xmath501 in yields the population of the level @xmath76 , i.e. , @xmath502_{11}.\ ] ] for this choice of @xmath418 and for both models , we have @xmath503 and @xmath504_{11 } -{{\rm e}}^{-\beta\hat\epsilon/2}[\rho_{{\rm s}}(0)]_{22 } .\ ] ] furthermore , @xmath505_{21 } \left\ { \begin{array}{l } { \left\langle{w^*(f_1)w(f_2)}\right\rangle}_{{{\rm r}}}\mbox{\quad ( collective)}\\ { \left\langle{w^*(f_1)}\right\rangle}_{{{\rm r}}_1 } { \left\langle{w(f_2)}\right\rangle}_{{{\rm r}}_2 } \mbox{\quad ( local ) } , \end{array } \right.\ ] ] and @xmath506 combining with - yields .
* dynamics of decoherence and proof of theorem [ thm2 ] . * choosing @xmath507 in yields the off - diagonal dimer density matrix element , @xmath508_{12}.\ ] ] for this choice of @xmath418 and for both models , we have @xmath509 and @xmath510 furthermore , @xmath511_{12 } \left\ { \begin{array}{l } { \left\langle{w^*(f_2)w(f_1)}\right\rangle}_{{{\rm r } } } \mbox{\quad ( collective)}\\ { \left\langle{w(f_1)}\right\rangle}_{{{\rm r}}_1 } { \left\langle{w^*(f_2)}\right\rangle}_{{{\rm r}}_2 } \mbox{\quad ( local)}. \end{array } \right.\ ] ] taking into account - , the expression for @xmath512 , and expansion , we obtain .
we give a theoretical analysis of a ( chlorophyll - based ) dimer interacting with collective ( spatially correlated ) and local ( spatially uncorrelated ) protein - solvent environments , as they appear in photosynthetic bio - complexes .
we formulate the problem in the language of a spin - boson system , in which two excited electron energy levels of two light - sensitive molecules ( such as chlorophylls or carotenoids ) are described by an effective spin @xmath12 . both spin levels ( excited electron states of donor and acceptor ) interact with a thermal environment , modeled by bosonic degrees of freedom ( quantum linear oscillators ) . in the case of a correlated thermal environment , a single set of bosonic operators , with the characteristic frequencies of the environment , is introduced . in the case of an uncorrelated thermal environment ,
two sets of bosonic operators are introduced , one for the donor and one for the acceptor . in either situation ,
we introduce two independent constants of interaction between the donor and the acceptor and the environment(s ) .
we develop a mathematically rigorous perturbation theory in which the direct matrix element of the donor - acceptor interaction is a small parameter .
this perturbation theory is different from the standard bloch - redfield theory where the interaction constant between the dimer and the environment(s ) is used as a small parameter .
our approach allows us to consider , in a controlled way , the case of strong interaction constants and ambient temperatures .
this is important for applications to real bio - systems .
we derive the explicit expressions for the thermal electron transfer rates , and demonstrate the differences with the standard marcus expression for the electron transfer rates in the so - called high - temperature regime .
we analyze the dynamics of decoherence depending on the parameters of the system .
in particular , we demonstrate how long - time coherences naturally occurs in the dimer .
the results of the paper are important for a better understanding of the complicated quantum dynamics a chlorophyll - based dimer undergoes in photosynthetic complexes , for a wide region of parameters and under controlled approximations .
experiments could be used to verify our theoretical predictions .
this work was carried out under the auspices of the national nuclear security administration of the u.s .
department of energy at los alamos national laboratory under contract no .
de - ac52 - 06na25396 .
h.s . have been supported by nserc through a discovery grant and a discovery accelerator supplement .
m.m . is grateful for the hospitality and financial support of lanl , where part of this work was carried out .
acknowledges support from the conacyt , grant no .
