article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the 1d optical lattice is generated by two independent , phase stabilized laser beams at @xmath79=830 nm , intersecting at an angle of 117@xmath80 in the @xmath81 plane , which results in a lattice spacing of @xmath27=486 nm along @xmath82 .
the radial dimensions are only weakly confined by an optical dipole trap of trap frequency @xmath83=18@xmath582hz and @xmath84=45 - 55hz .
we load a bose einstein condensate of rubidium 87 atoms into the lattice , which results in an occupation of around 80 pancake - like lattice sites .
the lattice depth is 10@xmath581@xmath85 , where @xmath85=2@xmath86=3.33khz .
this results in a bare tunneling rate of roughly 0.007@xmath85 .
the bec is in the weakly interacting regime where @xmath871 , with on - site interaction @xmath88 .
the lattice is accelerated along the @xmath82 axis by modulating the frequency of one of the lattice beams with an acousto - optical modulator .
we shake the lattice time - periodically with @xmath89 = 1khz , as the modulation frequency has to be fast with rsespect to the atomic motion but small compared to band excitation energies .
the inertial force acting on the atoms in the lattice frame consists of a train of sinusoidal pulses and is defined over a period as : @xmath90 the forcing parameter @xmath40 is defined as : @xmath91 where @xmath92 is the amplitude of the frequency modulation , ranging here from 0 to 30khz .
many different analytical functions may be chosen for the time - dependent inertial force . * ( a ) * absorption images taken after 27ms time - of - flight over two periods of forcing .
the center - of - mass oscillates under the effect of the forcing .
the quasi - momentum , on the contrary , remains at any time described by the effective hamiltonian . *
( b ) * plot of the fitted momentum of the background ( resembling in good approximation the wannier envelope ) , denoted @xmath93 , as a function of time.,width=302 ] in order to induce complex tunneling parameters , the function should break the relevant symmetries ( a ) and ( b ) described in the text . for this , the frequency spectrum of the force has to contain more than one component .
a major advantage of the function defined in ( [ eq : f ] ) is the control of the sharpness of the peierls phase dependency in @xmath40 and the minimal amplitude of the effective hopping rate via the value of the asymmetry ratio @xmath26 , where for the experiment @xmath941ms was chosen . to reveal the quasi - momentum distribution of the bec in the 1d lattice ,
the lattice and the dipole trap potential are suddenly turned off .
we measure the absolute position of the interference pattern after 27ms time - of - flight , with the quasi - momentum distribution obtained without forcing as a reference .
when taking those pictures in the laboratory frame , one observes the quasimomentum distribution of the state evolving according to the time - independent effective hamiltonian , enveloped by the momentum distribution of the lowest - band wannier - function that oscillates like @xmath95 .
[ fig_sm ] shows those oscillations over two periods of the forcing , corresponding to a peierls phase of 0.64@xmath4 . for the data presented in the text , the pictures were always taken at times within a period where @xmath96 and averaged over 3 to 7 experimental runs
|
we present a universal method to create a tunable , artificial vector gauge potential for neutral particles trapped in an optical lattice .
the necessary peierls phase of the hopping parameters between neighboring lattice sites is generated by applying a suitable periodic inertial force such that the method does not rely on any internal structure of the particles .
we experimentally demonstrate the realization of such artificial potentials , which generate ground state superfluids at arbitrary non - zero quasi - momentum .
we furthermore investigate possible implementations of this scheme to create tuneable magnetic fluxes , going towards model systems for strong - field physics .
first introduced in electromagnetism , gauge fields play a central role in the description of interactions in physics , from particle physics to condensed matter . currently there is a large interest to introduce gauge fields into model systems in order to study fundamental aspects of physics @xcite .
especially the emulation of synthetic electric and magnetic fields for of ultracold atomic systems is crucial in order to extend their proven quantum simulation abilities further , e.g. to quantum hall physics or topological insulators . in this context , the analogy between inertial and lorentz forces triggered the simulation of homogeneous artificial magnetic fields using rapidly rotating trapped ultracold gases @xcite .
recently , several proposals ( @xcite and references therein ) and experimental realizations focused on the simulation of a gauge vector potential @xmath0 either in a bulk system @xcite or in optical lattices @xcite .
the realized schemes exploit the berry phase which arises when the atomic ground state is split in several space - dependent sublevels , as in the presence of an electromagnetic field .
hence they rely on the coupling between internal and external degrees of freedom induced by laser fields . here
we demonstrate the generation of artificial gauge potentials for neutral atoms in an optical lattice without any requirements on the specific internal structure .
as the realized scheme only relies on the trapability of the particle , it can be very widely applied to many atomic systems , in principle also to molecules and other complex particles .
it is particularly interesting for fermionic systems , where in a many - body state governed by pauli principle , the use of internal degrees of freedom often lead to conflicts with the creation of gauge potentials . as an important additional benefit ,
the internal degrees of freedom of the particles can be addressed independently , e.g. by real magnetic fields or microwave excitations . in general
, the presence of a gauge vector potential modifies the kinetic part of the hamiltonian describing the system . in a lattice , an artificial field
can then be simulated by engineering a complex tunneling parameter @xmath1 , where @xmath2 is the peierls phase .
the central approach here is to control this phase via a suitable forcing of the lattice potential , acting at the single - particle level .
we describe the general scheme for the creation of an artificial gauge potential and experimentally demonstrate the realization of a tunable vector gauge potential in 1d .
we analyze the dynamical processes leading to the relaxation towards a superfluid ground - state at any desired , finite quasi - momentum . in a 2d lattice
the described forcing can be used to create artificial magnetic fluxes smoothly tunable between @xmath3 and @xmath4 .
this opens the possibility to emulate strong - field physics with cold atom systems , in a regime currently not accessible in condensed matter because of experimental limitations on the magnetic field strength .
for the general scheme , we consider a system of ultracold atoms in a deep optical lattice described by the hubbard - type tight - binding hamiltonian : @xmath5 here @xmath6 , @xmath7 and @xmath8 denote the creation , annihilation and number operator for a particle ( boson or fermion ) of mass @xmath9 localized at site @xmath10 at @xmath11 ; @xmath12 quantifies the tunneling between neighboring sites , and @xmath13 comprises time - independent on - site terms describing , e.g. interactions or a trapping potential .
the system is driven by fast , off - resonant time - periodic potential modulations @xmath14 of zero time average @xmath15 with period @xmath16 .
the case of a homogeneous inertial force @xmath17 , which is created by shaking the lattice along the periodic orbit @xmath18 in space ( see fig .
[ fig1](a ) ) , is described by @xmath19 . under the conditions described in ref .
@xcite , one can show that the driven system is to good approximation described by the effective time - independent hamiltonian : @xmath20 the important terms here are the complex tunneling parameters @xmath21 ( see fig . [
fig1](c ) ) , where @xmath22 . for sinusoidal forcing
, such a dynamic modification of tunneling is restricted to @xmath23 or @xmath4 @xcite .
it has recently been observed in several experiments @xcite . we will now show that by suitable driving , the peierls phases @xmath24 can in fact be smoothly tuned to any value ( @xmath25 $ ] ) .
this allows one to engineer vector potentials that can give rise to both artificial electric and magnetic fields .
similar ideas have recently been presented for a bose - fermi mixture in a triangular lattice @xcite .
periodic driving of the 1d lattice .
* ( a ) * time - periodic movement of the lattice in real space . *
( b ) * resulting periodic inertial force of zero mean .
@xmath26 is the asymmetry parameter . *
( c ) * realized complex tunneling elements on a 1d lattice , with spacing @xmath27 . *
( d ) * effective single - particle dispersion relation .
a complex phase induces a shift of @xmath28 towards @xmath29=@xmath30 ( solid line).,width=321 ] the peierls phase @xmath24 can be varied smoothly , even though @xmath31 .
indeed @xmath32 can be complex provided the forcing breaks two symmetries that are also known to prevent ratchet - type rectification in classical @xcite and quantum @xcite lattice systems , namely : ( a ) reflection symmetry for a suitable time @xmath33 , i.e. @xmath34 and ( b ) shift ( anti)symmetry , i.e. @xmath35 .
this requires forcing with more than one frequency .
the forcing function used in our experiment ( see fig .
[ fig1](b ) ) induces a monotoneously increasing peierls phase @xmath36 with the amplitude of the force .
the peierls substitution directly links this phase to a gauge vector potential @xmath0 via @xmath37 ( integrated along a straight path ) , which allows to create artificial electric or magnetic fields .
the modification of the peierls phase in time results in an artificial electric force @xmath38 .
it will be shown at the end of this manuscript that our scheme allows the emulation of a vector potential that gives rise to a finite artificial magnetic flux through the elementary plaquettes of 2d lattices .
creation of complex tunneling matrix elements . *
( a ) * absolute value of the tunneling parameter obtained from equation ( [ eq : complexj ] ) for our experimental parameters ( @xmath39=1ms and @xmath26=2.1 ) . *
( b ) * the measured peierls phases in a 1d driven optical lattice for different values of the forcing amplitude @xmath40 are depicted as circles .
the dashed red curve corresponds to the theoretically expected values ( equation ( [ eq : complexj ] ) ) . *
( c ) * quasi - momentum distribution of the bec after 27ms time - of - flight for different values of @xmath40 .
the peierls phase as a function of @xmath40 is deduced from the observed shifts of the interference patterns.,width=321 ] to demonstrate the power of this scheme , we have emulated a vector potential for a rubidium bose - einstein condensate ( bec ) in a one - dimensional lattice .
the trapped particles are accelerated via frequency modulation of one of the lattice beams , a technique which is experimentally straightforward .
the resulting inertial force is comprised of a train of sinusoidal pulses separated by periods of rest of periodicity @xmath16=@xmath39=1ms as depicted in fig .
[ fig1](b ) .
the zero mean value of this force prevents the transfer of a net - acceleration to the lattice .
this leads to a renormalized tunneling matrix element of the form : @xmath41 where @xmath26 is the asymmetry parameter ; @xmath40 is the forcing amplitude ( see appendix ) and @xmath42 the zero - order bessel function . from equation
( [ eq : complexj ] ) , the induced peierls phase @xmath2 in the effective tunneling element can be calculated . for @xmath43 ,
one recovers the harmonic driving introduced in @xcite which only allows to control magnitude and sign of @xmath44 . in the opposite limit of small @xmath45 , the tunneling amplitude @xmath46 is only marginally affected whereas the phase @xmath2 depends linearly on the forcing amplitude @xmath40 . in order to avoid unwanted excitations of the system induced by strong forcing ,
we have chosen an intermediate value @xmath47 for the experiment realized here .
note however that for such an asymmetry parameter the peierls phase depends non - linearly on @xmath40 ( see fig .
[ fig2](b ) ) .
the thereby introduced gauge potential @xmath48 manifests itself via a shift of the dispersion relation by @xmath30 as depicted in fig .
[ fig1](d ) .
this follows directly from the effective hamiltonian ( [ eq : heff ] ) , whose eigenstates are bloch waves with the dispersion relation @xmath49 .
the gauge potential therefore allows for the generation of superfluid ground - states ( with group velocity @xmath50 ) at finite and tunable quasi - momentum @xmath51 .
as will be detailed later on , we experimentally observe the relaxation of the condensate quasi - momentum towards the minimum of the effective dispersion relation .
therefore , the imprinted peierls phase can be directly read out from the quasi - momentum distribution revealed in time - of - flight after a sudden switch off of the lattice and the external potential . as a central result fig .
[ fig2](b ) shows the experimental data together with the theoretical predictions from equation ( [ eq : complexj ] ) . after increasing the forcing amplitude slowly ( in up to 120ms ) to the desired value
, the corresponding quasi - momentum distribution was recorded . from the obtained time - of - flight images , examples of which
are shown in fig .
[ fig2](c ) , we extract the peierls phases @xmath2 ( see appendix ) .
we observe an excellent agreement between experiment and theory , thus proving the controlled generation of an arbitrary vector gauge potential encoded into the peierls phase @xmath52 $ ] .
in addition , the experimental images demonstrate the large degree of coherence maintained in the atomic sample throughout the shaking process . as an additional feature , fig .
[ fig2](a ) shows that the peierls phase allows now to invert the sign of the effective tunneling element without crossing @xmath53=0 via the rotation in the complex plane . in the following
we will discuss the details of the relaxation of the system towards non - zero quasi - momenta superfluid states , allowing for the described direct measurement of the peierls phase .
note that for an homogeneous and non - interacting system , the initial bloch wave at @xmath54=0 remains an eigenstate of the effective hamiltonian .
thus no transfer to states with @xmath55@xmath560 is expected after the shaking is turned on .
however , since we are working with interacting bosons and an external harmonic confinement more effects come into play .
time - resolved measurements .
* ( a),(b ) * a quench to a peierls phase of -@xmath4/4 within 1ms induces an excitation of the dipole mode , leading to oscillations of the quasimomentum as depicted in * ( a ) * and observed in the time - resolved measurement of the quasi - momentum up to 0.5s.*(b)*. the plain curve is a fit to a sinusoid from which one obtains @[email protected] , in perfect agreement with the expected dipole frequency of the dressed condensate . the final value of the peierls phase is depicted as a dashed line . * ( c)-(f ) * shift of the quasi - momentum towards a final peierls phase of 3@xmath4/2 for a slow increase of the forcing amplitude over 160ms * ( c),(d ) * or a sudden quench within 1ms * ( e),(f)*. the left graphs depict the center of mass dynamics in @xmath55-space in both cases . on the right , series of time - of - flight images are reported as a function of time .
the time - steps are 10ms and 1ms , respectively.,width=321 ] when the gauge potential is ramped up from 0 to @xmath59 , the condensate acquires a non - zero group velocity , reflecting the presence of an artificial electric force @xmath60=-@xmath61 .
this velocity induces a displacement of the condensate s center of mass position @xmath62 in the harmonic potential of frequency @xmath63 ( see appendix ) .
the resulting restoring force induces oscillations both in position and momentum - space ( see fig . [ fig3](a ) ) . in fig .
[ fig3](b ) , we report a time - resolved measurement of the condensate quasi - momentum after a quench to a final peierls phase of -@xmath4/4 . the oscillations around the final quasi - momentum result from an excitation of the dipole mode : the measured frequency of [email protected] perfectly matches the expected dressed condensate frequency @xmath64 for particles having an effective mass @xmath65 in the lattice of 10@xmath581@xmath66 depth with a tunneling amplitude of 0.3 @xmath67 ( @[email protected] ) .
the coupling to non - zero quasi - momenta results thus from the underlying harmonic trapping potential .
in addition , this center of mass dynamics is subjected to several damping mechanisms induced by the trap anharmonicity or the lattice discreteness , which leads to a coupling to other collective modes and therefore to the relaxation of the bec towards the new equilibrium state .
therefore the duration of the ramp from 0 to @xmath59 has to be compared with the timescale of those relaxation mechanisms . in fig .
[ fig3 ] we compare time - resolved measurements of the quasi - momentum distribution for a slow ramp ( fig .
[ fig3](d ) ) of @xmath69 to a final peierls phase @xmath2=3@xmath4/2 , with a sudden quench ( fig .
[ fig3](f ) ) . as the gauge field
is slowly increased , the bec follows the shift of the dispersion relation minimum , as depicted in fig .
[ fig3](c ) . for the quench , on the contrary , for which the shift of the dispersion relation occurs within one ms ,
the system can not follow and thus relaxes into the nearest minimum of the effective band structure ( see fig [ fig3](e ) ) .
for the chosen value this minimum lies on the left with respect to the original @xmath55=0 peaks and we thus find the bec at @xmath55=-@xmath4/2@xmath27 .
this demonstrates clearly that in presence of these relaxation mechanisms , the forcing does not induce a net particle current in the lattice unlike for ratchets but allows the engineering of ground - state superfluids at arbitrary non - zero quasi - momenta .
our scheme to generate complex tunneling offers fascinating possibilities to emulate gauge fields in higher dimensions . in the following , few examples based on lattice shaking and modulated superlattices will be described .
an artificial magnetic field is characterized by a finite magnetic flux through an elementary plaquette @xmath70 of the lattice , given by the sum @xmath71 $ ] taken around @xmath70 .
homogeneous forcing , implemented by lattice shaking , can not give rise to a magnetic flux through plaquettes having pairwise parallel edges like square or hexagonal .
it can , however , lead to a non - zero flux @xmath72 through a triangular plaquette ; @xmath73 for @xmath74 the amplitude of the force oriented parallel to one edge . as depicted in fig .
[ fig4](a ) , continuously tunable staggered fluxes are realized ( @xmath75 for the inverted plaquette ) . creation of magnetic fluxes in triangular lattices .
* ( a ) * staggered magnetic fluxes obtained in a 2d triangular lattice . * ( b ) * dispersion relation for different values of the magnetic flux @xmath76 through a plaquette.,width=321 ] our method enables one to continuously control the degree of frustration from zero ( @xmath77 ) to maximum ( @xmath78 ) in lattice geometries like triangular or kagom @xcite .
this control can be used to adiabatically prepare and study exotic quantum phases that are expected to appear in the strongly frustrated limit .
increasing the flux in a triangular lattice , frustration is reflected in the fact that the dispersion relation develops two minima that eventually , for @xmath4-flux , become degenerate ( see fig . [ fig4](b ) ) . in a triangular lattice with @xmath4-flux spontaneous symmetry - breaking
has recently been observed in an experimental simulation of classical magnetism @xcite . with the presented scheme ,
one is now able to study the influence of a small symmetry - breaking perturbation and to observe the many - body dynamics initiated by preparing a system in the upper minimum .
tunable magnetic fluxes through e.g. square or hexagonal plaquettes can also be obtained via inhomogeneous forcing .
a simple way of doing this , is to create an oscillating super - lattice as will be described elsewhere . beyond the engineering of the hopping element ,
another essential feature of our scheme is the possibility to address the atomic internal degrees of freedom independently . here
a spin - dependent hexagonal lattice @xcite constitutes a promising system . in such a geometry the interplay of next - nearest neighbor coupling , complex tunneling matrix elements and spin are particularly interesting in connection with haldane s model for instance @xcite .
the very general method presented in this manuscript should allow for large sets of new gauge fields schemes for various particle classes in optical lattices .
we thank the deutsche forschungsgemeinschaft ( for 801 , grk 1355 , sfb 925 ) and the landesexzellenzinitiative hamburg , which is supported by the joachim herz stiftung , for funding .
p.h . and m.l .
are grateful for support through the european research council grant quagatua , aaii - hubbard and caixa manresa .
25ifxundefined [ 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 10.1080/00018730701223200 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.92.050403 [ * * , ( ) ] http://stacks.iop.org/1367-2630/12/i=6/a=065025 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.83.1523 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.102.130401 [ * * , ( ) ] link:\doibase 10.1038/nature08609 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.107.255301 [ * * , ( ) ] @noop http://stacks.iop.org/0295-5075/89/i=1/a=10010 [ * * , ( ) ] link:\doibase 10.1103/physrevb.34.3625 [ * * , ( ) ] link:\doibase 10.1209/0295 - 5075/18/7/001 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.69.351 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.95.260404 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.99.220403 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.100.190405 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.102.100403 [ * * , ( ) ] link:\doibase 10.1126/science.1207239 [ * * , ( ) ] http://arxiv.org/pdf/1112.0972 [ ] link:\doibase 10.1103/physrevlett.84.2358 [ * * , ( ) ] link:\doibase 10.1103/physreva.75.063424 [ * * , ( ) ] http://stacks.iop.org/1367-2630/12/i=6/a=065025 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.045305 [ * * , ( ) ] link:\doibase 10.1038/nphys1916 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.107.235301 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.101.246810 [ * * , ( ) ]
|
at a distance of 16.7 mpc , m87 is the dominant galaxy of the virgo cluster .
it is one of the nearest radio galaxies and was the first extragalactic x - ray source to be identified . because of its proximity
, many interesting astrophysical phenomena can be studied in more detail in m87 than in other comparable objects .
particularly remarkable is the prominent jet extending from the nucleus , visible throughout the electromagnetic spectrum .
the central regions of m87 , in particular the structure of the jet , have been studied and compared intensively at radio , optical , and x - ray wavelengths .
( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
compared to the available information at these wavelengths , our knowledge of m87 at far - infrared ( fir ) wavelengths is rather poor .
a controversial issue is the origin of the fir emission in m87 , i.e. , the question of whether the fir emission is caused entirely by synchrotron emission or whether there is an additional contribution from dust associated with either the global interstellar medium or a nuclear dust component .
several papers , based on iras , iso and spitzer observations , arrive at different conclusions @xcite . , beam sizes are indicated in the bottom right corner .
top right : the global sed of m87 from mid - infrared to radio wavelengths .
where no error bars are seen , they are smaller than the symbol size .
the solid line in the plot is the best - fit power law of the isocam , iras , mips , scuba , gbt , wmap , and vla data ; the dotted line has only been fitted to the scuba , gbt , wmap , and vla data .
bottom right : residual between data and the best - fit synchrotron model in the infrared - submm wavelength range .
the cyan line is a modified black - body model with @xmath0 k and @xmath1.,title="fig:",scaledwidth=56.0% ] , beam sizes are indicated in the bottom right corner .
top right : the global sed of m87 from mid - infrared to radio wavelengths . where no error bars are seen
, they are smaller than the symbol size .
the solid line in the plot is the best - fit power law of the isocam , iras , mips , scuba , gbt , wmap , and vla data ; the dotted line has only been fitted to the scuba , gbt , wmap , and vla data .
bottom right : residual between data and the best - fit synchrotron model in the infrared - submm wavelength range .
the cyan line is a modified black - body model with @xmath0 k and @xmath1.,title="fig:",scaledwidth=44.0% ] we investigate the nature of the fir emission of m87 using new fir data from the herschel space observatory , obtained as part of the science demonstration phase ( sdp ) observations of the herschel virgo cluster survey .
hevics is an approved open time key program , which has been awarded 286 h of observing time in parallel mode with the pacs and spire instruments .
we will ultimately map four @xmath2 square degree regions of the cluster at 100 , 160 , 250 , 350 and 500 @xmath3 m , down to the 250 @xmath3 m confusion limit of about 1 mjy sr@xmath4 .
the hevics sdp observations consisted of a single cross - scan of one @xmath2 deg@xmath5 field at the centre of the virgo cluster .
while these sdp observations comprise only 6% of the total hevics observations , the analysis based on these observations already gives a prelude to the primary hevics science goals including the effects of the environment on the dust medium of galaxies , the fir luminosity function , the complete seds of galaxies and a detailed analysis of the dust content of dwarf and early - type galaxies .
the observations also enabled us to study the intensity and nature of the fir emission of m87 .
the left part of figure [ rawdata.pdf ] shows the pacs and spire images of the central @xmath6 region of m87 , which is clearly detected in all five bands .
the top right panel in figure [ rawdata.pdf ] shows the integrated sed in the infrared - submm - radio region between 15 @xmath3 m and 100 cm , with the new herschel data as well as isocam , iras , mips , and scuba , gbt , wmap , and vla data gathered from the literature @xcite ; the actual flux densities can be found in .
the solid line is the best - fit power law for all literature data and has a slope @xmath7 ; the dotted line fits only the submm and radio data and has a slope @xmath8 .
the bottom right panel in figure [ rawdata.pdf ] shows the residual from the best - fit power law in the infrared - submm wavelength region ; clearly , the integrated herschel fluxes are in full agreement with synchrotron radiation .
the cyan line in this figure is a modified black - body fitted with @xmath0 k and @xmath1 .
this temperature is the mean dust equilibrium temperature in the interstellar radiation field of m87 , determined using the skirt radiative transfer code @xcite and based on the photometry from @xcite .
the dust mass was adjusted to fit the upper limits of the residuals .
it is clear that the sed of m87 is incompatible with dust masses higher than @xmath9 .
m. the left panel shows the synchrotron image at the model resolution , the central panel shows the same model convolved to the spire 250 @xmath3 m beam and pixel size .
the right panel shows the observed spire 250 @xmath3 m image . in all panels ,
the green lines are the contours of the synchrotron model at the model resolution . ]
although indicative , the analysis of the integrated sed does not definitively identify the origin of the fir emission in m87 .
approximating the global sed as a single power - law synchrotron model is indeed an oversimplification of the complicated structure of m87 .
several studies have shown that m87 contains three distinct regions of significant synchrotron emission , each with their own spectral indices : the nucleus , the jet and associated lobes in the nw region , and the se lobes ( e.g. , * ? ? ?
* ; * ? ? ?
? * ; * ? ? ?
we have constructed a synchrotron model for the central regions of m87 , based on a newly reduced mips 24 @xmath3 m map and archival mustang 90 ghz and 15 , 8.2 , 4.9 , 1.6 , and 0.3 ghz maps @xcite .
we fitted a second - order polynomial synchrotron model to each pixel of the mips + radio data cube and used this synchrotron model to predict the emission of m87 at 250 @xmath3 m ( the spire 250 @xmath3 m image provides the optimal compromise between s / n and spatial resolution ) .
figure [ h250.pdf ] shows the comparison between the synchrotron model prediction at 250 @xmath3 m and the spire observations . at the model resolution
, the three distinct components are visible , but when we convolve this synchrotron model image with the spire 250 @xmath3 m beam , the three different components merge into a single extended structure with one elongated peak slightly west of the nucleus . comparing the central and right panels of figure [ h250.pdf ]
, we see that the synchrotron model is capable of explaining the observed spire 250 @xmath3 m image satisfactorily .
we conclude that for both the integrated sed and the spire 250 @xmath3 m map , we have found that synchrotron emission is an adequate explanation of the fir emission .
we do not detect a fir excess that can not be explained by the synchrotron model .
in particular , we have no reason to invoke the presence of smooth dust emission associated with the galaxy interstellar medium , as advocated by @xcite . for a dust temperature of 23 k , which is the expected equilibrium temperature in the interstellar radiation field of m87
, we find an upper limit to the dust mass of @xmath10 .
our conclusion is that , seen from the fir point of view , m87 is a passive object with a central radio source emitting synchrotron emission , without a substantial diffuse dust component .
|
the origin of the far - infrared emission from the nearby radio galaxy m87 remains a matter of debate .
some studies find evidence of a far - infrared excess due to thermal dust emission , whereas others propose that the far - infrared emission can be explained by synchrotron emission without the need for an additional dust emission component .
we observed m87 with pacs and spire as part of the herschel virgo cluster survey ( hevics ) .
we compare the new herschel data with a synchrotron model based on infrared , submm and radio data to investigate the origin of the far - infrared emission .
we find that both the integrated sed and the herschel surface brightness maps are adequately explained by synchrotron emission . at odds with previous claims , we find no evidence of a diffuse dust component in m87 .
|
`` the nobel prize in physics 2011 was divided , one half awarded to saul perlmutter , the other half jointly to brian p. schmidt and adam g. riess for the discovery of the accelerating expansion of the universe through observations of distant supernovae . '' the nobelprize.org wrote .
two research teams , supernova cosmology project ( scp ) lead by saul perlmutter and high - z supernova search team ( hzt ) headed by brian schmidt , raced to map the universe by locating the most distant supernovae .
the two research teams found over 50 distant supernovae which were dimmer than expected - this was a sign that the expansion of the universe was accelerating [ 1 ] , [ 2 ] . theoretically , this was not new idea since albert einstein came out the idea of cosmological constant ( often marked lambda ) .
einstein did that because he was guided by the paradigm of the day that the universe was static . when he heard the edwin hubble discovery that the universe was actually expanding he declared that the inclusion of the cosmological constant was his `` biggest blunder '' .
since there was no observation for cosmological constant after that most scientists assumed that lambda is zero .
a series of papers were published in astrophysics and space science in the early 90 s years before the supernova publications calculated @xmath0 from observed data .
pal et al . [ 3 ] used the so called pencil beam survey [ 4 ] to find out whether the regularity found in the galaxy distribution is quasiperiodical or not .
it was found that @xmath1 was preferably negative [ 3 ] .
therefore a nonzero lambda term was needed .
the preferred value was @xmath0 equal 2/3 .
this is very close to the value that later was observed by the nobel prize winners . in the second paper [ 5 ]
a two parameter fit was made .
the positive cosmological constant ( negative @xmath1 ) found to be still needed . in the third paper [ 6 ] optical and radio quasars
were also used to find the preferred cosmological parameters .
figure 8 . in that paper [ 6 ] showed the results ( see figure ) .
as it was written the contour meant 80% confidence level .
the preferred region is similar that the supernovae analisys later suggested . for comparison
please see this www page http://vizion.galileowebcast.hu/hoi/comparation.jpg therefore those earlier suggestions also support the nobel prize winning results .
the author thanks for his supervisors , b. lukcs and g. pal , to being part of these researches .
unfortunately , g. pal died in 1992 .
this was surely affected the fact that these results were almost unrecognized .
|
in 2011 the nobel prize in physics was awarded for the 1998 discovery of the nonzero cosmological constant .
this discovery is very important and surely worth to receive the nobel prize .
however , years earlier several papers had been published ( pal , horvth , & lukcs 1992 ; holba et al .
1992 , holba et al .
1994 ) about a very similar discovery from observational data .
|
m86 ( ngc 4406 ) is a bright elliptical ( e3/s0 ) galaxy in the virgo cluster of galaxies .
it is the dominant member of one of the larger subgroups within virgo ( binggeli et al .
1993 ; bhringer et al . 1994 ; schindler et al .
its line - of - sight velocity relative to m87 , the dominant member of the virgo cluster , is -1550 km s@xmath6 , much higher than the average cluster velocity dispersion ( smith et al .
the x - ray surface brightness distribution is unusual , with a large `` plume '' extending to the northwest from m86 , which was first noticed as part of a survey of virgo cluster galaxies undertaken with the _ einstein observatory _ ( forman et al . 1979 ) .
the galaxy has an optical asymmetry that extends in a direction similar to the direction of the plume ( nulsen & carter 1987 ; mihos et al .
several authors have interpreted this plume as arising from ram pressure stripping due to strong interactions with the virgo icm ( forman et al .
1979 ; fabian et al . 1980
; takeda et al . 1984 ; knapp et al . 1989 ; bregman & roberts 1990 ; white et al . 1991 ; rangarajan et al . 1995 ) .
elmegreen et al . (
2000 ) find dust streamers in the core of m86 that connect to the nucleated dwarf galaxy vcc 882 , and suggest a recent interaction between the pair that may have contributed to the asymmetry of m86 s optical isophotes . at x - ray energies ,
m86 has been observed by
_ einstein _ ( forman et al.1979 ; white et al .
1991 ) , _ ginga _ ( takano et al .
1989 ) , _ exosat _ ( edge 1990 ) , _ asca _ ( matsushita et al .
1994 ) , and _ rosat _ ( bhringer et al . 1994 ; rangarajan et al.1995 ) .
recently , finoguenov et al . ( 2004 ) presented _ xmm - newton _ observations of m86 and concluded that its unusual morphology is due to an interaction with an x - ray filament rather than the virgo icm .
however , their conclusions were based on the large separation between m86 and m87 along the line of sight of @xmath7 mpc reported by neilsen & tsvetanov ( 2000 ) , which is inconsistent with the more recent result from mei et al . ( 2007 ) who find a separation of @xmath8 mpc .
we report here on a mosaic of nine _ chandra _ observations of m86 and the surrounding field , totaling 240 ksec of exposure .
the observations and data reduction techniques are described in [ sec : obs ] .
the x - ray image is presented in [ sec : img ] , and results on temperature and abundance structure from spectral analysis are given in
[ sec : spec ] . in
[ sec : discuss ] , we discuss some of the more interesting features of m86 , and place constraints on its orbit . in particular , we give new results on the extent of the stripped tail , and measure the density profile of an x - ray brightness edge seen to the southeast .
our results are summarized in [ sec : summary ] .
we assume a distance to the virgo cluster of 16 mpc throughout , consistent with the latest results from mei et al .
( 2007 ) , which gives a scale of 0.08 kpc/ for @xmath9 , @xmath10 , and @xmath11 km s@xmath6 mpc@xmath6 .
all error ranges are 90% confidence intervals , unless otherwise stated .
table [ tab : obs ] summarizes the nine _ chandra _ observations of m86 , which completely cover the center of the galaxy , as well as the plume and long tail .
all data were reprocessed from the level 1 events files using the latest calibration files ( as of ciao3.3 ) .
cti and time - dependent gain corrections were applied where applicable .
lc_clean was used to remove background flares .
after periods with obvious flares were removed , the mean rate was calculated for some observed quiescent time interval , and the program was re - run , forcing the global mean to equal this quiescent rate . for observations without obvious flares ,
time bins that were not within 3@xmath12 of the mean were discarded .
the final cleaned exposure times are given in column ( 5 ) of table [ tab : obs ] .
the emission from m86 and the surrounding virgo cluster fills the image field of view for each observation .
we therefore used the standard caldb blank sky background files appropriate for each observation , normalized to our observations in the 10 - 12 kev energy band .
we combined background maps for each pointing separately , and generated exposure maps to account for the pointing offsets and the different ccd responses . to generate exposure maps
, we assumed a mekal model with @xmath13 kev , galactic absorption , and abundance of 30% solar at a redshift @xmath14 , which is consistent with typical results from detailed spectral fits ( see [ sec : spec ] ) .
the exposure corrected , background subtracted , smoothed mosaic image is shown in figure [ fig : fullimg ] .
several patches of diffuse emission are seen in m86 , as is an impressive tail extending to the northwest . to enhance the visibility of the diffuse emission we created an image with bright point sources removed . for each
_ chandra _ pointing , regions containing point sources were `` filled in '' using a poisson distribution whose mean was equal to that of a local annular background region .
the resulting smoothed image ( figure [ fig : smoimg ] ) shows : * a long tail extending to the nw , first detected by forman et al.(1979 ) .
the tail is bifurcated at its base , with the brighter ( northern ) extension at a position angle of @xmath15 ( measured from north to east ) and the shorter extension at @xmath16 . at 18.4(88.4 kpc ) from the center of the bright emission at the base of the tail ( or 21.1 ( 101.5 kpc ) from the center of m86 ) , the tail turns directly north for 6.8 ( 32.5 kpc ) .
it then curves to the west at 26.7 ( 128.1 kpc ) from m86 , and finally is no longer detected at 30.6 ( 150 kpc ) .
there is evidence for a fainter , parallel tail along the northeastern edge of the tip of the bright tail , discussed below . * a large plume of emission , directly north of m86 , at the root of the long tail ( noted previously by forman et al .
the tail emanates from this clump , rather than from m86 itself as one would expect for a continuous stripping process . * a sharp boundary along the northeastern edge of the tail , and an apparent `` void '' in the diffuse emission just north of the tail .
this is in stark contrast to the southwestern edge of the tail , where the surface brightness falls off smoothly .
rangarajan et al .
( 1995 ) first noted this void , and suggested that it is an evacuated mach cone left from m86 s passage through the outer regions of the virgo icm .
* an extended halo of x - ray emission associated with m86 with an edge 13 ( 62.4 kpc ) to the southeast .
this edge also is visible in _ rosat _ observations ( see figure [ fig : rosat ] ) . *
an asymmetry in the diffuse emission near the core of m86 , with an extension to the south .
rangarajan et al .
( 1995 ) suggest that this is a high pressure region , formed as the leading edge of m86 plows into the virgo icm .
* several nearby galaxies .
the elliptical galaxy m84 , west of m86 , shows complicated structure in the diffuse emission in the core related to the radio outburst , and a tail extending to the south ( finoguenov & jones 2002 ) .
also visible are ngc 4388 ( south - southwest of m86 ; beckmann et al . 2004 ) and ngc 4438 ( east of m86 ; machacek et al . 2004 ; vollmer et al.2005 ) .
both also have diffuse tails , which likely indicate their projected direction of motion with respect to the virgo icm .
most tails split into two main streams .
a small patch of faint emission also can be seen in the region of the disk galaxy ngc 4402 ( crowl et al .
2005 ) . for comparison , the ( heavily binned ) _ chandra _ mosaic image is plotted alongside the _ rosat _ and _ dss _ images in figure [ fig : rosat ] .
the _ rosat _
image clearly shows the distribution of diffuse emission on larger scales . in particular , the contribution from the m87 halo is seen in the southeast , as is a large halo of diffuse emission around m86 itself , outside the brightest central regions .
these components are individually analyzed in our detailed spectral fits ( see [ sec : dspec ] ) .
results from these fits indicate that the extended m86 halo is composed of group gas associated with m86 , at a temperature of @xmath17 kev .
a close up view of the stripped tail , with scaling and binning chosen to show its structure more clearly , is presented in figure [ fig : tail ] .
the twisting and curving of the tail at the faint tip may be caused by turbulence or similar `` weather '' in the virgo icm .
the tail is clearly bifurcated at its base .
additionally , a fainter tail , split from the first , is barely visible along the northern edge of the faint end of the main tail . to test the significance of this feature
, we extracted the total count rate in evenly spaced bins across the width of the box shown in figure [ fig : tail ] .
figure [ fig : tailproj ] shows two distinct peaks . if we take an average count rate from the highest three bins in each peak and compare it with the average count rate from the lowest two bins in the gap , we find that the gap is significant at about 2.5@xmath12 . as described below in [ sec : displace ] , these double streams , as seen clearly at the base of the tail and possibly at the faint end of the tail , as well as in other virgo cluster galaxies in the field , may be explained as stripping from an inclined aspherical potential .
in summary , the x - ray images show three diffuse emission components : the virgo icm ( centered on m87 ) , the extended m86 halo , and the core and tail of m86 .
we generate a temperature map as a guide for detailed spectral fitting to disentangle the various components .
we assume a galactic absorption of @xmath18 @xmath19 throughout . allowing the absorption to vary did not significantly improve any of the spectral fits . the temperature map was derived using a method employed by osullivan et al .
( 2005 ) and maughan et al .
for each temperature map pixel ( 19.7/pix ) , we extracted a spectrum from a circular region containing 1000 net counts ( after subtracting the blank sky background and a component to account for emission from m87 , see [ sec : dspec ] ) , up to a maximum radius of 3.3 ( on the order of the chip size ) .
this large maximum radius was used to obtain sufficient source counts to extend the temperature map to faint regions of the tail .
the resulting spectrum was fit in the 0.6 2.0 kev range with an absorbed apec model using xspec .
data from back - side- ( bi ) and front - side - illuminated ( fi ) ccds were treated as separate data groups for each observation . for each fit ,
the abundance was fixed at 26% solar ( consistent with most of the detailed spectral fits ; see table [ tab : spectra ] ) .
the resulting temperature map is shown in figure [ fig : tmap ] ( middle ) .
regions that did not have at least 1000 source counts within the maximum 3.3 radius were excluded from the map , as were regions in the southeast that were dominated by virgo cluster emission . for comparison
, we also show the tessellated temperature map in figure [ fig : tmap ] .
each bin was fit using only counts from that area , so the extraction regions are well - defined .
the bins were generated using the algorithm provided by diehl & statler ( 2006 ) , which is a generalization of cappellari & copin s ( 2003 ) voronoi binning algorithm , and requiring roughly 1100 net counts per bin .
each bin was fit with a single apec model , without a component to model the virgo icm background , in contrast to the `` smoothed '' temperature map .
the size of the bins roughly indicate the size of the extraction regions for the smoothed temperature map pixels in the same area .
the tessellated and smoothed temperature maps are in good agreement , except in the regions far from m86 where the contribution from the virgo cluster is more important . in these temperature maps , the remarkably sharp boundary between the cool galaxy gas and that of the cluster on the northern edge of the tail is apparent ( figure [ fig : tmap ] ) .
at least four distinct cold clumps can be seen : one centered on m86 , one 1.41 ( 6.8 kpc ) south of m86 , one elongated clump 2.45 ( 11.7 kpc ) east of m86 , and a larger cool plume 3.63 ( 17.4 kpc ) to the north . each clump corresponds to a bright peak in the x - ray image .
additionally , the cooler tail extends to the northwest . to the southeast , in the region of the brightness edge
discussed in
[ sec : halo ] , the gas is somewhat hotter ( @xmath20 kev ) in the southern half of this feature than in the northeastern half at the same distance from m86 .
this temperature difference is detected in spectra from non - overlapping regions in these two areas , indicating that the difference is not an artifact of the technique used to generate the temperature map .
more detailed fits are discussed in [ sec : dspec ] . based on the derived temperature map and the x - ray image , we defined 23 regions for detailed spectral analysis ( see figure [ fig : regmap ] ) . to include data from a particular observation in the spectral analysis
, we required that it contribute at least 200 counts ( each data group was binned such that there were at least 40 counts per bin ) .
a summary of the regions and spectral model for each region are given in table [ tab : spectra ] .
r1 - r4 are centered on the individual clumps of emission near m86 , while r5 -r10 trace the long tail nw of m86 ( hereafter regions will be labeled as r# , with the number defined by column 1 of table [ tab : spectra ] ) .
r11 roughly corresponds to the `` shocked region '' indicated by finoguenov et al .
( 2004 ; see their figure 7 ) .
there are several regions to either side of the m86 tail .
r17 , r18 , r19 , & r23 measure the large scale diffuse components from m86 and m87 , outside the central region where the cool gas is observed ( note that r23 is a combination of five disjoint regions , see figure [ fig : regmap ] ) . finally , r14 - r16 contain m84 and its tail of diffuse emission .
each region was initially fit with an absorbed apec model over the 0.6 - 5.0 kev range , with an additional apec component to model emission from the virgo icm as in [ sec : tmap ] , with the abundance allowed to vary .
additional components were added , as needed , to improve the fits ( see table [ tab : spectra ] .
for example , a single temperature model gave large residuals at high energies when fit to the regions corresponding to the cores of m86 and m84 ( r1 and r14 , respectively ) .
since each of these regions is expected to have a contribution from unresolved x - ray point sources , a power law component was added to the model , which greatly improved the fits ( and gave a photon index consistent with @xmath21 , as expected , see e.g. sarazin et al .
2003 ) . for regions where this power law component could not be well - constrained , we fixed the photon index at @xmath22 .
nine regions showed an improved fit when another apec model was added ( in such cases the abundances of each component were constrained to vary together ) .
data from each pointing were treated as separate data groups , with normalizations for each pointing allowed to vary independently . as an example , the two temperature plus power law fit to the spectra for the core of m86 ( r1 ) is shown in figure [ fig : m86spec ] . each of the three main components of the diffuse emission ( the virgo icm , m86 group gas , and the cooler central m86 gas ) was detectable in some subset of the regions shown in figure [ fig : regmap ] . in r13
it was possible to fit all three components simultaneously .
the virgo and m86 halos cover most , if not all of the fov .
it was therefore necessary to include these components when doing detailed spectral fits , even in regions where they could not be modeled directly due to inadequate statistics . to account for the varying virgo contribution
, we took the normalizations of the best fit apec models to r18 and r23 and linearly interpolated across the field based on the distance from the center of m87 .
although this only gives a rough estimate , we note that it does not affect our results significantly . for regions closer to m87 ( e.g. , r19 )
the interpolation gives a value very close to that measured from r18 , which is an appropriate local background .
similarly , regions near the faint end of the m86 tail ( e.g. , r10 ) have normalizations close to that of r23 , which again , is an appropriate local background for this area .
for most regions of interest , m86 is bright enough that the contribution from virgo is negligible , and the accuracy of interpolating is therefore sufficient .
a similar approach was used to model the contribution from the halo around m86 .
although , in general , the normalization of this component decreased with distance from the core of m86 , the variation across the field was more complicated than for virgo , which is expected since this gas is in the process of being stripped from m86 .
we included a fixed component , with @xmath23 kev , abundance set equal to that of the main component , and variable normalization , in regions where the m86 halo temperature could not be determined directly ( see column 2 of table [ tab : spectra ] ) .
the effects of accounting for this component can be seen , for example , by comparing lines 1 & 2 in table [ tab : spectra ] . including the hotter 1.2 kev component results in a lower best - fit temperature for the cooler gas , and gives a better statistical fit to the data . in general , our results are consistent with finoguenov et al.(2004 ) , who derive x - ray temperatures and abundances for the central region , including m86 and the plume from _ xmm - newton _ observations .
we note that for the cores of m86 and m84 ( r1 and r14 in table [ tab : spectra ] ) the abundances are poorly determined , despite the large number of net counts . using a vapec model to allow for non - solar abundance ratios did not improve the fits
. the failure of these models may be due to abundance variations within the defined regions .
for instance , the abundance map given by finoguenov et al .
( 2004 ) shows significant variation near the core of m86 , even on relatively small scales .
additionally , the models do not allow for multiphase gas , although the temperature most likely varies significantly in the cores . using the detailed spectra
, we can derive the approximate mass in cool gas for the base of the tail ( the `` plume '' ) and the core of m86 .
using the temperature and abundance measured in r4 , and assuming a prolate spheroid geometry with major and minor axis lengths of 12.3 kpc and 7.9 kpc , we find a density of @xmath24 @xmath25 and a mass @xmath26 .
similarly , in the core of m86 , we find @xmath27 @xmath25 and @xmath28 within a sphere of radius 9.6 kpc . assuming a cylindrical geometry for the tail , and considering only the cool gas components in r5 - r9
, we find a typical density of @xmath29 @xmath25 and a total gas mass of @xmath30 .
the mass in stripped gas is therefore more than three times the gas mass in the core , which supports the scenario where a significant fraction of the m86 ism is removed rapidly due to ram pressure stripping ( takeda et al . 1984 ) .
rangarajan et al .
( 1995 ) find similar results for the plume and the core .
however , we find consistent abundance values in the core and in the plume within the errors , while rangarajan et al .
( 1995 ) find a higher abundance in the plume .
this disagreement is likely due to the multiphase model we use , which includes a separate component for the 1.2 kev m86 halo emission .
the most striking feature in figure [ fig : smoimg ] is the long tail of emission extending to the nw from m86 .
m86 has a line - of - sight velocity of -244@xmath31 km s@xmath6 , while m87 s is 1307@xmath32 km s@xmath6 ( smith et al .
2000 ) . therefore , m86 is traversing the virgo cluster at @xmath33 km s@xmath6 ( about mach 2 for @xmath34 kev ) .
the tail naturally forms due to ram pressure stripping of the m86 corona by the virgo cluster icm .
finoguenov et al .
( 2004 ) suggested that the tail formed due to interactions with a filament rather than the virgo icm , though their result was based on older distance measurements that placed m86 outside the virial radius of the virgo cluster .
if we assume that m86 is bound to the virgo cluster , its large line - of - sight relative velocity allows us to constrain the separation between m86 and m87 . using the m87 mass profile detailed in
[ sec : orbit ] , we find that a free - fall velocity of 1500 km s@xmath6 corresponds to a separation from m87 of 0.5 mpc ( the projected separation is 0.35 mpc ) .
this is an upper - limit , since the total relative velocity must be at least as large as the line - of - sight relative velocity .
this separation is consistent with recent surface brightness fluctuation distances of virgo cluster galaxies , which give a distance between m86 and m87 of @xmath35 mpc ( mei et al.2007 ) .
we also can place a lower - limit on the length of the long stripped tail using the mass profile of m87 and the line - of - sight velocity .
the maximum free - fall velocity of m86 from infinity at the projected separation of 0.35 mpc is about 1680 km s@xmath6 . using this as an upper - limit on its current 3d velocity , we find that the angle between the direction of motion of m86 and the line - of - sight is @xmath36 . assuming that the stripped tail is aligned with m86 s current direction of motion , and given that the length of the tail in the plane of the sky is 150 kpc ( 0.51@xmath37 )
, we find a lower - limit on the actual length of the tail of @xmath38kpc , making this the longest ram pressure stripped tail presently known .
the long stripped tail originates in the plume of diffuse emission located directly north of m86 . for a continuous stripping process
, one would expect the tail to extend from m86 itself .
this separation between m86 and the plume can be explained by rapid recent stripping , in which a significant fraction of the remaining gas in m86 is rapidly stripped when the ram pressure stripping condition is met .
a detailed discussion of this process is given in
[ sec : displace ] as seen from figure [ fig : regmap ] and table [ tab : spectra ] , the stripped tail shows a general trend of cooler gas in and near the plume ( with temperatures in the 0.80.85 kev range ) and warmer gas at the tip ( in the 0.91.2 kev range ) .
the temperature structure of the tail is consistent with a ram pressure stripping model , where , as m86 falls into the virgo cluster , the hotter , higher entropy group gas is stripped first due to interactions with the virgo icm , followed by the cooler , lower entropy m86 ism , which is removed rapidly once the stripping condition is met .
the _ chandra _ and _ rosat _ x - ray images ( figure [ fig : rosat ] ) show an extended halo associated with m86 , with a sharp brightness edge to the southeast in the direction of m87 . since m86 is traversing the virgo cluster supersonically ( see [ sec : tail ] ) we expect a shock to be driven in the virgo icm , possibly producing a brightness edge similar to that seen in the southeast .
we attempted to measure the density jump associated with the shock in the following way .
we extracted the _
chandra _ 0.52.0 kev surface brightness profile of this edge in two roughly equal sectors since the morphology of this feature is irregular in the south ( see figure [ fig : rosat ] ) , and since the gas temperature differs in these two regions ( see [ sec : tmap ] ) .
the emission measure profile for the northern sector is shown in figure [ fig : nedge ] .
distance is measured from the center of curvature of the apparent edge , which is 10 kpc ( 2.16 ) east of m86 .
this profile was fit with a spherical gas density model consisting of two power laws .
the free parameters were the normalization , the inner ( @xmath39 ) and outer ( @xmath40 ) slopes , the position of the density discontinuity ( @xmath41 ) , and the amplitude of the jump ( @xmath42 ) .
we assumed that the gas is isothermal with @xmath13 kev and that the abundance is constant at 30% solar , consistent with results from spectral fits ( see [ sec : spec ] ) .
for the best fit model ( see figure [ fig : nedge ] ) , we find @xmath43 , @xmath44 , @xmath45 kpc , and @xmath46 .
similar results are found from the southern sector , but with larger errors .
the lack of a well - defined edge is consistent with our findings for the orbit of m86 ( see [ sec : orbit ] ) . in particular , for all of the likely orbits , m86 is moving to the southeast , such that our lines of sight pass through the mach cone . to see a sharp edge ,
our line of sight must be tangent to the shock front .
knowing the orbit of m86 is the key to understanding its interaction with the virgo cluster .
the ram pressure stripped tail reveals the motion of the galaxy on the sky .
the length and direction of the tail , together with the large line - of - sight speed of m86 , constrain its orbit .
the large luminosity of m86 suggests that it dominates the associated in - falling subgroup ( schindler et al .
1999 ) . in the following ,
we assume m86 is bound to the virgo cluster .
for the purpose of calculation , the gravitational potential of the virgo cluster is treated as a spherical nfw potential ( navarro , frenk & white 1997 ) , with a virial radius of 1.3 mpc ( evrard , metzler & navarro 1996 ) and a concentration parameter of 4.5 ( neto et al.2007 ) , appropriate for a @xmath47 kev cluster .
the gravitating mass was normalized to match the total mass within 320 kpc of @xmath48 ( schindler et al .
1999 ; scaled to a distance of 16 mpc ) . for this potential , the escape speed from a radius of 351 kpc ( the projected separation of m86 from m87 ) is 1677 @xmath3 , not much larger than m86 s line - of - sight speed of 1550 @xmath3 relative to m87 .
the empirical mass distribution of schindler ( 1999 ) gives the same escape speed at 351 kpc as our model potential .
we note that the gravitational potential of the virgo cluster is not well constrained at distances from the cluster center comparable to or greater than its virial radius .
furthermore , the truncated nfw potential is highly simplified , ignoring mass beyond the virial radius , departures from spherical symmetry , and the dynamic state of the cluster due to continuing in - fall .
since some of the limits we derive here are sensitive to the poorly known potential at large radii , they should be treated as indicative rather than quantitative ( outside the context of the model ) .
the material in this section is supplemented by a more detailed discussion in appendix [ ap : orbit ] .
we consider radial orbits first , as suggested by previous studies ( e.g. , forman et al .
1979 ; white et al . 1991 ) . in order for the gas tail to point away from the cluster center
, the galaxy must be inbound .
it can easily be shown that the line - of sight velocity , @xmath49 , is maximized for some radius @xmath50 greater than the observed separation @xmath51 ( see appendix [ ap : orbit ] ) . for a marginally bound ( zero energy ) radial orbit in the potential described above
, the maximum value of @xmath52 occurs when m86 is close to twice its projected distance from the cluster center , giving @xmath53 , less than the observed value of @xmath54 . thus ,
unless m86 is significantly unbound from the virgo cluster ( or the potential is incorrect ) , its line - of - sight speed is inconsistent with radial and nearly radial orbits .
it is convenient to specify more general orbits in terms of their inner and outer turning radii , @xmath55 and @xmath56 , respectively .
we constrain the possible orbits of m86 by placing limits on these parameters .
coarse limits can be placed on these radii by considering energy arguments alone .
assuming that m86 is bound to the virgo cluster , we find @xmath57 kpc and @xmath58 mpc , or @xmath59 times the virial radius of the virgo cluster ( see appendix [ ap : orbit ] ) .
these limits can be further restricted by considering the range of possible viewing directions for each point on an orbit .
there can be zero , two , or four possible viewing directions that would place m86 at the observed separation from the cluster and give it the observed line - of - sight velocity ( see appendix [ ap : orbit ] ) .
figure [ fig : mb ] shows a range of marginally bound orbits ( @xmath60 ) , with regions that meet these conditions marked in color .
the range of an orbit where these conditions are met shrinks as @xmath56 decreases , i.e. , as the orbit becomes more tightly bound . for the marginally bound orbits , the full range of @xmath55 for which @xmath49 can attain its observed value is @xmath61 kpc ( the range represented in figure [ fig : mb ] ) . for more tightly bound orbits , the acceptable range of @xmath55
is reduced .
the lower limit on @xmath56 is increased if we consider the extent of the ram pressure stripped tail .
the projected orbit must extend to at least the distance ( @xmath62 kpc ) that the tail projects beyond m86 in the direction away from the cluster center .
locations on the orbits where the line - of - sight speed can attain @xmath63 and these conditions also are met are shown in red in figure [ fig : mb ] .
we see that the range of possible locations for m86 on these orbits is tightly constrained .
we repeated this analysis for a number of values for the outer turning radius .
the range of potential orbits and locations for m86 shrinks with decreasing @xmath56 and no orbits were found to meet these criteria for @xmath64 mpc ( 6.3 virial radii ) .
figure [ fig : projmb ] shows projections onto the sky of the marginally bound orbits ( corresponding to the midpoints of the red regions of figure [ fig : mb ] ) overlaid on the _ chandra _ 0.5 2 kev image , with m86 at the representative location for each orbit .
it is evident that the possible orbits for m86 can be pruned further .
for example , the orbit on the lower right in figure [ fig : projmb ] does not reach far enough north to produce the remote part of the gas tail . at the other limit , the position angle of m87 measured from m86 is @xmath65 ( east from north ) , generally eastward of these orbits .
overdense stripped gas tends to fall towards the cluster center , in that direction , so that some stripped gas can reasonably lie to the east of the orbit .
stripped gas can not lie west of the orbit .
thus the orbits that cross the region to the east of the tail and north of m86 , where there is no sign of stripped gas , are unlikely candidates for the orbit of m86 .
note that the orbits in figure [ fig : projmb ] are not simply related to their inner turning radii .
the orbit corresponding to the smallest value of the inner turning radius appears second from the right at the top of the figure . at first , orbits for increasing values of @xmath55
lie to the left of this , but at a value of @xmath55 approaching ( but less than ) @xmath51 , the acceptable viewing direction flips from outward to inward ( i.e. , the acceptable sign of @xmath66 changes , see appendix [ ap : orbit ] ) . at this point , the projected position of the orbit shifts from leftmost to rightmost at the top of figure [ fig : projmb ] . whereas the apparent location where the orbit passes through the virial radius was moving outward
, it now moves inward as @xmath55 increases , coming into the field of view for the largest values of @xmath55 here .
this results in disjoint ranges of possible orbits for m86 . thus the marginally bound orbits with
inner turning radii close to @xmath67 kpc and those with inner turning radii in the range @xmath68 kpc are best suited to model the orbit of m86 .
applying the same criteria to the more tightly bound orbits yields the collection of possible orbits shown in figure [ fig : m86orb ] .
we have not attempted to be highly selective or exhaustive , since that requires a well defined model for the formation of the gas trail . for values of the outer turning radius in the lower end of the acceptable range ,
the shrinking range of possible orbits and locations excludes orbits with inner turning radii @xmath69 .
all of the acceptable orbits for m86 are weakly bound to the virgo cluster , with the most tightly bound having outer turning radii at @xmath70 mpc .
all of the possible locations for m86 lie only a little farther from m87 than m86 does in projection .
all are close to the plane of the sky , ranging from 167 kpc closer than m87 to 263 kpc farther than m87 from the sun at the extremes .
m86 must also be close to the pericenter of its orbit .
the direction of motion of m86 is close to our line of sight , within @xmath71 @xmath72 of it , for all of the cases illustrated .
m86 is traversing the virgo cluster supersonically ( @xmath73 @xmath3 alone is almost twice the sound speed @xmath74 , which is taken to be 850 @xmath3 ) .
therefore , we expect a shock to be driven in the virgo icm . for the orbits
we derive , the angle of inclination is smaller than the opening angle of the mach cone ( @xmath75 ) and we should not expect shock fronts to be visible on the plane of the sky ( cf .
rangarajan et al .
this is illustrated in figure [ fig : mach ] , which shows the marginally bound orbit with @xmath76 kpc .
each circle is drawn centered on a point where m86 was at a time @xmath77 in the past , with a radius of @xmath78 .
this is done at equally spaced times to indicate the shape of the mach cone .
the absence of caustics in figure [ fig : mach ] shows that we would not see the shock front in projection .
( while this statement is accurate , except possibly in the small region where the shock front is highly supersonic , the diagram is schematic .
for example , if m86 were coming directly towards the earth , radii of sections of the mach cone would be @xmath79 larger than drawn . )
our lines of sight through the compressed gas behind the shock are longest where the circles pile up to the southeast of m86 , consistent with the enhancement in x - ray surface brightness seen @xmath80 southeast of m86 ( figure [ fig : mach ] ) .
the prominent plume of gas lying 3 4(@xmath81 kpc ) north of m86 , noted previously ( e.g. , forman et al .
1979 ) , appears to be physically separated from the dense gas remaining in the galaxy and it is centered some distance east of the orbits illustrated in figure [ fig : m86orb ] .
the removal of a significant fraction of the ism in a single blob is expected from rapid ram pressure stripping ( takeda et al .
1984 ) . a highly simplified model for ram pressure stripping
treats the plume as a single particle subject to gravity and drag due to its motion through the surrounding gas . allowing for buoyancy , the net force on the plume due to gravity is @xmath82 , where @xmath83 is its density
, @xmath84 is its volume , @xmath85 is the density of the ambient icm , and @xmath86 is the local acceleration due to gravity .
the drag force on the plume is @xmath87 , where @xmath88 is the drag coefficient , @xmath42 is the plume cross section , @xmath89 is its velocity , and @xmath90 .
thus , the net acceleration of the plume is @xmath91 { { \bf g}}- [ { c_{\rm d}}{\rho_{\rm e}}a / ( \rho v ) ] v { { \bf v}}$ ] .
two of the main parameters of this model are the density contrast , @xmath92 , and the factor @xmath93 . for the density of the icm , @xmath85 ,
we use the beta model of schindler et al .
( 1999 ; @xmath94 , core radius @xmath95 ) . on the grounds that the density of the plume is likely to be lower now ( due to its stripping and ejection from the confining potential of m86 ) , we set its density contrast at the projected radius of m86 to be 18 ( cf .
@xmath81 from the numbers above , see
[ sec : dspec ] ) and treat the density of the plume as constant .
the other factor is determined by treating the plume as a sphere with a constant radius of 10 kpc , with @xmath96 .
as above , this radius is a little larger than the observed radius . to simulate stripping ,
a gas blob is placed at the location and velocity of m86 at the time it is released from the cluster virial radius .
the orbit of the blob is followed , subject to the gravitational acceleration of the cluster and the moving galaxy .
this model can not account for the current location of the plume if the gravitational potential of m86 is spherical .
the path of the blob is determined by the competition between gravity and drag .
if the drag is large , the blob is ejected from m86 early on its orbit , and slows quickly until it is falling towards the cluster center at its terminal speed , and ultimately lies farther away from m86 than observed .
its early ejection also leaves it too far back along the orbit of m86 . reducing the relative significance of the drag causes the blob to be ejected later , bringing it closer to its observed position but not far enough from the orbit in the direction of the cluster center .
this issue can be resolved if the gravitational potential of m86 is aspherical .
consider a small gas blob in an aspherical galaxy moving at an inclined angle through the icm , illustrated schematically in figure [ fig : inclined ] .
while ram pressure is insufficient to eject the blob , it is driven to an equilibrium position where ram pressure is balanced by gravity .
since the direction of the gravitational field must oppose the drag , the external flow must be perpendicular to the equipotential surface .
the equilibrium position therefore lies at a location away from the axis passing through the center of the galaxy and parallel to the flow ( e.g. , near the point labeled `` equipotential '' in figure [ fig : inclined ] ) .
as the drag starts to overwhelm gravity and displaces the blob in the direction of the flow , the equipotentials the blob encounters tilt , so that it is subject to a component of the gravitational force directed away from the axis .
because forces in the direction of the flow are nearly balanced , this off - axis component of gravity drives the blob farther away from the axis of the flow .
thus , a blob stripped from such a galaxy tends to emerge away from the axis of the flow . treating the gas as a fluid , interstellar gas that is pushed inward by ram pressure along the leading edge of the galaxy flows out preferentially along the long axis of the potential , in the direction of the weakest gravitational force .
the model for the aspherical gravitational potential of m86 used here has the form @xmath97 , where @xmath98 / w$ ] is a minor modification of the standard nfw form ( @xmath99 / w$ ] in the same notation ) .
the extra factor of @xmath100 , where @xmath101 is the concentration parameter , makes the total mass converge , avoiding the need to truncate the mass distribution at the virial radius ( which would create problems for our fourth order integrator ) .
the coordinate @xmath102 , where @xmath103 is the nfw potential scale length and @xmath104 is chosen to give the desired ellipticity .
the additional 1 under the square root flattens the potential at small @xmath50 , providing a better model for the gravitational force on an extended gas blob ( of size comparable to @xmath103 ) when it is close to the center of the galaxy .
this modification also makes the integrator behave better near the center of the galaxy .
the virial radius of m86 was set to 100 kpc and its concentration parameter to 8 , roughly the values expected for a massive galaxy .
the normalizing factor , @xmath105 , was expressed as @xmath106 , so that @xmath12 is a rough measure of the line - of - sight velocity dispersion for m86 .
the value used below is @xmath107 .
we note that for the model what matters is the ratio of the drag force on the blob to the binding force of the m86 potential , allowing for a trade off between the choice of potential and the density ratio and size of the blob . as a result ,
the choice of potential is not very critical .
the two remaining parameters are the axial ratio , @xmath108 , and the orientation of m86 .
consistent with its s0/e3 classification , optical images show that m86 is highly flattened on scales comparable to the projected distance of the plume ( e.g. @xmath109 for the isophote with @xmath110 in the deep image of nulsen & carter 1987 ) . based on the discussion above , the potential of m86 would need to have a long axis pointing roughly northward from our line - of - sight .
therefore , we assume it is oblate , using @xmath111 .
the orientation of m86 is then determined by the direction of its minor axis . from the optical images ,
this has a position angle of @xmath112 on the sky .
however , its tilt with respect to the plane of the sky is unknown . we have set the minor axis to point towards us in the north , at an angle of @xmath113 from the plane of the sky
this is roughly the orientation that maximizes the transverse displacement of the blob .
figure [ fig : blob ] shows the path of the blob on the sky for this set of model parameters for the orbit with @xmath56 = 9.1 mpc and @xmath55 = 314 kpc . in this model ,
the blob is currently slower than m86 by 355 @xmath3along our line of sight and trails it by 38 kpc . around the edges of the gas halo in m86 , where the shear in the external flow is strong ( figure [ fig : inclined ] ) ,
if the effective viscosity is high , viscous stresses pull the interstellar gas out of the galaxy . alternatively ,
if the viscosity is low , shear instabilities mix the interstellar gas with the icm , also stripping it from the galaxy ( nulsen 1982 ) .
this stripping is aided by the low pressure due to the bernoulli effect around the edges of the galaxy , which tends to pull gas into the path of the flow .
this process works around all edges of the inclined galaxy , as seen from the direction of the flow .
however , it is expected to be greatest at the leading and trailing edges .
the shear in the external flow is expected to be greatest near the leading edge , favoring stripping there .
the large ram pressure at the leading edge displaces the edge of the interstellar gas deeper into the potential of the galaxy , pushing it towards the trailing edge of the galaxy , where the pressure is lower .
thus , gas at the trailing edge sits higher in the gravitational potential of the galaxy , favoring its removal from the galaxy . this may account for the apparent double streams of gas seen trailing m86 and several other galaxies in the composite image .
`` viscous '' stripping from the main body of m86 as well as the plume can explain the broad features of the gas trail , though modeling the finer features , such as the smaller blobs of gas lying to the east of the galaxy , requires a more detailed treatment of the gas dynamics ( e.g. , numerical simulations ) and is beyond the scope of this paper .
the proton mean free path due to coulomb collisions is given approximately by @xmath114 pc , where @xmath115 is the gas temperature in kev and the electron density is @xmath116 . in terms of this
, the reynolds number is @xmath117 , where @xmath118 is the flow speed , @xmath51 is the sound speed , and @xmath119 is a relevant length scale . taking @xmath120 and @xmath121
kpc gives @xmath122 in the 2.4 kev icm , but @xmath123 for the gas in the 0.77 kev plume ( and interstellar gas ) if it is exposed directly to the external flow .
these values suggest that the external flow can be largely laminar , while the flow in the cool interstellar gas is relatively turbulent . further complicating matters , the effective mean free path may be significantly smaller than the coulomb mean free path ( e.g. , schekochinin et al .
the reynolds number for this flow is therefore not well defined , due to the range in gas temperature and uncertainty in the mean free path .
we have argued that the plume and long tail of m86 formed due to ram pressure stripping forces generated as m86 falls into the virgo cluster .
several studies have found a similar interpretation for the formation of this feature ( e.g. , forman et al .
1979 ; fabian et al.1980 ; rangarajan et al .
1995 , however see bregman & roberts 1990 ; finoguenov et al .
we concentrate on these main results : * the plume and long tail observed in the diffuse emission are created by ram pressure stripping as m86 falls into the virgo cluster .
the tail is 150 kpc in projection ( a simple estimate , which assumes free - fall velocity for m86 and an nfw potential for m87 , gives a lower - limt on the true length of the tail of 380 kpc ) , making this the longest ram pressure stripped tail presently known . *
based on the x - ray spectra , we detect three distinct components associated with the m86/virgo cluster system : the virgo icm , with @xmath124 kev ; the extended halo of m86 , with @xmath125 kev ; and the cooler central and stripped gas of m86 , with @xmath126 kev .
the temperature structure of the tail is consistent with ram pressure stripping , where the higher entropy m86 halo gas is stripped first and deposited in the tip of the tail , and the lower entropy m86 ism is stripped more recently , constituting the base of the tail and the plume . * the large line - of - sight velocity of m86 , and its position relative to the virgo cluster ,
tightly constrain its orbit , especially if it is assumed that the gas tail traces the orbit . in particular , the observations are inconsistent with a radial orbit .
we show that m86 is at best only marginally bound to the virgo cluster , with an inner turning radius on the order of 300 kpc as expected from its recent in - fall .
our best - fitting orbital model requires that m86 be close to m87 , less than 167 kpc closer than or 263 kpc farther than m87 along our line of sight , which is consistent with the most recent distance estimates based on surface brightness fluctuations ( mei et al .
2007 ) which give a line - of - sight separation of @xmath127 mpc . *
the prominent plume of gas lying 3 4north of m86 appears to have been rapidly driven from m86 by ram pressure stripping .
the projected position of the plume , which does not lie directly on our best - fit model orbit for m86 , can be understood if m86 has an aspherical potential ( as suggested by optical isophotes ) . if m86 moves through the virgo icm at an inclination angle relative to the local flow , the gas at the trailing edge is more easily stripped , thereby displacing the gas from the nominal orbit of m86 itself .
this model may also explain the apparent double streams of gas seen trailing m86 , as well as those in other virgo galaxies .
* the apparent brightness edge to the southeast seen in _ rosat _
observations is also seen in the _ chandra _ images .
the edge is well fit with a two power law gas density model , with an abrupt jump in density by a factor of @xmath128 at the edge ( consistent with no jump ) . assuming that this brightness edge is the shock generated as m86 supersonically falls into the virgo cluster , the lack of a well - defined density jump is consistent with what is expected from our orbital model , which suggests that the orientation of the mach cone would make it difficult to detect the shock edge .
the financial support for this work was provided by nasa contracts nas8 - 39073 , nas8 - 38248 , nas8 - 01130 , nas8 - 03060 , the chandra science center , and the smithsonian institution .
beckmann , v. , gehrels , n. , favre , p. , walter , r. , courvoisier , t. j .-
l . , petrucci , p .- o .
, & malzac , j. 2004 , apj , 614 , 641 binggeli , b. , popescu , c. c. , & tammann , g. a. 1993 , a&as , 98 , 297 birzan , l. , rafferty , d. a. , mcnamara , b. r. , wise , m. w. , & nulsen , p. e. j. 2004 , apj , 607 , 800 bhringer , h. , briel , u. g. , schwarz r. a. , voges , w. , hartner , g. , & trmper , j. 1994 , nature , 368 , 828 bregman , j. n. , & roberts , m. s. 1990 , apj , 362 , 468 cappellari , m. , & copin , y. 2003 , mnras , 342 , 345 crowl , h. h. , kenney , j. d. p. , van gorkom , j. h. , & vollmer , b. 2005 , aj , 130 , 65 diehl , s. , & statler , t. s. 2006 , mnras , 368 , 497 edge , a. c. 1990 , phd thesis , univ .
leicester fabian , a. c. , schwarz , j. , & forman , w. 1980 , mnras , 192 , 135 finoguenov , a. , & jones , c. 2002 , apj , 574 , 74 finoguenov , a. , pietsch , w. , aschenbach , b. , & miniati , f. 2004 , a&a , 415 , 415 forman , w. , schwarz , j. , jones , c. , liller , w. , & fabian , a. c. 1979 , apj , 234 , l27 machacek , m. e. , jones , c. , & forman , w. r. 2004 , apj , 610 , 183 matsushita , k. et al .
1994 , apjl , 436 , 41 maughan , b. j. , ellis , s. c. , jones , l. r. , mason , k. o. , crdova , f. a. , & priedhorsky , w. 2006 , apj , 640 , 219 mei , s. , et al .
2007 , apj , 655 , 144 mihos , c. , harding , p. , feldmeier , j. , & morrison , h. 2005 ; apjl , 631 , 41 navarro , j. s. , frenk , c. s. , & white , s. d. m. 1995 , mnras , 275 , 720 neilsen , e. h. , & tsvetanov , z. i. 2000 , apj , 536 , 255 neto et al .
2007 , mnras , 381 , 1450 nulsen , p. e. j. 1982 , mnras , 198 , 1007 nulsen ,
p. e. & carter , d. 1987 , mnras , 225 , 939 osullivan , e. , vrtilek , j. m. , kempner , j. c. , david , l. p. , & houck , j. c. 2005 , mnras , 357 , 1134 rangarajan , f. v. n. , white , d. a. , ebeling , h. , & fabian , a. c. 1995 , mnras , 277 , 1047 sarazin , c. l. , kundu , a. , irwin , j. a. , sivakoff , g. r. , blanton , e. l. , & randall , s. w. 2003 , apj , 595 , 743 schekochinin , a. a. , cowley , s. c. , r. m. kulsrud r. m. , rosin , m. s. , & heinemann , t. 2007 , preprint ( astro - ph/0709.3828 ) schindler , s. , binggeli , b. , & bhringer , h. 1999 , a&a , 343 , 420 smith , r. j. , lucey , j. r. , hudson , m. j. , schlegel , d.j . , & davies .
r. l. 2000 , mnras , 313 , 469 takano , s. , awaki , h. , koyama , k. , kunieda , h. , tawara , y. 1989 , nautre , 340 , 289 takeda , h. , nulsen , p. e. j. , & fabian , a. c. 1984 , mnras , 208 , 261 vollmer , b. , braine , j. , combes , f. , sofue , y. 2005 , a&a , 441 , 473 white , d. a. , fabian , a. c. , forman , w. , jones , c. , & stern , c. 1991 , apj , 375 , 35 lcccc 318&2000 - 04 - 07&m86 & s3 , s2 , i2 , i3&12.964 + 803&2000 - 05 - 19&m84 & s3 , s2 , i2 , i3&26.699 + 963&2000 - 04 - 07&m86 & s3 , s1 , s2 , i2 , i3&13.220 + 1619&2001 - 06 - 08&ngc4388 & s3 , s2 , i2 , i3&19.705 + 2882&2002 - 01 - 29&ngc4438 & s3 , s2 , i2 , i3&24.891 + 5908&2005 - 05 - 01&m84 & s3 , s1 , s2 , i2 , i3&36.299 + 5912&2005 - 03 - 09&se of m86 & i0 , i1 , i2 , i3 , s2&31.370 + 5913&2005 - 03 - 19&m86 tail tip & i0 , i1 , i2 , i3 , s2&34.744 + 6131&2005 - 11 - 07&m84 & s3 , s1 , s2 , i2 , i3&37.855 + lccccc 1&0.724@xmath129//(2.4)&0.42@xmath130&1.71@xmath131&108/91=1.18&6235 + 1&0.677@xmath132/1.291@xmath133/(2.4)&0.64@xmath134&(1.5)&90/89=1.01&6235 + 2&0.697@xmath135/(1.2)/(2.4)&0.31@xmath136&&56/57=0.98&3086 + 3&0.653@xmath137/(1.2)/(2.4)&0.50@xmath138&(1.5)&88/83=1.06&4273 + 4&0.773@xmath139/(1.2)/(2.4)&0.47@xmath140&&109/84=1.29&8630 + 5&0.854@xmath141/(1.2)/(2.4)&0.40@xmath142&&94/93=1.01&4461 + 6&0.849@xmath143/(1.2)/(2.4)&0.31@xmath144&&89/88=1.01&5059 + 7&0.860@xmath145/(1.2)/(2.4)&0.43@xmath146&&87/74=1.18&2843 + 8&0.785@xmath147/(1.2)/(2.4)&0.45@xmath148&&44/40=1.09&1749 + 9&0.924@xmath149/(1.2)/(2.4)&0.50@xmath150&&30/29=1.03&1043 + 10&1.191@xmath151/(2.4)&0.42@xmath152&&20/25=0.73&739 + 11&0.865@xmath153/1.290@xmath154/(2.4)&0.55@xmath155&&163/176=0.92&8795 + 12&0.871@xmath156/1.176@xmath157/(2.4)&0.62@xmath158&&240/187=1.29&8914 + 13&0.675@xmath159/1.256@xmath160/2.723@xmath161&0.72@xmath162&&343/293=1.17&11033 + 14&0.625@xmath163/(1.2)/(2.4)&1.17@xmath164&1.82@xmath165&459/327=1.40&42002 + 15&0.790@xmath166/(1.2)/(2.4)&0.60@xmath162&&93/90=1.03&3666 + 16&1.355@xmath167/(2.4)&0.49@xmath168&&57/59=0.96&1746 + 17&1.091@xmath169/2.422@xmath170&0.40@xmath171&&53/45=1.18&3054 + 18&1.085@xmath172/2.107@xmath173&0.22@xmath174&&118/120=0.98&4643 + 19&0.651@xmath175/1.065@xmath176/(2.4)&0.25@xmath177&&147/128=1.15&7325 + 20&1.319@xmath178/(2.4)&0.39@xmath179&&30/23=1.29&1087 + 21&1.023@xmath180/(2.4)&0.22@xmath181&&147/105=1.39&2912 + 22&0.881@xmath182/1.236@xmath183/(2.4)&0.32@xmath136&&77/75=1.03&2646 + 23&1.186@xmath184/2.712@xmath185&0.45@xmath186&&214/236=0.91&5193 + = .31 .
the model is a two - temperature thermal plasma with fitted temperatures and abundance ( given in the second row of table [ tab : spectra ] ) , plus an unresolved power law component with photon index of 1.5 and a fixed thermal plasma component to model emission from the virgo icm , all with galactic absorption .
the data are from two s3 observations , obs - ids 318 and 963 ( see table [ tab : obs ] ) .
[ fig : m86spec ] ] ) .
the x - axis gives the radius from the apparent center of curvature defined by the feature .
the best fit two power law density jump model is given by the solid line .