15349 and partial support during his visit from the biology division , b-11 , at lanl .
g.p.b , s.g . , and r.t.s .
acknowledge support from the ldrd program at lanl .
let @xmath513 and @xmath514 be the orthogonal projections onto the eigenspaces of @xmath515 associated to the eigenvalues zero and @xmath516 , where @xmath517 ( see also proposition [ propunitrans ] . ) explicitly , @xmath518 and @xmath519 , @xmath520 , where we recall that @xmath521 is given in . to each unperturbed eigenvalue of @xmath522
, we associate a level shift operator .
the one associated to the eigenvalue zero is ( the @xmath11 matrix ) @xmath523 where @xmath524 is the interaction operator , @xmath525 or @xmath526 , see , .
the level shift operators associated to the eigenvalues @xmath527 are both one - dimensional , @xmath528 [ lso0prop ] * ( 1 ) * consider the model with the collective environment and set @xmath529 where @xmath278 , @xmath237 are given in . in the basis @xmath530
, we have @xmath531 the eigenvalues of @xmath532 are @xmath533 and @xmath534 , where @xmath84 is given in .
moreover , the one - dimensional level shift operators @xmath535 have the expressions @xmath536 * ( 2 ) * consider the model with the two local reservoirs .
the level shift operator @xmath537 has the form , where now @xmath538 } { { \rm e}}^{-\frac{\lambda_2 ^ 2}{\pi } \big[q_2^{(2)}(t)-{{\rm i}}q_1^{(2 ) } ( t)\big ] } { { \rm d}}t \label{m17'}\ ] ] and @xmath539 , @xmath540 are given in .
the eigenvalues of @xmath541 are @xmath533 and @xmath542 , where @xmath85 is given in .
moreover , the one - dimensional level shift operators @xmath535 have the expression @xmath543 _ proof of proposition [ lso0prop ] . _ * ( 1 ) * in the case of a single reservoir , the calculation of the level shift operator @xmath537 , is an easy transcription of that performed in proposition 3.5 of @xcite , where the case @xmath226 was considered .
we do not present the details . *
( 2 ) * once we have the form , it is readily seen that the vector @xmath544 is in the kernel of @xmath537
. therefore , the eigenvalues of @xmath537 are @xmath533 and @xmath545 .
the relation @xmath546 ( see ) together with gives @xmath547 .
our task now is to show .
let @xmath548 and write @xmath549 instead of @xmath550 and @xmath551 for @xmath552 , where the @xmath553 are given in . then @xmath554 if we want to stress the dependence of @xmath537 on @xmath100 and @xmath555 we wirte @xmath556
. then gives @xmath557 with @xmath558 we use the relation @xmath559 ( for @xmath560 ) to obtain @xmath561 with the canonical commutation relations @xmath562 and the thermal average @xmath563 we arrive at @xmath564 then , remembering the definition of @xmath553 , we arrive at @xmath565 which is the expression for @xmath566 given in .
note that @xmath567 is invariant under a sign change of either @xmath81 or @xmath82 .
next , we find an expression for @xmath568 . starting with the definition and replacing the resolvent by an integral over the propagator , as above , yields the expression @xmath569 where @xmath570 with @xmath571 .
relation shows that @xmath572 is invariant under changing signs of either of @xmath100 , @xmath81 and @xmath82 .
note also that @xmath572 is real .
the symmetry @xmath573 together with and the fact that @xmath574 yields @xmath575 we show below that @xmath576 together with the invariance of @xmath572 under flipping the sign of @xmath100 , this gives @xmath577 this shows the form , modulo showing relation .
we prove now .
set @xmath578 and consider , for @xmath560 , @xmath579 \chi}\right\rangle}. \label{m16}\end{aligned}\ ] ] since @xmath580 is well defined , we can separate the terms in .
we have used in the last step in that @xmath581 ( by the properties of the modular conjugation @xmath148 and the modular operator @xmath582 ) .
then , by , @xmath583 next , by the functional calculus , @xmath584 \chi}\right\rangle } & = & \int_{\mathbb r}\frac{r\ , ( x-\hat\epsilon)}{(x-\hat\epsilon)^2+r^2}\frac{1 -{{\rm e}}^{-\beta(x-\hat\epsilon)/2}}{(x-\hat\epsilon)}{{\rm d}}\mu_\chi(x),\ \ \
\qquad \label{m19}\end{aligned}\ ] ] where @xmath585 is the spectral measure of @xmath586 in the state @xmath587 .
one readily sees that the integrand satisfies the bound @xmath588 independently of @xmath589 .
the right side is integrable w.r.t . @xmath585 since @xmath587 is in the domain of definition of @xmath590 .
then , since the integrand in converges to zero as @xmath591 , for all @xmath592 , and since @xmath593 has measure zero w.r.t
. @xmath585 ( as @xmath594 is not an eigenvalue of @xmath586 ) , we can apply the dominated convergence theorem to conclude that @xmath595 \chi}\right\rangle } = 0 .