[ fig : nedge ] ] towards us .
locations on the orbits where the past orbit would also appear to project 100 kpc outside the current position of m86 are shown in red ( solid lines ; the midpoint of each of these regions is marked with an asterisk ) .
both regions shrink for orbits with smaller outer turning radii . as plotted , all orbits pass through the virial radius at @xmath187 kpc on the @xmath188 axis . ] ( marked with an asterisk ) .
orbits are plotted from the point where they cross the virial radius ( sometimes outside the field shown ) . ] , one representative orbit is shown projected onto the smoothed 0.5 2 kev image of m86 for the remaining plausible orbits .
colors indicate the outer turning radius as follows : @xmath189 mpc , yellow ; @xmath190 mpc , blue ; @xmath191 mpc , green ; @xmath60 , red . ] kpc , one of the better candidates for the orbit of m86 . centered on the orbit , at equally spaced times
, circles are plotted on the plane of the sky with radii @xmath192 , where @xmath77 is the time before the present when m86 was at that location and @xmath193 is the sound speed ( @xmath194 ) . ]
in cylindrical polar coordinates , with the origin at the cluster center and the orbit ( instantaneously ) in the @xmath195
@xmath196 plane , the position of m86 can be expressed as @xmath197 and its velocity as @xmath198 , where @xmath199 , @xmath200 , @xmath201 are the coordinate basis vectors .
when m86 is viewed along the unit vector , @xmath202 , its location projected onto the plane of the sky is @xmath203 .
expressing @xmath202 in terms of the coordinate vectors as @xmath204 , the projected separation , @xmath51 , of m86 from the cluster center is given by @xmath205 , which requires @xmath206 ( and @xmath207 ) . since @xmath202 is a unit vector , @xmath208 , further requiring @xmath209 .
the remaining parameter , @xmath210 , is the cosine of the polar angle of the viewing direction ( i.e. , the orbit inclination ) . with the two values of @xmath66 above , the parametrization @xmath211 , for @xmath212
, specifies the complete range of viewing directions that project m86 at the observed distance from the cluster center . in this notation ,
the velocity of m86 along our line - of sight is @xmath213 . for radial orbits with m86 inbound ( @xmath214 ) , @xmath215 .
this vanishes for @xmath216 ( where a radial orbit must be perpendicular to our line - of - sight ) , while the radial speed , @xmath217 , decreases with increasing @xmath50 , so that @xmath52 is maximized for some @xmath207 .
the inner and outer turning radii , @xmath55 and @xmath56 , respectively , are related to the specific energy of an orbit by @xmath218 , where @xmath219 is the gravitational potential at @xmath55 and @xmath220 is that at @xmath56 .
the specific angular momentum of the orbit , @xmath221 , is given by @xmath222 . if m86 is bound to the virgo cluster , then , since @xmath223 and its speed , @xmath224 , which requires @xmath225 .
taking @xmath226 , this gives an upper limit on the inner turning radius of @xmath227 kpc . to obtain a lower limit on @xmath56 , note that @xmath228 , where the kepler speed at @xmath56 is given by @xmath229 and the speed at the outer turning radius is @xmath230 .
the net acceleration at the outer turning radius is inward , requiring @xmath231 , so that @xmath232 is an increasing function of @xmath56 . interchanging @xmath55 and @xmath56 in this argument and noting that the net acceleration is outward at the inner turning radius ( @xmath233 ) shows that @xmath232 is also an increasing function of @xmath55 . since @xmath234 , for a fixed energy , maximizing @xmath55 minimizes @xmath56 . for m86 , @xmath235 and @xmath236 ,
so that its energy @xmath237 .
to obtain @xmath238 with @xmath239 requires @xmath240 mpc , or @xmath59 times the virial radius of the cluster . for values of @xmath55 larger than @xmath51
, the total energy would need to match or exceed @xmath241 .
the value of @xmath56 required to attain this energy is an increasing function of @xmath55 , unless the mean density of the cluster increases with radius .
thus , @xmath242 provides a lower limit on the outer turning radius of m86 s orbit . at a fixed point on an orbit ,
the requirement @xmath243 determines the possible viewing directions , if any , that could model m86 . for each of the two solutions @xmath244 ,
the requirement @xmath245 gives @xmath246 .
if either of these gives @xmath247 in the range @xmath248 $ ] , then it corresponds to two possible viewing directions , with @xmath249 ( symmetrically placed above and below the plane of the orbit ) .
thus , for each point on an orbit , there can be zero , two , or four potential viewing directions that would place m86 at the observed separation from the cluster center and give it the observed line - of - sight velocity .
figure [ fig : mb ] shows a range of marginally bound orbits ( @xmath60 ) , with regions that meet these conditions marked in color . since latexmath:[$|{v_{\rm
{ v_{\rm los , max}}= \sqrt{1 - s^2/r^2 } if the trail of stripped gas lies along the orbit of m86 , then the past orbit must project away from the cluster center at least as far as the gas tail extends .
from above , @xmath251 is a unit vector in the plane of the sky that points from the cluster center towards m86 . for each potential location of m86 on an orbit ,
we have computed @xmath252 for values of @xmath253 ranging over the past orbit .
if stripping is to account for the location of the gas tail , @xmath254 must increase initially as @xmath253 moves backward along the orbit and the maximum value of @xmath255 must exceed the distance , @xmath62 kpc , that the tail projects beyond m86 in the direction away from the cluster center .
locations on the orbits where the line - of - sight speed can attain @xmath63 and these conditions are also met are shown in red in figure [ fig : mb ] .
|
we present results from a mosaic of nine _ chandra _ observations of m86 and the surrounding field .
we detect three main diffuse components : the virgo icm at @xmath02.4 kev , the extended halo of m86 at @xmath01.2 kev , and the cooler central and stripped gas of m86 at @xmath00.8 kev .
the most striking feature is a long tail of emission , which consists of a plume @xmath0 4north of m86 and two main extensions emanating from the plume .
based on the morphology and temperature structure of the tail , we conclude that it is formed by ram pressure stripping of m86 as it falls into the virgo cluster and interacts with the virgo icm , in agreement with earlier work .
the tail is 150 kpc in projection , and a simple estimate gives a lower limit on the true length of the tail of 380 kpc , making this the longest ram pressure stripped tail presently known .
the total gas mass in the plume ( @xmath1 ) and tail ( @xmath2 ) is about three times that in the core of m86 , which supports the scenario where most of the gas was stripped rapidly and recently .
the projected position of the plume can be understood if m86 has an aspherical potential , as suggested by optical isophotes .
ram pressure stripping from an aspherical potential can also explain the split `` double tails '' seen in m86 and in other virgo cluster galaxies in the field .
the large line - of - sight velocity of m86 ( 1550 @xmath3with respect to m87 ) , its position relative to the virgo cluster , and the orientation of the tail tightly constrain its orbital parameters . the data are inconsistent with a radial orbit , and imply inner and outer turning radii of @xmath4 kpc and @xmath5 mpc , indicating that m86 is , at best , only weakly bound to the virgo cluster .
|
due to its unique properties , topological insulators ( ti ) @xcite are likely to play a major role as a component material in different types of heterostructures . for instance , with a view towards spintronics applications,@xcite heterostructures involving ferromagnetic ( fm ) materials or magnetic impurities have been studied both theoretically @xcite and experimentally.@xcite underlying the many applications of magnetic heterostructures involving tis is the so - called axion electrodynamics,@xcite which was shown to distinguish the electromagnetic response of tis from ordinary insulators in an essential way.@xcite quite generally , it was shown in ref . that the lagrangian describing the electromagnetic response of _ all _ three - dimensional insulators is given by , @xmath5 where @xmath6 is in general a scalar field , the so - called axion,@xcite and @xmath7 is the fine - structure constant . in ordinary insulators
@xmath6 vanishes , but this is not the case in tis.@xcite in its simplest variant the axion field is uniform and assumes the value @xmath8 for a bulk time - reversal ( tr ) invariant ti.@xcite for uniform @xmath6 the axion term becomes a surface term , leaving therefore the maxwell equations unaffected.@xcite despite being a surface term when @xmath6 is uniform , the axion term still plays an important role in finite samples .
indeed , if we imagine a semi - infinite ti sample extending over @xmath9 up to the surface @xmath10 , we can use a covariant formalism to obtain , @xmath11 with the latin indices running over four - dimensional spacetime .
application of gauss theorem yields , @xmath12 where the greek indices run over @xmath13 , @xmath14 , and @xmath15 .
the above axion action at the surface actually represents a chern - simons ( cs ) term.@xcite as the chern - simons term does not depend on the metric , _
i.e. _ on the geometry of the sample , its presence can be considered as a manifestation of the topological insulator .
when some symmetry breaking is induced on the topological surface , the axion term may cause significant modifications on the dynamics of order parameters .
for example , if the ti is in contact with a fm material and a proximity - induced magnetization arises on the topological surface , the magnetization dynamics is modified @xcite due to the so - called topological magnetoelectric ( tme ) effect,@xcite consisting of an electric field - induced magnetization caused by the quantum spin - hall effect . although the axion term with @xmath6 uniform does not modify the maxwell equation , it does modify the landau - lifshitz equation for the magnetization precession on the topological surface .
@xcite there are also situations where a non - uniform @xmath6 is relevant , like for example in the case of magnetic fluctuations coupled to the electromagnetic field.@xcite another example is when two topological surfaces of the material are gapped and an external magnetic field induces multichannel edge states.@xcite also in effective theories of topological superconductors a dynamical axion field plays an important role.@xcite in all these cases the maxwell equations are modified as well and in addition a dynamical field equation for the axion arises . in order to generate an electromagnetic response featuring an axion term , the helical states have to gap .
this may be achieved by an out - of - plane exchange field which may be induced by proximity effect .
this means that the dirac fermions on the ti surface become massive and integrating them out generates a cs term , eq .
( [ eq : axion - cs ] ) , having @xmath8.@xcite thus , in this case tr and parity symmetries are broken on the ti surface , but are still preserved in the bulk.@xcite to understand why the mass term breaks the tr and parity symmetries , observe that the qed - like theory emerging from the proximity - induced ferromagnetism on the surface of three - dimensional ti ( see sect .
ii ) features two - component dirac fermions and , for this reason , does not have a chiral symmetry , since @xmath16-like matrices can only be defined for representations featuring four - component spinors.@xcite indeed , for @xmath17 @xmath18-matrices it is not possible to find an additional matrix that anticommutes with all of them .
hence , the massless case corresponding to the case of in - plane magnetization has no _ internal _ symmetry that would prevent the addition of a mass term . on the other hand , a mass term breaks discrete spacetime symmetries .
this case corresponds to an out - of - plane magnetization , which indeed is associated to mass term that breaks parity and tr symmetries . in particular , in 2 + 1 dimensions parity is realized in terms of a reflection ( mirror symmetry ) , for example , @xmath19 . note that inversion of both @xmath20 and @xmath21
does not work , since this is equivalent to a rotation by @xmath22 . in this case
the dirac fermions transform under parity like @xmath23 , @xmath24 , and @xmath25 .
the mass term is therefore not invariant under parity.@xcite in addition , the tr symmetry , defined by @xmath26 , @xmath27,@xcite is also broken once the mass term is introduced .
this breaking of parity and tr symmetries by massive two - component dirac fermions causes a chern - simons ( cs ) term @xcite to be generated upon integrating them out .
this generation of a cs term is related to the tme effect if these dirac fermions in 2 + 1 dimensions are viewed as surface states of a three - dimensional ti.@xcite it has been shown recently that the generation of the cs term by fermionic quantum fluctuations significantly affects the magnetization dynamics.@xcite however , when only in - plane exchange is present , the dirac fermions on the ti surface remain massless , thus not violating tr or parity .
consequently , in this case a cs term is not expected to be generated on the ti surface . an interesting question to be asked is whether masses for the dirac fermions simultaneously with the cs term can be spontaneously generated by some symmetry breaking mechanism .
we recall that there are several examples of _ dynamical _ mass generation in qed in 2 + 1 dimensions @xcite and related theories , including some condensed matter models for graphene @xcite where the coulomb interaction is taken into account,@xcite and theories for the pseudogap in high-@xmath28 ( cuprate ) superconductors.@xcite however , the latter theories feature an even number of dirac cones , allowing the use of four - component dirac spinors .
therefore , they have a chiral symmetry,@xcite since in this case two @xmath16-like matrices anticommuting with all @xmath18-matrices can be defined ( see sect .
[ sec : seff - o - o - p ] ) , and this is simply not possible with an odd number of dirac cones arising in tis.@xcite thus , in none of the mentioned models a cs term can be generated when masses for the dirac fermions are dynamically generated . in this paper
we analyze what happens for an interacting ti having an odd number of dirac fermions in the proximity to a fm inducing an in - plane exchange . in particular , we show that in the presence of a screened coulomb interaction , a mass for the dirac fermions is dynamically generated only if the interaction strength exceeds some critical value . under the same conditions a cs term
is also generated . as a result , the dynamical generation of the mass due to screened coulomb interaction in the case of ti in proximity to in - plane fm yields a tme effect similar to the case of an out - of - plane magnetization . in agreement with earlier calculations in the context of qed,@xcite
we also show that for an even number of dirac fermions there is mass generation , but parity and tr are overall preserved and no cs term arises .
the plan of the paper is as follows . in sect .
ii we define the qed - like model used in this paper and discuss its effective action in sects .
[ sec : seff - o - o - p ] and [ sect : seff - i - p ] .
[ sec : dynmass ] contains the main results , i.e. , the solution of the gap equation , showing that a semi - metal insulator transition occurs for a large enough value of the coupling constant .
[ sect : discussion ] discuses the relation of our results to recent experiments and in sect .
[ sect : conclusion ] we present the conclusions of this work .
three appendices contain additional technical information about the calculations .
in first - quantized form the hamiltonian for a topological surface with strong spin - orbit coupling in contact with a thin fm layer can be written in a form including a rashba - like term and an anisotropic exchange energy,@xcite @xmath29 where @xmath30 is the fermi velocity , @xmath31 , and @xmath32 and @xmath33 are the in - plane and out - of - plane exchange energies coupling to the magnetization @xmath34 , respectively .
for an uniform magnetization , the hamiltonian is easily diagonalized , yielding the generally gapped energy spectrum , @xmath35 where @xmath36 . for a vanishing out - of - plane exchange we have a gapless spectrum with a dirac point at @xmath37 .
thus , while for an out - of - plane magnetization the dirac spectrum is gapped , it is gapless in the case of in - plane exchange ; see fig .
[ fig : precession ] . in order to see
whether for @xmath38 a mass can be dynamically generated , we have to consider the quantum fluctuations of the magnetization on the ti surface and , in addition , the coulomb interaction . if @xmath39^t$ ] , the schrdinger equation , @xmath40 , for a vanishing out - of - plane exchange in the absence of coulomb interaction reads , @xmath41 where @xmath42 plays the role of a vector potential .
the above schrdinger equation has actually the form of a dirac equation in the presence of an electromagnetic field .
thus , the lagrangian of the ti surface proximate to a fm thin film inducing a planar magnetization on it is given by , @xmath43\psi,\ ] ] where @xmath44 , @xmath45 , and @xmath46 .
the above lagrangian has a qed - like form in @xmath47 dimension with a vector potential @xmath42 , and no time component for the gauge field .
a time component for the gauge field is introduced if we assume a screened coulomb interaction on the ti surface with interaction hamiltonian density @xmath48 where @xmath49 and @xmath50 as usual .
then , the full lagrangian acquires the following form @xmath51 which can be rewritten in terms of an auxiliary field , @xmath52 , via a hubbard - stratonovich ( hs ) transformation to obtain , @xmath53 where we have used the standard dirac slash notation , @xmath54 .
note that the vector field @xmath55 is not dynamical at the lagrangian level , since the only quadratic term in the gauge field @xmath55 is a term proportional to @xmath56 with a time - independent coefficient .
this term implies that the gauge symmetry is broken in the temporal direction .
we are disregarding the long - range contribution to the coulomb interaction because it has been shown to be irrelevant in the long wavelength limit in theoretical studies of interacting graphene @xcite and a similar reasoning also applies here . from a model point of view
, our lagrangian corresponds to a restricted thirring model,@xcite in the sense that only the zeroth component of the current @xmath57 appears squared in the interaction . one important consequence of the term quadratic in @xmath52 ,
is that @xmath0 does not renormalize .
this follows from the gauge symmetry of the fermionic sector and can be easily proved using ward identities.@xcite an easy way of seeing this without making explicit use of the ward identity is to introduce renormalized fields @xmath58 and @xmath59 and observe that gauge invariance of the fermionic sector implies that the renormalized exchange coupling is @xmath60 , otherwise the form of the covariant derivative would not be preserved by a gauge transformation .
furthermore , current conservation implies that any fluctuation correction for @xmath61 is transverse , and therefore terms quadratic in @xmath61 which are not gauge invariant do not renormalize , yielding @xmath62 and consequently @xmath63 . therefore , @xmath0 is a good tuning parameter in our theory .
the fact that @xmath0 does not renormalize will be important in our subsequent analysis .
note that our lagrangian does not include an intrinsic dynamics for the magnetization .
although the fm above the ti surface has its own dynamics , we are assuming a minimal model on the topological surface where the only exchange interaction is the one between the electronic spin and the surface magnetization .
thus , the whole magnetization dynamics on the topological surface will be generated by the quantum fluctuations of the dirac fermions .
it is certainly important to include other exchange effects , like it was done in refs . and .
however , our main aim here is to study the dynamical mass generation and the spontaneous breaking of parity and tr symmetries . for this purpose our minimal exchange model ( [ eq : ti - qed-1 ] ) already exhibits this feature and has the advantage of being analytically more tractable .
let us first recall the situation for @xmath64 that is when the ferromagnet has the out - plane component of the magnetization . since we are assuming that the fm above the ti surface is in the broken symmetry state , we can write @xmath65 , where @xmath66 . which can be either positive or negative , and @xmath67 are small transverse fluctuations . in this case
the lagrangian ( [ eq : ti - qed-1 ] ) becomes @xmath68 where @xmath69 .
as discussed in the introduction , the mass term _ explicitly _ breaks parity and time - reversal symmetry .
let us assume that we have @xmath2 dirac fermion species and work within an imaginary time formalism . in this case
after integrating out the @xmath2 fermionic degrees of freedom , we obtain , @xmath70 now we expand the above effective action up to quadratic order in the vector field @xmath55 , which in momentum space reads , @xmath71,\ ] ] where @xmath72 and @xmath73 is the one - loop vacuum polarization , which is evaluated in detail in appendix [ app : vacpol ] .
the result is @xmath74\left(\delta_{\mu\nu}-\frac{p_\mu p_\nu}{p^2}\right ) \nonumber\\ & -&2nj^2mi(p)\epsilon_{\mu\nu\lambda}p_\lambda,\end{aligned}\ ] ] where @xmath75 thus , in the long wavelength regime @xmath76 [ see eq .
( [ eq : large - mass ] ) in appendix a ] the effective action in real time is given by , @xmath77 , \label{cs - term}\ ] ] where @xmath78 . in ref .
this result was used for the case where @xmath79 .
we note also that the first ( maxwell ) term in eq .
( [ cs - term ] ) contains a dimensionfull coefficient as it depends on @xmath80 .
thus , this term is non - universal and receives corrections from all orders in perturbation theory .
the second term ( the cs term ) is universal and its prefactor @xmath81 is just a sign .
moreover , being independent of the metric , it is not expected to be modified by the scale transformations , so it does not renormalize .
the coleman - hill theorem @xcite on the non - renormalization of the cs term provides a more precise statement of this argument .
observe that the number of dirac fermions , @xmath2 , must be necessarily odd , otherwise no cs term is generated .
although this point was noticed before in ref .
( see sect .
iv - d there ) , we would like to revisit it in the framework of our calculations .
in order to see the effect , let us now assume that each of the @xmath2 dirac fermions has a mass , @xmath82 ( @xmath83 ) .
it is straightforward to see that the low - energy form of the cs term is now given by , @xmath84 now , if the number of dirac fermions is even , we can rewrite the dirac lagrangian in terms of @xmath85 four - component dirac fermions using @xmath86 @xmath18-matrices . in this case
a mass term @xmath87 is invariant under both parity and time - reversal transformations , just like in the case of qed in four - dimensional spacetime .
namely , when four - component dirac fermions are introduced in 2 + 1 dimensions , it is possible to introduce @xmath86 dirac matrices of the form,@xcite @xmath88 @xmath89 where @xmath90 , @xmath91 , and @xmath92 are the pauli matrices . with the above representation
the chiral symmetry can be defined via the following two matrices,@xcite @xmath93 anticommuting with all @xmath18-matrices ( [ eq : gamma ] ) , where @xmath94 is a @xmath17 identity matrix .
the lagrangian for massless four - component dirac fermions has therefore the invariance under the chiral transformations @xmath95 and @xmath96 .
since under these transformations @xmath97 and @xmath98 , the current @xmath99 is invariant , but @xmath100 is not .
thus , for massless qed in 2 + 1 dimensions with four - component dirac fermions the chiral symmetry prevents the addition of a mass term.@xcite indeed , a term @xmath101 in the lagrangian would explicitly break the chiral symmetry , but not parity and tr . in our particular case
this implies that we must have @xmath102 in the representation where @xmath103 two - component dirac fermions are used , since this is equivalent to @xmath104 four - component dirac fermions .
thus , the spectrum must consist of @xmath85 masses having the opposite sign of the remaining @xmath85 ones .
we conclude that @xmath2 must be odd in order to generate a cs term , and we can write @xmath105 ( @xmath106 ) .
this is consistent with the fact that a ti features an odd number of dirac fermions .
therefore , the cs action originating from the coupling of dirac fermions to an out - of - plane exchange can be written in the form , @xmath107 which emphasizes the role of an odd number of dirac fermions .
note that the apparent `` quantization '' of the cs coefficient in eq .
( [ eq : cs - action ] ) does not have in the present context the same origin as the hall conductivity in the quantum hall effect , as it arises from integrating out dirac fermions coupled to a vector field .
however , it is consistent with the analysis of qi _
et al._@xcite of the axion term , since we find here that the axion field has the constant value @xmath8 .
indeed , for a single dirac fermion ( @xmath108 ) , eq .
( [ eq : cs - action ] ) coincides with eq .
( [ eq : axion - cs ] ) for this value of @xmath6 and identifying @xmath109 , so the in - plane exchange coupling squared plays the role of the fine - structure constant in this case .
it is worth to mention in this context yet another aspect of the problem discussed recently in ref . , namely , the dependence of the hall conductivity at the edges of a finite sample for a general value of @xmath6 , not necessarily @xmath13 or @xmath22 .
this case is relevant if there is an external magnetic field . in this situation
it can be shown that the hall conductivity is given by @xmath110 $ ] [ in units where @xmath111 ; here we are assuming that the lowest value of @xmath104 is zero , just like in eq .
( [ eq : cs - action ] ) ] , and quantization will hold if @xmath6 changes by @xmath112 on loops containing @xmath113 edges channels.@xcite it is important to emphasize here the difference between a non - abelian cs action , where the cs coupling is quantized due to the requirement of gauge - invariance,@xcite so that the integer arising there is actually a winding number .
note that the derivation of the axion term in the bulk leads to an expression for @xmath6 given in terms of a non - abelian berry connection in momentum space , which is derived in terms of the actual band structure on the bulk.@xcite this is a topological invariant generalizing the thouless - kohmoto - nightingale - nijs invariant,@xcite which features a berry abelian curvature in momentum space , to higher dimensions . in our calculations done in the 2 + 1 dimensions
@xmath6 is also quantized in the sense that it may have only two values , @xmath22 or 0 , where the latter refers to the absence of the cs term .
rewriting the cs term in explicitly in terms of components allows us to analyze its physical content relative to the magnetization dynamics on the topological surface:@xcite @xmath114 where @xmath115 yields the electric field associated to the screened coulomb potential and @xmath116 .
we observed that the cs action contains an induced berry phase associated to the precession of the magnetization.@xcite if we neglect for a moment the contribution of the maxwell term in the effective action , we obtain simply , @xmath117 which is the expected result for a spin - hall response . in order to obtain the full magnetization precession
, we have also to consider the fluctuations in @xmath118 around its expectation value @xmath119 .
this was done in ref . .
the result is a landau - lifshitz equation where in addition to the usual torque @xmath120 yielding the precession around the effective magnetic field @xmath121 , a magnetoelectric torque @xmath122 arises .
when @xmath38 the dirac fermions are massless .
thus , in this regime the effective action becomes , @xmath123,\end{aligned}\ ] ] where @xmath124 is the usual vacuum polarization for massless dirac fermions in 2 + 1 dimensions . from eq .
( [ eq : s - eff ] ) we derive the propagator ( see appendix b ) , @xmath125\frac{p_\mu p_\nu}{\omega^2 } -\frac{(p_\mu\delta_{\nu 0}+p_\nu\delta_{\mu 0})}{\omega } \right\}.\end{aligned}\ ] ] interestingly , this result shows that a vector field with a mass term only along the temporal direction is not gapped . with an isotropic mass term of the form @xmath126
, we would obtain instead @xmath127^{-1}[\delta_{\mu\nu}+(g / j^2)\pi(p)p_\mu p_\nu]$ ] , which is clearly gapped . as we will see shortly , this difference is important , as in our case two of the components of the vector field relate to the magnetization and magnetic excitations are supposed to be gapless .
the propagator ( [ eq : gauge - prop ] ) does not smoothly connect to the strongly coupled regime , @xmath128 .
this is a typical behavior for massive vector fields @xcite which is also reflected here , although our vector field is only massive along the temporal direction .
note , however , that in the strongly coupled regime our model reduces to a qed model in 2 + 1 dimensions with two - component dirac fermions .
as mentioned earlier , no cs term is generated in this case when the dirac fermions become gapped.@xcite the purely magnetic effective action is finally obtained by integrating out @xmath52 in the effective action ( [ eq : s - eff ] ) .
this yields , @xmath129 a_i(p)a_j(-p)\nonumber\\ & = & \frac{1}{2}\int\frac{d^3p}{(2\pi)^3}\pi(p)\left\{\left[p^2-f(p)v_f^2{\bf p}^2\right]{\bf n}(p)\cdot{\bf n}(-p ) \right.\nonumber\\ & + & \left.f(p)v_f^2[{\bf p}\cdot{\bf n}(p)][{\bf p}\cdot{\bf n}(-p ) ] \right\},\end{aligned}\ ] ] where @xmath130 going back to real time , the magnetic susceptibility @xmath131 , where @xmath132 , is determined from eq .
( [ eq : s - eff-1 ] ) as , @xmath133\right\}.\end{aligned}\ ] ] from the pole of @xmath134 we infer that the spin - wave velocity is identical to the fermi velocity .
this is the consequence of our approximation as we ignored the bare spin dynamics of the ferromagnet at the interface and our spin excitations are itinerant excitations due to dirac fermions . in this case
, we also see that by comparing with the scaling behavior @xmath135^{\eta/2 - 1}$ ] for @xmath136 near @xmath137 , yields an anomalous scaling dimension @xmath138 .
this induced anomalous dimension on the topological surface is very different from the one of a two - dimensional planar fm at @xmath139 corresponding to a three - dimensional ( @xmath47 ) xy universality class , having @xmath140 .
next , we consider the fermionic propagator . within an imaginary time formalism , the fermion propagator @xmath141 is given in general form by @xmath142 where @xmath143 is the vertex function .
it is understood that the dirac matrices above are the imaginary time counterparts of the real time ones defined earlier .
they are assumed to satisfy the clifford algebra @xmath144 . in order to determine @xmath141 approximately
, we make the decomposition @xmath145 and assume the lowest order form for the vertex function , @xmath146 .
furthermore , we will set @xmath147 for @xmath141 inside the integral in eq .
( [ eq : f - prop ] ) . a mass will be generated if @xmath148 does not vanish .
note that a non - vanishing @xmath149 implies that @xmath150 .
since @xmath151 , mass generation implies also an emergent third component of the magnetization
. our strategy will be to make an approximation in which @xmath152 is uniform , @xmath153 , and see whether there is a solution to eq .
( [ eq : f - prop ] ) with @xmath154 .
we will solve eq .
( [ eq : f - prop ] ) under the assumption that @xmath155 , where @xmath156 is an ultraviolet cutoff , which here is naturally given by @xmath157 , similarly to qed in 2 + 1 dimensions,@xcite where the cutoff is determined by the charge squared times the number of fermion components .
the fermion mass modifies the vacuum polarization , and now a term odd under parity may arise , so that the photon self - energy becomes , @xmath158 however , under the assumption @xmath155 , the contribution that is even under parity and time - reversal , corresponding to the transverse term in eq .
( [ eq : s - eff ] ) , remains unchanged , and we obtain @xmath159 . in order to investigate the gap equation ,
we follow ref . and assume that @xmath160 fermions acquire a positive mass @xmath161 , while the remaining @xmath162 fermions acquire a negative mass @xmath163 .
thus , the effective action ( [ eq : s - eff ] ) for the vector field receives the following additional contribution odd under parity and time - reversal [ see eq .
( [ eq : small - mass ] ) in appendix a ] , @xmath164 this leads in turn to an additional term in the vector field propagator given by @xmath165 .
thus , the following self - consistent equation for @xmath82 is obtained , @xmath166\int\frac{d^3k}{(2\pi)^3}\frac{1}{|k|(k^2+m^2 ) } \right.\nonumber\\ & + & \left.\int\frac{d^3k}{(2\pi)^3}\frac{|k|}{\omega^2(k^2+m^2 ) } \right\ } + g\int\frac{d^3k}{(2\pi)^3}\frac{k^2}{\omega^2(k^2+m^2)}.\nonumber\\\end{aligned}\ ] ] the second and third integrals above require some care , although they are not difficult to solve , and are calculated in the appendix c. after performing all integrals , the gap equation for @xmath167 becomes , @xmath168 while for @xmath169 , we obtain , @xmath170 to have the solution for @xmath171 one sees that the gap equations ( 31)-(32 ) are only compatible with each other if @xmath2 is even , i.e. , @xmath172 . for the odd number of fermions @xmath173 which we discuss below . for an even number of dirac fermions we introduce
the dimensionless quantities @xmath174 and @xmath175 and obtain @xmath176 where @xmath177 in terms of the dimensionful coupling constant @xmath0 , we can explicitly write @xmath178 note that @xmath1 has dimension of length squared and relates in terms of energy scales of the lattice model via @xmath179 , where @xmath180 is the lattice spacing , and @xmath181 and @xmath182 are the hubbard interaction and hopping , respectively . in field - theoretic units , @xmath183 , @xmath0
, it has of course dimension of length , since the dirac fermions have in this case dimension of ( length)@xmath184 and the action has to be dimensionless . thus , we obtain that if @xmath2 is even a gap is generated provided @xmath185 . in this case
there is no generation of cs term .
therefore , parity and time - reversal symmetries remain preserved .
the above result distinguishes itself from the qed case @xcite due to a complete cancellation of the logarithm . for the odd number of dirac fermions , which is the situation corresponding to a ti
, we now search for the solution @xmath186 for all @xmath187 and @xmath173 . in this case the logarithmic term survives and dominates for @xmath188 over the linear term in @xmath80 .
thus , the gap equation is given by eq .
( [ eq : gap-1 ] ) with @xmath173 and where the term proportional to @xmath189 is neglected
. then we obtain , @xmath190,\ ] ] where @xmath191 is the same as before , given by eq .
( [ eq : gc ] ) .
( [ eq : sol - odd ] ) only makes sense for @xmath185 , otherwise it does not decrease with increasing @xmath2 , which would at large @xmath2 contradict the condition @xmath155 , in a situation reminiscent from the qed case.@xcite however , in our case it is possible to overcome the difficulty encountered there and obtain in addition the generation of a cs term . in other words , we find that the dynamic generation of the mass due to the screened coulomb interaction in a ti / fm heterostructure where the fm has in - plane components has similar consequences for the electrodynamics of the ti in contact to a fm with out - of - plane magnetization .
namely , the topological magnetoelectric term arises in the former case when values of the interaction above @xmath192 is reached .
the latter is determined by the bare value of @xmath0 in the topological insulator , multiplied by the ratio of @xmath193 ; see eq .
( [ eq : gc-1 ] ) .
this means that in the ti / fm heterostructure with in - plane magnetization the dynamic mass generation will be proportional to the absolute value of the in - plane exchange coupling in the fm .
this is interesting as it points towards experimental realizability of the observed effect by varying the fm substrate of the heterostructure .
we note that the case of in - plane exchange coupling on a topological surface is highly nontrivial with respect to the case of out - of - plane exchange . indeed , in the case of an out - of - plane exchange a simple mean - field theory would generate a gap for arbitrarily small values of the coupling constant @xmath0 , and
no semi - metal - insulator transition would take place in this case.@xcite this situation is reminiscent from the metal - insulator transition in the hubbard model , where a mean - field theory at half - filling yields a gap @xmath194 , where @xmath181 is the on - site coulomb interaction.@xcite it is well known that this result is not correct for values of @xmath181 smaller than energy scales of the order of the bandwidth.@xcite note , that on a ti surface the mean - field result also leads to a phase transition if a momentum dependence of the interaction induced by projecting the bulk hamiltonian on the edge states is taken into account but for some critical value @xmath195.@xcite in our case the out - of - plane exchange is generated dynamically from the interplay between quantum planar magnetic and charge fluctuations , which is characterized by a competition between the in - plane exchange coupling @xmath32 and the screened coulomb interaction @xmath0 .
this leads to a dynamical mass generation accompanied by the generation of a cs term , which implies a coupling between the in - plane magnetization and the electric field @xmath115 , and to a berry phase governing the precession dynamics of the magnetization .
note that the inclusion of the charged channel in the hubbard - stratonovich decoupling is crucial to obtain a cs term .
the case of in - plane magnetization is of great experimental relevance . recently
a thin film of fm insulator eus has been successfully grown on the surface of bi@xmath3se@xmath4,@xcite making the surface of bi@xmath3se@xmath4 ferromagnetic by proximity effect with magnetization at the interface being different from the bulk eus values .
there were several features which point out towards a strong interaction between the magnetic moments of eus and bi@xmath3se@xmath4 .
one of them refers to the fact that the dependence of the planar magnetoresistivity of the interface bi@xmath3se@xmath4 shows an effectively lower curie temperature than that of the bulk eus which could be the result of the quantum fluctuations due to presence of surface dirac fermions.@xcite another effect is even more interesting as it reports the significant out - of - plane magnetization of the magnetic moments at the fm - ti interface while the bulk eus has the in - plane orientation of the magnetic moments.@xcite our results show that the out of plane orientation of the moments will be indeed generated at the interface by the interaction among the dirac fermions although on the experimental side further mechanisms related to the crystalline anisotropy at the grown interface can be also in play .
in addition , the direct gapping of the dirac spectrum of the surface electrons was reported very recently at the eus - bi@xmath3se@xmath4 interface.@xcite in particular , it was found that below curie temperature there is a negative megnetoresistance near zero field which is believed to be the consequence of gap opening in the dirac spectrum due to proximity to the ferromagnet with out - of - plane magnetization.@xcite according to our calculations this effect must be dependent on the temperature and on the strength of the out - plane component of the magnetization , induced by the interaction .
this would be interesting to test experimentally .
in conclusion , we have shown that in topological insulators with a proximity - induced in - plane magnetization a gap can be spontaneously generated by tuning the local electronic interaction above a critical value , leading in this way to a semi - metal - insulator transition . in particular , considering the @xmath2 number of dirac fermions we find that when @xmath2 is even , the masses are generated in pairs @xmath196 and no cs term is generated , so that parity and time - reversal is overall preserved . on the other hand , for @xmath2 odd which
the case of ti all generated masses are equal and positive and as a result , the cs term with tme effect is generated .
in particular , we find that the critical dimensionless value of @xmath192 for generating this term is also proportional to the value of the in - plane exchange coupling of the fm making this effect to depend on the choice of the fm substrate in the experiment . that no cs term is generated for @xmath2 even is physically reasonable , since in this case we can change to a representation where there are @xmath85 four - component dirac spinors , in which case the model may be reinterpreted as some model for graphene , a material featuring an even number of dirac cones .
interestingly , in such a graphene model the gap generation is associated to a mass spectrum containing masses @xmath196 , a scenario not considered so far in interacting models for graphene where the `` vector field '' has only the time component as compared to qed.@xcite tis , on the other hand , have an odd number of dirac cones . in this context , recent experiments on eus / bi@xmath3se@xmath4 heterostructures open the possibility that the experimentally elusive gap generation in qed - like theories may finally be observed in the near future .
we would like to thank alex altland , igor herbut , jagadeesh moodera , achim rosch , jrg schmalian , and peng wei for interesting discussions .
we thank j. s. moodera for sending us his preprint prior publication .
the authors would like to thank the deutsche forschungsgemeinschaft ( dfg ) for the financial support via the collaborative research center sfb tr 12 .
for pedagogical reasons , we review in this appendix in detail the calculation of the one - loop vacuum polarization in ( 2 + 1)-dimensional qed with massive two - component dirac fermions.@xcite we will perform the calculation in euclidean space ( imaginary time ) . in this case
the dirac matrices satisfy the same algebra as the pauli matrices , having anticommutator @xmath197 and a commutator @xmath198=2i\epsilon_{\mu\nu\lambda}\gamma_\lambda$ ] . from this algebra
it follows the traces of product of @xmath18-matrices necessary to calculate the vacuum polarization , @xmath199 @xmath200
@xmath201 the vacuum polarization is represented by the feynman diagram shown in fig .
[ fig : vacpol ] .
it corresponds to the photon self - energy and is analytically given by , @xmath202,\ ] ] where @xmath203 is the ( matrix ) fermion propagator , @xmath204 and the sign of the mass can be positive or negative . in the field theory literature
@xmath205 , the electric charge . using the trace formulas ( [ eq : tr2gammas],[eq : tr3gammas],[eq : tr4gammas ] ) , we can express eq .
( [ eq : qedvacpol ] ) in the form , @xmath206 } -2nj^2m\epsilon_{\mu\nu\lambda}p_\lambda i(p),\ ] ] where @xmath207}.\ ] ] current conservation implies that eq .
( [ eq : vacpol-1 ] ) can be cast in the form , @xmath208 indeed , it is not difficult to show that @xmath209 .
we will use dimensional regularization in this appendix , with the understanding that this is done only in the evaluation of integrals , while @xmath18-matrices and levi - civitta tensor remain as defined above for the ( 2 + 1)-dimensional case . taking the trace yields , @xmath210}.\ ] ] by writing @xmath211-m^2-\frac{p^2}{2},\ ] ] we can express @xmath212 in the form , @xmath213 the rules of dimensional regularization imply,@xcite @xmath214 such that eq .
( [ eq : s - integral ] ) becomes , @xmath215 where we have introduced the standard notation @xmath216 used in qed .
the integral @xmath217 can be evaluated explicitly , @xmath218 and @xmath219 we note therefore the important limit cases , @xmath220 and @xmath221 therefore , in the small mass limit the photon self - energy becomes , @xmath222 which in terms of the one - loop vacuum polarization for massless fermions as defined in the main text , @xmath124 becomes , @xmath223 in the large mass limit , on the other hand , we obtain , @xmath224 the above form was used in ref . to derive the effective lagrangian for the magnetization dynamics on the surface of a three - dimensional ti .
in order to find the propagator ( [ eq : gauge - prop ] ) of the vector field @xmath55 from eq .
( [ eq : s - eff ] ) , we have simply to solve the matrix equation , @xmath225 where @xmath226 with @xmath227 . in order to solve eq .
( [ eq : matrixeq ] ) , we decompose @xmath228 in the form , @xmath229 the unknown coefficients @xmath230 , @xmath231 , @xmath232 , @xmath233 , and @xmath234 are easily determined from eq .
( [ eq : matrixeq ] ) , so that eq .
( [ eq : gauge - prop ] ) follows
. the calculation can be easily generalized to the case where a cs term is present and eq .
( [ eq : mmunu ] ) is replaced by @xmath235 where we have assumed that the masses are small ( see appendix [ app : vacpol ] ) . in this case
the decomposition of @xmath228 has to include the additional tensors @xmath236 , @xmath237 , and @xmath238 . the result for the propagator is thus , @xmath239}{(p\cdot l)^2 } p_\mu p_\nu\right.\nonumber\\ & -&\left.\frac{p_\mu l_\nu+p_\nu l_\mu}{p\cdot l}-\frac{4}{np^2}\sum_i m_i\epsilon_{\mu\nu\lambda}p_\lambda\right\},\end{aligned}\ ] ] which in the regime @xmath240 used to solve the gap equation and to write eq .
( [ eq : sodd ] ) is approximated simply by , @xmath241}{(p\cdot l)^2 } p_\mu p_\nu\right.\nonumber\\ & -&\left.\frac{p_\mu l_\nu+p_\nu l_\mu}{p\cdot l}-\frac{4}{np^2}\sum_i m_i\epsilon_{\mu\nu\lambda}p_\lambda\right\}.\end{aligned}\ ] ]
here we calculate two integrals appearing in eq .
( [ eq : self ] ) : @xmath242 and @xmath243 where we recall the notation @xmath244 and @xmath245 . here
we can simply set @xmath246 .
the integrand of both integrals @xmath247 and @xmath32 are singular at @xmath248
. however , we can solve this singularity by assuming a regularization where a finite value is obtained through the principal values of @xmath247 and @xmath32 .
let us consider first the integral @xmath247 .
using partial integration , we can cast the integral in @xmath136 in the form , the integral @xmath32 can be rewritten as @xmath253 , where @xmath254 and @xmath255 in the limit @xmath251 we can trivially evaluate @xmath256 : @xmath257 for @xmath258 we can again regularize the singularity for @xmath248 , @xmath259 therefore , @xmath260 y. s. hor , p. roushan , h. beidenkopf , j. seo , d. qu , j. g. checkelsky , l. a. wray , d. hsieh , y. xia , s .- y .
xu , d. qian,, m. z. hasan , n. p. ong , a. yazdani , and r. j. cava , phys . rev .
b * 81 * , 195203 2010. t. appelquist and l.c.r .
wijewardhana , phys .
* 60 * , 2575 ( 1988 ) ; d. nash , phys .
. lett . * 62 * , 3024 ( 1989 ) ; k .-
kondo and p. maris , phys .
d * 52 * , 1212 ( 1995 ) ; p. maris , phys .
d * 54 * , 4049 ( 1996 ) ; w. armour , j. b. kogut , and c. strouthos , phys .
d * 82 * , 014503 ( 2010 ) .
|
a thin film of ferromagnetically ordered material proximate to the surface of a three - dimensional topological insulator explicitly breaks the time - reversal symmetry of the surface states . for an out - of - plane ferromagnetic order parameter on the surface , the parity is also broken , since the dirac fermions become massive .
this leads in turn to the generation of a topological chern - simons term by quantum fluctuations .
on the other hand , for an in - plane magnetization the surface states remain gapless for the non - interacting dirac fermions . in this work
we study the possibility of spontaneous breaking of parity due to a dynamical gap generation on the surface in the presence of a local , hubbard - like , interaction of strength @xmath0 between the dirac fermions .
a gap and a chern - simons term are generated for @xmath0 larger than some critical value , @xmath1 , provided the number of dirac fermions , @xmath2 , is odd . for an even number of dirac fermions
the masses are generated in pairs having opposite signs , and no chern - simons term is generated .
we discuss our results in the context of recent experiments in eus / bi@xmath3se@xmath4 heterostructures .
|
it is well known that on rather coarse lattices with lattice spacings @xmath3 0.1 fm the wilson gauge action produces quite large cut - off effects . on the other hand ,
decreasing the lattice spacing below 0.1 fm becomes increasingly difficult for simulations including fermions . therefore using a better
, i.e. improved gauge action might well be important .
one possible way of constructing non - perturbatively improved lattice gauge theories is provided by the fixed point ( fp ) action approach @xcite .
fp actions are classically perfect , i.e. they have no lattice artefacts on classical configurations . in particular , they possess exactly scale invariant instanton solutions .
together with fermion fp actions they form appealing formulations of non - perturbatively improved lattice qcd @xcite . here
we present a new parametrization for fp gauge actions which we apply to su(3 ) .
the new parametrization is much more flexible and has a richer structure than the loop parametrizations studied so far @xcite and therefore allows to better reproduce the classical properties of the theory . in order to check the scaling behaviour of the parametrized fp action
we perform a series of simulations in which we determine the critical temperature of the deconfining phase transition , the static @xmath4 potential and glueball masses .
this paper summarizes some of the results @xcite .
we consider su(@xmath5 ) pure gauge theory in four dimensional euclidean space - time on a periodic lattice .
for asymptotically free theories the determination of the fp action reduces to a saddle point problem encoded in the fp equation @xcite @xmath6 where @xmath7 is the fp action , @xmath8 is the blocking kernel of a renormalization group ( rg ) transformation and @xmath9 and @xmath10 are the gauge fields on the fine and coarse lattice , respectively .
we use the rg transformation of ref .
@xcite where the parameters of the transformation have been optimized for a short interaction range of the fp action and improved rotational invariance .
( [ eq : fp_equation ] ) represents an implicit equation for the fp action which , in principle , contains infinitely many couplings . in order to use it in mc simulations one
has to find an appropriate parametrization .
the main novelty in the new parametrization of the fp gauge action is the use of smeared ( `` fat '' ) links which are used to build simple wilson loop plaquettes @xcite . to build a plaquette in the @xmath11-plane from smeared links we introduce asymmetrically smeared links : instead of summing up all staples contributing to a smeared link we are suppressing the staples lying in the @xmath11-plane relative to those in the orthogonal planes @xmath12 , @xmath13 . in this way we are able to effectively generate long and complicated loops which , however , are still very compact . from these `` smeared '' loops
@xmath14 and from the unsmeared standard wilson loop @xmath15 we construct smeared and unsmeared plaquette variables @xmath16 and @xmath17 , respectively , @xmath18 which in turn are used in a mixed polynomial ansatz for the fp action @xmath19=\frac{1}{n_c}\sum_{x,\mu<\nu}\sum_{k , l } p_{kl}\,u_{\mu\nu}(x)^k\ , w_{\mu\nu}(x)^l\,.\ ] ] the linear parameters @xmath20 and the non - linear ones entering the smearing are determined through eq .
( [ eq : fp_equation ] ) which is used recursively connecting coarse configurations to smoother and smoother ones . on the smoothest configuration we use the analytically calculable quadratic approximation to the fp action as the starting point @xcite .
an additional novelty is the use of local information stored in the fine configurations minimizing the r.h.s . of eq .
( [ eq : fp_equation ] ) for the fitting procedure .
obviously the minimizing configurations contain much more information than just the total value of the action . to explore this information
we calculate the derivatives of the fp action with respect to the gauge links in a given colour direction at each site and , together with the action values , include them in the fitting procedure for determining the parameters .
the final parametrization @xcite obtained in this way contains five non - linear parameters entering the smearing and 14 linear ones @xmath21 describing the mixed polynomial in eq .
( [ eq : polynomial ansatz ] ) .
using the parametrized fp action we perform a large number of simulations on coarse lattices with lattice spacings @xmath22 between 0.1 and 0.33 fm in order to explore the scaling behaviour of different physical quantities . in all simulations
we generate independent gauge configurations using a mixture of metropolis and overrelaxation updates . wherever it is applicable we make use of smearing techniques and
employ variational techniques to increase the overlap of wave functions with the ground state @xcite .
we first measure the critical temperature of the deconfining phase transition using the polyakov loop @xmath23 as an order parameter . to do
so we determine the critical couplings on lattices with temporal extent @xmath24 and 4 and various spatial volumes from the location of the peak in the polyakov loop susceptibility , @xmath25 the location of the peak can be determined very precisely by a spectral density reweighting @xcite which combines the data of many simulations at values of the gauge coupling near the critical value . in order to perform the thermodynamic limit we resort to the finite scaling behaviour linear in @xmath26 as expected for a first order phase transition .
the results from these extrapolations are @xmath27 and @xmath28 .
finally , the critical temperature is determined through @xmath29 .
one way of examining the scaling of the parametrized fp action is by measuring the static @xmath30 potential @xmath31 in terms of the hadronic scale @xmath0 and comparing the quantity @xmath32 versus @xmath33 at several values of @xmath34 ( see figure [ fig : all_pot ] ) .
potential @xmath31 expressed in terms of the hadronic scale @xmath0 .
, width=283 ] we would like to stress that on coarse lattices this a non - trivial scaling test .
the quality of our measurements can be seen for example by comparing our fit ( dashed line ) of the form coulomb plus linear term with the result ( dotted line ) from measurements with an anisotropic tadpole / tree level symanzik improved action @xcite .
the hadronic scale used above is determined from the force @xmath35 between two static quarks .
we use @xcite @xmath36 where @xmath37 .
however , since eq .
( [ eq : def_hadronic_scale ] ) needs an interpolation of the force in between lattice sites , on coarse lattices this definition of the scale is plagued by ambiguities stemming from the discreteness of the lattice points .
another physical quantity of interest is the string tension @xmath1 which describes the asymptotic long range behaviour of the potential .
note , that here we do not aim at testing the rotational invariance of the potential .
for the rgt used here this has been done at finite temperature , however , using a different parametrization of the fp action @xcite .
we are now in a position to investigate the scaling behaviour of the parametrized fp action by means of the rg invariant quantities @xmath38 and @xmath39 . for different actions.,width=283 ]
we find that these quantities scale even on lattices as coarse as @xmath40 fm @xcite . as an example we display the results for @xmath41 in figure [ fig : sigmatc ] together with results obtained with different other actions @xcite .
in the pure gauge glueball spectrum the lowest - lying @xmath42 state shows particularly large cut - off effects when measured with the wilson action .
it therefore provides an excellent candidate for sizing the improvements achieved with the parametrized fp action .
to do so we perform simulations at three different lattice spacings in the range @xmath43 and spatial volumes between ( 1.4 fm)@xmath44 and ( 1.8 fm)@xmath44 .
the scale is set by @xmath0 as determined in the previous section .
we measure all length eight loops on five smearing levels in order to construct a wave function with a large overlap to the ground state . in figure [ fig:0pp_final ]
we compare the results for the lowest lying @xmath42 glueball mass measured with the wilson action @xcite , the anisotropic tadpole / tree level symanzik improved action @xcite and the fp action .
glueball.,width=283 ] performing the continuum limit and using @xmath45 fm we obtain a mass of @xmath46 mev for the @xmath42 glueball .
finally , we also determine the @xmath47 glueball mass at a lattice spacing @xmath48 fm and obtain @xmath49 mev @xcite .
measuring glueball states on rather coarse lattices immediately calls for anisotropic gauge actions .
the construction of such an action using the fp approach is under way @xcite .
the new parametrization presented here provides a method for approximating fp gauge actions which is very general and flexible but still simple .
it is possible to describe the fp actions accurately enough so that their classical properties are preserved .
however , it is necessary to check the range of validity of the parametrized fp action . in view of future applications of the action in connection with a parametrized fp dirac operator @xcite
this is of crucial importance .
in addition , one has to assure that no pathologies are introduced through the parametrization .
for the quantities and lattice spacings investigated so far we find that this is indeed the case .
in particular , we find that these physical quantities scale even on lattices as coarse as @xmath50 fm .
99 p. hasenfratz and f. niedermayer , nucl . phys . *
b414 * , 785 ( 1994 ) , hep - lat/9308004 .
see e.g. p. hasenfratz , in `` non - perturbative quantum field physics , peniscola 1997 '' , hep - lat/9803027 , and references therein .
t. degrand , a. hasenfratz , p. hasenfratz and f. niedermayer , nucl . phys . *
b454 * , 587 ( 1995 ) , hep - lat/9506030 ; nucl . phys . *
b454 * , 615 ( 1995 ) , hep - lat/9506031 ; phys . lett . *
b365 * , 233 ( 1996 ) , hep - lat/9508024 ; m. blatter and f. niedermayer , nucl . phys .
* b482 * , 286 ( 1996 ) , hep - lat/9605017 .
f. niedermayer , p. rfenacht and u. wenger , ( 2000 ) , hep - lat/0007007 .
a. m. ferrenberg and r. h. swendsen , phys .
* 61 * , 2635 ( 1988 ) ; phys .
lett . * 63 * , 1195 ( 1989 ) .
k. j. juge , j. kuti and c. j. morningstar , nucl .
* 63 * , 326 ( 1998 ) , hep - lat/9709131 .
r. sommer , nucl .
b411 * , 839 ( 1994 ) , hep - lat/9310022 .
y. iwasaki , k. kanaya , t. kaneko and t. yoshie , phys .
* d56 * , 151 ( 1997 ) , hep - lat/9610023 . b. beinlich , f. karsch , e. laermann and a. peikert , eur .
j. * c6 * , 133 ( 1999 ) , hep - lat/9707023 .
d. w. bliss , k. hornbostel and g. p. lepage , ( 1996 ) , hep - lat/9605041 . g. boyd et .
al . , nucl . phys . *
b469 * , 419(1996 ) , hep - lat/9602007 .
r. g. edwards , u. m. heller and t. r. klassen , nucl . phys . *
b517 * , 377 ( 1998 ) , hep - lat/9711003 .
p. de forcrand et .
al . , nucl . phys . *
b577 * , 263 ( 2000 ) , hep - lat/9911033 .
m. teper , ( 1998 ) , hep - th/9812187 .
g. s. bali et .
al . , phys . lett . *
b309 * , 378 ( 1993 ) , hep - lat/9304012 .
a. vaccarino and d. weingarten , phys
* d60 * , 114501 ( 1999 ) , hep - lat/9910007 . c. j. morningstar and m. peardon , phys
. rev . * d56
* , 4043 ( 1997 ) , hep - lat/9704011 ; phys . rev .
* d60 * , 034509 ( 1999 ) , hep - lat/9901004 . c. liu , ( 2000 ) , hep - lat/0004018 .
p. rfenacht and u. wenger , these proceedings , hep - lat/0010057 .
p. hasenfratz et .
al . , hep - lat/0003013 ; these proceedings , hep - lat/0010061 .
|
we present a new parametrization of a su(3 ) fixed point ( fp ) gauge action using smeared ( `` fat '' ) gauge links .
we report on the scaling behaviour of the fp action on coarse lattices by means of the static quark - antiquark potential , the hadronic scale @xmath0 , the string tension @xmath1 and the critical temperature @xmath2 of the deconfining phase transition .
in addition , we investigate the low lying glueball masses where we observe no scaling violations within the statistical errors .
|
long flexible polymer macromolecules in a good solvent possess configurational statistics , which is perfectly captured by the model of self - avoiding random walks ( saw ) on a regular lattice @xcite .
this corresponds to the _ regime of polymer coils _ which holds provided that the temperature @xmath3 is above the so - called @xmath4-temperature . in this regime ,
the mean end - to - end distance of an @xmath5-step chain scales as @xmath6 , where @xmath7 is an universal exponent , which depends on space dimension @xmath8 only . with lowering the temperature
, the effect of monomer - monomer attraction grows and the polymer radius shrinks . at @xmath9 the effective repulsion due to the volume exclusion constraint
is exactly balanced by attractive interactions . at this particular temperature
a crossover occurs from high temperature saw behavior to @xmath4-statistics with exponent @xmath10 @xcite , @xmath11 . below the @xmath4-temperature ,
the entropic effects , which make the polymer chain swell , are overcome by interaction energy and a collapse to the _ globule regime _ ( with collapsed size exponent @xmath12 ) occurs .
it is generally accepted @xcite that the coil - globule transition is of second order , in the sense that the density of an infinite globule is zero at @xmath9 and increases continuously when @xmath3 is lowered further ; more presicely , it is a tricritical point with the upper critical dimension @xmath13 .
the coil - globule transition is one of the fundamental problems in polymer physics @xcite , being deeply connected with problems like protein folding , dna condensation and chromatin organization .
the properties of polymers in the vicinity of the @xmath4-point can be successfully studied on the basis of self - attractive self - avoiding walks ( sasaw ) , where a nearest - neighbor interaction is included .
estimates of the transition temperature @xmath2 for flexible polymers have been so far the subject of numerous studies @xcite .
note that the @xmath4-temperature is a non - universal quantity , depending on the lattice type ; in particular it appears to be proportional to the connectivity constant @xmath14 of a saw on a given type of lattice ( recent estimates of @xmath14 and @xmath4-temperatures on simple square and cubic lattices are given in table 1 ) .
the recent progress in experimental techniques employing optical tweezers @xcite , atomic force microscopy @xcite and soft microneedles @xcite makes it possible to monitor the behavior of various polymers under tension and stress .
in particular , applying a force on an isolated protein , the mechanism of force - driven phase transitions was studied , such as unfolding of giant titine protein @xcite , the stretching and unzipping of collapsed dna molecules @xcite .
of special interest in biophysics is the stretching of a collapsed polymer , i.e. , of a polymer in a poor solvent below the @xmath4-temperature .
unfolding proteins in this way could give important information on their spontaneous folding pathways .
an intriguing question is whether intermediate stages observed in different protein denaturating processes under applying force would always correspond to folding pathways in an enforced environment @xcite .
note that varying temperature alone , a polymer can not acquire the conformation of the completely stretched state with size exponent @xmath15 .
hence force not only influences the elastic , mechanical and structural properties of polymers , but also introduces a new stretched state ( _ regime of string _ ) which is otherwise not accessible .
the properties of force - induced globule - coil and coil - string transitions in @xmath0 , @xmath1 have been studied intensively both analytically @xcite and applying numerical simulations @xcite .
the existence of many intermediate states was found when unfolding under an applied force , which are not the same as the thermodynamically stable intermediate states .
the phase diagrams of the globule - coil and coil - string coexistence when varying temperature and stretching force were constructed .
new challenges have been raised recently in studying protein folding in the natural cellular environment @xcite .
real biological cells can be described as a very crowded environment built of the biochemical species , which occupy a large fraction of the total volume ; the volume exclusion " arises due to the steric repulsion between molecules @xcite . in the language of lattice models , the crowded environment with structural obstacles can be considered as a disordered lattice , where some amount of randomly chosen sites contains defects . of particular interest is the case , when the concentration @xmath16 of lattice sites allowed for the saws equals the critical concentration @xmath17 ( @xmath18 @xcite , @xmath19 @xcite ) and the lattice becomes percolative .
an incipient percolation cluster appears in the system , on which infinitely long saws can reside . in this regime ,
the scaling law for the end - to - end distance holds with a new exponent @xmath20 @xcite . a hint to the physical understanding of this phenomenon
is given by the fact that the percolation cluster itself is a fractal object with fractal dimension @xmath21 dependent on @xmath8 . in this way , scaling law exponents of saws change with the dimension @xmath21 of the ( fractal ) lattice on which the walk resides .
note that studying the scaling of saws on a percolative lattice , one is interested rather in the backbone of the percolation cluster , which is defined as follows .
assume that each site of the cluster is a resistor , the neighbour sites are connected by conducting bonds and an external potential drop is applied at two ends of the cluster .
the backbone is the subset of the cluster consisting of all sites through which the current flows ; i.e. , it is the structure left when all dangling ends " are eliminated from the cluster .
the saws can be trapped in dangling ends " , therefore infinitely long chains can only exist on the backbone of the cluster , which is characterized by its own fractal dimension @xmath22 ( @xmath23 @xcite , @xmath24 @xcite ) . whereas the scaling behavior of saws on percolative lattices served as a subject of numerous studies since the early 80th ( see , e.g. , ref .
@xcite for a recent review ) , less attention has been paid to peculiarities of the influence of the fractal structure of the underlying lattice on properties of the coil - globule transition .
the upper critical dimension shifts to @xmath25 for both saw and @xmath4-point statistics on a disordered lattice at the percolation threshold .
the scaling of polymer size at the coil - globule transition point is governed by @xmath26 for @xmath27 ( e.g. , @xmath28 @xcite , @xmath29 @xcite ) .
it is established that the value of the @xmath4-temperature is lowered due to the presence of disorder @xcite , but estimates for @xmath30 were found up to now only for the case of bond percolation .
the existing estimates are : @xmath31 @xcite , @xmath32 @xcite ( note , that corresponding values of connectivity constant for saw on bond - diluted percolative lattices read : @xmath33 and @xmath34 for @xmath0 and 3 , respectively @xcite ) .
the response of a polymer in crowded media to the stretching force modelled by the sasaw model on percolative lattice has been considered so far only in @xmath0 for relatively short chains by exact enumeration @xcite .
however , much more important is studying this problem in three dimensions , which describes real polymer systems .
this still needs a careful analysis and clarification . also , a quantitative description of the globule - coil transition under applying force , in particular the estimate of @xmath4-temperatures in disordered environment under stretching , still remains an open question .
the aim of the present study is to apply state - of - the - art numerical simulations to analyze the pecularities of sasaws on site - diluted lattices at the percolation threshold ( modelling flexible polymers in crowded environment ) under applied external stretching force in space dimensions @xmath0 , @xmath1 .
we estimate the shift of the @xmath4-temperature of the globule - coil transition under the influence of stretching and analyze the effect of applied force on the phase transitions between collapsed , extended and stretched phases .
the outline of the rest of the paper is as follows . in the next section
we describe the details of the numerical algorithm used in our study . in section
iii we present our results of coil - globule transition peculiarities for saws on percolative lattices , and in section iv we analyze the influence of stretching force on properties of the transition between collapsed and extended states .
we end up by giving conclusions and an outlook in section v.
we consider site percolation on regular lattices of edge lengths up to @xmath35 in dimensions @xmath36 , respectively .
each site of the lattice was assigned to be occupied with probability @xmath37 and empty otherwise . to obtain the backbone of a percolation cluster on a given disordered lattice
, we apply an algorithm consisting of the following two steps : first finding the percolation cluster based on the site - labeling method of hoshen and kopelman @xcite and then extracting the backbone of this cluster @xcite ( the algorithm is explained in detail in our previous papers @xcite ) .
we constructed 1000 clusters in each space dimension .
self - avoding walks on the backbone of a percolation cluster in @xmath0 under stretching force @xmath38.,width=177 ] to study sasaws on the backbone of percolation clusters , we apply the pruned - enriched rosenbluth method ( perm ) @xcite , taking into account that a sasaw can have its steps only on the sites belonging to the backbone of the percolation cluster . perm is based on the original rosenbluth - rosenbluth ( rr ) method @xcite and enrichment strategies @xcite .
the polymer grows step by step , i.e. , the @xmath39th monomer is placed at a randomly chosen empty neighbor site of the last placed @xmath40th monomer ( @xmath41 , where @xmath5 is the total length of the chain ) .
the growth is stopped , if the total length of the chain is reached . in order to obtain correct statistics , any attempt to place
a monomer at an already occupied site would result in discarding the entire chain .
this leads to an exponential attrition " ( the number of discarded chains grows exponentially with the chain length ) .
the bias due to avoiding this case is corrected in the rr algorithm by means of giving a weight @xmath42 to each sample configuration at the @xmath39th step , where @xmath43 is the number of free lattice sites to place the @xmath44th monomer .
this method is particularly useful for studying @xmath4-polymers , since the rosenbluth weights of the statistically relevant chains approximately cancel against their boltzmann probability .
population control in perm suggests pruning configurations with too small weights , and enriching the sample with copies of high - weight configurations @xcite .
these copies are made while the chain is growing , and continue to grow independently of each other . pruning and enrichment
are performed by choosing thresholds @xmath45 and @xmath46 depending on the estimate of the partition sums of the @xmath39-monomer chain .
if the current weight @xmath47 of an @xmath39-monomer chain is less than @xmath45 , a random number @xmath48 is chosen ; if @xmath49 , the chain is discarded , otherwise it is kept and its weight is doubled .
thus , low - weight chains are pruned with probability @xmath50 .
if @xmath47 exceeds @xmath46 , the configuration is doubled and the weight of each copy is taken as half the original weight .
for updating the threshold values we apply similar rules as in @xcite : @xmath51 and @xmath52 , where @xmath53 denotes the number of created chains having length @xmath39 , and the parameter @xmath54 controls the pruning - enrichment statistics . after a certain number of chains of total length @xmath5
is produced , the iteration is finished and a new tour starts .
we adjust the pruning - enrichment control parameter such that on average 10 chains of total length @xmath5 are generated per each iteration @xcite , and perform @xmath55 iterations .
one end of the chain is subjected to an external force @xmath38 acting in a chosen direction , say @xmath56 ( see fig . [ sawonpcf ] ) , while the other end ( the starting point ) is kept fixed .
the stretching energy @xmath57 arising due to the applied force for an @xmath39-step trajectory is given by : @xmath58 where @xmath59 is the @xmath56-component of distance from the starting point @xmath60 .
the rosenbluth weight factor @xmath47 is thus taken to be : @xmath61 where @xmath62 denotes the energy of the @xmath44-step chain ( @xmath63 with @xmath64 being an attractive energy between two nearest neighbors and @xmath65 the number of nearest neighbors contacts for a given chain ) and @xmath66 is the boltzmann constant . in what follows , we will assume units in which @xmath67 .
when a chain of total length @xmath5 is constructed , a new one starts from the same starting point , until the desired number of chain configurations is obtained .
the configurational averaging , e.g. , for stretching in @xmath56-direction , is then given by @xmath68 where @xmath69 and @xmath70 denote @xmath56-coordinates of start- and end - points , respectively , and @xmath71 is the weight of an @xmath5-monomer chain in a given configuration .
note , that studying saws on disordered lattices , we have to perform two types of averaging : the first average @xmath72 is performed over all saw configurations on a single backbone of percolation cluster ; the second average @xmath73 is carried out over different realizations of disorder , i.e. , over many backbone configurations .
statistical fluctuations of the energy @xmath74 of a polymer chain , expressed by the behaviour of the specific heat @xmath75 , signalize thermodynamic activity in the system , and thus the peak structure of @xmath75 as a function of temperature indicates transitions or crossovers between physically different states . in the case of a polymer system
, this corresponds to the transition between globule and coil regimes .
@xmath75 can be expressed via energy fluctuations as follows : @xmath76 it is worthwile first to discuss the behaviour of the energy distribution @xmath77 with varying temperature of the system , as presented in fig .
[ enerf0 ] .
since the energy distribution shows one peak only , the transition could be denoted as being second - order like .
the width of the distribution grows with increasing temperature , until it has reached its maximum broadening value .
this happens in the vicinity of the @xmath4-point . at higher temperatures ,
the distribution becomes narrower again .
left : energy distribution for @xmath78-step sasaws on a pure lattice in @xmath0 at different temperatures : @xmath79 ( squares ) , @xmath80 ( diamonds ) , @xmath81 ( triangles ) .
the broadening of the distribution curve emerges close to @xmath82 .
right : energy distribution for @xmath78-step sasaws on the backbone of percolation clusters in @xmath0 at different temperatures : @xmath83 ( squares ) , @xmath84 ( diamonds ) , @xmath85 ( triangles ) .
the broadening of the distribution curve emerges close to @xmath86.,title="fig:",width=158 ] left : energy distribution for @xmath78-step sasaws on a pure lattice in @xmath0 at different temperatures : @xmath79 ( squares ) , @xmath80 ( diamonds ) , @xmath81 ( triangles ) .
the broadening of the distribution curve emerges close to @xmath82 .
right : energy distribution for @xmath78-step sasaws on the backbone of percolation clusters in @xmath0 at different temperatures : @xmath83 ( squares ) , @xmath84 ( diamonds ) , @xmath85 ( triangles ) .
the broadening of the distribution curve emerges close to @xmath86.,title="fig:",width=158 ] specific heat per monomer as a function of temperature for a saw in @xmath0 on a pure lattice ( left ) and backbone of percolation cluster ( right ) .
squares : @xmath87 , diamonds : @xmath88 , triangles : @xmath89.,title="fig:",width=158 ] specific heat per monomer as a function of temperature for a saw in @xmath0 on a pure lattice ( left ) and backbone of percolation cluster ( right ) .
squares : @xmath87 , diamonds : @xmath88 , triangles : @xmath89.,title="fig:",width=158 ] specific heat per monomer as a function of temperature for a sasaw in @xmath90 on a pure lattice ( left ) and backbone of percolation cluster ( right ) .
squares : @xmath91 , diamonds : @xmath92 , triangles : @xmath93.,title="fig:",width=158 ] specific heat per monomer as a function of temperature for a sasaw in @xmath90 on a pure lattice ( left ) and backbone of percolation cluster ( right ) .
squares : @xmath91 , diamonds : @xmath92 , triangles : @xmath93.,title="fig:",width=158 ] for comparison with available results and testing our methods , we performed also simulations on pure lattices
. figures [ cv2f0 ] , [ cv3f0 ] show the specific - heat behavior of sasaws on a pure lattice and the backbone of percolation clusters in @xmath94 for several different chain lengths .
note , that the maximum of the specific heat per monomer grows with @xmath5 for saws on a pure lattice , whereas for the case of the backbone of percolation clusters it descreases with increasing @xmath5 . for finite chain length @xmath5 ,
the temperature defined by position of specific heat maximum @xmath95 is well below the collapse transition @xmath96-temperature .
this finite - size deviation of @xmath95 from @xmath2 obeys scaling behavior with @xmath5 .
[ cvmaxf0 ] presents the chain - length dependence of the specific - heat peaks for pure and percolative lattices . for the case of pure lattice
, @xmath90 is the upper critical dimension for @xmath4-transition , and the mean - field flory - huggins theory @xcite suggests : @xmath97 in the general case , the approach of @xmath95 to its limiting value obeys : @xmath98 where @xmath99 are constants , @xmath100 is the size exponent of a saw at @xmath4-point : @xmath101 peak temperatures of the specific heat of a sasaw in @xmath0 ( left ) and @xmath90 ( right ) at different chain length @xmath5 .
squares : pure lattice , diamondes : backbone of percolation cluster.,title="fig:",width=158 ] peak temperatures of the specific heat of a sasaw in @xmath0 ( left ) and @xmath90 ( right ) at different chain length @xmath5 .
squares : pure lattice , diamondes : backbone of percolation cluster.,title="fig:",width=158 ] note , that we consider the special case of collapse transition on the site - diluted percolative lattice , which was not studied before . left : energy distribution for @xmath78-step sasaws on pure @xmath0 lattice at @xmath84 under a force @xmath38 acting in the environment : @xmath102 ( squares ) , @xmath103 ( diamonds ) , @xmath104 ( triangles ) .
right : energy distribution for @xmath78-step sasaws on the backbone of percolation clusters at @xmath105 under force acting in the environment : @xmath102 ( squares ) , @xmath103 ( diamonds ) , @xmath104 ( triangles).,title="fig:",width=158 ] left : energy distribution for @xmath78-step sasaws on pure @xmath0 lattice at @xmath84 under a force @xmath38 acting in the environment : @xmath102 ( squares ) , @xmath103 ( diamonds ) , @xmath104 ( triangles ) .
right : energy distribution for @xmath78-step sasaws on the backbone of percolation clusters at @xmath105 under force acting in the environment : @xmath102 ( squares ) , @xmath103 ( diamonds ) , @xmath104 ( triangles).,title="fig:",width=158 ] specific heat per monomer of a sasaw with @xmath106 steps in @xmath0 under stretching force @xmath38 as a function of temperature . left : pure lattice , right : backbone of percolation cluster .
squares : @xmath107 , diamonds : @xmath108 , triangles : @xmath109.,title="fig:",width=158 ] specific heat per monomer of a sasaw with @xmath106 steps in @xmath0 under stretching force @xmath38 as a function of temperature . left : pure lattice , right : backbone of percolation cluster .
squares : @xmath107 , diamonds : @xmath108 , triangles : @xmath109.,title="fig:",width=158 ] specific heat per monomer of a sasaw with @xmath106 steps in @xmath90 under stretching force @xmath38 as a function of temperature . left : pure lattice , right : backbone of percolation cluster .
squares : @xmath107 , diamonds : @xmath108 , triangles : @xmath109.,title="fig:",width=158 ] specific heat per monomer of a sasaw with @xmath106 steps in @xmath90 under stretching force @xmath38 as a function of temperature . left : pure lattice , right : backbone of percolation cluster .
squares : @xmath107 , diamonds : @xmath108 , triangles : @xmath109.,title="fig:",width=158 ] our estimates for @xmath30 are obtained by least - square fitting of ( [ tcv ] ) , taking into account ( [ phi2pc ] ) and ( [ phi3pc ] ) , and presented in table 1 . as it was already mentioned , the collapse transition temperature is a non - universal property , which depends on the lattice type and , in particular , on the connectivity constant . as expected , the values of @xmath30 appear to be smaller than the pure lattice values , as far as relation @xmath110 holds .
in addition , due to the fact that @xmath111 decreases with @xmath8 , the same tendency is reflected in the transition temperature behaviour : @xmath112 .
this is in contrast to the values on the pure lattice , but coincides with the tendency of bond - diluted percolation values , cited in the introduction .
let us note , finally , that our values for @xmath30 of site percolation are larger than that of bond - percolation , which can be explained again by the difference of connectivity constants of these two fractal lattice structures .
let now act an additional stretching force @xmath38 in the @xmath56-direction of sasaw trajectory , fixed at its starting point ( see fig .
the energy distributions @xmath77 at fixed temperature and varying @xmath38 are presented in fig .
[ enert ] for the cases of the pure lattice and the backbone of percolation clusters for comparison .
we have chosen the value of @xmath3 in both cases to be well below the corresponding @xmath4-temperatures , so that at @xmath113 we restore the energy distribution of the globular state . with increasing @xmath38 , the averaged energy of the chain decreases applied force stretches the polymer globule .
the width of the distribution changes at increasing @xmath38 , and reaches its maximum broadening in the vicinity of the @xmath4-point . the value of the transition temperature is now shifted by the presence of force . at higher values of @xmath38 ,
the distribution becomes narrower again the chain is an in extended state .
peak temperatures of the specific heat of a sasaw as a function of chain length @xmath5 .
upper row : @xmath0 , left : pure lattice , right : backbone of percolation cluster .
lower row : @xmath90 , left : pure lattice , right : backbone of percolation cluster .
squares : @xmath113 , diamonds : @xmath107 , triangles : @xmath108 , pluses : @xmath109 .
, title="fig:",width=162 ] peak temperatures of the specific heat of a sasaw as a function of chain length @xmath5 .
upper row : @xmath0 , left : pure lattice , right : backbone of percolation cluster .
lower row : @xmath90 , left : pure lattice , right : backbone of percolation cluster .
squares : @xmath113 , diamonds : @xmath107 , triangles : @xmath108 , pluses : @xmath109 . , title="fig:",width=158 ] peak temperatures of the specific heat of a sasaw as a function of chain length @xmath5 .
upper row : @xmath0 , left : pure lattice , right : backbone of percolation cluster .
lower row : @xmath90 , left : pure lattice , right : backbone of percolation cluster .
squares : @xmath113 , diamonds : @xmath107 , triangles : @xmath108 , pluses : @xmath109 . , title="fig:",width=158 ] peak temperatures of the specific heat of a sasaw as a function of chain length @xmath5 .
upper row : @xmath0 , left : pure lattice , right : backbone of percolation cluster .
lower row : @xmath90 , left : pure lattice , right : backbone of percolation cluster .
squares : @xmath113 , diamonds : @xmath107 , triangles : @xmath108 , pluses : @xmath109 .
, title="fig:",width=158 ] phase diagrams of the stretching of a sasaw under applied force @xmath38 .
upper row : @xmath0 , left : pure lattice , right : backbone of percolation cluster .
the solid line presents results of exact enumeration @xcite .
lower row : @xmath90 , left : pure lattice , right : backbone of percolation cluster .
the filled triangle present result of @xcite @xmath114 .
, title="fig:",width=158 ] phase diagrams of the stretching of a sasaw under applied force @xmath38 .
upper row : @xmath0 , left : pure lattice , right : backbone of percolation cluster .
the solid line presents results of exact enumeration @xcite .
lower row : @xmath90 , left : pure lattice , right : backbone of percolation cluster .
the filled triangle present result of @xcite @xmath114 .
, title="fig:",width=158 ] phase diagrams of the stretching of a sasaw under applied force @xmath38 .
upper row : @xmath0 , left : pure lattice , right : backbone of percolation cluster .
the solid line presents results of exact enumeration @xcite .
lower row : @xmath90 , left : pure lattice , right : backbone of percolation cluster . the filled triangle present result of @xcite @xmath114 .
, title="fig:",width=158 ] phase diagrams of the stretching of a sasaw under applied force @xmath38 .
upper row : @xmath0 , left : pure lattice , right : backbone of percolation cluster .
the solid line presents results of exact enumeration @xcite .
lower row : @xmath90 , left : pure lattice , right : backbone of percolation cluster .
the filled triangle present result of @xcite @xmath114 .
, title="fig:",width=161 ] to study the @xmath4-transition of sasaws , when the external stretching force is acting in the environment , we are working in the constant - force " ensemble .
fixing the value of @xmath38 , we study the specific - heat behavior ( figs .
[ cvf2 ] , [ cvf3 ] ) . analyzing the peak structure of the specific heat
, we immediately conclude , that increasing the value of @xmath38 leads to decreasing the transition temperature .
[ cvf ] presents the chain - length dependence of the specific - heat peaks for a pure and percolative lattices at several different values of @xmath38 . for finite chain length @xmath5 , the temperature defined by the position of the
specific - heat maximum @xmath95 is well below the collapse transition @xmath96-temperature .
our estimates for @xmath30 in the presence of force are obtained by least - square fitting of ( [ tcv ] ) with ( [ phi2 ] ) , ( [ phi2pc ] ) , ( [ phi3pc ] ) .
results are presented in fig .
[ cvdiagram ] in the form of a phase diagram of transitions from globule to extended state , and listed in table 2 . for the pure lattice in @xmath0 ,
we compare our results with an exact enumeration study of kumar et al .
@xcite , where sasaws of length up to @xmath115 under stretching force were studied .
our results also appear to be in good correspondence with that of ref .