\label{m20}\end{aligned}\ ] ] it follows from , and that @xmath596 relation shows that @xmath566 is invariant under a change of the sign of @xmath81 and @xmath82 ( independently ) , so interchanging @xmath597 and @xmath598 in the expression for @xmath599 does not change the value of the expression .
thus we have @xmath600 where we have used in the last step ( with @xmath73 replaced by @xmath601 ) . the definition for @xmath602 shows that the expression on the right side is @xmath603 .
therefore we have proven the relation @xmath604 , which yields since @xmath605 .
this completes the proof of . finally ,
the forms of @xmath535 given at the end of the proposition are checked directly by a simple calculation .
this completes the proof of proposition [ lso0prop ] .
in the setting of the collective reservoir , let @xmath606 where @xmath607 . a direct calculation shows that @xmath608 where @xmath185 is given in and @xmath609 is the raising operator . to calculate the equilibrium state with @xmath78
, we note that @xmath610 now follows readily from and .
in the situation of the local reservoirs model , set @xmath611 , where @xmath612 , and proceed as above to show . using the same notation as in section [ subsectb1 ] we have @xmath613_{12 } = { \rm tr}\left ( { { \rm e}}^{-{{\rm i}}t uh_{{\rm col}}u^ * } u ( \rho_{{\rm s}}(0)\otimes\rho_{{\rm r } } ) u^ * { { \rm e}}^{{{\rm i}}t uh_{{\rm col}}u^ * } u ( |\varphi_2\rangle \langle\varphi_1|\otimes{\mathchoice { \rm 1\mskip-4mu l } { \rm 1\mskip-4mu l } { \rm
1\mskip-4.5mu l } { \rm 1\mskip-5mu l}}_{{\rm r } } ) u^*\right).\ ] ] using the form and that @xmath614 ( and the suitable expressions for the local reservoirs model ) , we arrive at the following expression , for both the local and collective reservoirs models , @xmath615_{12 } = { \cal d}(t ) \ { { \rm e}}^{-{{\rm i}}t\hat \epsilon } [ \rho_{{\rm s}}(0)]_{12},\ ] ] where @xmath616 here , @xmath617 . then , using the canonical commutation relations @xmath618 and the thermal average @xmath619 , follows directly from and the definitions of the functions @xmath620 ( see , ) .
we now show proposition [ prop4 ] .
the reservoir alone has the property of ` return to equilibrium ' , which can be expressed as follows .
[ lemma1 ] let @xmath418 , @xmath457 and @xmath90 be observables of the reservoir and set @xmath621 .
then we have @xmath622 _ proof of lemma [ lemma1 ] . _ for any @xmath623 there exists a @xmath624 which commutes with all observables and which satisfies @xmath625 this is simply the separability of @xmath626 .
therefore , @xmath627 where @xmath628 .
next , since @xmath624 commutes with @xmath629 and since @xmath630 , we have @xmath631 with @xmath632 relation follows from the fact that @xmath633 in the weak sense as @xmath125 ( which in turn is implied by the fact that @xmath634 has absolutely continuous spectrum except for a simple eigenvalue at zero , with eigenvector @xmath626 . ) also , the term @xmath635 in equals @xmath636 , with @xmath637 , by .
we finally obtain @xmath638 with @xmath639 and @xmath640 satisfying .
relation implies .
it follows from lemma [ lemma1 ] that @xmath641 where @xmath512 is given in . for the model with the local reservoirs ,
the relations and @xmath642 are derived in the same way .
this finishes the proof of proposition in the basic resonance expansion , for @xmath643 , all terms on the right side except the one with @xmath644 are @xmath126 .
thus @xmath645_{12 } & = & { { \rm e}}^{-\gamma t/2}{{\rm e}}^{-{{\rm i}}t(\hat\epsilon + x_{\rm ls } ) } { \left\langle { ub^*\psi_0 } , { \big ( |\varphi_{21}\rangle\langle\varphi_{21}|\otimes|\omega_{{\rm r}}\rangle\langle\omega_{{\rm r}}| \big ) ua\omega_{\beta,{\vec{\lambda}},v}}\right\rangle}\nonumber\\ & & + o(v)+r(t ) .
\label{-01}\end{aligned}\ ] ] the projection @xmath646 is the ( weak ) limit @xmath647 and so the scalar product term on the right side of is @xmath648 here , @xmath649 is given in and we have made use of the fact @xmath650 and relation .
one readily sees that @xmath651 the last equality follows from the explicit form of @xmath496 .
using in shows that @xmath652_{12 } + o(v ) .