@xcite . averaged extention in @xmath56-direction , giving information about the internal structure of the polymer configuration under applying force , is presented in fig .
[ exten ] for the cases of pure lattice and backbone of percolation cluster . at small forces ,
a polymer chain is still in the compact folded state and slightly oriented along the force direction . at larger forces ,
the polymer chain has a conformation similar to the extended ( swollen ) structure .
note , that completely stretched states , corresponding to @xmath116 can be obtained only on the pure case and are not accessible on the percolative lattices due to complicated fractal structure on the underlying percolation cluster .
our estimates of @xmath117 in @xmath0 appear to be in a good correspondence with that of available exact enumeration studies @xcite .
upper row : averaged extentions of a sasaw under applying force acting in a @xmath0 environment at @xmath83 ; left - pure lattice , right - backbone of percolation cluster , squares : @xmath118 , diamonds : @xmath87 , triangles : @xmath88 .
lower row : averaged extentions of a sasaw under applying force acting in a @xmath90 environment ; left - pure lattice ( @xmath119 ) , right - backbone of percolation cluster ( @xmath84 ) , squares : @xmath118 , diamonds : @xmath87 , triangles : @xmath88 . , title="fig:",width=151 ] upper row : averaged extentions of a sasaw under applying force acting in a @xmath0 environment at @xmath83 ; left - pure lattice , right - backbone of percolation cluster , squares : @xmath118 , diamonds : @xmath87 , triangles : @xmath88 .
lower row : averaged extentions of a sasaw under applying force acting in a @xmath90 environment ; left - pure lattice ( @xmath119 ) , right - backbone of percolation cluster ( @xmath84 ) , squares : @xmath118 , diamonds : @xmath87 , triangles : @xmath88 .
, title="fig:",width=151 ] upper row : averaged extentions of a sasaw under applying force acting in a @xmath0 environment at @xmath83 ; left - pure lattice , right - backbone of percolation cluster , squares : @xmath118 , diamonds : @xmath87 , triangles : @xmath88 .
lower row : averaged extentions of a sasaw under applying force acting in a @xmath90 environment ; left - pure lattice ( @xmath119 ) , right - backbone of percolation cluster ( @xmath84 ) , squares : @xmath118 , diamonds : @xmath87 , triangles : @xmath88 .
, title="fig:",width=151 ] upper row : averaged extentions of a sasaw under applying force acting in a @xmath0 environment at @xmath83 ; left - pure lattice , right - backbone of percolation cluster , squares : @xmath118 , diamonds : @xmath87 , triangles : @xmath88 . lower row : averaged extentions of a sasaw under applying force acting in a @xmath90 environment ; left - pure lattice ( @xmath119 ) , right - backbone of percolation cluster ( @xmath84 ) , squares : @xmath118 , diamonds : @xmath87 , triangles : @xmath88 . , title="fig:",width=151 ]
we studied the self - attracting self - avoding walks on disordered lattices in space dimensions @xmath0 , @xmath1 , modelling flexible polymer macromolecules in porous environment .
we considered the special case , when the concentration of disorder is exactly at the percolation threshold , so that an incipient percolation cluster of sites , allowed for saws , emerges on the lattice . in our study , sasaws reside only on the backbone of the percolation cluster , which has a fractal structure . in the first part of our study
, attention has been paid to the influence of the fractal structure of the underlying lattice on properties of the coil - globule transition . applying the pruned - enriched rosenbluth algorithm
, we obtain estimates of the collapse transition temperature @xmath30 of sasaws on site - diluted percolative lattices in @xmath0 and @xmath1 dimensions ( note , that so far only estimates for the bond - percolation case have been found ) .
the values of @xmath30 appear to be smaller than the pure lattice value .
this can be explained , remembering that the collapse transition temperature is proportional to the connectivity constant @xmath14 of saws on a given lattice .
in addition , due to the fact that @xmath111 decreases with @xmath8 , the same tendency is reflected in the anomalous transition temperature behaviour : @xmath112 . next , keeping one end of a sasaw trajectory on the backbone of percolation cluster fixed , we applied a stretching force , acting in some chosen direction ( say , @xmath56 ) .
especially interesting was to study this problem in @xmath90 , which corresponds to real polymer systems and was not considered before .
we estimated the shift of the @xmath4-temperature of the globule - coil transition under the stretching and constructed phase diagrams of collapsed and extended states coexistence in @xmath94 . as expected , the presence of stretching force in environment leads to a decreasing @xmath4-temperature value .
we thank sanjay kumar for interesting discussions and communicating ref . @xcite prior to publication .
v.b . is grateful for support through the alexander von humboldt foundation and the schsische dfg - forschergruppe for877 .
m. rief , m. gautel , f. oesterhelt , j.m .
fernandez , and h.e .
gaub , science * 276 * , 1109 ( 1997 ) ; m.s.z .
kellermayer , s.b .
smith , h.l .
granzier , and c. bustamante , science * 276 * , 1112 ( 1997 ) ; l. tskhovrebova , j. trinick , j.a .
sleep , and r.m .
simmons , nature * 387 * , 308 ( 1997 ) .
baumann , v. bloomfield , s.b .
smith , c. bustamante , m. wang , and s.m .
block , biophys .
j. * 78 * , 1965 ( 2000 ) ; c. danilowicz , y. kafri , r.s .
conroy , v.w .
coljee , j. weeks , and m. prentiss , phys .
rev . lett . * 93 * , 078101 ( 2004 ) .
d. gritz , s. kreitmeier , and m. wittkop , j. macromol .
* 35 * 615 ( 1996 ) ; m. wittkop , s. kreitmeier , and d. gritz , phys . rev .
e * 53 * 838 ( 1996 ) ; s. kreitmeier , m. wittkop , and d. gritz , ibid . * 59 * , 1982 ( 1999 ) .
d. marenduzzo , a. maritan , and f. seno , j. phys .
a * 35 * , l233 ( 2002 ) ; d. marenduzzo , s.m .
bhattacharjee , a. maritan , e. orlandini , and f. seno , phys .
* 88 * , 028102 ( 2001 ) ; d. marenduzzo , a. maritan , a. rosa , and f. seno , phys . rev . lett . * 90 * , 088301 ( 2003 ) .
s. kumar , d. giri , and s.m .
bhattacharjee , phys .
e. * 71 * 051804 ( 2005 ) ; s. kumar and d. giri , phys .
e. * 72 * , 052901 ( 2005 ) ; h. zhou , j. zhou , z.c .
ou - yang , and s. kumar , phys .
lett . * 97 * , 158302 ( 2006 ) ; s. kumar and d. giri , phys .
lett . * 98 * 048101 ( 2007 ) ; s. kumar , i. jensen , j.l .
jacobsen , and a.j .
guttmann , phys .
lett . * 98 * , 128101 ( 2007 ) .
|
we study flexible polymer macromolecules in a crowded ( porous ) environment , modelling them as self - attracting self - avoiding walks on site - diluted percolative lattices in space dimensions @xmath0 , @xmath1 .
the influence of stretching force on the polymer folding and properties of globule - coil transitions are analyzed . applying the pruned - enriched rosenbluth chain - growth method ( perm )
, we estimate the transition temperature @xmath2 between collapsed and extended polymer configurations and construct the phase diagrams of the globule - coil coexistence when varying temperature and stretching force . the transition to a completely stretched state , caused by applying force ,
is discussed as well .
|
with the increase in both wireless channel bandwidth and the computational capability of wireless devices , wireless networks now can be used to support time critical applications such as video streaming or interactive gaming .
such time critical applications require the data content to reach the destination node(s ) in a timely fashion , i.e. , a delay deadline is imposed on packet reception , beyond which the reception becomes useless ( or invalid ) @xcite .
recently , network coding becomes a promising approach to improve wireless network performance @xcite .
specifically , the work in @xcite proposed the first network coding based packet forwarding architecture , named _
cope _ , to improve the throughput of wireless networks . with cope
, each node opportunistically overhears some of the packets transmitted by its neighbors , which are not intended to itself .
the relay node can then intelligently xor multiple packets and forward it to multiple next hops with only one transmission , which results in a significant throughput improvement . in most recent works for wireless network coding
, network nodes always transmit packets at a fixed rate
. however , most wireless systems are now capable of performing adaptive modulation to vary the link transmission rate in response to the signal to interference plus noise at the receivers .
transmission rate diversity exhibits a rate - range tradeoff : the higher the transmission rate , the shorter the transmission range for a given transmission power @xcite . to aid overhearing
, one may use the lowest transmission rate , so as to successfully deliver packet to more receivers / overhearing nodes .
although this may increase the coding opportunity , it may not yield good performance , especially for time critical applications , as the arrival times of packets may be delayed . in the literature ,
only a few works studied the relationships between adapting the transmission rate and the network coding gain @xcite .
the work in @xcite showed that compared with pure network coding scheme , joint rate adaptation and network coding is more effective in throughput performance .
they also proposed a joint rate selection and coding scheme to minimize the sum of the uplink and the downlink costs in star network topology .
the work in @xcite mathematically formulated the optimal packet coding and rate selection problem as an integer programming problem , and proposed an efficient heuristic algorithm to jointly find a good combination of coding solution and the transmission rate .
however , there are only a few works considered the delay guarantee of packet receptions , which is especially important for time critical applications .
so far , only @xcite considered the delay constraint of packet reception , and proposed a coding scheme to minimize the number of packets that miss their deadlines .
however , they assume that the transmission rates on all the links are fixed and the same . in this paper , by considering the impact of both transmission rate and network coding on the packet reception delay , we design a joint rate selection and network coding ( rsnc ) scheme for wireless time critical applications , so as to minimize the total number of packets that will miss their deadlines at the destination nodes .
the main contributions of our paper can be concluded as follows . * we propose a novel graph model , which considers both the heterogenous transmission rates and the deadlines of the packet receptions during the graph construction . based on the graph model
, we mathematically formulate the problem of minimizing the total number of packets that miss their deadlines by joint rate selection and network coding , as an integer programming problem . *
we propose a metric to determine the coded packet and the transmission rate for each packet transmission . by considering the impact of the transmission rate on both delay and network coding gain
, we also design an efficient algorithm to optimize the proposed metric .
* we compare the performance of the proposed rsnc scheme with some existing algorithms .
simulation results show that the proposed scheme can significantly reduce the packet deadline miss ratio .
the rest of the paper is organized as follows .
we define our problem in sec .
[ sec.formulation ] .
[ sec.solution ] gives the graph model and problem formulation .
the algorithm design for each transmission is given in sec .
[ sec.algorithm ] .
we show the simulation results in sec .
[ sec.simulation ] , and conclude the paper in sec .
[ sec.conclusion ] .
[ fig.rsnc ]
in this section , we first illustrate the motivation of our problem . we then give the problem description and its complexity .
we now give an example to show how joint rate selection and network coding affect the time critical applications .
take fig .
[ fig.rsnc ] as an example , where source / transmitting node node @xmath0 needs to transmit packet @xmath1 to node @xmath2 respectively .
[ fig.rsnc](a ) gives the set of overheard packets @xmath3 at destination @xmath4 .
suppose that the size of each packet is @xmath5 , and the maximum transmission rates from @xmath0 to @xmath2 are @xmath6 and @xmath7 , respectively .
[ fig.rsnc](b ) shows the reception deadline of each wanted " packet at its destination . for the current transmission , according to the work in @xcite , @xmath0 will send the encoded packet @xmath8 , as the most number of destinations can decode it .
however , there is a problem for selecting the transmission rate at @xmath0 .
if @xmath9 is selected , @xmath10 can not successfully receive the packet , as the maximum transmission rates from @xmath0 to them are both @xmath7 .
if @xmath7 is selected , although all of the three receivers @xmath2 can receive and decode one wanted " packet , @xmath11 will miss its deadline at @xmath12 , as its arrival time is @xmath13 . as an alternative
, we may choose to first send packet @xmath11 with transmission rate @xmath9 , where destinations @xmath12 will obtain a wanted " packet in @xmath14 . after this transmission , the encoded packet @xmath15 can be sent with transmission rate @xmath7 , where destination @xmath16 and @xmath17 will obtain a wanted " packet after @xmath18 ( including the waiting time of the first transmission ) . obviously , the latter solution is better than the first one , as no packet will miss their deadline . in this paper
, we consider the application of network coding in wireless networks .
each network node knows the overheard / routed packets that its neighbors have such that it can perform network coding operations .
such information can be achieved by using _
reception reports _ , as introduced in @xcite .
we also assume that the forwarding / relaying node knows the deadlines of the packet receptions at its receivers . specifically , we consider the transmission scheme within a single hop since multi hop can be regarded as multiple single hops . as in cope
@xcite , only xors coding is performed at the node in our work . without loss of generality ,
let @xmath0 be the current source node , and @xmath19 be the set of packets required to be transmitted from @xmath0 .
suppose that @xmath20 is the set of @xmath0 s neighbors which requires packets in @xmath21 sent from node @xmath0 .
let @xmath22 be the set of wanted " packets at @xmath23 , and @xmath3 be the set of overheard packets at @xmath23 , where @xmath24 . for each @xmath25 ,
let @xmath26 be the reception deadline of packet @xmath27 at node @xmath23 .
we also assume that @xmath28 is the maximum transmission rate on link @xmath29 , and only if the transmission rate from @xmath0 to @xmath23 is less than @xmath28 , the packet sent from @xmath0 can be successfully received by @xmath23 @xcite .
we also assume that the size of each packet is @xmath30 .
our problem is that given the set of overheard packets at each node @xmath23 , @xmath3 , the set of packets required by @xmath23 , @xmath22 , the deadline of required packet @xmath27 at node @xmath23 , @xmath26 , and the maximum transmission rate @xmath28 on the link from @xmath0 to @xmath23 , design the encoding strategy of the packets and select the transmission rate for each propagation , such that the total number of packets that miss their deadlines at each destination is minimized .
let @xmath31 be @xmath32 if packet @xmath27 misses its deadline at @xmath23 , otherwise , let it be @xmath33 , where @xmath25 .
thus , our objective is to minimize @xmath34 in this paper , we refer such a problem of joint rate selection and network coding ( rsnc ) for time critical applications as rsnc problem .
the rsnc problem is np - hard .
we can consider a special case of the rsnc problem : @xmath35 and the maximum transmission rates on all the links are the same . then
, this special case is equivalent to finding a maximum weight clique problem as in @xcite , which is known as an np - hard problem .
thus , the rsnc problem is also np - hard .
although the graph model in @xcite works well for the case where the transmission rates on all the links are the same and fixed , it can not be used directly for our rsnc problem . here
, we construct a novel graph model @xmath36 , which considers both the transmission rates and the packet reception deadlines .
we define @xmath37 as the minimum transmission rate that can be used to meet the deadline of @xmath25 at @xmath23 .
we add a vertex @xmath38 in @xmath39 , only if the following two conditions can be met .
\(1 ) @xmath25 ; \(2 ) @xmath40 . note that , if @xmath41 , packet @xmath27 will definitely miss its deadline at @xmath23 .
thus , conditions ( 1 ) and ( 2 ) ensure that we add a vertex @xmath38 in @xmath39 only if the wanted " packet @xmath27 will not miss its deadline at @xmath42 .
that is , @xmath43 .
then , for any two different vertices @xmath44 , there is a link @xmath45 if all the following conditions can be satisfied .
\(a ) @xmath46 ; \(b ) @xmath47 or @xmath48 and @xmath49 ; \(c ) @xmath50 and @xmath51 . for any clique @xmath52 in @xmath53 ,
let @xmath54 .
according to the work in @xcite , if node @xmath55 successfully receives the encode packet @xmath56 , where @xmath57 , @xmath23 can decode a wanted " packet @xmath27 , where @xmath58 .
next , we will use an example to show the novelty of our graph model as compared to others in the literature , e.g. , @xcite .
still take fig .
[ fig.rsnc ] as an example .
the graph constructed by @xcite is shown in fig .
[ fig.example ] ( a ) . according to @xcite ,
any clique in the graph represents a feasible encoded packet .
thus , @xmath8 can be sent and its intended next hops are @xmath2 , because @xmath59,@xmath60,@xmath61 forms a clique . as described in sec .
[ sec.formulation.motivation ] , it is not a good choice .
however , with our graph model shown in fig .
[ fig.example](b ) , @xmath62 will not be encoded as vertices @xmath63,@xmath60,@xmath64 do not form a clique in the graph .
in addition , for the current transmission , the encoded packet derived from any clique in the graph can be sent without missing the deadlines at its intended destinations .
for example , if @xmath15 , which is derived from the clique @xmath65 , is sent with the minimum of the maximum transmission rates among @xmath66 and @xmath67 , @xmath7 , its intended next hops @xmath10 can successfully decode the packets @xmath68 respectively , without missing their deadlines .
[ fig.example ] with the graph @xmath36 , we have the following lemma .
[ lemma_graph ] for the current packet transmission , if the encode packet @xmath56 , where @xmath69 , is sent with the transmission rate @xmath70 , it will be received by all the nodes in @xmath71 . in addition , for each @xmath58 , the packet @xmath27 will be decoded by @xmath23 without missing its deadline .
firstly , we can easily obtain that with transmission rate @xmath70 , all the receivers in @xmath71 can successfully receive the sending packet .
this is because the transmission rate @xmath72 must be lower than the maximum transmission rate from @xmath0 to @xmath55 .
secondly , our graph is the subgraph of that constructed in @xcite . according to @xcite ,
if @xmath73 is successfully received by @xmath74 , @xmath23 can decode its wanted " packet @xmath27 , where @xmath58 .
thus , any receiver @xmath55 can obtain a wanted " packet @xmath27 from @xmath75 with transmission rate @xmath72 , where @xmath58 .
thirdly , according to the condition ( c ) , we have @xmath76 so , its arrival time at receiver @xmath23 is @xmath77 in other words , the arrival time of the packet @xmath78 will not miss its deadline at its receiver @xmath55 , where @xmath58 . as in lemma
[ lemma_graph ] , a clique @xmath79 in the graph represents a feasible transmission solution for the current propagation , with the encoded packet @xmath75 , transmission rate @xmath70 , intended next hops in @xmath71 , and the propagation delay @xmath80 .
while lemma [ lemma_graph ] ensures that any encoding strategy based on any clique in the graph will be delivered within deadline for the current packet transmission , the transmission orders of the encoded packets , represented by the cliques in @xmath36 , also affects the timely packet receptions at their destinations . as shown in fig .
[ fig.example](b ) , if we first schedule packet @xmath11 with transmission rate @xmath9 , represented by clique @xmath81 , and then schedule packet @xmath15 with transmission rate @xmath7 , represented by clique @xmath65 , all the packets will be received at their destinations without missing deadlines . however ,
if we first schedule packet @xmath15 , and then packet @xmath11 , packet @xmath11 will miss its deadline at @xmath12 .
thus , the next task for us is to find a set of cliques in the graph and schedule the transmissions of the encoded packets represented by each clique , so as to minimize the number of missed packets for the whole transmission process .
suppose that @xmath82 is a clique in the graph , and the encoded packet represented by it is sent as the @xmath83-th transmission at node @xmath0 .
we also assume that @xmath84 , @xmath85 .
thus , the @xmath83-th transmission at @xmath0 is @xmath86 where @xmath87 , and the transmission rate is @xmath88 .
let @xmath89 be the transmission delay of the @xmath83-th transmission .
we firstly define the following variant .
@xmath90 then , we can formulate the rsnc problem based on the graph model as follows .
@xmath91 subject to@xmath92 where @xmath93 is a sufficient large constant . in the above formulation , the term of the objective represents the number of packets that miss their deadlines at the receivers , which needs to be minimized . constraint ( [ ch.1 ] ) denotes that each vertex in the graph can only belong to one clique .
constraint ( [ ch.2 ] ) means that if there is no link between vertex @xmath38 and @xmath94 , vertices @xmath95 can not be in the same clique .
constraint ( [ ch.3 ] ) gives the transmission delay for the @xmath83-th transmission , which is equal to the transmission delay with the minimum transmission rate among the rates from @xmath0 to all intended receivers .
the sufficient large constant @xmath93 is used to guarantee that if @xmath96 , @xmath31 must be @xmath32 , as denoted in constraint [ ch.4 ] , and if @xmath97 , @xmath31 must be @xmath33 , as denoted in constraint [ ch.5 ] . note that the arrival time of the packet in the @xmath83-th transmission should consist of both the waiting time of the previous @xmath98 transmissions and the transmission time of the @xmath83-th transmission , i.e. , @xmath99 .
thus , constraint ( [ ch.4 ] ) and ( [ ch.5 ] ) show that only if the arrival time of @xmath27 at @xmath23 , i.e. , @xmath100 , is no more than the reception deadline of @xmath27 at @xmath23 , @xmath31 can be @xmath33 . with the above integer nonlinear programming , we can get the optimal solution of rsnc problem . however , the computational complexity for the above integer nonlinear programming is too high when the graph is large .
thus , we need to find an efficient algorithm to solve it .
since each clique in the graph represents a feasible transmission strategy for the current transmission , instead of determining the whole transmission sequence at once , we first design the algorithm to determine the encoding strategy and rate selection scheme for each packet propagation , by selecting a clique at a time . the whole transmission process
consists of multiple packets transmission / cliques selection .
first of all , in order to measure the goodness " of transmitting an encoded packet at a specific transmission rate for each packet propagation , it is necessary for us to adopt a reasonable metric which should take into account the impact of the transmission rate and the packet reception deadlines . in this section , we shall design a metric , which not only satisfies as more requests as possible , but also minimizes the number of packets missing the deadlines after the current transmission .
for the current transmission , given an encoded packet and a selected transmission rate , let @xmath101 be @xmath32 if @xmath27 is decoded / received by @xmath23 from the current propagation without missing its deadline , otherwise , let it be @xmath33 .
in addition , as described in fig .
[ fig.rsnc ] , the current encoding strategy and transmission rate also affect the timely receptions of the packets in the following propagations .
let @xmath102 be 1 if @xmath27 will definitely miss its deadline at @xmath23 after the current propagation , otherwise , let it be @xmath33 .
later , we will introduce how to calculate @xmath101 and @xmath102 for a given encoded packet and transmission rate .
let @xmath72 be the transmission rate selected for the current propagation .
our metric can be defined as follows . for a coding solution @xmath103 , define the metric @xmath104 when using the transmission rate @xmath72 as follows : where @xmath105 is the parameter , which can be defined as the benefit ( e.g. , importance ) of the packet @xmath27 .
firstly , for a given encoded packet @xmath103 , @xmath101 is @xmath32 only if all the following conditions are met : 1 ) @xmath106 , which means @xmath23 can successfully receive the sending packet ; 2 ) @xmath25 , which means @xmath27 is required at @xmath23 ; 3 ) all the other packets combined in the encoded packet except @xmath27 are available at @xmath23 , which denotes the decoding requirement of @xmath27 at @xmath23 ; 4 ) @xmath107 , which shows the requirement of the reception deadline .
secondly , for each packet @xmath25 which is not successfully received / decoded by @xmath23 from the current transmission without missing deadlines ( @xmath108 ) , @xmath102 is @xmath32 only if @xmath109 here , @xmath80 is the transmission delay of the current transmission , and @xmath110 denotes the minimum delay to meet @xmath27 s deadline at @xmath23 in the next transmission . if the sum of the current transmission delay and the next minimum transmission delay is larger than the deadline of @xmath27 at @xmath23 , @xmath27 will definitely miss its deadline , i.e. , @xmath111 .
thus , given an encoded packet and its transmission rate , @xmath101 and @xmath102 are both determined .
hence the meaning of metric @xmath104 in ( [ equation.u ] ) can be explained as follows . the first term @xmath112 denotes the benefit obtained from the packets that are received without missing their deadlines from the current transmission . the second term @xmath113 represents the lost due to packets that will definitely miss their deadlines after the current transmission .
so , the metric @xmath104 denotes the net benefit obtained from the current encoded packet and the transmission rate .
thus , for each packet propagation , we aim to determine an encoded packet and select the transmission rate @xmath72 , that will maximize the metric @xmath104 .
note that , the problem of maximizing the defined metric @xmath104 is also np - hard
. we can prove it by considering its special case : the transmission rates on all the links are the same , the reception deadlines for all the packets are the transmission time of one packet , and each packet has the same benefit @xmath105 .
the special case of maximizing the defined metric @xmath104 becomes to maximize the total number of the receivers that can decode one wanted " packet from the current encoded packet , which has been proved to be np - hard in @xcite .
although maximizing the defined metric @xmath104 is also np - hard , we can easily obtain the following observations , based on which we can design the heuristic algorithm .
p1 : maximizing the first term of the metric @xmath104 is equal to find a maximum weight clique in the graph , where the weight at vertex @xmath38 is defined as the benefit @xmath105 .
p2 : the transmission rate is a parameter that adjusts the trade - off between delay and network coding gain . if @xmath0 uses a low transmission rate , more receivers can successfully receive the sending packet , and the current transmission may satisfy more receivers requirements , denoted by the first term in @xmath104 . however , low transmission rate means high transmission delay , which may cause more packets to miss their deadlines in the following transmission , denoted by the second term in @xmath104 . based on the above observations , we then design a heuristic algorithm for each packet propagation , by gradually increasing the transmission rate .
initially , the transmission rate is set to be no less than the lowest one from @xmath0 to its receivers .
let @xmath114 be the set of available transmission rates from @xmath0 to all the destinations , and let @xmath115 be the @xmath116-th lowest rate in @xmath117 . as in sec .
[ graph.model ] , we construct the auxiliary graph with the given information .
we also assign the weight @xmath118 in vertex @xmath38 for @xmath119 to denote the benefit of @xmath27 . in the @xmath116-th step ,
we restrict that the transmission rate used at @xmath0 must be no less than @xmath115 . for @xmath23 ,
if its maximum transmission rate from @xmath0 is less than @xmath115 , it can not successfully receive the sending packet .
this restriction can be realized by omitting the vertex @xmath38 in @xmath36 if @xmath120 .
then , we find the maximum weight clique in the subgraph @xmath121 , and adopt the transmission rate represented by the found clique .
each vertex @xmath38 in the found clique denotes that @xmath27 will be successfully obtained by @xmath23 without missing its deadline , for the given transmission rate . for each of the other packets that can not be obtained at their receivers from the current transmission
, we then judge whether it will definitely miss its deadline at its destinations , by ( [ eq.miss ] ) .
thus , in each step , we calculate @xmath104 .
such process continues until all the rates in @xmath117 are considered .
finally , we compare the values of @xmath104 obtained from each step and adopt the one with the largest value as solution .
note that , if there are more than one solution with the maximum value of @xmath104 , we will choose the one with the smaller lost represented by the second term of ( [ equation.u ] ) .
the detailed of the algorithm is shown in algorithm 1 of fig .
[ alg ] . while algorithm 1 in sec .
[ sec.algorithm.design ] describes the encoding of packets and selection of rate for every transmission , the whole transmission process will consist of multiple of such single process .
we will first construct the graph @xmath36 based on model in sec .
[ graph.model ] , and the graph will be updated by removing the selected vertices in the found clique by algorithm 1 , and the vertex @xmath38 if @xmath27 will definitely miss its deadline at destination @xmath23 .
the packet reception deadlines for the packets also need to be updated after each transmission .
the whole transmission process continues until the vertices set @xmath122 of @xmath53 becomes empty .
the detail algorithm for the whole transmission is given in algorithm 2 of fig .
in this section , we demonstrate the effectiveness of our rsnc scheme through simulations . we randomly generate a set of available packets in @xmath3 and the wanted " packets in @xmath22 at destination @xmath123 , where @xmath124 . the maximum transmission rate from @xmath0 to @xmath23 is randomly selected in @xmath125 $ ] , and
the packet reception deadline is randomly generated in @xmath126 $ ] . for comparison purpose ,
we include two baseline algorithms , namely , _
dsf ( deadline smallest first ) coding _ algorithm @xcite and _ sin-1 _ algorithm @xcite .
dsf coding algorithm does not consider the heterogenous transmission rates on the links , and in each time slot , always finds the maximum weight clique in the defined graph .
sin-1 algorithm always sends the packet with the minimum sin-1 " in each transmission , where sin-1 " of packet @xmath27 is defined as the ratio of the duration from the current time to the deadline of the most urgent request for packet @xmath27 , to the number of requests for @xmath27 . in the simulation
, we compare the deadline miss ratio under different transmission schemes , where deadline miss ratio is defined as the ratio of the number of packets missing their deadlines to the total number of requests . for each setting , we present the average result of 100 samples .
we first investigate the impact of the transmission rate on the performance of random one packet propagation during the whole transmission process . given the rate for the current transmission , we run the maximum weight clique algorithm in the graph model to find the maximum number of packets that can be obtained at their destinations without missing deadline , i.e. , satisfied requests , based on which we derive the number of packets that will definitely lose their deadlines in the next transmissions according to ( [ eq.miss ] ) , i.e. , failed requests .
we set @xmath127 , @xmath128 , and @xmath129 .
[ sim.tradeoff ] as shown in fig .
[ sim.tradeoff ] , with the increase of the transmission rate , the number of packets that can be successfully received / decoded by their receivers from the current transmission decreases .
the reason is that with the increase of the transmission rate , higher number of destinations may not receive the packets due to shorter transmission range , which thus decreases the encoding opportunity at @xmath0 .
we can also see that with the increase of the transmission rate , the number of packets that will definitely miss their deadlines at the receivers decreases .
this is because , higher transmission rates incur less transmission delay , which gives more chances for the other packets received timely in the following transmissions .
the above observation motivates the algorithm design in sec .
[ sec.algorithm.design ] , by deciding the tradeoff between transmission rate and network encoding strategy .
we then investigate the impact of the transmission rate on the performance of the whole transmission .
we set @xmath130 and vary the scale of the transmission rates , i.e. , @xmath131 $ ] . as shown in fig .
[ sim.rate ] , the deadline miss ratio with our rsnc scheme is much lower than with other schemes . in addition , the deadline miss ratio also decreases with the increase of the transmission rate .
this is because higher transmission rates incur less transmission delay , which satisfies more successful transmissions .
[ sim.rate ] [ sim.m ] we then investigate the impact of the number of destinations @xmath132 and the transmission rates on the deadline miss ratio . in this case , we set @xmath133 by varying @xmath132 in @xmath134 $ ] for @xmath135 and @xmath136 . as shown in fig .
[ sim.m ] , the deadline miss ratio with our rsnc scheme is much lower than with other schemes . with the increase of @xmath132 , the gain of our rsnc scheme increases .
we can also see that the dsf algorithm does not show significant gain over sin-1 algorithm .
this is because , although with network coding in dsf , more packets can be combined together , the encoded packet may still miss its deadline at some destinations , due to inappropriate transmission rate used . from fig .
[ sim.m ] , we see that , with the increase of @xmath132 , the deadline miss ratio increases .
the reason is that there are more packets to be sent at @xmath0 within the same deadline scale .
[ sim.n ] finally , we investigate the impact of the total number of packets @xmath137 and the reception deadlines on the deadline miss ratio .
we set @xmath138 by varying @xmath137 in @xmath139 $ ] for the cases of @xmath140 and @xmath141 .
again , from fig
. [ sim.n ] , we can see that our proposed rsnc scheme has the lowest deadline miss ratio .
in addition , with the increase of @xmath137 , the deadline miss ratio increases .
this is because more packets need to be transmitted at node @xmath0 . from fig .
[ sim.n ] , it is easy to see that the deadline miss ratio is smaller when @xmath142 , compared with @xmath143 .
it is reasonable because with the increase of the deadlines , less packet will lose its deadline .
in this paper , we propose a novel joint rate selection and network coding ( rsnc ) scheme for time critical applications .
we first prove that the proposed problem is np - hard , and design a novel graph model to model the problem . using the graph model ,
we mathematically formulate the problem .
we also propose a metric , based on which we design an efficient algorithm to determine transmission rate and coding strategy . finally , simulation results demonstrate the proposed rsnc algorithm effectively reduces the packet deadline miss ratio .
j. xu , x. tang , and w. lee , `` time - critical on - demand data broadcast : algorithms , analysis , and performance evaluation , '' _ ieee trans . on parallel and distributed systems _ , vol .
17 , no . 1 , pp . 3 14 , 2006 .
k. chi , x. jiang , and s. horiguchi , `` joint design of network coding and transmission rate selection for multihop wireless networks , '' _ ieee trans . on vehicular technology _ ,
2435 2444 , 2010 .
t. kim , s. vural , i. broustis , d. syrivelis , s. krishnamurthy , and t. la porta , `` a framework for joint network coding and transmission rate control in wireless networks , '' in _
proceedingsof 2010 ieee infocom _ , mar .
2010 , pp . 1 9 .
|
in this paper , we dynamically select the transmission rate and design wireless network coding to improve the quality of services such as delay for time critical applications . with low transmission rate , and hence longer transmission range , more packets may be encoded together , which increases the coding opportunity .
however , low transmission rate may incur extra transmission delay , which is intolerable for time critical applications .
we design a novel joint rate selection and wireless network coding ( rsnc ) scheme with delay constraint , so as to minimize the total number of packets that miss their deadlines at the destination nodes .
we prove that the proposed problem is np - hard , and propose a novel graph model and transmission metric which consider both the heterogenous transmission rates and the packet deadline constraints during the graph construction . using the graph model ,
we mathematically formulate the problem and design an efficient algorithm to determine the transmission rate and coding strategy for each transmission .
finally , simulation results demonstrate the superiority of the rsnc scheme .
|
the idea of a fully background independent formulation of the microscopic structure of quantum space - time has recently found a tentative realisation in the ( tensorial ) group field theory ( gft ) approach to quantum gravity @xcite .
two main lines of research converge in this approach : tensor models @xcite and loop quantum gravity @xcite .
the first , a higher - dimensional generalisation of the matrix models formulation of 2d quantum gravity , provide the combinatorial backbone and several key mathematical tools , the second suggests the needed algebraic data that allow the definition of a hilbert space structure and of richer , more interesting quantum geometric models for 3d and 4d quantum gravity . in these notes , as in our talk at the workshop , we illustrate briefly how this enrichment of tensor models through the input of loop quantum gravity and spin foam models works , focusing in particular on data and tools from non - commutative geometry , and what it leads to , reviewing some recent results .
we thus complement the mostly combinatorial treatment of tensor models and tensorial group field theories provided by prof .
rivasseau at the same workshop @xcite .
tensor models @xcite aim at a definition of quantum gravity in terms of random discrete spaces .
the basic dynamical object is an array of @xmath0 ( complex ) numbers @xmath1 ( for simplicity , we consider @xmath2 ) , where each index takes values in some finite set @xmath3 , which can be depicted as a ( d-1)-simplex . abusing slightly of terminology ( because we do not yet assume any specific transformation property of this object ) , we call it a tensor .
a choice of action functional @xmath4 where we have separated a quadratic ( kinetic ) part with kernel @xmath5 from the higher order ( interaction ) part with kernel @xmath6 , in the simple case of a single type of interaction , weighted by coupling constant @xmath7 , specifies then the dynamics of the system , and the quantum amplitudes obtained from the perturbative expansion ( in @xmath7 ) of the partition function @xmath8[d\bar{t}_{i_1 ... i_d } ] e^{-\ , s(t,\bar{t } ) } \ , = \ , \sum_{\gamma } \frac{\lambda^{n_{\gamma}}}{sym(\gamma)}\ , \mathcal{a}_{\gamma } \qquad\ ] ] with @xmath9 the number of interaction vertices of the feynman diagram @xmath10 , and @xmath11 the order of its eventual symmetries . obviously , the precise choice of action ( that is , of kinematic and interaction kernels ) specifies the detailed model one is studying and the structure of its feynman diagrams @xmath10 .
the characteristic feature of tensor models ( and of gfts ) is that the interaction kernels are based on a non - local pairing of tensor indices .
models with simplicial interactionslike ( in @xmath12 ) @xmath13 ( and complex conjugate ) , where repeated indices indicate complete summation over the index sets , have feynman diagrams that can be represented by stranded diagrams ( with each strand corresponding to an index of the tensor @xmath14 or @xmath15 ) dual to simplicial complexes , i.e. gluings of d - simplices ( corresponding to interaction kernels , e.g. a tetrahedron in the example ) across shared ( d-1)-simplices .
one can turn @xmath14 into a proper tensor by assuming it transforms under the unitary ( orthogonal , in the real case ) group @xmath16 where each component transforms a given index of the tensor @xmath14 .
a natural set of interaction kernels is then given by polynomial tensor invariants @xmath17 labelled by closed coloured graphs @xmath18 obtained as follows : a ) draw a white ( resp .
black ) vertex for each tensor @xmath14 ( resp .
@xmath15 ) appearing in the polynomial ; b ) associate @xmath19 different colours the @xmath19 indices of a tensor and draw a coloured link between a black and a white vertex corresponding to the index that is contracted between a tensor @xmath14 and a tensor @xmath15 in the polynomial ( no contraction between two @xmath14 or two @xmath15 is allowed ) .
one can also associate to the coloured graph a triangulation made of a number of ( d-1)-simplices equal to the order of the polynomial , glued along shared ( d-2)-faces according to the pattern of links in the graph , and consider the polynomial to correspond to a d - cell bounded by such triangulation . in this case , the feynman diagrams correspond to cellular complexes made out of glueings of such -d - cells along boundary ( d-1)-simplices . in turn , any such d - cell can be triangulated by inserting a single vertex inside it and connecting this with the boundary vertices ( so that the d - cell is formed by @xmath20 d - simplices glued around such central vertex ) .
the proposed definition of quantum gravity given by tensor models is therefore purely combinatorial : a sum over cellular complexes ( dual to ) @xmath10 weighted by an amplitude @xmath21 that only depends on the combinatorial structure of @xmath10 itself .
one can then endow the cellular complexes with a notion of discrete metric given by the graph distance .
then the sum over cellular complexes admits the interpretation of a sum over discrete geometries and the amplitude @xmath21 can be re - interpreted ( after appropriate re - interpretation of the constants @xmath22 and @xmath7 ) as the exponential of the regge action for the equilateral triangulation corresponding to @xmath10 .
we thus have a lattice gravity path integral analogous to the one defining the dynamical triangulations approach to quantum gravity , now generated by the field theory for @xmath14 and @xmath15 . by now , an impressive wealth of mathematical results have accumulated for this type of models , the understanding of the combinatorics of these models , and the analytic control over their partition function is remarkable , comprising a powerful large - n expansion , some understanding of critical behaviour and double scaling limits , results on the schwinger - dyson equations of several models , nice universality results , and much more .
for all this , we refer to the literature .
what we focus on here , instead , is the models that are obtained by enriching this already rich combinatorial models via group - theoretic data , namely group field theories . by this
we mean models of the same tensor type but with index spaces replaced by lie group manifolds ( or subspaces thereof ) , as we will describe , so that one has a ( complex ) field @xmath23 instead of the finite tensor @xmath14
. why is this enrichment useful at all
? one may indeed fear that it would correspond to an unnecessary complication of already complicated models , at the same time losing the beauty of a purely combinatorial definition of the fundamental structure of space and time ( which is , by the way , successful in 2d as matrix models for liouville quantum gravity show ) .
the main advantages are the following : \1 ) the definition of a proper renormalisation group flow , in turn very useful for handling the quantum dynamics of such field - theoretic models , requires the soft breaking by the kinetic term of the invariances of the interactions ( be them based on the simplicial or the tensor invariance principles ) ; this is conveniently obtained with @xmath5s involving differential operators which in turn require the domain of the field @xmath14 to be a proper differentiable manifold ; ( lie ) groups are an obvious choice ; \2 ) turning the index set into a group manifold turns the tensor model into a proper quantum field theory with ( a priori ) an infinite number of degrees of freedom , greatly widening not only the possibilities for model building but also the set of quantum field theory tools that can be used to analyse them , and leading to richer phase structure , symmetries , etc ; on the one hand , there is no proof that this widening is necessary , of course ; on the other hand , it is reasonable to expect that it is , given the greater complexity of geometry and physics in dimensions higher than 2 , where the purely combinatorial matrix models suffice ; to face this complexity , it is nicer to have a larger variety of them at our disposal ; \3 ) most important , the additional data allow the tensor models framework to make direct contact with other promising approaches to quantum gravity , that is loop quantum gravity and spin foam models , and thus with another large set of physical insights and mathematical results ; obviously , these approaches also independently suggest that the group - theoretic data that characterise group field theories are a necessary ingredient of a fundamental description of space and time in a full quantum gravity context .
thanks to the additional data , the connection with discrete geometry and lattice gravity path integrals can also be refined , leading to further progress . and
further progress can be expect via the use of tools from non - commutative geometry , which are made available by the additional data .
we will now turn to a brief description of group field theories and of the above connections .
as anticipated , group field theories are field theories over a lie group manifold @xmath24 , in such a way that tensor models can be understood as a special case in which the lie group is replaced by a finite group , e.g. @xmath25 , or even its underlying finite set of n elements , neglecting the group structure entirely .
we thus have the ( complex ) field @xmath26 as our basic variable . as such ,
the underlying phase space for each gft quantumis the cotangent bundle @xmath27 .
in particular , for quantum gravity applications , as we will see , the most relevant groups are @xmath28 and @xmath29 ( or , in the lorentzian signature , which will not concern us here , @xmath30 ) .
the connection with loop quantum gravity ( lqg ) and with lattice gravity at the kinematical level is then immediate to sketch .
canonical variables in this phase space are thus @xmath19 pairs of group elements and lie algebra elements @xmath31 , understood as classical configuration and moment variables .
given this choice of canonical variables , one can expect two basic representations for quantum states : one in terms of functions of group elements , and one in terms of conjugate lie algebra elements .
as we will discuss , the latter representation requires new mathematical tools from non - commutative geometry to be defined , due to the fact that the lie algebra of @xmath32 is , in general , a non - commutative space .
another equivalent representation arises instead from the peter - weyl expansion of functions on the group in irreducible representations of the group itself .
the classical lqg phase space is , roughly speaking , composed of a copy of the same cotangent bundle @xmath33 for each link of all possible graphs @xmath34 .
correspondingly , quantum states of the gravitational field in loop quantum gravity @xcite are associated to all possible graphs @xmath34 in such a way that they form a hilbert space of functions @xmath35 , with respect to the haar measure , where @xmath36 is the number of links in the graph @xmath34 . restricting in particular to d - valent graphs
, one can embed each of the above hilbert spaces @xmath37 into the space @xmath38 , where @xmath20 is the number of vertices in the graph @xmath34 . in more physical language
, one can consider any lqg state associated to a graph @xmath34 as a specific many - particlestate in which each particleis represented by an open spin network vertex , that is a graph vertex with @xmath19 open links attached to it , and with @xmath19 group elements associated to them , so that the original graph is obtained by gluingthe open spin network vertices to one another along their links . considering the hilbert space for all possible number of vertices @xmath20 @xmath39 , one obtains a hilbert space large enough to contain all possible graph - based lqg states ( the exact hilbert space structure is different than in standard lqg , due to the different scalar product and the avoidance of any cylindrical consistency condition and projective limit , here ) .
gauge invariance conditions can be imposed in the usual way with respect to the group action at the vertices of the graphs .
this last hilbert space can be recast in 2nd quantised language as a fock space , using the standard construction , and under the assumption of bosonic statistics for the open spin network vertices(other choices are of course possible ) .
the resulting fock space is the kinematical space of quantum states of group field theories based on this choice of group , and in absence of additional conditions on the gft field or on its dynamics .
more details can be found in @xcite .
still at the kinematical level , one realises immediately that each gft quantum can be equally depicted as a ( d-1)-simplex dual to the d - valent open spin network , similarly to tensor models , but now with its ( d-2)-faces labelled by group elements @xmath40 or by lie algebra elements @xmath41 .
generic states are then going to be dual to ( d-1)-simplicial complexes labelled by the same algebraic data .
a first interpretation is that these are exactly the classical data of topological bf theories discretized on a simplicial complex . indeed ,
when appropriate gauge invariance conditions are imposed , gfts produce the same classical phase space of simplicial bf theories , generalised to arbitrary simplicial complexes and superpositions of them at the quantum level .
these data can be understood already in terms of discrete simplicial geometry in @xmath12 , where bf theory coincides with 3d gravity in the first order formalism . in this case , the lie algebra variables correspond to discretised triad fields and the group elements to elementary parallel transports of the gauge connection ( before metricity is imposed by the equations of motion ) .
we are going to describe the gft formulation of 3d gravity in some more detail in the following . in higher dimensions , @xmath42 in particular ,
bf variables can be understood in geometric terms only after appropriate geometricity conditions , called simplicity constraintsare imposed @xcite .
the imposition of them in the gft models describing bf theory in 4d , assuming this is done correctly , results in models with the right quantum geometric data to describe 4d quantum gravity .
we are going to see an example of this construction in the following .
the correspondence of gft with lqg and discrete gravity carries on to the dynamical level . indeed , given any canonical operator equation to be imposed on lqg states and encoding the quantum dynamics of the theory , i.e. some version of the hamiltonian constraint operator , one can write down a corresponding second quantised equation in fock space for the gft states ; this then leads to the definition of a qft partition function for the corresponding gft model , assuming that one relaxes the requirement that only quantum states satisfying the quantum constraints are included in the partition function , thus working in the analogue of a grand canonical ensemble in which both energy and particle number are allowed to fluctuate ( with solutions of the hamiltonian constraint equation corresponding to zero energystates , and the number of particlesbeing the number of graph vertices in lqg states ) .
we have then a direct correspondence between the canonical formulation of lqg and gfts at the dynamical level @xcite .
the correspondence with the covariant formulation of lqg follows suit .
given the gft partition function so obtained , its perturbative expansion defines transition amplitudes for quantum states in terms of a sum over cellular complexes @xmath10 , weighted by amplitudes @xmath21 which are now themselves integrals over the group ( or the lie algebra ) of functions of group or lie algebra elements ( or irreps of the group ) associated to the cellular complex .
the latter amplitudes are , when written in terms of irreps of the group , so - called spin foam models , defining the covariant form of the dynamics of lqg . when written instead in terms of lie algebra elements
, the same amplitudes take the form of lattice gravity path integrals , that is of path integrals for simplicial gravity ( like regge calculus ) , written in first order form , i.e. in the classical variables of discrete ( constrained ) bf theory .
we will see these forms of the quantum amplitudes in the following , for specific examples .
the crucial point to note , here , is that this important connection with lqg and with discrete gravity , providing a well - defined hilbert space as well as interesting quantum dynamical amplitudes , is possible only because of the additional group - theoretic data that enrich group field theories with respect to the simpler tensor models .
the quantum geometry underlying both gfts and lqg is manifest in the lie algebra representation of their quantum states and amplitudes . because in this context the lie algebra elements are interpreted as fluxes of the gravitational triad field , this is also called _
flux representation_. this uses some new tools from non - commutative geometry , namely a new notion of non - commutative fourier transform .
let us introduce these mathematical tools , before showing a few specific applications in quantum gravity .
we follow the presentation in @xcite , to which we refer for further references .
our phase space of interest is the cotangent bundle @xmath43 , for a lie group @xmath32 , the extension to several copies of this phase space being straightforward @xcite . with the canonical symplectic structure
, we get the poisson algebra @xmath44 , and for any functions @xmath45 we obtain the brackets @xmath46 where @xmath47 are lie derivatives with respect to an orthonormal basis of right - invariant vector fields , @xmath48 are euclidean coordinates on @xmath49 , @xmath50 , @xmath51 the structure constants , and repeated indices are summed over .
we now seek to quantize a _
maximal subalgebra _
@xmath52 of this poisson algebra , as an abstract operator @xmath53-algebra @xmath54 .
we define a _ quantization map _
@xmath55 such that @xmath56 for all @xmath57 , and @xmath58 , satisfying @xmath59 = 0\,,\quad [ \hat{x}_i,\hat{f } ] = i\widehat{\mathcal{l}_i f } \in
\mathfrak{a}_g\,,\quad [ \hat{x}_i,\hat{x}_j ] = i c_{ij}^{\phantom{ij}k}\hat{x}_k\,,\quad \forall \hat{f},\hat{g}\in \mathfrak{a}_g\,,\ ] ] where @xmath60 is the subalgebra of @xmath61 of functions constant in the second argument , and @xmath62 . in general , global coordinates can not be defined , but can be approximated arbitrarily well by elements in @xmath63 , and we may define coordinate operators @xmath64 not necessarily in @xmath65 corresponding to a set of coordinates @xmath66 , satisfying @xmath67 = 0 \ , \quad [ \hat{x}_i,\hat{\zeta}^j ] = i\widehat{{\mathcal{l}}_i \zeta^j } \ , \quad [ \hat{x}_i,\hat{x}_j ] = i c_{ij}^{\phantom{ij}k}\hat{x}_k\ , .\ ] ] assuming @xmath68 and @xmath69 , we can define @xmath70 , where @xmath71 are constant coefficients specific to the chosen coordinates .
furthermore , @xmath72 is endowed with a natural hopf algebra structure with coproduct @xmath73 , counit @xmath74 , and antipode @xmath75 , which can be used to define a hopf algebra structure for @xmath76 , which will be crucial to define what we called the algebra representation of the quantum system .
given the algebra of observables , the task becomes that of finding representations of it on suitable hilbert spaces .
the simplest to define is the _ group representation _
@xmath77 on @xmath78 is defined as the one diagonalizing all the operators @xmath79 : @xmath80 , for all @xmath81 such that @xmath82 .
the resulting function @xmath83 will not in general lie in @xmath78 for all @xmath84 , but we may restrict the domain of @xmath85 to be the subspace of @xmath60 of smooth compactly supported functions @xmath86 on @xmath32 dense in @xmath78 , so that @xmath87 for all @xmath88 .
lie algebra operators @xmath89 are represented then as @xmath90 , with similar remarks about the domain of @xmath91 .
the commutation relations are then correctly reproduced .
the inner product is given for @xmath92 by @xmath93 , where @xmath94 is the right - invariant haar measure on @xmath32 .
to prove this , one relies on the compatibility between the pointwise multiplication , used in the representation of operators , and the coproduct on the algebra .
we now move on to define a representation in terms of functions of the classical dual space @xmath95 .
obviously , the route taken to obtain the group representation can not used because @xmath96 do not commute .
the strategy we adopt is then to _ deform _ the action @xmath97 , giving the needed freedom to satisfy the commutation relations
. we will denote it by a star - product @xmath98 , and define for all @xmath99 @xmath100 , in such a way that the commutator @xmath101 = i c_{ij}^{\phantom{ij}k}\hat{x}_k$ ] turns into @xmath102 .
in fact , we impose on the star - product the stronger condition @xmath103 , for all @xmath104 such that @xmath105 .
this guarantees that @xmath106 has the interpretation of the function which upon quantization gives @xmath107 , and so establishes a connection between the classical phase space structure and the quantum operators .
thus , @xmath108 for all @xmath109 .
therefore , the @xmath98-product and the quantization map @xmath110 are related by @xmath111 which is the idea of star - products defined in the context of deformation quantization and the choice of quantization map determines uniquely the @xmath98-product to be used in representing the quantum algebra in terms of functions on @xmath95 .
now , given a star product @xmath98 and some coordinate operators @xmath64 , we define the representation of the operators @xmath64 and @xmath89 acting on the space of smooth compactly supported functions @xmath112 on @xmath95 to be @xmath113 where we denote @xmath114 .
now , in order to show that we have a representation of the quantum algebra , the only non - trivial part is to show that the commutator @xmath115 $ ] is correctly reproduced , and this amounts to showing the compatibility between the star product and the coproduct on the observable algebra .
this compatibility represent then a condition on the quantisation map to allow for an algebra representation as constructed here .
the details can be found in @xcite .
the next task is to find a unitary map between the two representations @xmath77 and @xmath116 of @xmath54 .
we assume that this correspondence takes the form of an integral transform , generalising the standard fourier transform , @xmath117 : @xmath118 where @xmath119 , and we denote by @xmath120 the integral kernel of the transform .
then , the goal is to identify the defining equations for the kernel @xmath120 , using the fact that the intertwined functional spaces define a representation of the same quantum algebra , and applying the action of @xmath54 in the different representations .
its actual existence has then to be verified once an explicit choice of quantization map has been made .
we find that the kernel @xmath120 si given by : @xmath121 with @xmath122 taken from any given branch of the logarithm .
thus , given a deformation quantization @xmath98-product , this formula gives the general expression for the integral kernel @xmath120 .
the same kernel should satisfy also @xmath123 the prefactor @xmath124 depends on the quantization map @xmath110 chosen . for a given @xmath98-product ,
determining coordinates for which both forms are satisfied might be difficult and , in general , there is no guarantee that such coordinates exist .
it is a requirement for the existence of the non - commutative fourier transform .
some properties of these plane waves are : @xmath125 where the right - hand side is the dirac delta distribution with respect to the right - invariant haar measure on @xmath32 .
this completes the definition of the integral transform @xmath126 intertwining the representations @xmath77 and @xmath116 : @xmath127 where @xmath122 is taken in the principal branch .
the @xmath128-product is extended by linearity to the image of @xmath126 more interesting general properties of this non - commutative fourier transform can be found in @xcite .
let us now give a few concrete examples of star products and plane waves following from specific choices of quantisation maps .
we limit to the case of @xmath129 .
a generic element @xmath130 can be written as @xmath131 , @xmath132 , where @xmath133 are the pauli matrices , while for any group element @xmath134 we have @xmath135 .
another convenient parametrization of @xmath136 can be written as @xmath137 .
these two parametrizations are thus related by the change of coordinates @xmath138 , where @xmath139 , or @xmath140 according to @xmath141 , @xmath142 respectively , and @xmath134 assumes the form @xmath143 .
the so - called _ freidel - livine - majid map _ , is defined , for exponentials @xmath144 , as @xmath145 and can be basically seen as the symmetrization map in conjunction with a change of parametrization . accordingly : @xmath146 we may simply write @xmath147 , but the coordinates @xmath148 only cover the upper ( or lower ) hemisphere @xmath149 , and the resulting transform is applicable only for functions on @xmath150 .
one also finds : @xmath151 the factor @xmath152 , introduced by the projection , is the sign of @xmath153 , which is 1 if both @xmath154 are close to zero or one of them is infinitesimal , and @xmath155 when the addition of two upper hemisphere vectors ends up in the lower hemisphere ( thus projecting the result to its antipode on the upper hemisphere ) .
the @xmath156-monomials read @xmath157 the non - commutative fourier transform is then @xmath158 the _ duflo map _ is instead given by @xmath159 where @xmath160 is the symmetric quantisation map and @xmath161 is the following function on @xmath162 @xmath163 where the last expression holds for @xmath164 .
when applied to exponentials , it gives @xmath165 which can be inverted to @xmath166 on monomials , we get : @xmath167 for the non - commutative plane wave we again have the corresponding projected star - product @xmath168 , which satisfies @xmath169 where @xmath170 is the lie algebra element resulting from the bch formula .
the explicit form of the non - commutative fourier transform is thus @xmath171 we are now ready to show some applications of the algebra / flux representation to gft , discrete gravity , spin foam models and lqg , highlighting the connections between these formalims .
the main advantage of using the flux / algebra representation for gft fields and amplitudes is that it brings the quantum geometry underlying them to the forefront . indeed , as mentioned above , the algebra variables have the interpretation of discrete triad fields associated to the edges of the simplicial complexes defining the quantum states and the amplitudes , in 3d gravity , or of discrete b fields associated to triangles of simplicial complexes in 4d bf models , that become invertible for a discrete tetrad field once geometricity constraints are imposed
. we will now show how this is the case .
we first consider the group field theory formulation of 3d riemannian gravity @xcite .
the variables are fields @xmath172 on @xmath173 satisfying the invariance : @xmath174 the dynamics is given by the action with simplicial ( tetrahedral ) interactions : @xmath175 ^ 3 \,\varphi^2_{123 } - \frac{\lambda}{4 ! } \ !
\int [ dg]^6 \,\varphi_{123}\varphi_{345}\varphi_{526}\varphi_{641}\ ] ] the feynman diagrams @xmath10 are thus dual to 3d triangulations
@xmath176 : the combinatorics of the field arguments in the interaction vertex is that of a tetrahedron , while the kinetic term dictates the gluing rule for tetrahedra along triangles . by peter - weyl expansion of @xmath177 into irreps of @xmath178
, the field can be pictured as an open 3-valent spin network vertex or as a triangle , with the three field arguments associated to its edges @xmath179 . in this representation
, the interaction term can be written in terms of 6j - symbols , and the feynman amplitude gives the ponzano - regge spin foam model : @xmath180 where the sum is over irreps of @xmath178 labelled by integers @xmath181 associated to the edges of the triangulation , and one has a 6j - symbol for each tetrahedron in the same triangulation , function of the six representations associated to its six edges . in group
variables the same amplitudes takes the form of a lattice gauge theory imposing flatness at each edge of the simplicial complex , which is indeed the geometric content of 3d gravity and bf theory .
we now make this connection even clearer by working out the algebra representation of the model @xcite .
we use the flm quantisation map and the associated star - product .
fourier transform and @xmath98-product extend to functions of several variables like the gft field as @xmath182 ^ 3\ , \varphi_{123 } \ , e_{g_1}(x_1 ) e_{g_2}(x_2 ) e_{g_3}(x_3)\ ] ] the invariance under group action becomes the ` closure constraint ' for the variables @xmath183 , @xmath184 where the non - commutative delta function has the obvious definition in terms of non - commutative plane waves ; this confirms their interpretation as edge vectors ( discrete triad fields ) of the corresponding triangle .
the action becomes : @xmath185 ^ 3 \,\hat{\varphi}_{123 } \star \hat{\varphi}_{123 } - \frac{\lambda}{4 ! } \ !
\int [ dx]^6 \,\hat{\varphi}_{123 } \star \hat{\varphi}_{345 } \star \hat{\varphi}_{526 } \star \hat{\varphi}_{641}\ ] ] where @xmath98-products relate repeated indices .
the feynman amplitudes for closed diagrams , in this representation , read : @xmath186 where we have used the explicit form of the flm plane waves .
this is the simplicial path integral of first order 3d gravity ( or 3d bf theory ) .
the variables @xmath187 corresponds to the parallel transport ( discrete connection ) between the two tetrahedra ( vertices of the feynman diagram @xmath10 ) sharing the triangle @xmath188 ( dual to a link of @xmath10 ) ; @xmath189 is the holonomy ( discrete curvature ) around the boundary of @xmath190 ( face of @xmath10 , dual to an edge @xmath179 of the simplicial complex ) , calculated from a chosen reference tetrahedron frame .
the lie algebra variables @xmath191 , one per edge of the simplicial complex ( dual to faces of @xmath10 ) , play the role of discrete triad @xcite .
a similar result can easily be obtained for the duflo map , leading to the same type of variables and interpretation , but to a slightly different definition of the discretized bf action , due to the different choice of coordinates on the group appearing in the plane waves ( and an additional term in the measure ) . for generic open diagrams ,
the amplitudes are given again by the simplicial 3d gravity path integral with the appropriate boundary terms ( for fixed discrete triad on the boundary ) .
beside making the link with discrete geometry explicit , the flux representation is then also a useful tool for semiclassical analysis of the same amplitudes , given the fact that they take the form of standard discrete path integrals .
such analysis has been performed , for generic simplicial complexes , with metric boundary conditions , in @xcite , extending previous results in the literature .
we sketch now the results of this analysis , before moving on to the 4d case .
the explicit form of the feynman amplitudes for fixed metric data on the boundary is : @xmath192 \big[\frac{dy_{ji}}{(2\pi\hbar\kappa)^3}\big ] \big[\prod_ldh_l\big ] \big[\frac{d y_{e}}{(2\pi\hbar\kappa)^3}\big ] \left [ \prod_{e \notin \partial\delta } \eta(h_{e}(h_l ) ) \right ] \left [ \prod_{\substack{(i , j)\in\mathcal{n}\\i < j } } c(g_{ij } h_{j}^{-1 } k_{ji}(h_l ) h_{i } g_{ji}^{-1 } ) \right ] \nonumber \\ & \times & \exp\left\{\frac{i}{\hbar } \left[\sum_{e \notin \partial\delta } y_e \cdot \zeta(h_{e}(h_l ) ) + \!\!\ ! \sum_{\substack{(i , j)\in\mathcal{n}\\i < j } } y_{ji}\cdot \zeta(g_{ij } h_{j}^{-1 } k_{ji}(h_l ) h_{i } g_{ji}^{-1 } ) + \!\!\ ! \sum_{(i , j)\in\mathcal{n } } x_{ij}\cdot \zeta(g_{ij}^{-1 } ) \right]\right\ } \qquad\end{aligned}\ ] ] where : the set of ordered pairs of labels associated to neighboring boundary triangles is @xmath193 ; @xmath194 is the group element associated to the dual half - link going from the boundary triangle @xmath195 to the center of the bulk tetrahedron with triangle @xmath195 on its boundary ; @xmath196 is the holonomy along the bulk dual links from the center of the tetrahedron with triangle @xmath161 to the center of the tetrahedron with triangle @xmath195 ; @xmath197 is the discrete triad ( edge vector ) shared by the triangles @xmath198 on the boundary , as seen from the frame of reference of the triangle @xmath161 ; @xmath199 is the non - commutativity parameter entering star products and plane waves ( so the abelian limit of the phase space corresponds to the no - gravity limit @xmath200 ) , the @xmath201 are discrete triad variables associated to edges in the bulk of the triangulation , while @xmath202 are auxiliary metric variables associated to boundary edges ( they end up being identified with the @xmath197 by the dynamics , as it can verified immediately by performing some of the group integrations ) ; @xmath203 are real functions on the group depending on the choice of quantisation map ( one has @xmath204 for the flm map and @xmath205 for the duflo map , as we have seen ) .
the amplitudes are directly suited for studying the semiclassical limit through a saddle point analysis of the discrete gravity action : @xmath206 in first order variables @xmath201 and @xmath207 , plus boundary data @xmath197 , @xmath202 and @xmath208 ( only the first are fixed at the boundary , the others being subject to variations ) .
we expect the classical limit to be governed by the equations of motion of this action : geometricity constraints imposing flatness of holonomies around dual faces and closure of edge vectors for all triangles ( up to parallel transport ) .
this is exactly what happens .
standard variations of this action give the equations : @xmath209 for all @xmath210 , i.e. , the flatness of the connection around the dual faces in the bulk .
@xmath211 for all @xmath212 , i.e. , the triviality of the connection around the dual faces to @xmath213 .
@xmath214 where @xmath215 implements the parallel transport from the frame of @xmath201 to the frame of @xmath216 , and @xmath217 accounts for the orientation of @xmath207 with respect to the holonomy @xmath218 and thus the relative orientations of the edge vectors , and @xmath219 parallel transports the edge vectors @xmath220 to the frame of the boundary triangle @xmath221 , and @xmath222 again accounts for the relative orientation .
these impose the closure constraint for the three edge vectors of each bulk triangle @xmath223 in the frame of @xmath216 , and the closure of the boundary integration variables @xmath220 .
these same conditions give the metric compatibility of the discrete connection , which in turn , if substituted back in the classical action , before considering the other saddle point equations , turn the discrete 1st order action into the 2nd order regge action for the triangulation @xmath176 .
@xmath224 where we denote @xmath225 ; these are a _ deformed _ closure constraint for the boundary edge variables @xmath197 , and a _ deformed _ identification , up to a parallel transport , of the boundary edge variables @xmath197 and @xmath226 . the amplitudes become @xmath227 \big[\prod_{v\in\partial\delta } \delta(h_v(g_{ij } ) ) \big ] \left [ \prod_{f_i \in \partial\delta } \delta_\star\big(\!\!\!\sum_{\substack{f_j\in\partial\delta\\(i , j)\in\mathcal{n } } } \!\!\ !
d^{\zeta}(g_{ij } ) x_{ij}\big ) \right ] \nonumber \\ & \star \left [ \prod_{\substack{(i , j)\in\mathcal{n}\\i < j } } \delta_\star\big(\textrm{ad}_{g_{ij } } ( d^\zeta(g_{ij})x_{ij } ) + \textrm{ad}_{g_{ji } } ( d^\zeta(g_{ji})x_{ji})\big ) \right ] \star\ \exp\left\{\frac{i}{\hbar } \sum_{(i , j)\in\mathcal{n } } x_{ij}\cdot \zeta(g_{ij}^{-1 } ) \right\ } \big(1 + \mathcal{o}(\hbar)\big ) \qquad\label{eq:0thpramp}\end{aligned}\ ] ] where the delta functions impose the constraints on boundary data discussed above .
in particular , @xmath228 are the holonomies around the boundary vertices @xmath229 , whose triviality follows from the triviality of the bulk holonomies .
notice that one must write the integrand in terms of @xmath98-products and @xmath98-delta functions in order for the constraints to be correctly imposed , since the amplitude acts on wave functions through @xmath98-multiplication .
the non - commutative nature of the variables actually has to be taken into account in a more subtle way .
in fact , it turns out @xcite , that one needs to consider the deformation of phase space structure and take a _ non - commutative variation _
@xmath230 of the action @xmath231 in the amplitude via @xmath232 , where the @xmath98-product acts on the fixed boundary variables @xmath197 , @xmath233 refers to terms higher than first order in the variations , and @xmath234 is the varied action .
it is easy to see that the non - commutative variation so defined undeforms the identification of @xmath197 and @xmath226 ( up to parallel transport ) , simply because we have @xmath235 for any @xmath236 and @xmath237 implementing the variation of @xmath238 .
we obtain the geometric identification @xmath239 , and the undeformed closure condition for boundary @xmath197 .
these undeformed relations replace the deformed ones in the above semiclassical approximation of the amplitudes .
we recover exactly the geometric relations for the boundary variables , regardless of the choice of a quantization map , and the standard equations of motion for 3d gravity in the bulk .
once more , the algebra representation of the theory allows for a straightforward , geometrically transparent analysis .
we now turn to 4d quantum gravity and describe the application of the non - commutative geometry tools introduced above to the construction of interesting gft models of it .
we will see another instance of the general duality between spin foam models and simplicial gravity path integrals , and of the way the algebra / flux representation brings the quantum geometry of such models to the forefront .
more details can be found in @xcite .
classical continuum 4d gravity can be expressed as a constrained @xmath240 theory @xcite @xmath241 for @xmath242 valued 1-form connection @xmath243 and 2-form field @xmath244 ( also lie algebra valued ) , where @xmath245 are ( so - called ) simplicity constraints and @xmath7 is some lagrange multiplier .
the constraints force @xmath244 to be a function of a tetrad 1-form field @xmath246 , turning bf to the palatini action for gravity in the first order formalism .
the immirzi parameter @xmath34 , which plays a crucial role in lqg , is introduced by changing variables @xmath247 in the bf term .
the constraints lead then to the holst action , classically equivalent to palatini gravity and the classical starting point of lqg .
the general strategy in group field theory and spin foam models @xcite has been , therefore , to start from a formulation of simplicial 4d bf theory and impose a discrete counterpart of the simplicity constraints to obtain a model for 4d gravity .
the starting point is therefore the straightforward 4d generalisation of the gft model for 3d gravity described above : @xmath248 ^ 4 \varphi^2_{1234 } + \frac{\lambda}{5 ! } \int [ dg_i]^{10 } \varphi_{1234 } \,\varphi_{4567 } \,\varphi_{7389}\ , \varphi_{962\,10 } \,\varphi_{10\,851}\ ] ] where @xmath249 and @xmath250 .
the perturbative expansion in @xmath7 of the partition function generates a sum over ( closed ) feynman diagrams which are cellular complexes dual to simplicial 4d complexes , to which the theory assigns amplitudes which , once more , can be written down equivalently as a bf lattice gauge theory ( in the group representation ) , as simplicial path integrals for discrete bf ( in the algebra representation ) and as a spin foam model ( in the spin representation ) .
the algebra representation is obtained by extending the non - commutative @xmath178 fourier transform to @xmath251 ^ 4 $ ] ( in the following , we use the flm quantisation map ) , and leads to a function of four @xmath242 elements @xmath252 interpreted as the discrete b fields of bf theory , associated to the four triangles belonging to the tetrahedron corresponding to the gft field .
the gauge invariance condition becomes again the closure @xmath253 .
an entirely equivalent model is obtained by adding an additional variable @xmath254 , interpreted as unit vectors normal to the tetrahedron corresponding to the gft field , relaxing the gauge invariance condition to the covariance : @xmath255 , @xmath256 ( using the selfdual / anti - selfdual decomposition of @xmath257 , @xmath258 is the normal rotated by @xmath259 ) and defining the action : @xmath260 = \frac{1}{2}\int [ dg_i]^4 dk \ , \varphi^2_{k1234 } + \frac{\lambda}{5 ! } \ !
\int [ dg_i]^{10 } [ dk_i]^5 \ , \varphi_{k_11234 } \varphi_{k_24567 } \varphi_{k_37389 } \varphi_{k_4962\,10 } \varphi_{k_510\,851 } \ ; .\ ] ] the discrete counterpart of the simplicity constraints is , classically @xcite : @xmath261 where @xmath262
. a geometrically clear way of imposing these constraints in the model is to use non - commutative delta functions , thus effectively constraining the measure on the bivectors .
we thus introduce the function @xmath263 } e^{i \beta \mbox{\footnotesize tr } [ u x^+]}= \int_{{\mathrm{su}}(2 ) } \!\!{\mathrm{d}}u \ , e_{\mathbf{u}^k_{\beta}}(x)$ ] where : @xmath264 , with @xmath265 and @xmath266 such that @xmath267 , we introduced @xmath268 , and @xmath269 .
the action of @xmath270 is well - defined on gauge invariant fields , as it commutes with the gauge transformations @xmath271 = e_h \cdots e_h \star ( \widehat{s}^\beta\triangleright\hat{\varphi})_{h^{-1 } \triangleright k } $ ] these relations ensure that rotating a bi - vector which is simple with respect to a normal @xmath272 gives a bi - vector which is simple with respect to the rotated normal , as geometrically required .
a geometrical gft model will be defined by constraining the field @xmath273 , by acting on it by @xmath98-multiplication by the product @xmath274 of four simplicity functions : @xmath275 for generic values of @xmath262 , the operator @xmath276 is not a projector . depending on whether it is inserted in the propagator , in the vertex , or in both , in a single or in multiple copies , one gets slightly different feynman amplitudes .
we choose to constrain the field in the interaction of the extended bf model . defining @xmath277 ,
we consider : @xmath278 ^ 4 \,dk \ , \widehat{\varphi}_{k 1 2 3 4 } \star \widehat{\varphi}_{k 1 2 3 4 } \ ; + \frac{\lambda}{5 ! } \int [ d^6 x_i]^{10 } \ , \widehat{\psi}_{1 2 3 4 } \star \widehat{\psi}_{4 5 6 7}\star \widehat{\psi}_{7 3 8 9}\star \widehat{\psi}_{9 6 2\,10 } \star \widehat{\psi}_{10\,8 5 1}\ ] ] where the star product pairs repeated indices .
we can now compute the feynman amplitudes of this model .
one sees that , for each triangle @xmath188 , the constraints impose the linear simplicity condition of @xmath279 with respect to the normals of all the tetrahedra @xmath280 sharing @xmath188 .
the square is present because both 4-simplices @xmath281 sharing the tetrahedron @xmath282 contribute a factor @xmath283 . for closed graphs ,
the normals @xmath284 drop from the amplitude and we obtain : @xmath285[d^6 x_t ] \left[\prod_t \bigstar_{j=0}^{n_t } \delta^{\star 2}_{- \bar{h}_{0j } x^- \bar{h}_{0j } ^{-1}}(\beta x^+)\right ] \star e^{i \sum_t tr \ , x_t h_t}\ ] ] where we wrote @xmath286 for the squared function @xmath287 , and @xmath288 .
the simplicity operators for the same triangle , appearing in different frames corresponding to different tetrahedra sharing it , can be split into a single constraint expressed in a given frame , and a modified measure term for the discrete connection , encoding the conditions on the connection that have to be satisfied to ensure the correct parallel transport of the simplicity constraint across simplicial frames .
the feynman amplitudes of this theory thus take the form of simplicial path integrals for a constrained bf theory of holst - plebanski type with immirzi parameter @xmath34 , with linear simplicity constraints .
as noted , the specific form of the modification of the bf measure , encoding the simplicity constraints , depends on how exactly one has inserted the simplicity operator in the gft action .
the same amplitudes can be written down in pure gauge theory form ( in the group representation ) and in pure spin foam form ( expanding in irreducible representations ) . in the latter form
can be compared more easily with other spin foam models for 4d gravity proposed in the literature @xcite .
we refer to @xcite for more details .
we have presented a brief review of the role of group - theoretic data in group field theories , focusing on the non - commutative flux representation of gravity models . while we believe that this could be of interest to mathematicians for its own sake ,
we have shown how useful it is from a quantum gravity point of view .
it makes manifest the quantum geometric content of gfts , and realises an explicit duality between spin foam models and simplicial gravity path integrals at the level of gft amplitudes , providing an rich , promising framework for quantum gravity .
a. baratin , d. oriti , phys .
lett . * 105 * 221302 ( 2010 ) , arxiv:1002.4723 [ hep - th ] ; a. baratin , b. dittrich , d. oriti , j. tambornino , class .
quant . grav .
* 28 * , 175011 ( 2011 ) , arxiv:1004.3450 [ hep - th ]
|
we review briefly the motivations for introducing additional group - theoretic data in tensor models , leading to the richer framework of group field theories , themselves a field theory formulation of loop quantum gravity .
we discuss how these data give to the gft amplitudes the structure of lattice gauge theories and simplicial gravity path integrals , and make their quantum geometry manifest .
we focus in particular on the non - commutative flux / algebra representation of these models .
|
according to the internal - external shocks model the prompt gamma - ray burst ( grb ) is produced by internal shocks within a relativistic flow while the afterglow is produced by external shocks between this flow and the surrounding matter .
the early afterglow appears during the transition from the prompt @xmath0-ray emission to the afterglow . during this transition
the relativistic flow , ejected by the source , interacts directly with the circum burst medium .
this interaction can be used to pin down the nature of the relativistic flow ( baryonic or poynting flux ) . in a baryonic
flow the reverse shock ( rs ) that propagates into the ejecta produces both optical and radio emission .
a poynting flux flow is expected to produce only the higher energy forward shock emission . while other sources of early optical and radio emission may exist also in poynting flux
, the rs emission has a very robust optical and radio observable signatures that are very unlikely to be imitated by other phenomena ( @xcite ) .
-ray emission ( the flux is in arbitrary units ) resulting from the collisions of four shells . the initial lorentz factor and width of these four shells
is depicted in the inset . *
bottom * : the optical light curve of the reverse shock as simulated numerically .
the initial hydrodynamic profile of the simulation appears in the top panel inset . the total energy ( isotropic equivalent ) of the ejecta is @xmath1 erg and the external density is a constant @xmath2
. the energy equipartition parameters of the electrons and the magnetic field @xmath3 and @xmath4 respectively .
the electrons spectral index is @xmath5 and the burst is at a redshift @xmath6 .
note that the time axis is logarithmic . ] in grbs the reverse shock is mildly relativistic ( @xcite ) .
this implies that during one crossing the shock dissipates almost all the bulk motion energy of the outflow to internal energy . after a single crossing of the outflow a rarefaction wave
is reflected and only the forward shock remains , forming the shocked external medium into a blandford - mckee self - similar solution . the original ejecta expands and cools at the tail of the shocked external medium .
the reverse shock emission can be divided to two distinctive phases , the rs crossing phase and the expanding - cooling phase .
observationally , the two are separated by the peak of the optical flash .
the evolution before the time of the optical peak , @xmath7 , is highly sensitive to the initial profile of the ejecta , namely to the density and the lorentz factor of the ejecta at the beginning of the rs crossing .
thus the light curve before and at @xmath7 can be used as a diagnostic tool of these properties . on the other hand , the rs that crosses the shell erases , to a large extend , the initial shell profile .
moreover , the evolution during the expanding and cooling phase depends only weakly on this profile ( @xcite )
. therefore , the behavior after @xmath7 depends weakly on the initial conditions and as such it provides a very unique signature of an rs emission .
swift is expected to provide detailed optical observation of the early afterglow . in some cases
the observation is expected to start before @xmath7 and thus open a unique window into the properties of the ejecta ( in case that the ejecta is baryonic ) . moreover , here we show that if the prompt emission results from internal shock then the light curve of the early optical afterglow at @xmath8 is expected to be correlated with the prompt @xmath0-ray emission ( see fig .
after the reverse shock crosses the ejecta , the shocked plasma cools and expand at the tail of the shocked external medium .
kobayashi and sari ( @xcite ) explored this phase numerically and they have shown that the evolution of the hydrodynamical parameters in the cooling ejecta depends weekly on the strength of the reverse shock ( that in turn depends on the initial conditions of the ejecta ) . as a result the light curve during this phase depends on the initial conditions only by the relative values of the three reverse shock break frequencies at @xmath7 ( @xmath10 , the self - absorbtion frequency , @xmath11 , the synchrotron frequency and @xmath12 the cooling frequency ) . over a wide range of the parameter space ( assuming an ism @xcite . in this case
the light curve evolution at @xmath9 is different than the evolution in an ism . ] ) these frequencies satisfy @xmath13 , where @xmath14 and @xmath15 are the observed radio and optical frequencies respectively ( @xcite ) . in this case
the optical light curve decays at @xmath9 as @xmath16 ( @xcite ) .
in contrast to the optical emission , the radio continues to rise at @xmath9 and it peaks at a later time , @xmath17 , when @xmath18 . over a wide range of initial conditions ( when the shock is not ultra relativistic ) @xmath19hz . in this case
the radio emission is expected to rise first as @xmath20 until @xmath21 and then as @xmath22 until @xmath17 . at @xmath23 it is expected to decay , similarly to the optical emission , as @xmath16 .
this behavior implies that : @xmath24 where @xmath25[@xmath26 is the peak optical [ radio ] flux ( at @xmath7[@xmath17 ] ) .