\label{-05}\end{aligned}\ ] ] the last equality is obtained from , since @xmath653 is exactly the @xmath654 matrix element of the dimer density matrix at time @xmath47 , evolving according to the evolution with @xmath78 .
we combine with and to reach @xmath655_{12 } = { { \rm e}}^{-\gamma t/2}{{\rm e}}^{-{{\rm i}}t(\hat\epsilon + x_{\rm ls } ) } \big\ { \lim_{t\rightarrow\infty } { \mathcal d}(t)\big\ } \ , [ \rho_{{\rm s}}(0)]_{12 } + o(v)+r(t ) .
\label{-04}\ ] ] * upshot . * the rigorous derivation the formula given above can be put into heuristic words , uncovering the mechanism making @xmath656 appear in , as follows .
the resonance approximation ( c.f . ) consists in replacing the propagator @xmath657 , where @xmath658 are complex resonance energies and @xmath487 are projections .
while the @xmath659 depend on @xmath74 , the projections @xmath487 are lowest oder approximations ( i.e. @xmath660 ) and hence independent of @xmath74 .
the observable @xmath643 selects a single projection , @xmath661 , in that all other terms are @xmath126 , @xmath662 due to the dispersiveness of the reservoir(s ) , we have @xmath663 , so that @xmath644 is the long - time limit of the dynamics with @xmath78 , i.e. , @xmath664 .
thus , @xmath665 this is the form , which can be rewritten as ( see ) @xmath666_{21 } = { { \rm e}}^{-\gamma t/2 } { { \rm e}}^{-{{\rm i}}t(\hat\epsilon + x_{\rm ls } ) } \lim_{t\rightarrow\infty } { { \rm e}}^{{{\rm i}}t\hat\epsilon } \big[\rho_{{\rm s}}(t , v=0)\big]_{12 } + o(v)+r(t),\ ] ] where @xmath667 is the dimer density matrix at time @xmath47 , evolving under the dynamics coupled with the reservoir(s ) with @xmath78 . in , @xmath668_{12}$ ] can be considered as a ` shifted initial condition ' for the off - diagonal dimer density matrix .
the analogous analysis holds for the populations , e.g. for @xmath669_{1,1}$ ] .
however , since the populations are stationary under the coupled dimer - reservoir(s ) evolution with @xmath78 , we have @xmath670_{11}=[\rho_{{\rm s}}(0)]_{11}$ ] for all times .
thus the shifted initial condition coincides with the true one and the analogue of the factor @xmath155 is just the factor @xmath76 for the populations .
we limit our discussion here to the situation of a collective reservoir and symmetric coupling , i.e. , @xmath226 .
consider the dimer coupled to the collective reservoir with @xmath78 , and where @xmath217 .
the hamiltonian is ( c.f . )
@xmath671 one readily verifies that the exact solution for the off - diagonal dimer density matrix element is , where @xmath672 is with @xmath673 , and then write the evolution for @xmath674 using a time - dependent generator for the propagator . ]
@xmath675_{12 } = { { \rm e}}^{-{{\rm i}}\epsilon t } [ \rho_{{{\rm s}}}(0)]_{12}\ , { \rm tr}_{{\rm r}}(\rho_{{\rm r}}\ , { { \rm e}}^{-2{{\rm i}}\lambda\int_0^t\phi({{\rm e}}^{{{\rm i}}\omega s}g)}).\ ] ] let @xmath676 and denote @xmath677 then @xmath678 the process @xmath679 is ( non - commutative ) gaussian .
namely , the @xmath0-point correlations functions are expressed solely using two - point correlations according to wick s theorem " ( c.f .
@xcite for example ) , @xmath680 and the correlation functions of odd order vanish . here
, the sum in is over the @xmath681 pairings of the indices .
now @xmath682 where @xmath683 we use in and get @xmath684 a direct calculation shows that @xmath685 where @xmath237 is given in .
relation is thus the same as and @xmath134 is expressed via @xmath686 , the double integral over the correlation function , as @xmath687 instead of considering a quantum mechanical noise as in the previous section , one may introduce a time - dependent hamiltonian ( including an `` external noise '' ) as @xmath688 where @xmath689 is a commutative stochastic process . the exact solution for the off - diagonal density matrix element for each realization of the noise @xmath689 is @xmath690_{12 } = { { \rm e}}^{-{{\rm i}}\epsilon t } [ \rho_{{{\rm s}}}(0)]_{12}\ , { { \rm e}}^{-2{{\rm i}}\lambda\int_0^t\xi_s ds}$ ] and taking the average @xmath691 over the noise gives @xmath692_{12 } = { { \rm e}}^{-{{\rm i}}\epsilon t } [ \rho_{{{\rm s}}}(0)]_{12}\ , \langle { { \rm e}}^{-2{{\rm i}}\lambda\int_0^t\xi_s ds}\rangle.\ ] ] we have again @xmath693 and
if @xmath689 is a gaussian process , then by the gaussian moment theorem ( wick s theorem ) , just as in the quantum case , @xmath694 and the correlation functions of odd order vanish .
we then get in the same way as for the quantum case @xmath695 comparing and shows that decoherence in the classical and the quantum noise case is _ exactly the same _ if the noises have the same characteristics , i.e. , if @xmath689 is gaussian with zero average and two - point function @xmath696 . for either the classical or quantum noise , consider the correlation function @xmath697 for stationary processes , @xmath698 and so @xmath699 .
then @xmath700 we have @xmath701 only the decay asymptotics of the correlation function plays a role to determine if @xmath702 or not .
assume @xmath703 then @xmath704 for large @xmath47 , so we have full decoherence exactly if @xmath705 ( for @xmath706 , we have @xmath707 ) .
this holds equally well for quantum and classical noises . * discussion for the quantum thermal noise .
* one has @xmath708 where @xmath184 is the spectral density of noise . to find the decay of the correlation function , write @xmath709
assume the form @xmath710 and that @xmath184 decays at least as @xmath711 for large @xmath189 . by integrating by parts to transfer the action of the @xmath189-derivatives onto the function @xmath712
, one easily finds that @xmath713 according to the discussion after , the critical value for full decoherence is @xmath714 , so if @xmath184 vanishes more quickly than @xmath715 for small @xmath189 , then we do not have full decoherence . in this way , the decay speed of correlations is governed by the low frequency behavior of the spectral density of noise and therefore , this behavior determines whether we have or do not have full decoherence .
we can sum this up : [ note : in @xcite , ( p. 577 ) , it is mentioned that the effect of non - full decoherence is due to ... the suppressed influence of low - frequency fluctuations ... " , which coincides with the picture we uncover here . ]
we specify the precise regularity requirements which lead to and .
the spectral function of the reservoir is linked to the form factor by @xmath716 and the form corresponds to @xmath717 where @xmath718 is a ( real valued ) function which satisfies @xmath719 for @xmath720 . here , @xmath721 is the spherical representation of vectors @xmath24 .
there are two origins of the infrared regularity needed here .
one is the application of a ` polaron transformation ' ( c.f .
proposition [ propunitrans ] ) , which is indispensable in order to deal with arbitrary values of @xmath81 and @xmath82 and which requires only the milder regularity @xmath72 .
the other origin is that we use a ` dynamical resonance theory ' developed in @xcite , which gives a proof of under the assumptions that @xmath722 where @xmath723 in the case of the collective reservoir model , and @xmath724 and @xmath725 for the local reservoirs model .
we recall where @xmath726 is defined .
the square root factor @xmath727 in is infinitely many times differentiable for @xmath728 and bounded above by ( a constant times ) @xmath729 .
a form factor @xmath54 such that @xmath723 satisfies needs to obey @xmath730 ( recall that @xmath731 ) and it needs to decay at infinity .
let @xmath145 be the strength of the zero and @xmath732 the speed of decay , namely , @xmath733 , for some function @xmath718 which is four times differentiable w.r.t .
its radial component ( @xmath734 and satisfies @xmath735 for @xmath720 .
assume that @xmath54 is real valued ( this is not necessary , but it slightly simplifies the exposition ) .
close to @xmath736 , the singularity structure of @xmath737 is @xmath738 . to have local square integrability of @xmath739 at @xmath736 we thus need that either @xmath740 is an integer @xmath741 , or that @xmath742 .
hence @xmath743 , @xmath744 or @xmath745 ( as @xmath720 ) .
the integrability at @xmath746 is guaranteed for @xmath747 .
note that the above infrared condition is stronger than the infrared condition necessary to apply the polaron transformation , which would only ask for @xmath748 ( so that @xmath749 , which is the condition used to be able to deal with arbitrary sizes of @xmath81 , @xmath82 with the help of the polaron transformation , see proposition [ propunitrans ] ) .