@xmath28 is a factor of the order unity that arises due to uncertainty in the hydrodynamics ( @xcite ) .
this entire evolution provides several independent tests to verify that we observe a reverse shock emission .
it provides also an independent measurement of @xmath29 : @xmath30 detailed radio observations at @xmath31 that identify the break during the rising phase , when @xmath32 would enable a determination of @xmath33 as well .
the reverse shock emission during the time that the shock crosses the ejecta depends strongly on the strength of the shock , and in particular on the value of @xmath34 where @xmath35 is the density of the ism and @xmath36 [ @xmath37 is the lorentz factor [ density ] of the ejecta .
the density profile of the ejecta before the internal shocks is not smoothed by the shocks , on the contrary , a density contact discontinuity is produced whenever an internal shock occurs within the ejecta .
therefore at the end of the internal shocks the density profile of the ejecta reflects the initial shells that produced the internal shocks .
since the _ instantaneous _ emission of the rs depends strongly on the density of the ejecta it would vary strongly whenever the rs crosses a density discontinuity that remains from the internal shocks . as a result of the angular smoothing ( @xcite ) the _ observed _
rs emission is an average of the _ instantaneous _ emission over a range of radii .
this smoothing limits the variability time scale of the rs to be of the order of the observed time since the beginning of the prompt emission . therefore , if the prompt @xmath0-ray emission results from internal shocks , the rs optical light curve at @xmath8 is expected to resemble a smooth version of the prompt @xmath0-ray emission . in order to demonstrate this relation between the prompt @xmath0-ray emission and the reverse shock light curve at @xmath8 we have carried out a one dimensional hydrodynamic numerical simulation , starting before the internal shocks and ending long after the reverse shock crossed the ejecta .
the initial conditions are four shells with lorentz factors that vary between 60 and 240 with constant energy ( see the inset in the top panel of fig .
the shells collide at a radius of @xmath38 cm and produce a prompt @xmath0-ray emission composed of two pulses with durations of approximately a second each .
an illustration of the prompt @xmath0-ray emission is depicted at the upper panel of fig .
1 . having the hydrodynamic evolution of the ejecta and the external medium we used a detailed synchrotron radiation code to calculate the rs emission .
the resulting optical light curve is depicted in the bottom panel of fig 1 .
the similarity between the two light curves is clear .
note also that in this specific case the optical reverse shock emission peaks long after the prompt @xmath0-ray emission ends .
this delay is expected when the ejecta is thin . in this case
the ejecta expands linearly with the radius between the internal shocks radius ( @xmath39 cm ) and the external shock radius ( @xmath40 cm ) .
figure 1 shows also that after the reverse shock crosses the ejecta , @xmath9 , the optical light curve is insensitive to the complex initial profile of the ejecta and it decays as @xmath41 .
|
we discuss the optical and radio early afterglow emission of the reverse shock that crosses a baryonic ejecta as it interacts with the external interstellar medium ( ism ) .
we show that the peak of the optical flash divides the light curve of the reverse shock into two distinctive phases .
the emission after the peak depends weakly on the initial conditions of the ejecta and therefore it can be used as an identifiable signature of a reverse shock emission . on the other hand , the emission before the optical peak is highly sensitive to the initial conditions and therefore can be used to investigate the initial hydrodynamic profile of the ejecta . in particular ,
if the prompt @xmath0-ray emission results from internal shocks , the early reverse shock emission should resemble a smoothed version of the prompt @xmath0-ray light curve .
|
diluted magnetic semiconductors ( dms ) are considered to be potential candidates for present and future technological applications in semiconductor spintronics @xcite . during the last decade ,
there have been numerous experimental and theoretical studies of ii - vi , iii - v and iv - vi dms . among the iii - v dms , mn
doped gaas system has been studied rigorously for the last few years .
this system shows a curie temperature ( @xmath1 ) of 110 k for a mn doping concentration of 10 @xmath2 . despite several attempts , @xmath1
could nt be raised beyond 175 k. recently , there have been reports of some room temperature dms .
they include mn doped gap @xcite , mn doped chalcopyrite cdgep@xmath3 @xcite , mn doped gan @xcite etc .
the origin of ferromagnetism in these compounds is still under debate @xcite .
dietl _ et al._@xcite predicted theoretically a high curie temperature ( @xmath4 400 k ) for mn doped gan ( 5 @xmath2 mn ) .
their theory was based on a mean field model of hole mediated ferromagnetism . as the curie temperature for a mn doped gaas system is comparatively lower , the theoretical prediction for a higher t@xmath5 drew much attention .
_ @xcite grew wurtzite mn doped gan films by the molecular beam epitaxy method .
magnetic measurements showed a very high curie temperature of about 940 k. they ruled out the possibility of phase segregation of some ferromagnetic compound e.g. mnga and mn@xmath6n which also have high curie temperatures .
room temperature ferromagnetism in mn doped gan was also observed by reed _
deep level optical spectroscopy measurements @xcite show that mn forms a deep acceptor level at 1.42 ev above the valence band maximum for gan doped with small concentration of mn .
it is to be noted that mn forms an acceptor level at 0.11 ev above the valence band maximum in the case of mn doped gaas .
so , the overlap of the mn d - states with the valence band is rather strong in mn doped gaas compared to mn doped gan .
recently , there have been a few first principles electronic structure calculations of mn doped gan systems .
et al._@xcite performed electronic structure calculations of fe and mn doped gan using the tight - binding linearized muffin tin orbital ( tb - lmto ) method .
sato and katayama - yoshida @xcite performed kkr - cpa ( korringa - kohn - rostoker - coherent potential approximation ) calculations to study the relative stabilities of ferromagnetic and spin glass phases .
they showed that for a low concentration of mn , ferromagnetism is favored whereas for the high concentration , the spin glass phase is stable .
the disordered local moment model was assumed to describe the spin glass phase .
they explained the origin of ferromagnetism in these systems by a competition between double exchange and superexchange interactions .
@xcite studied electronic , magnetic and optical properties of zinc - blende ( mn , ga)n for different concentrations of mn by the tb - lmto method in a supercell approach .
anomalous exchange interactions in iii - v dms were found from calculations by schilfgaarde and mryasov @xcite .
they predicted aggregation of magnetic nanoclusters inside the iii - v host .
@xcite considered ( mn , ga)n in wurtzite structure and performed electronic structure calculations using a plane wave pseudopotential method . in a recent preprint @xcite , the different origins of ferromagnetism in ( mn , ga)as and ( mn , ga)n systems were discussed .
the authors pointed out from self - interaction corrected ( sic ) pseudopotential calculations that ( mn , ga)n is characterized by localized mn 3@xmath0 states with a strong self - interaction . in ( mn , ga)as , @xmath0 states are weakly correlated and are rather delocalized being strongly hybridized with as @xmath7 states . in this communication
, we attempt to understand the electronic structure and magnetic interactions in mn doped gan and gaas systems .
the motivation of this paper is twofold : ( a ) to investigate electronic structure and magnetism of mn doped gan system in detail and ( b ) to have a comparison with mn doped gaas system .
the paper is organized as follows : in the next section , we describe the computational details . in the results section
, a subsection describes the electronic structure and magnetism of the ( mn , ga)n with varying mn concentration in the local spin density approximation ( lsda ) .
then we present results from lsda+u calculations .
finally , we show a comparison of exchange interactions in ( mn , ga)n and ( mn , ga)as systems .
gan can be grown both in the zinc - blende and wurtzite structures .
but , usually , mn is doped in a wurtzite gan host @xcite . in our calculations
, we have considered the wurtzite structure . also ,
for a comparison , we show calculations for the zinc - blende structure .
experimental lattice parameters such as a=3.189 and
c=5.185 with a c / a ratio of 1.626 were taken for the calculations in the wurtzite structure .
results from the atomic relaxations revealed that the nearest neighbor bond - lengths between mn and n change only by 3 @xmath2 compared to that of the bulk gan .
this is in agreement with the results of kronik _ et al .
_ @xcite . in general ,
mn doping in substitutional site results in a small relaxation @xcite of the nearest neighbor anions around it .
calculations have been performed by an _ ab - initio _ plane wave code ( vasp ) @xcite .
vanderbilt @xcite type ultrasoft pseudopotentials were used for the lsda calculations .
lsda+u calculations were done in the projector augmented wave ( paw ) @xcite method as implemented by bengone _
ga 3d orbitals were included in the basis set of the ga pseudopotential . a kinetic
energy cut - off of 350 ev was used for the plane waves included in the basis set .
ceperley and alder @xcite exchange - correlation functional parameterized by perdew and zunger @xcite was considered within lsda .
we have also checked that the results obtained within gga ( generalized gradient approximation ) @xcite are similar .
a 8x8x6 k - points grid was used in the monkhorst pack scheme @xcite for small supercells . for the largest supercell
considered , a 2x2x1 grid was used .
local properties such as local density of states and local magnetic moments were calculated by projecting the wave functions onto spherical harmonics @xcite .
the radii chosen for the projection were 1.31 , 1.21 and 0.74 for mn , ga and n respectively .
we have modeled the system using different supercell sizes to simulate different mn concentrations . for wurtzite structure ,
supercells having 4 , 8 , 16 , 32 , 72 and 108 atoms were used to model a composition mn@xmath8ga@xmath9n for @xmath10=0.5 , 0.25 , 0.125 , 0.0625 , 0.028 and 0.018 respectively . for zinc - blende structure ,
we used a 64 atom - cell to simulate a mn concentration of @xmath10=0.03125 .
it has been found experimentally that mn occupies the ga - site @xcite .
therefore , in our calculational unit cell , 1 ga atom was substituted by a mn atom .
in a iii - v semiconductor , a cation vacancy creates 3 holes in the valence band leaving anion dangling bonds .
when mn occupies the cation site , it donates 3 electrons to fulfill the bonding .
mn is left with 4 unpaired @xmath0 electrons which give rise to 4 @xmath11/mn atom . in a realistic situation , there can be compensating donors e.g. as antisites and interstitial mn atoms @xcite , present in the system to increase or decrease the magnetic moments .
but as we are dealing with ideal uncompensated systems , we always obtain a magnetic moment of 4 @xmath11/mn atom for substitutional mn .
ga@xmath9)n in the wurtzite structure .
data are shown for ( a ) pure gan ( b ) 6.25 @xmath2 mn in gan ( c ) 12.5 @xmath2 mn ( d ) 25 @xmath2 mn ( e ) 50 @xmath2 mn . in ( a )
, total dos / cell as well as the @xmath7-dos of n have been shown whereas in ( b ) to ( e ) , @xmath0-dos of mn ( in shade ) , @xmath7-dos of n and total dos have been plotted .
energies are plotted with reference to valence band maximum ( vbm ) of gan in ( a ) and fermi energies ( @xmath12 ) in ( b ) to ( e).,scaledwidth=45.0% ] in fig .
1(a - e ) , we show the density of states ( dos ) for various concentrations(@xmath10 ) of mn in mn@xmath8ga@xmath9n in the wurtzite structure . in ( a ) , the dos of the undoped gan is presented .
the large band gap is evident from the figure .
the calculated direct band gap at the @xmath13 point is 1.9 ev which is underestimated compared to the experimental band gap of @xmath4 3.4 ev .
this well known underestimation is inherent in the formulation of density functional theory and is well documented in existing literature . from fig .
1(b ) to 1(e ) , we show the doss with increasing @xmath10 .
the mn impurity @xmath0 peak in the band gap is away from the valence band for small @xmath10 . as the concentration increases
, this peak is broadened due to the overlap of mn @xmath0 wavefunctions and the gap between the impurity peak and valence band edge vanishes .
the total dos / cell gradually regains the shape of the dos of undoped system as @xmath10 is reduced .
however , up to @xmath10=0.25 , the fermi level cuts only the spin up doss . as there is no state at fermi level for the spin down channel , we obtain a half metallic solution giving rise to an integer magnetic moment of 4 @xmath11/mn atom . for @xmath10=0.5 shown in fig .
1(e ) , the system behaves like a ferromagnetic metal with a reduced magnetic moment .
one noticeable difference between mn doped gaas and mn doped gan systems is the position of mn impurity band in the gap . in ( mn , ga)as ,
mn impurity band almost merges with the top of the valence band ( approximately 0.1 ev above the valence band edge ) whereas in ( mn , ga)n , the mn impurity band is separated from the valence band by 0.56 ev .
the width of this impurity band is 0.94 ev .
as the spin polarized mn impurity band is distinctly in the gap , the valence band is less spin polarized in ( mn , ga)n compared to ( mn , ga)as .
spin polarization ( @xmath14 ) of the valence and conduction bands at the @xmath13 point is 0.05 ev and -0.42 ev respectively for ( mn@xmath15ga@xmath16)n .
@xmath17 is the eigenvalue and vbm and cbm represent valence band maximum and conduction band minimum respectively .
. projected charges and magnetic moments ( in @xmath11 ) .
mn - d@xmath18 indicates charge in the the sphere around mn for d electrons . mn@xmath19 and n@xmath19 indicate projected magnetic moments inside mn and n spheres respectively . [ cols="<,<,<,<",options="header " , ] we have done lsda+u calculations to investigate the effect of electron correlations .
firstly , these calculations do not exist in literature and secondly , it is interesting to compare this simple technique with more rigorous sic calculations . in fig .
3(a - d ) , we show the doss obtained from lsda and lsda+u calculations for 6.25 @xmath2 mn . as the value of u for mn @xmath0 states
is not obtained self - consistently from first principles calculations , we varied u from 4 ev to 7 ev treating it as a parameter . in all cases ,
the exchange parameter j was considered to be 1.0 ev . increasing u results in a slow shift of the spin - up impurity band towards the valence band .
the small peak around 1.5 ev below the fermi level ( @xmath12 ) is diminished gradually in magnitude whereas the peak around -4.5 ev below @xmath12 increases in magnitude . even for u=10 ev ( not shown here ) , the localized peak remains pinned close to @xmath12 and is _ not _ merged with the delocalized valence band states .
the spin down dos shifts almost rigidly away from the fermi level towards higher energy with increasing u. local charges and magnetic moments of mn and nearest neighbor n atoms are listed in table ii .
magnetic moment of mn increases with u due to increase in localization of @xmath0 states .
the induced moment of n atoms also increase . in fig .
3(c ) , a comparison between lsda and lsda+u calculations is shown .
the redistribution of weights of mn @xmath0 peaks with the inclusion of u is visible . in lsda+u result ,
the sharp peak is very close to the fermi level in the impurity band whereas it is in the valence band in the lsda calculation .
this peak is of @xmath20-character and does nt take part in the bonding with the neighboring n atoms .
these results are in agreement with the more sophisticated sic calculations reported by filippetti _
they also found a flat band with @xmath21 character at @xmath12 .
the nature of the states close to @xmath12 from our calculation is shown later in fig .
5 . to our knowledge ,
the valence band photoemission spectra of ( mn , ga)n is not available in the literature .
so , the extent of validity of lsda approach can not be tested .
our lsda+u findings can be compared with future angle - resolved photoemission experiments on ( mn , ga)n to verify the existence of the localized peak close to @xmath12 . a comparative study with mn doped gaas
is shown in fig .
the main broad mn @xmath0 peak around 2.8 ev below the fermi level in an lsda calculation is shifted 4 ev below the fermi level with a smaller band width .
the dos at the fermi level is also decreased compared to that of an lsda calculation . in both lsda and lsda+u calculations , the hybridization between mn @xmath0 and as @xmath7 states are seen .
so the holes in the valence band have hybridized @xmath22 character . .
zb and w represent zinc - blende and wurtzite structures.,scaledwidth=30.0% ] in fig .
4 , we show the integrated magnetization density around various atoms in the unit cell as a function of the radius of integration . it has been calculated as @xmath23 where @xmath24 and @xmath25 are the spin - up and spin - down charge densities respectively and @xmath26 is the magnetic moment obtained for a radius @xmath27 . for ( mn , ga)n and ( mn , ga)as in the zinc - blende structure , the magnetic moment reaches the value 4 @xmath11 at a distance of 5 far from the mn center .
but , in the wurtzite structure of ( mn , ga)n , this value is reached at a smaller distance .
this again shows a more localized character of mn @xmath0 states in wurtzite gan .
the integration around nearest neighbor anions reveal that n has a positive contribution in ( mn , ga)n whereas , as has a negative contribution . to @xmath12 - 0.5
ev for ( left ) ( mn , ga)n in the wurtzite structure , ( middle ) ( mn , ga)as in the zinc - blende structure and ( right ) ( mn , ga)n in the zinc - blende structure having 6.25 @xmath2 mn in all cases . in x - axis , the plotted states are in the order from left : @xmath28 . see text for details . for the middle panel ,
both spin up and spin down contributions are shown whereas for the left and right panels , there is no state in the spin down channel . in the left column ,
the empty bars represent lsda+u values . for each panel ,
data are shown for mn(top ) , four nearest neighbor ( nn ) anions ( middle ) and the collective contributions from all other atoms ( bottom ) in the unit cell.,scaledwidth=35.0% ] between ferromagnetic and antiferromagnetic configurations vs. mn - mn distance in the unit cell .
data for zinc - blende(zb ) gan , zb gaas and wurtzite ( w ) gan are shown .
the filled triangle represents @xmath29 for [ 001 ] direction in zb gaas.,scaledwidth=35.0% ] in fig . 5
, we show the character of the states within an energy interval close to the fermi level .
the states close to @xmath12 in the electronic structure are important in characterizing the origin of ferromagnetism .
the projection of the wavefunction onto spherical harmonics around each atom has been calculated as mentioned in ref . @xcite . in the figure
, @xmath30 is defined as : @xmath31 , where @xmath32 and @xmath33 are the spin and atom indices respectively and @xmath34 is the index for the bands within the specified energy interval .
it is clear from the figure that for ( mn , ga)n in the wurtzite structure ( left panel ) , the @xmath21 component of mn @xmath20-orbital and the @xmath35 component of nearest neighbor n p orbital are the dominant states close to @xmath12 .
on the other hand , mn @xmath36 and nearest neighbor as @xmath7 states hybridize strongly in ( mn , ga)as .
this is also true for ( mn , ga)n in the zinc - blende structure .
symmetry of the crystal structure and the splitting of d - states under the corresponding crystal fields determine the position of @xmath36 and @xmath20 states .
for the rest of the atoms in the unit cell , states having @xmath37 and @xmath7 character are the dominant .
to determine the interatomic exchange interactions , we followed a simple model . in the unit cell ,
two mn atoms were placed in various positions . for each mn - mn separation , ferromagnetic ( fm ) and antiferromagnetic ( afm ) alignments of mn spins
were considered .
the total energy difference @xmath29 ( @xmath29=@xmath38-@xmath39 ) between these two alignments is a measure of interatomic exchange interaction . in fig .
6 , we plot @xmath29 as a function of mn - mn separation @xmath0 for both mn doped gaas and mn doped gan systems . for mn
doped gan , both zinc - blende and wurtzite structures were considered for these calculations with two mn atoms in unit cells having 64 atoms and 72 atoms respectively .
for mn doped gaas , a 64 atom unit cell was chosen .
ferromagnetic interaction between mn spins is favored for all the cases considered here . in ( mn , ga)n , the first nearest neighbor ( nn ) exchange interaction is the strongest .
the value of @xmath29 is increased a little bit compared to the same for ( mn , ga)as .
so , for a defect - free calculation , there is no indication that ( mn , ga)n should have a much higher @xmath1 than ( mn , ga)as .
it suggests that the formation of other phases during the growth is responsible for very high @xmath1 observed in certain experiments .
it is also seen from the figure that @xmath29 decreases sharply with @xmath0 for mn doped gan systems .
it shows that ferromagnetic exchange interaction in ( mn , ga)n is short - ranged .
this indicates that the formation of mn clusters within a short radial distance might lead to high values of @xmath1 . for mn
doped gaas , the exchange interaction is long ranged and does nt decrease rapidly .
the proper range of ferromagnetic interactions can be studied with a bigger supercell . in summary ,
the results for exchange interactions indicate that the ferromagnetic interaction between mn spins in ( mn , ga)as is mediated by delocalized valence band holes whereas the origin of ferromagnetism in ( mn , ga)n may result from a double - exchange mechanism @xcite involving the hopping of mn @xmath0 electrons .
the other probable mechanism can be the formation of zhang - rice magnetic polaron @xcite .
another interesting observation is the anisotropy of the exchange interactions in different crystallographic directions . in ( mn ,
ga)as , ferromagnetic coupling is stronger either in the bonding direction e.g. [ 111 ] or in a direction where the two mn spins are connected by as bonds e.g. in [ 110 ] direction
. the coupling decreases for the [ 001 ] direction where there is no as atom in between ( shown as a filled triangle in fig .
the strong ferromagnetic interactions in [ 110 ] or [ 111 ] directions result from strong @xmath22 hybridization between mn @xmath0 and as @xmath7 states . as the interaction
is mediated by the delocalized states , they are sufficiently long - ranged . in [ 001 ] direction , the exchange interaction between the mn spins can not be mediated by the @xmath22 hybridization .
so , the value of @xmath29 is decreased compared to the other cases . in ( mn , ga)n , the exchange interaction is not mediated by the delocalized valence band states and the ferromagnetic interactions decrease sharply with the mn - mn separation .
so , the anisotropy in the exchange interactions is not significantly observed . we can conclude that in general , this anisotropy should be present for all mn doped semiconductors where ferromagnetic long range interaction is mediated by delocalized valence band states .
we also studied the exchange interactions in ( mn , ga)n within the lsda+u scheme in a similar way described above .
u=4 ev and u=7 ev were considered for the calculations .
we found that @xmath29 for nearest neighbor mn - mn distance increases from 170 mev / mn to 240 mev / mn while going from u=4 ev to u=7 ev .
ferromagnetic interactions between mn spins become stronger as the localization of the mn @xmath0-states is increased .
we have studied the electronic structure and magnetism of mn doped gan systems for a wide concentration range of mn .
the deep acceptor level of mn lies distinctly in the gap of gan , separated from the valence band of host gan .
this is in contrast to the case of ( mn , ga)as where the mn forms shallow acceptor level close to the valence band of gaas .
ferromagnetic interactions are short - ranged in ( mn , ga)n systems whereas they have delocalized itinerant character for ( mn , ga)as . also , in ( mn , ga)as , exchange interactions are significantly anisotropic in different crystallographic directions . this has been explained in terms of anisotropic @xmath22 hybridization .
the results presented here are for ideal systems having no contribution from the native defects formed during the non - equilibrium growth . to have a more realistic picture
, one should take into account these effects .
also , mn can occupy interstitial positions in the lattice and the spin interactions can become quite complicated . we are presently calculating the formation energies of the defects and subsequently the magnetic interactions in presence of them .
results will be reported in future communications .
we are grateful to the swedish foundation for strategic research ( ssf ) , the swedish natural science research council ( nfr ) , and the gran gustafsson foundation for financial support .
we thank j. hafner and g. kresse for giving us the possibility to use vasp .
is grateful to dr .
lars nordstrm and prof .
olle eriksson for valuable suggestions .
also , we acknowledge computational support from national supercomputer center ( nsc ) at linkping , sweden .
|
mn doped semiconductors are extremely interesting systems due to their novel magnetic properties suitable for the spintronics applications .
it has been shown recently by both theory and experiment that mn doped gan systems have a very high curie temperature compared to that of mn doped gaas systems . to understand the electronic and magnetic properties , we have studied mn doped gan system in detail by a first principles plane wave method .
we show here the effect of varying mn concentration on the electronic and magnetic properties . for dilute mn concentration , @xmath0 states of mn form an impurity band completely separated from the valence band states of the host gan .
this is in contrast to the mn doped gaas system where mn @xmath0 states in the gap lie very close to the valence band edge and hybridizes strongly with the delocalized valence band states . to study the effects of electron correlation
, lsda+u calculations have been performed .
calculated exchange interaction in ( mn , ga)n is short ranged in contrary to that in ( mn , ga)as where the strength of the ferromagnetic coupling between mn spins is not decreased substantially for large mn - mn separation .
also , the exchange interactions are anisotropic in different crystallographic directions due to the presence or absence of connectivity between mn atoms through as bonds .
|
the existence of dark matter halos surrounding large , bright galaxies is well established ( e.g. , @xcite , @xcite , @xcite , @xcite and references therein ) , and in the standard cold dark matter ( cdm ) paradigm , the halos of large field galaxies are expected to extend to virial radii of @xmath5 kpc to @xmath6 kpc and have masses of @xmath7 ( e.g. , @xcite , @xcite , @xcite ) . until very recently
, however , direct observational constraints on the nature of the dark matter halos of field galaxies have not been especially strong . in particular , it has been challenging to address the question as to whether the halos of observed galaxies are consistent with the halos that one would expect in a cdm universe .
the lack of a keplerian fall off in the rotation curves of the disks of most spiral galaxies ( e.g. , @xcite ) indicates that the dark matter halos extend far beyond the visible radii of the galaxies .
therefore , in order to place constraints on the total mass distribution , it is necessary to use tracers of the halo potential that exist at large projected radii ( @xmath8 kpc ) .
two such tracers of the large scale potential are satellite galaxies that are in orbit about isolated host galaxies , and photons emitted by distant galaxies that , on their way to the observer , happen to pass through the potential wells of more nearby galaxies at small impact parameters ( i.e. , gravitational lensing ) .
`` strong '' gravitational lensing , in which multiple and highly distorted images of a source occur , is a rare phenomenon because it requires nearly perfect alignment of the lens and source galaxy ( e.g. , @xcite ) .
`` weak '' gravitational lensing , in which multiple images and significant image distortion do not occur , is , however , commonplace in the universe ( e.g. , @xcite , @xcite , @xcite , @xcite ) , and it is on this extremely mild regime of gravitational lensing that i will focus for this discussion .
weak lensing of background galaxies by foreground galaxies ( `` galaxy galaxy '' lensing ) and the motions of satellite galaxies about host galaxies are phenomena that can only lead to constraints on halo potentials through ensemble averages over statistically large samples .
that is , for any given foreground galaxy , the distortion that it induces in the images of background galaxies due to weak lensing is so small that the signal can not be detected convincingly for any one foreground lens galaxy .
similarly , isolated host galaxies are typically found to have 1 to 2 satellite galaxies on average and , so , the potential of any one host galaxy can not be constrained at all well by the motions of its own satellites .
both galaxy galaxy lensing and satellite dynamics , therefore , lead to statistical constraints on the halo population as a whole and by their very nature they require large samples of galaxies in order to obtain such constraints . until several years ago , galaxy galaxy lensing and satellite dynamics were both tantalizingly close ( or frustratingly close , depending on one s point of view ) to being able to fulfill their theoretical promise to map out the gravitational potentials of the halos of field galaxies . with the advent of routine availability of wide field imagers and the completion ( or near completion ) of large redshift surveys , however , both galaxy galaxy lensing and satellite dynamics are now yielding sufficiently strong constraints on the dark matter halos of galaxies that the observations can be used to test the theoretical predictions ( i.e. , cdm ) at a substantive level .
there are distinct advantages and disadvantages of galaxy galaxy lensing versus satellite dynamics when it comes to constraining halo potentials . a clear advantage of galaxy
galaxy lensing is that it can be applied to _ all _ foreground galaxies and , since gravitational lensing is affected only by the total mass along the line of sight and not its dynamical state , the halos of the foreground lens galaxies need not be virialized . a complicating factor in galaxy
galaxy lensing is that it is not correct to assume that each background galaxy has been lensed solely by one foreground galaxy ( e.g. , @xcite , @xcite , @xcite ) .
instead , photons emitted by the distant galaxies are deflected by _ all _ mass along the line of sight , including individual galaxies , groups , and clusters ( e.g. , figure [ multiple ] ) .
that is , galaxy galaxy lensing is inherently a multiple deflection problem and care must be taken when using observations of galaxy galaxy lensing to constrain the halos of a given subset of lens galaxies ( i.e. , the halos of early type galaxies versus late type galaxies , or the halos of high luminosity galaxies versus low luminosity galaxies ) .
therefore , a computation of the weak lensing signal about the white lenses in figure [ multiple ] above is _ not _ identical to a measurement of weak lensing signal _ produced by _ the white lenses since the black lenses also contribute to the net shape of the final image .
that is not to say that galaxy
galaxy lensing can not be used to probe the potentials of halos surrounding lenses of differing types ; it most certainly can , but the presence of multiple deflections in the data must be taken into account when modeling the observed signal . in the case of relatively shallow data ( @xmath9 )
, most sources will have been lensed by only one foreground galaxy ( e.g. , @xcite ) , but in deep data sets ( @xmath10 ) most source galaxies will have been lensed at a significant and comparable level by two or more foreground galaxies ( e.g. , @xcite , @xcite )
. a further disadvantage of galaxy galaxy lensing is that the signal is very small ( systematic image distortions of @xmath11% or in the image ellipticities ) , so the images of millions of background galaxies must be obtained and , in general , be meticulously corrected for the presence of anisotropic , spatially
varying point spread functions . finally , it is possible that newtonian tidal distortions of genuine satellites of the lens galaxies could masquerade as a weak lensing signal .
happily , such distortions appear to be at most a very small contributor to the observed weak lensing signal ( e.g. , @xcite , @xcite .
@xcite , @xcite ) .
an advantage to using dynamics of satellite galaxies to probe the potentials of the halos of isolated host galaxies is that , unlike deep weak lensing data , the only important potential well in the problem is that of the host galaxy . in principle , this is a `` cleaner '' , more straightforward probe of the halo potential which is intentionally restricted to the physical scales that one would expect to characterize the halos of individual large galaxies .
however , there are a number of arguments against the use of satellite dynamics to probe the mass distributions of host galaxies : ( 1 ) satellites must be found at large projected radii in order to probe the halo potential on the very largest scales , ( 2 ) noise is introduced by the presence of `` interlopers '' ( i.e. , galaxies that are selected as satellites but which are , in fact , not associated dynamically with the host galaxy ) , and ( 3 ) the relaxation times of these systems are large compared to the age of the universe .
the first argument is much less compelling now than it was in the past simply because of the availability of large redshift surveys ( i.e. , the data bases are now sufficiently big that although it is rare to find satellites at , say , a projected radius of 500 , the large number of redshifts that are now available makes it possible to compile statistically significant samples ) .
the second argument has also become much less compelling with the realization that it is straightforward to account for the effects of interloper galaxies on the determination of the velocity dispersion ( see below ) .
the third argument can still be compelling , since it makes little sense to apply a virial type mass estimator to systems which are not relaxed . however , an assumption of virialization is not necessary _ a priori _ , and the use of secondary infall models can be used to bypass this assumption ( e.g. , @xcite ) .
for the sake of a certain amount of brevity and , at the very least , an attempt at providing some level of coherent argument , i will focus here on only the most recent results that are directly relevant to the halos of field galaxies and which have been obtained from four large surveys : the combo17 survey , the red sequence cluster survey ( rcs ) , the sloan digital sky survey ( sdss ) , and the two degree field galaxy redshift survey ( 2dfgrs ) .
even with this restriction , it is simply not possible to discuss all of the most recent results from these studies in great detail , and the reader should consult the source literature for further information .
finally , i should hasten to add that all errors or omissions in this article in regards to my colleagues work are entirely unintentional and entirely my own fault .
i can only hope that my colleagues will be kind enough to forgive me .
the acronym combo17 stands for `` classifying objects by medium
band observations in 17 filters '' @xcite , @xcite , @xcite .
the combo17 survey consists of high quality imaging data with the ability to obtain both rest frame colors and accurate photometric redshifts ( @xmath12 for @xmath13 , @xmath14 for @xmath15 , and @xmath16 for @xmath17 ) .
the survey consists of 5 fields , including an extended region in the location of the chandra deep field south ( cdfs ) .
the observations were carried out using the wide field imager at the 2.2m mpg / eso telescope .
the field of view of the camera is @xmath18 and the 17band filter set covers a wavelength range of @xmath19 .
the latter allows for a rough determination of the spectral energy distributions of the objects , which in turn leads to both reliable classification of the objects into galaxies , quasars , and stars , as well as the ability to determine accurate photometric redshifts .
a catalog containing astrometry , photometry in all 17 bands , object classification , and photometric redshifts for the 63,501 objects in the extended cdfs is publicly
available @xcite ( see http://cdsweb.u-strasbg.fr/cgi-bin/qcat?j/a+a/421/913 ) . the combo17 results that will be discussed here consist of efforts to use galaxy galaxy lensing to study dark matter halos .
the data set is particularly well suited to this task because of the reliability with which background galaxies ( i.e. , lensed sources ) can be separated from foreground galaxies ( i.e. , the lenses ) .
note , too , that although the full combo17 survey covers 5 fields , the results shown here come from only 3 of the fields ( a field centered on the cluster a901 , the cdfs field , and a random field @xcite ) .
the rcs ( @xcite , @xcite ) is a somewhat shallow ( 5@xmath20 point source detection limits of @xmath21 and @xmath22 ) , wide field ( @xmath23 sq .
imaging survey that was designed primarily to search for galaxy clusters out to redshifts of @xmath24 .
the images for the complete survey were obtained with the cfht and ctio 4m telescopes using mosaic cameras , and consist of 22 widely separated patches of @xmath25 .
the rcs results that will be discussed here consist of galaxy galaxy lensing studies and were obtained from @xmath26 sq .
deg . of northern rcs data . without spectroscopic or photometric redshift information , the rcs galaxy galaxy results had to be obtained from a rough separation of lenses and sources that was based upon apparent magnitude cuts ( i.e. , galaxies with `` faint '' apparent magnitudes are on average background objects while galaxies with `` bright '' apparent magnitudes are on average foreground objects ) .
although the foreground background distinction between a given pair of galaxies in the rcs data is by no means as secure as in the combo17 data , the rcs is nevertheless a superb data set for galaxy galaxy studies simply because of the area covered ( @xmath27 times larger than combo17 for the weak lensing work ) . given that weak lensing is primarily a statistical game , this is a good example of how well the galaxy galaxy lensing signal can be detected and also used to constrain the nature of dark matter halos given only minimal distance information and a tremendous number of candidate lenses and sources .
the sdss is a combined photometric and spectroscopic survey that will ultimately map roughly one quarter of the sky above @xmath28 and provide redshifts of @xmath29 galaxies and @xmath30 quasars with @xmath31 .
the sdss is a fully digital survey and makes use of 5 broad optical bands ( @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 ) for photometry .
the data for the sdss are being acquired at the apache point observatory in sunspot , new mexico using a 2.5m telescope , as well as three , smaller subsidiary telescopes for the purposes of photometric calibration , monitoring of the seeing , and scanning for clouds .
the rms galaxy redshift errors are @xmath37 to @xmath38 ( e.g. , @xcite , @xcite ) .
a technical summary of the sdss can be found in york et al .
@xcite , information about the main galaxy sample is given by strauss et al .
@xcite , and information about the photometric system and photometric calibration is given by fukugita et al .
@xcite , hogg et al .