we note that in @xcite , only a dynamical resonance theory for model with a single reservoir is considered .
one must slightly adapt the arguments for the case of two reservoirs to be able to describe our local reservoirs model . an easy way to do
this is to use a well - known property of fock space , the isometric isometry @xmath750 under this mapping , the formalism of two reservoirs turns into one of a single reservoir ( albeit on a different one - particle space ) .
the arguments of @xcite leading to the resonance expansion of the propagator ( c.f . ) can then be adapted in a straightforward way to this new single - reservoir setting .
we do not present the details of the analysis in the present work .
the large coupling results of @xcite our resonance approach in the current work is based on are stated as follows . given arbitrary @xmath751 , there is a constant @xmath298 such that if @xmath752 , then the results hold .
the upper bound @xmath753 depends on the fixed @xmath129 .
this dependence can be found by tracing the parameters @xmath19 through the arguments of @xcite .
we set @xmath754 where @xmath13 and @xmath100 are given in and , respectively .
note that @xmath755 and @xmath100 depend on @xmath81 and @xmath82 .
the following are constraints used to prove the validity of the dynamical resonance theory . * collective reservoir model . *
the interaction depends only on the difference @xmath756 of the coupling constants .
define @xmath757 let @xmath90 be a constant which does not depend on @xmath758 .
* validity of the _ born approximation _ , @xmath759 has spectrum in the lower complex half plane , c.f .
lemma 4.3 of @xcite ; the main dynamics can then be expressed on @xmath760 . ] * separation of resonances , @xmath761 , @xmath762 , and @xmath763 .
rescaled by @xmath764 , their separation ( smallest distance among them ) is @xmath765 . even if the real part @xmath766 of @xmath762 is large , shifting those two resonances horizontally far away from the origin , one resonance ( @xmath767 ) stays at a distance @xmath173 from @xmath763 .
the separation of resonances ( rescaled by @xmath764 ) must persist for perturbations small in @xmath74 , hence the above upper bound on @xmath74 .
[ technical remark : this condition is used to guarantee that @xmath768 is close to the level shift operator in lemma 3.1 of @xcite . ] * no backreaction on the dimer , latexmath:[\[\label{need3 } states of the uncoupled dimer - reservoir system are ( superpositions of ) states with the dimer in a dimer - energy eigenstate and the reservoir in equilibrium .
the above condition ensures that away from these states , the coupled dynamics is dispersive ( local observables decay sufficiently quickly in time ) .
the dispersiveness estimates are obtained separately for different stationary dimer bohr energies , so they involve their separation , @xmath100 ( on which @xmath770 depends ) .
[ technical remark : this condition is used to prove the limiting absorption principles of the interacting resolvent , in particular when reduced to @xmath771 , theorem a.1 in @xcite . ] * discussion .
* consider spectral functions of the form , @xmath772 , with @xmath75 as in .
* consider @xmath773 .
we now show that @xmath774 for small @xmath775 , we have @xmath776 , see section 5 of @xcite . taking into account the form of the spectral density , we obtain @xmath777 consider two cases : + \(a ) @xmath778 hence @xmath779 . the conditions
become @xmath780 and from we obtain the first estimate in .
+ \(b ) @xmath781 , so that @xmath782 . the conditions
become @xmath783 so @xmath784 . *
consider @xmath785 .
the conditions become @xmath786 . *
consider @xmath787 .
from we expect that @xmath788 for some @xmath789 . the conditions
become @xmath790 and therefore @xmath791 is extremely tiny .
* conclusion .
* sizeable values of @xmath84 will be obtained in the regime @xmath785 only , that is , when the difference of @xmath81 and @xmath82 is not too large nor too small . for @xmath792 ,
the dynamical process is suppressed ( weakly , @xmath84 is proportional to a power of @xmath793 ) because for @xmath169 there is no relaxation or decoherence at all . for @xmath794
the dynamical process is is strongly suppressed ( @xmath795 , which is an effect analogous to anderson localization .
if we use the expression for @xmath84 in the marcus regime , we have @xmath796 . the above constraints on @xmath74
give an upper bound on @xmath84 , @xmath797 these upper bounds are qualitatively correct ( they coincide qualitatively with the ones obtained from the true expression for @xmath84 , even though for small @xmath775 , the power of the upper bound is different ) . * local reservoirs model . * here the bounds - need to be verified with @xmath798 and @xmath799 .
a reservoir of vibrational modes is typically modeled by a collection of ( independent ) quantum oscillators with hamiltonian ( @xmath183 ) @xmath800 with @xmath801 ( kronecker symbol ) . in the context of a _ ` molecular reservoir ' _ , one has @xmath2 ( protein , solvent ... ) atoms in the environment , each one being modeled by three degrees of freedom , i.e. , by a three - dimensional linear quantum oscillator with frequencies @xmath802 , @xmath803 , @xmath804 .
the hamiltonian of the ( finite mode ) reservoir could then be written as in with either @xmath2 replaced by @xmath805 or it could be written as a sum over @xmath806 . in a _ ` structured environment ' _ ,
say , given by a periodic structure , by phonons or photons , the index @xmath807 is the wave vector and has associated frequency @xmath808 .
this fixes a relation between @xmath24 and @xmath189 , the dispersion relation ( e.g. @xmath809 for acoustic phonons or photons ) .
however , the frequencies present in a molecular reservoir may be more ` random ' .
one can list them increasingly as @xmath810 , but there is no inherent relation between the index @xmath811 and the value @xmath812 .
the defining property for a molecular reservoir is the _
fraction among all oscillators having a given frequency_. it is determined by a function @xmath813 such that for any set @xmath814 , the quantity @xmath815 is the proportion of all oscillators having frequencies lying in @xmath816 .
in particular , @xmath817 .
the function @xmath818 is the _ frequency density of the molecular environment_. mathematically , the treatment of molecular and structured reservoirs is the same .
our method does not rely on the presence or absence of a dispersion relation .
the only quantity that matters is the spectral function @xmath184 , which is defined via the correlation function of the reservoir , c.f . .
the _ explicit relation _ between the spectral function @xmath148 and the form factor @xmath54 will depend on the nature of the reservoir ( dispersion relation or frequency density ) .
for instance , for three dimensional phonons or photons with @xmath819 , we have @xmath820 ( where @xmath54 is represented in spherical coordinates of @xmath28 ) . for a molecular reservoir with frequency density @xmath821 , @xmath822 however , in either case , the relevant physical characteristics of the reservoir are condensed into the function @xmath184 , which is typically of the form or .
m. aghtar , j. strmpfer , c. olbrich , k. schulten , and u. kleinekathfer : _ different types of vibrations interacting with electronic excitations in phycoerythrin 545 and fenna - matthews - olson antenna systems _ , j. phys .
, * 5 * , 3131 - 3137 ( 2014 ) d.d .
eads , e.w .
castner jr . , r.s .
alberte , l. mets , and g.r .
fleming : _ direct observation of energy transfer in a photosynthetic membrane : chlorophyll b to chlorophyll a transfer in lhc _ , j. phys .
, * 93 * , 8271 - 8275 ( 1989 ) m. merkli , g.p .
berman , a. redondo : _ application of resonance perturbation theory to dynamics of magnetization in spin systems interacting with local and collective bosonic reservoirs _ ,
a : math . theor .
* 44 * , 305306 - 305330 ( 2011 ) c. p.
van der vegte , j.d .
prajapati , u. kleinekathfer , j. knoester , and t. l. c. jansen : _ atomistic modeling of two - dimensional electronic spectra and excited - state dynamics for a light harvesting 2 complex _ , j. phys .
b , * 119 * 1302 - 1313 ( 2015 ) d. xu , k. schulten : _ coupling of protein motion to electron transfer in a photosynthetic reaction center : investigating the low temperature behavior in the framework of the spin - boson model _ , chem . phys .
* 182 * , 91 - 117 ( 1994 )
|
we present a theoretical analysis of exciton transfer and decoherence effects in a photosynthetic dimer interacting with collective ( correlated ) and local ( uncorrelated ) protein - solvent environments .
our approach is based on the framework of the spin - boson model .
we derive explicitly the thermal relaxation and decoherence rates of the exciton transfer process , valid for arbitrary temperatures and for arbitrary ( in particular , large ) interaction constants between the dimer and the environments .
we establish a generalization of the marcus formula , giving reaction rates for dimer levels possibly individually and asymmetrically coupled to environments .
we identify rigorously parameter regimes for the validity of the generalized marcus formula .
the existence of long living quantum coherences at ambient temperatures emerges naturally from our approach .
la - ur-15 - 29509
|
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.