@xcite , and smith et al .
all of the sdss data , including astrometry , photometry , redshifts , and spectra , are available via the sdss website ( http://www.sdss.org ) using structured queries that can search and combine the individual data bases .
the third sdss data release occurred on september 27 , 2004 and includes spectra of 374,767 galaxies , spectra of 51,027 quasars , and photometry of 141 million unique objects .
the sdss results that will be discussed here consist of both galaxy galaxy lensing studies and studies of the satellites of large , isolated galaxies .
the 2dfgrs is a spectroscopic survey in which the target objects were selected in the @xmath39 band from the automated plate measuring ( apm ) galaxy survey ( @xcite , @xcite ) and extensions to the original survey . a detailed discussion of the survey and the data base is given by colless et al .
the observations , which are now complete , were carried out at the anglo
australian telescope using the two degree field ( 2df ) multifiber spectrograph .
the final data release occurred on june 30 , 2003 @xcite and includes reliable redshifts of 221,414 galaxies with extinction corrected magnitudes of @xmath40 , covering an area over @xmath41 square degrees .
galaxies with reliable redshifts have an rms uncertainty of 85 @xcite .
all data , including spectroscopic catalogs ( 245,591 objects ) , photometric catalogs ( 382,323 objects ) , and fits files containing the spectra , are publicly available from the 2dfgrs website ( http://msowww.anu.edu/au/2dfgrs ) .
the 2dfgrs data base is fully
searchable via structured queries , and on line documentation is available on the 2dfgrs website .
the photometric transformation from the sdss band passes to @xmath39 is @xmath42 ( e.g. , @xcite ) .
the 2dfgrs results that will be discussed here consist of investigations into the nature of the dark matter halos of large , isolated galaxies that are orbited by one or more satellite galaxies .
general relativity tells us that any mass will cause a curvature of spacetime in its vicinity . therefore , any mass located along the line of sight to a distant luminous object will act as a gravitational lens by deflecting light rays emanating from the object as they propagate through the universe .
the most striking instances of gravitational lensing ( e.g. , multiple images , rings , arcs ) are examples of rare phenomena caused by strong gravitational lenses , which greatly distort the images of distant galaxies . in contrast to this , weak gravitational lenses distort the images of distant galaxies very little but produce a net coherent pattern of image distortions in which there is a slight preference for the lensed galaxies to be oriented tangentially with respect to the direction vector that connects their centroids with the center of the gravitational potential of the lens .
while weak lenses do not give rise to stunning individual images , they are detectable in a statistical sense via ensemble averages over many mildly distorted images ( e.g. , @xcite , @xcite , @xcite , @xcite ) . provided the distance traveled by the light ray is very much greater than the scale size of the lens , it is valid to adopt the `` thin lens approximation '' in order to describe a gravitational lens . consider a lens with an arbitrary 3-dimensional potential , @xmath43 . in the thin lens approximation a conveniently scaled 2-dimensional potential for the lens ( i.e. , the 3-dimensional potential of the lens integrated along the optic axis ) is given by @xmath44 where @xmath45 is the location of the lensed image on the sky , measured with respect to the optic axis , and @xmath46 , @xmath47 , and @xmath48 are angular diameter distances between the lens and source , observer and lens , and observer and source , respectively ( e.g. , @xcite ) .
it is then straightforward to relate the gravitational potential of the lens to the two fundamental quantities that characterize the lens : the convergence ( @xmath49 ) and the shear ( @xmath50 ) .
the convergence , which describes the isotropic focusing of light rays , is given by @xmath51 the shear describes tidal gravitational forces acting across a bundle of light rays and , therefore , the shear has both a magnitude , @xmath52 , and an orientation , @xmath53 . in terms of @xmath54 ,
the components of the shear are given by @xmath55\ ] ] and @xmath56.\ ] ] the effect of convergence and shear acting together in a gravitational lens is to distort the images of distant objects . consider a source galaxy which is spherical in shape . in the absence of a gravitational lens , an observer would see an image of the galaxy which is truly circular .
if a gravitational lens is interposed along the line of sight to the distant galaxy , the observer will see an image which , to first order , is elliptical and the major axis of the ellipse will be oriented tangentially with respect to the direction vector on the sky that connects the centroids of the image and the lens .
that is , the circular source is distorted into an ellipse , and to first order the distortion consists of both a tangential stretch of @xmath57 and a radial compression of @xmath58 ( e.g. , @xcite ) . in the weak lensing regime , both the convergence and shear are small ( @xmath59 and @xmath60 ) .
the fundamental premise in all attempts to detect weak lensing is that , in the absence of lensing , galaxy images have an intrinsically random ellipticity distribution .
gravitational lensing then introduces a shift in the ellipticity distribution that , in the mean , manifests as a tangential alignment of background sources around foreground lenses .
the image of a distant galaxy can be approximated an ellipse with complex image ellipticity given by @xmath61 where @xmath62 and @xmath63 are the major and minor axes , respectively , and @xmath64 is the position angle .
the complex image ellipticity is often referred to as the `` image polarization '' ( e.g. , @xcite ) and is computed in terms of flux weighted second moments , @xmath65 where @xmath66 is the intensity at a given pixel and @xmath67 is a weighting function .
the real and imaginary components of the image polarization are then given by : @xmath68 the _ observed _ image polarization for any one source is , of course , a combination of its intrinsic ellipticity and any ellipticity that is induced by lensing . in the limit of weak lensing , the observed image polarization , @xmath69 , is related to the intrinsic image polarization , @xmath70 through a shift in the complex plane . although we can not determine @xmath70 for any one particular source galaxy , we have that the mean intrinsic ellipticity distribution for an ensemble of source galaxies is @xmath71 since the galaxies should be randomly oriented in the absence of lensing .
an estimator for the shear induced by weak lensing is then @xmath72 ( e.g. , @xcite ) .
this simple estimator does not reflect the fact that the way in which the shear alters the shape of a source depends upon its intrinsic ellipticity , and in practice this is generally taken into account when computing the shear .
see , e.g. , @xcite , @xcite , and @xcite for discussions of the `` shear polarizability '' and `` shear responsivity '' of sources .
in addition , it is worth noting that , while it is common practice to approximate image shapes as ellipses , there will be some images that have been sufficiently distorted by galaxy
galaxy lensing that a mild bending , or `` flexion '' , of the images will occur and such images can not be accurately represented as ellipses . in principle , flexion of images can be used to detect weak lensing with a signal to noise that is increased over the common practice of fitting equivalent image ellipses @xcite , @xcite . a preliminary application of this technique
@xcite has been carried out with the deep lens survey @xcite , and it will be interesting to see how the technique is further developed and implemented in practice . the first attempts to detect systematic weak lensing of background galaxies by foreground galaxies ( @xcite , @xcite ) were met with a certain degree of skepticism because the apparent distortion of the source galaxy images was rather smaller than one would expect based upon the typical rotation velocities of the disks of large spiral galaxies .
the situation changed when brainerd , blandford & smail @xcite measured the orientations of 506 faint galaxies ( @xmath73 ) with respect to the locations of 439 bright galaxies ( @xmath74 ) and found that the orientation of the faint galaxies was inconsistent with a random distribution at the 99.9% confidence level .
the faint galaxies showed a clear preference for tangential alignment with the direction vector on the sky that connected the centroids of the faint and bright galaxies , in agreement with the expectations of systematic weak lensing of the faint galaxies by the bright galaxies . almost immediately , a number of similar investigations followed in the wake of brainerd , blandford & smail @xcite ( @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) .
these studies made use of a wide variety of data and analysis techniques , and all were broadly consistent with one another and with the results of brainerd , blandford & smail @xcite ( see , e.g. , the review by brainerd & blandford @xcite ) . the first truly undeniable detection of galaxy galaxy lensing was obtained by fischer et al .
@xcite with 225 sq .
deg . of early commissioning data from the sdss , and it was this result in particular that helped to make the study of galaxy galaxy lensing into a respectable endeavor , whereas previously many had considered the whole field rather dodgy at best .
fisher et al .
@xcite demonstrated conclusively that even in the limit of somewhat poor imaging quality , including the presence of an anisotropic point spread function due to drift scanning , galaxy galaxy lensing can be detected with very high significance in wide field imaging surveys . in the last few years , detections of galaxy galaxy lensing and the use of the signal to constrain the dark matter halos of field galaxies has improved dramatically ( @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) owing to a number of factors that include such things as very large survey areas , sophisticated methods for correcting image shapes due to anisotropic and spatially varying point spread functions , and the use of distance information for large numbers foreground lens galaxies in the form of either spectroscopic or photometric redshifts .
figure [ henk : gammat ] shows one example of the high statistical significance with which weak lensing due to galaxies is now being routinely detected .
the result comes from an analysis of the distortion of the images of @xmath75 source galaxies due to @xmath76 lens galaxies in the rcs @xcite , where the lens and source populations were separated solely on the basis of their apparent magnitudes .
the top panel of figure [ henk : gammat ] shows the mean tangential shear _ computed _ about the lens centers which , because of the clustering of the lens galaxies , is not simply interpreted as the tangential shear _ due to _ individual lens centers .
instead , it is a projected ( i.e. , 2dimensional ) galaxy mass cross correlation function , and in order to compute the average properties of the halos of the lens galaxies it is necessary to , e.g. , make use of monte carlo simulations that include all of the multiple weak deflections that the sources have undergone .
the bottom panel of figure [ henk : gammat ] shows a control statistic in which the tangential shear about the lens centers is computed after rotating the images of the sources by @xmath77 .
if the signal in the top panel of figure [ henk : gammat ] is caused by gravitational lensing , the control statistic in the bottom panel of figure [ henk : gammat ] should be consistent with zero ( and indeed it is ) .
note that , although the tangential shear about the rcs lenses persists to scales of order @xmath78 , the shear on such large scales is not indicative of the masses of individual lens galaxies ; rather it reflects the intrinsic clustering of the lenses .
it is also worth noting that less than decade ago observers were struggling to measure a tangential shear of @xmath79 with a modest degree of confidence .
now , however , confident detection of tangential shears of @xmath80 is effectively `` routine '' in these extremely large data sets .
the mean tangential shear , @xmath81 , in an annulus of projected radius @xmath82 is related to the projected surface mass density of the lens through @xmath83 where @xmath84 is the mean surface mass density interior to the projected radius @xmath82 , @xmath85 is the projected surface mass density at radius @xmath82 ( e.g. , @xcite , @xcite , @xcite ) , and @xmath86 is the so called critical surface mass density : @xmath87 where @xmath88 is the velocity of light and @xmath48 , @xmath47 , and @xmath46 are again angular diameter distances @xcite .
the quantity @xmath89 above is , therefore , a mean excess projected mass density . shown in figure [ erin : fig6 ]
is the mean excess projected surface mass density in physical units of @xmath90 for @xmath91 lens galaxies in the sdss for which spectroscopic redshifts are known @xcite .
in addition to spectroscopic redshifts for the lenses , photometric redshifts were used for @xmath92 source galaxies . moreover , because the redshifts of the lens galaxies are known , @xmath93 can be computed as a function of the physical projected radius at the redshift of the lens ( rather than an angular scale ) . in figure
[ erin : fig6 ] , @xmath93 has been corrected for the clustering of the sources around the lenses via a function which is effectively a weighted cross
correlation function between the lenses and sources @xcite .
having obtained a measurement of @xmath94 , or equivalently @xmath89 , constraints can then be placed on the nature of the dark matter halos of the lens galaxies by modeling the observed signal . as mentioned earlier
, quite a bit of care has to be taken in doing this if the goal is to constrain the halo parameters as a function of , say , the host luminosity , color , or morphology ( see , e.g. , @xcite ) . in the past few years , however , good constraints on the mass of an `` average '' halo associated with an @xmath95 galaxy , as well as fundamental differences between the halos of @xmath95 ellipticals versus @xmath95 spirals , have emerged from galaxy galaxy lensing studies and it is those studies which are summarized below .
in order to use satellite galaxies to probe the potentials of host galaxies , one needs to _ define _ an appropriate sample of host and satellite galaxies . unlike cosmology simulators who are blessed with full 6-dimensional phase space information ,
observers are , of course , limited to 3 dimensions ( ra , dec , and redshift ) . given this limited information , then , one must base the selection criteria on projected radii ( evaluated at the redshift of the host ) and relative radial velocities , @xmath96 , of the candidate hosts and satellites . to guarantee that the dynamics of the satellites are determined solely by their host galaxy , the hosts must be determined to be `` isolated '' in some sense .
that is , if another large , bright galaxy is too close to a candidate host galaxy to guarantee that the satellite orbits are affected solely by the candidate host , that candidate host and its satellites are rejected from the sample .
satellites must , necessarily , be fainter than their host , be found within some reasonable projected radius of the host , and have some reasonable line of sight velocity with respect to the host .
there are a number of different selection criteria that have been used in the recent literature , and three sets of selection criteria that have been used in more than one investigation are summarized below : 1 .
hosts must be at least 8 times brighter than any other galaxy that is within @xmath97 kpc and @xmath98 km sec@xmath99 .
in addition , hosts must be at least 2 times brighter than any other galaxy that is within @xmath100 mpc and @xmath98 km sec@xmath99 .
satellites must be at least 8 times fainter than their host , must be found within @xmath97 kpc , and must have @xmath101 km sec@xmath99 . here @xmath102 has been adopted ( @xcite , @xcite ) .
2 . hosts must be at least 2 times brighter than any other galaxy that falls within @xmath103 mpc and @xmath104 km s@xmath99 .
satellites must be at least 4 times fainter than their host , must be found within @xmath105 kpc and must have @xmath98 km s@xmath99 .
here @xmath106 has been adopted ( @xcite , @xcite , @xcite , @xcite ) .
3 . hosts must be at least 2.5 times brighter than any other galaxy that is within a projected radius of @xmath107 kpc and a relative radial velocity difference of @xmath98 km sec@xmath99 .
satellites must be at least 6.25 times fainter than their host , must be found within @xmath97 kpc , and the host satellite velocity difference must be @xmath101 km sec@xmath99 . here
@xmath106 has been adopted ( @xcite , @xcite ) .
although the above criteria may seem lax or even somewhat arbitrary , in the case of the first two sets of criteria , both the milky way and m31 would be excluded from the sample of hosts .
that is , these particular selection criteria give rise to samples of unusually isolated host galaxies .
in addition , both prada et al .
@xcite and brainerd @xcite adopted a number of different selection criteria in their investigations of the satellites of sdss galaxies and concluded that there were no statistical differences between results that were obtained with different selection criteria .
in other words , provided sufficiently `` reasonable '' criteria are adopted for selecting isolated hosts and their satellites , the results of the investigations are stable to modest differences in the details of those selection criteria .
no matter what selection criteria are adopted , however , there will always be `` interlopers '' in the satellite data .
interlopers are galaxies that are falsely identified as satellites ; that is , they pass the formal selection criteria , but they are not , in fact , dynamically associated with the host galaxy .
the presence of interlopers will artificially inflate any measurement of the velocity dispersion of genuine satellites , and recent investigations of satellite dynamics ( @xcite , @xcite , @xcite , @xcite ) have corrected for the effects of interlopers by modeling the distribution of host satellite velocity differences as the sum of a gaussian distribution ( due to the genuine satellites ) and a constant offset ( due to the interlopers ) .
prada et al.@xcite used numerical simulations to show that this is a sensible way in which to correct for the effects of interlopers . moreover , both brainerd & specian @xcite and prada et al .
@xcite have pointed out that an accurate determination of the velocity dispersion profile , @xmath108 , for satellite galaxies depends on a proper determination of the interloper fraction as an explicit function of the projected radius .
that is , by purely geometrical effects , the interloper fraction is necessarily an increasing function of @xmath82 .
an example of fitting a `` gaussian plus offset '' to the distribution of velocity differences for late type galaxies and early type galaxies in the 2dfgrs is shown in figure [ dv : fit ] .
one can clearly see from this figure that the velocity dispersion of the satellites is a function of the morphology of the host galaxy ( being larger for early type hosts than late type hosts ) , and that the interloper fraction increases with projected radius .
the above `` gaussian plus offset '' fit to the distribution of host satellite velocity differences accounts for the fact that the number of interlopers is a function of projected radius , and it assumes _ a priori _ that the number of interlopers at a given projected radius is constant with @xmath109 .
recently , however , van den bosch et al . @xcite used simulations of galaxy redshift surveys to investigate this and found a sharp increase in the number of interlopers for small relative velocities .
van den bosch et al .
@xcite note , however , that the value of @xmath110 that is determined from a simple `` gaussian plus offset '' fit is not strongly affected by the fact that the number of interlopers varies with @xmath109 .
this is because the best
fitting value of @xmath110 is rather insensitive to the precise value of the interloper fraction .
brainerd @xcite also finds that the number of interlopers is larger for small values of @xmath109 than it is for large values of @xmath109 , but that the effect is not nearly as pronounced as found by van den bosch et al .
given the size of the error bars on the distribution of host satellite velocity differences in the current observational samples , then , it would appear that the simple `` gaussian plus offset '' fit to the distribution of velocity differences is more than adequate to the task of estimating @xmath108 .
high resolution cdm simulations have established the existence of a `` universal '' density profile for dark matter halos which results from generic dissipationless collapse ( e.g. , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) .
this density profile fits objects that span roughly 9 orders of magnitude in mass ( ranging from the masses of globular star clusters to the masses of large galaxy clusters ) and applies to physical scales that are less than the `` virial '' radius , @xmath111 .
conventionally , @xmath111 is defined to be the radius at which the spherically averaged mass density reaches 200 times the critical mass density ( e.g. , @xcite , @xcite , @xcite ) .
navarro , frenk & white @xcite , @xcite , @xcite showed that the universal density profile for dark matter halos was fitted well by a function of the form @xmath112 and halos having such a density profile are generally referred to as `` nfw '' halos . here
@xmath113 is the critical density of the universe at the redshift , @xmath114 , of the halo , @xmath115 is hubble s parameter at that same redshift , and @xmath116 is newton s constant .
the scale radius @xmath117 is a characteristic radius at which the density profile agrees with the isothermal profile ( i.e. , @xmath118 ) , @xmath88 here is a dimensionless number known as the concentration parameter , and @xmath119 is a characteristic overdensity for the halo .
formally , the above fitting function for the radial density profiles of cdm halos converges to a steep , cuspy profile : @xmath120 .
the nfw fitting formula , however , was never intended to be extrapolated to very small radii ( i.e. , radii smaller than the practical resolution limits of the simulations ) and much fuss has been made over whether observed galaxies actually show such cuspy inner density profiles ( e.g. , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) .
more recent numerical work has shown that the density profiles of cdm halos do not , in fact , converge to a well defined asymptotic inner slope ( e.g. , @xcite , @xcite @xcite , @xcite ) , and it has become increasingly clear that fair and direct comparisons of simulated galaxies with observed galaxies on very small physical scales is an extremely challenging thing to do ( e.g. , @xcite , @xcite ) .
weak lensing and satellite dynamics do not have the ability to provide any information whatsoever on the cuspiness ( or lack thereof ) in the central regions of galaxies . instead ,
both are governed by the large scale properties of the halos ( i.e. , the regime in which the nfw profile is known to be an excellent description of the density profiles of cdm halos ) and , at least in principle , both have the potential to discriminate between nfw halos and simpler singular isothermal sphere halos .
the radial density profile of a singular isothermal sphere halo is given by @xmath121 ( e.g. , @xcite ) , where @xmath110 is the velocity dispersion .
the isothermal sphere is characterized by the single parameter @xmath110 , which is constant as a function of radius . a key prediction for nfw halos , however , is that the radial velocity dispersion will have a strong dependence upon the radius and this , of course , is inconsistent with the constant value of the velocity dispersion that characterizes singular isothermal spheres .
specifically , on sufficiently small scales @xmath122 should _ increase _ with radius , and on large scales @xmath122 should _ decrease _ with radius .
hoeft , mcket & gottlber @xcite have shown that the radial velocity dispersion of nfw halos can be fitted by a function of the gravitational potential , @xmath123 , of the form : @xmath124^{1/2 } .
\label{nfw : sigmav}\ ] ] note that @xmath125 above is not the `` line of sight '' velocity dispersion , since @xmath126 is a true 3dimensional radius in eqn .
( [ nfw : sigmav ] ) .
the parameters @xmath62 and @xmath49 have values of @xmath127 and @xmath128 , and @xmath129 is the outer potential of the halo .
therefore , we expect the dynamics within an nfw halo to differ fundamentally from the dynamics within an isothermal sphere halo . in the case of weak lensing ,
nfw halos give rise to a distortion in the images of distant galaxies that differs somewhat from the distortion that would be yielded by an isothermal sphere halo ( e.g. , @xcite , @xcite ) .
the radial dependence of the shear for the isothermal sphere is given by : @xmath130 ( e.g. , @xcite ) .
here @xmath88 is the velocity of light and @xmath48 , @xmath47 , and @xmath46 are again angular diameter distances . in the case of nfw halos , the radial dependence of the shear
is given by : @xmath131 & \mbox{$\left(x = 1\right)$ } \\ & \\ \frac{r_{s}\delta_{c}\rho_{c}}{\sigma_{c } } g_{>}(x ) & \mbox{$\left(x > 1\right)$ } \end{array } \right .
\label{gamma : nfw}\ ] ] where @xmath132 , @xmath86 is the critical mass density for gravitational lensing given by eqn .
( [ sigmac ] ) , and the functions @xmath133 are explicitly independent of the cosmology : @xmath134 ( e.g. , @xcite , @xcite ) . in the following sections i summarize the most recent attempts to study the dark matter halos of field galaxies through satellite dynamics and weak lensing , including attempts to distinguish between isothermal and nfw potentials on the basis of the velocity dispersion profile and on the weak lensing shear .
at best , galaxy galaxy lensing and satellite dynamics have the potential to constrain the dependence of the line of sight velocity dispersion on the projected radius , @xmath108 .
determining @xmath108 has proven to be quite a challenge to galaxy galaxy lensing studies , in large part because the shear profiles of nfw lenses and isothermal sphere lenses are not dramatically different , except on the very smallest ( @xmath135 ) and very largest ( @xmath136 ) scales @xcite . to date , only one tentative measurement of @xmath108 has been made from observations of galaxy galaxy lensing @xcite .
kleinheinrich et al .
@xcite modeled the lens galaxies in the combo17 survey as singular isothermal spheres with velocity dispersions that scaled with luminosity as @xmath137 where @xmath138 is the line of sight velocity dispersion of the halo of an @xmath139 galaxy .
kleinheinrich et al .
@xcite fixed @xmath140 to be 0.35 and determined best - fitting values of @xmath138 for projected radii in the range @xmath141 . when they considered all lenses in their sample , kleinheinrich et al .
@xcite found @xmath142 for @xmath143 kpc , @xmath144 for @xmath145 kpc , and @xmath146 for @xmath147 kpc .
this suggests a velocity dispersion profile that rises at small radii , reaches a maximum , then decreases at large radii .
however , the formal error bars on these measurements show that all of these values of @xmath138 agree to within one to two standard deviations .
in addition , it should be kept in mind that each of these measurements of @xmath138 is not independent ( as they would be if a differential measurement of @xmath148 were made ) , so the data points and their error bars are all correlated with one another .
considerably stronger constraints on the dependence of the halo velocity dispersion with projected radius have come from the most recent investigations of the motions of satellites about host galaxies . in particular , both prada et al .
@xcite and brainerd @xcite have measured decreasing velocity dispersion profiles for the satellites of host galaxies in the sdss and 2dfgrs , respectively . although they used different data sets and different host
satellite selection criteria , both prada et al .
@xcite and brainerd @xcite used the same technique to make measurements of the velocity dispersion profiles .
that is , the distribution of velocity differences , @xmath149 , for satellites found within projected radii of @xmath150 was modeled as a combination of a gaussian and an offset due to interlopers . in both studies , the interloper fraction was determined separately for each of the independent radial bins . prior to correcting for the contamination of interlopers , prada
et al.@xcite found a velocity dispersion profile , @xmath108 , that increased with projected radius .
after the removal of the interlopers , however , prada et al .
@xcite found decreasing velocity dispersion profiles in both cases .
the corresponding velocity dispersion profiles are shown in figure [ prada : sigmav ] .
moreover , their corrected velocity dispersion profiles were fitted well by the velocity dispersion profiles of nfw halos with virial masses of @xmath151 ( hosts with absolute magnitudes @xmath152 ) and @xmath153 ( hosts with absolute magnitudes @xmath154 ) . since prada et al .
@xcite adopted a value of @xmath106 and since the absolute magnitude of an @xmath95 galaxy is @xmath155 , these results suggest that the virial mass of the halo of an @xmath95 galaxy is @xmath156 .
brainerd @xcite selected hosts and satellites from the final data release of the 2dfgrs using criteria identical to those of sample 3 in prada et al . @xcite .
in addition , she used these same criteria to select hosts and satellites from the present epoch galaxy catalogs of the flat , @xmath157dominated the gif simulation @xcite .
this is a publicly available simulation which includes semi
analytic galaxy formation in a cdm universe .
brainerd @xcite restricted her analysis to hosts with luminosities in the range @xmath158 , and found a roughly similar number of hosts and satellites in both the 2dfgrs ( 1345 hosts , 2475 satellites ) and the gif simulation ( @xmath159 hosts , @xmath160 satellites , depending upon the viewing angle ) .
like prada et al .
@xcite , brainerd @xcite obtained a decreasing velocity dispersion profile for the satellites of the 2dfgrs galaxies once the effects of interlopers were removed . in addition , excellent agreement between @xmath108 for the 2dfgrs galaxies and @xmath108 for the gif galaxies was found , showing consistency between the motions of satellites in the 2dfgrs and the expectations of a @xmath157dominated cdm universe .
see figure [ 2df : sigmav ] .
further , brainerd @xcite divided her sample of 2dfgrs host galaxies into thirds based upon the spectral index parameter , @xmath140 @xcite , and computed the dependence of the velocity dispersion profile on host spectral type .
the subsamples corresponded to hosts which are expected to have morphologies that are approximately : ( i ) e / s0 , ( ii ) sa , and ( iii ) sb / scd .
the median luminosities of the hosts in the subsamples were all fairly similar : ( i ) @xmath161 , ( ii ) @xmath162 , and ( iii ) @xmath163 .
the velocity dispersion profiles of all three samples decreased with radius and , moreover , @xmath108 was found to have a much higher amplitude and steeper decline for the satellites of early type hosts than it did for the satellites of late type hosts .
see figure [ 2df : svspec ] .
although there is some difference in the median luminosities of the hosts in the subsamples , the difference is too small to have a significant effect on the velocity dispersion profiles .
therefore , the results of brainerd @xcite seem to indicate that early type galaxies have deeper potential wells ( and hence more massive halos ) than late type galaxies .
previous work on the dependence of @xmath110 with projected radius using sdss galaxies @xcite and 2dfgrs galaxies @xcite concluded that @xmath108 was consistent with an isothermal profile ; i.e. , @xmath164 . in both of these investigations , the hosts and satellites
were selected in a manner that was identical to that of sample 3 in prada et al.@xcite . in both previous analyses , however , the number of hosts and satellites was significantly smaller than the more recent studies , and the formal error bars were correspondingly larger .
in addition , the original analysis of sdss host satellite systems @xcite neglected to account for the fact that the interloper fraction increases with radius , which would have biased measurements of @xmath110 at large @xmath82 towards values which are higher than the actual satellite velocity dispersion at those radii . even more recently , conroy et al . @xcite used satellites of @xmath165 host galaxies in the deep2 survey to investigate @xmath108 .
deep2 ( deep extragalactic evolutionary probe 2 ) is being carried out with the deimos spectrograph at the keck
ii telescope , and will ultimately collect spectra of @xmath166 galaxies with redshifts of @xmath167 to a limiting magnitude of @xmath168 @xcite . unfortunately , the survey is still far from complete and only 61 isolated host galaxies ( having a total of 75 satellites ) were found in the current deep2 data . because of this ,
the errors on @xmath108 are large , and formally @xmath108 for the deep2 galaxies is fitted well by a constant value : @xmath169 , @xmath170 , @xmath171 . therefore , isothermal halos for the deep2 galaxies can not be ruled out at the moment .
conroy et al .
@xcite show , however , that their velocity dispersion measurements are consistent with expectations for nfw halos with virial masses in the range @xmath172 .
this is in good general agreement with the results of prada et al .
@xcite , especially considering that the deep2 hosts are of order one magnitude brighter than the sdss hosts ( i.e. , the virial mass implied for the halos of the brightest galaxies in the sdss sample is @xmath173 ) . at the moment , however , the deep2 data are too severely limited by small number statistics to place strong constraints on the nature of the dark matter halos of galaxies with redshifts of order unity .
although it ought to be straightforward and even easy to compare the halo masses and galaxy mass to light ratios that are obtained from different studies , in practice it is rather like comparing persimmons to tomatoes ; i.e. , they are vaguely similar on the inside and outside , but they are definitely not interchangeable .
the fundamental problem is that it is simply not possible to measure the `` total '' mass of a galaxy halo ( since it is not possible to say where such a halo `` ends '' ) and , hence , all halo masses are simply masses that are contained within some physical radius of the center of the halo . along those same lines , and given that velocity dispersion profiles of nfw halos decrease with radius , if one wants to compare the results of two investigations which have measured a velocity dispersion averaged over some large scale , it is important that those scales be _
identical_. that is , suppose a single measurement of @xmath174 is made by averaging over scales @xmath175 kpc in one study and a single measurement of @xmath174 is made by averaging over scales @xmath176 kpc in another .
if the second measurement of @xmath174 is lower than the first by some significant amount , that does not necessarily mean that the values are in disagreement . they would be in disagreement if both halos were isothermal spheres , but if the halos are nfw objects , then it is only to be expected that the second measurement would be lower than the first .
a more subtle problem is the definition of the `` virial radius '' in the context of nfw halos . while @xmath111 was originally proposed as the radius at which the interior mass density is 200 times the critical mass density ( e.g. , @xcite , @xcite , @xcite ) , it is not at all uncommon to find that investigators who have fit nfw models to their data have defined the virial radius as the radius at which the interior mass density is 200 times the mean mass density of the universe
therefore , what is meant by a `` virial mass '' in the context of an nfw fit to data can ( and does ) vary from investigation to investigation , and a certain amount of care has to be taken when comparing such results . despite the difficulties of comparing the conclusions of different studies , i will forge ahead because it is becoming clear that a consistent picture really is emerging on the topic of the masses of the halos of field galaxies , and their corresponding mass to light ratios .
the weak lensing studies yield results that are by and large consistent with each other , and the dynamical studies seem to be in general agreement with the trends in the weak lensing data : the halos have masses that are consistent with expectations for galaxy sized halos in cdm , and there are real , physical differences between halos surrounding ( i ) early type and late type galaxies and ( ii ) high luminosity and low luminosity galaxies . in the case of galaxy galaxy lensing , it is not possible at the moment to discriminate between shear profiles that are caused by nfw versus isothermal galaxy halos .
therefore , investigators will often choose one or the other to constrain the properties of the halos that are producing the lensing signal .
in the case of isothermal sphere halos , the velocity dispersions of the lens galaxies used to model the observed signal are often chosen to scale as in eqn .
( [ tf ] ) above , @xmath179 , where again @xmath110 is the velocity dispersion of a halo that contains a galaxy of luminosity @xmath180 , and @xmath138 is the velocity dispersion of the halo of an @xmath95 galaxy .
hoekstra et al .
@xcite used this approach with their rcs data , as did kleinheinrich et al .
@xcite with their combo17 data .
when all lenses and sources were used in the investigations , and when the lensing signal was averaged over an identical scale ( @xmath181 kpc ) , both the rcs and combo17 results are in very good agreement with each other .
in particular , hoekstra et al .
@xcite find @xmath182 for an adopted value of @xmath183 , and kleinheinrich et al .
@xcite find @xmath184 and @xmath185 .
further , kleinheinrich et al .
@xcite find that there are clear differences in the halos surrounding `` blue '' galaxies ( rest frame colors of @xmath186 $ ] ) and those surrounding `` red '' galaxies ( the remainder of the sample ) .
that is , the red combo17 lens galaxies have a higher velocity dispersion than the blue combo17 lens galaxies , but both have a similar value of the index @xmath140 above .
see figure [ martina : sis ] .
in addition , guzik & seljak @xcite , hoekstra et al.@xcite , and kleinheinrich et al .
@xcite have all used nfw halos to model their lens galaxies , and all find very reasonable fits to their lensing signals .
further , the derived values of the nfw virial masses of the halos of @xmath95 galaxies are in quite good agreement amongst these studies when they are determined in similar band passes ( e.g. , @xmath126 ) and with identical definitions of the virial radius @xcite : @xmath187 @xcite , @xmath188 @xcite , and @xmath189 @xcite .
these are also in remarkably good agreement with the virial mass implied for the halos of @xmath95 galaxies by the dynamical analysis of prada et al .
@xcite ( e.g. , @xmath190 ) . shown in figure [ henk : nfw ]
are 1@xmath20 , 2@xmath20 and 3@xmath20 confidence limits on a joint
parameter fit of the circular velocity at @xmath111 , @xmath191 , and scale radius , @xmath192 , for the lenses in the rcs data @xcite .
note that in the analysis of the rcs data , @xmath191 and @xmath192 were allowed to vary freely , while , to within some scatter , these parameters are strongly correlated in the nfw theory ( i.e. , the nfw model is in essence specified by a single parameter ) .
the dashed line in figure [ henk : nfw ] therefore shows the prediction for a strict adherence to the nfw theory ( i.e. , @xmath191 and @xmath192 are correlated appropriately ) , and the fact that the theoretical nfw line passes so well through the contours gives a certain amount of confidence that the nfw model is a very good fit to the data .
kleinheinrich et al .
@xcite find good fits of the nfw model to their data and , moreover , find that both the virial radii of the halos and the parameter @xmath140 are dependent upon the rest frame colors of the galaxies , with red galaxies having a somewhat larger virial radius ( and , hence , larger virial mass ) than blue galaxies .
see figure [ martina : nfw ] . here
@xmath140 is defined not as in eqn .
( [ tf ] ) , since the velocity dispersion is a function of projected radius in the nfw model , but rather it is defined as : @xmath193 in analogy to the tully fisher and faber jackson relations ( see @xcite ) . in this case , @xmath194 is the virial radius of the halo of an @xmath95 galaxy , defined at 200 times the mean mass density of the universe .
the variation of @xmath140 with galaxy color and its implications for the mass to light ratios of the galaxies will be discussed below .
a particularly detailed study of the masses of lensing galaxies as a function of their color was carried out by guzik & seljak @xcite for @xmath195 lenses and @xmath196 sources in the sdss .
all of the lens galaxies have spectroscopic redshifts in this case , and all of the halos were modeled as nfw objects in the context of the `` halo model '' . in all 5 of the sdss band passes , guzik & seljak @xcite find that the virial masses of @xmath95 ellipticals exceed those of @xmath95 spirals though , unsurprisingly , the amount by which the masses of the ellipticals exceeds those of the spirals is a strong function of the band pass . in the redder bands , the masses of the ellipticals
exceed those of the spirals by a factor of @xmath197 to @xmath198 , while in @xmath33 the difference is a factor of @xmath199 and in @xmath32 the difference is close to an order of magnitude .
although it is difficult to make direct comparisons between the two studies ( because of the differing definitions of the virial radius and the different definitions of the subsamples of galaxies ) , there is good general agreement between the results of guzik & seljak @xcite and kleinheinrich et al .
@xcite : when the galaxy galaxy lensing signal is detected red band passes ( e.g. , @xmath200 , @xmath34 ) and the lenses are modeled as nfw objects , the virial masses of red / early type galaxies exceed those of blue / late type galaxies by a factor of order 2 .
in addition to the halos of early type lenses having more mass than those of late type lenses , the weak lensing work of sheldon et al .
@xcite indicates that , again , in all 5 sdss band passes , the projected excess surface mass density increases with the luminosity of the lens .
sheldon et al .
@xcite separated their @xmath91 lenses into 3 magnitude bins ( high , middle , and low luminosity ) , and the magnitude cuts differ for the different band passes .
( see table 2 of sheldon et al .
@xcite for a complete list of the magnitude cuts as a function of band pass . ) in the case of the @xmath34 data , the `` high '' luminosity galaxies have a mean absolute magnitude of -22.5 , the `` middle '' luminosity galaxies have a mean absolute magnitude of -21.9 , and the `` low '' luminosity galaxies have a mean absolute magnitude of -20.5 .
these mean luminosities correspond roughly to @xmath201 ( `` high '' ) , @xmath202 ( `` middle '' ) and @xmath203 ( `` low '' ) in the @xmath34 band . in all cases
, @xmath89 for the `` high '' luminosity galaxies exceeds that of the `` medium '' and `` low '' luminosity galaxies , and for @xmath204 mpc , the difference corresponds to an approximately constant multiplicative factor . specifically at @xmath205 mpc , however , @xmath206 for the high luminosity lenses in sheldon et al.@xcite exceeds that for the low luminosity lenses by a factors of @xmath207 in @xmath32 , @xmath208 in @xmath33 , @xmath208 in @xmath34 , @xmath209 in @xmath35 , and @xmath209 in @xmath36 ( e.g. , figure 14 of sheldon et al .
similar trends ( i.e. , higher projected excess surface mass density for more luminous lenses ) were found by seljak et al.@xcite in their galaxy galaxy lensing analysis of sdss data .
lastly , although there is reasonable agreement regarding the relative increase in mass for the halos of early type lens galaxies versus late type lens galaxies at fixed luminosity ( i.e. , @xmath95 ) , there is some disagreement over the dependence of the mass to light ratio on the luminosity of the host . specifically , in their redder bands guzik & seljak @xcite find that the mass to light ratio goes as @xmath210 for @xmath211 , suggestive of a mass to light ratio that increases with luminosity .
kleinheinrich et al .
@xcite , however , find that @xmath178 for their sample of lenses is more consistent with a constant value : @xmath212 . both guzik & seljak @xcite and kleinheinrich et al .
@xcite agree , however , that the mass to light ratio of red / early type @xmath95 lens galaxies exceeds that of blue / late type @xmath95 lens galaxies by a factor of @xmath197 to @xmath198 in the redder bands . in the 1990 s , zaritsky et al .
@xcite and zaritsky & white @xcite used the velocity differences between a small number of isolated spiral galaxies and their satellites to show that the halos of the spirals were massive and extended to large radii : @xmath213 to @xmath214 .
moreover , zaritsky et al .
@xcite found a somewhat curious result : the velocity difference between their 115 satellites and 69 hosts was independent of the inclination corrected h - i line width of the host and was , therefore , independent of the luminosity of the host ( through , e.g. , the tully fisher relation ) . at fixed large radius ,
then , this would imply that @xmath178 for the spiral hosts decreased as @xmath215 .
more recent investigations of halo masses and corresponding mass to light ratios from satellite dynamics have led to rather a large assortment of conclusions .
mckay et al .
@xcite and brainerd & specian @xcite used the dynamics of the satellites of sdss galaxies and 2dfgrs galaxies , respectively , to constrain the dynamical masses of the halos of the host galaxies interior to a radius of @xmath216 kpc .
both used an isothermal mass estimator of the form @xmath217 where @xmath110 is the line of sight velocity dispersion .
both felt this assumption was justified because both found that their velocity dispersion profiles were consistent with a constant value . in the case of mckay et al .
@xcite , however , no correction for an increasing number of interlopers with projected radius was made and this may have led to an incorrect conclusion that @xmath218 was independent of @xmath82 . in the case of brainerd & specian @xcite , the increasing number of interlopers at large @xmath82 was taken into account , but only galaxies from the 100k data release of the 2dfgrs were used ( i.e. , roughly half as many galaxies as in the full data release ) , and although @xmath219 was consistent with a constant value in their data , the later analysis by brainerd @xcite showed that this was simply due to the rather large error bars in brainerd & specian @xcite .
this being the case , the mass to light ratios published by these two studies are suspect at some level , but it is unclear at the moment just how suspect they may actually be .
that is , while it is true that the velocity dispersion profile of nfw halos decreases with radius , the fall off in @xmath108 is not particularly sharp and it is not obvious how badly isothermal mass estimates of the form in eqn .
( [ mdynam : iso ] ) , which are based on an average value of @xmath110 , will compare to proper nfw mass estimates .
formally , mckay et al .
@xcite found that in all 5 sdss band passes , @xmath220 was roughly constant for @xmath221 , and that the value of @xmath220 was a strong function of the band pass ( being systematically higher in the blue bands than in the red bands ) .
brainerd & specian @xcite found that for @xmath222 , @xmath220 was a constant for dynamical analyses that included ( i ) all 809 hosts in their sample and ( ii ) 159 hosts that had been visually classified as early type ( e / s0 ) .
however , much like the results of zartisky et al .
@xcite , brainerd & specian @xcite found that @xmath220 decreased as @xmath223 for 243 hosts that had been visually classified as spirals .
this latter result remains puzzling , and is certainly in need of further investigation with larger data sets . in their analysis of the dynamics of the satellites of sdss host galaxies , prada et al .
@xcite found that the velocity dispersion of the satellites scaled with host luminosity as @xmath224 ( i.e. , in good agreement with the local @xmath225band tully fisher relationship @xcite ) for satellites with projected radii @xmath226 kpc .
( recall , too , that in this study @xmath108 was specifically corrected for the increase in interlopers at large @xmath82 . ) in addition , prada et al .
@xcite found that for satellites at large projected radius , @xmath227 , the velocity dispersion scaled with luminosity as @xmath228 ( i.e. , steeper than expected from the tully fisher relation ) .
see figure [ prada : tf ] . similar to prada et al .
@xcite , brainerd @xcite also computed the dependence of the small scale velocity dispersion of satellites on host luminosity .
see figure [ 2df : tf ] .
like prada et al .
@xcite , brainerd @xcite corrected for the fact that the interloper fraction is an increasing function of projected radius and overall , she found excellent agreement between the velocity dispersions of satellites with projected radii @xmath229 kpc in the 2dfgrs and gif simulations .
the velocity dispersions of the 2dfgrs satellites were , however , seen to scale with host luminosity as @xmath230 , which is only marginally consistent with the results of prada et al .
@xcite and the local @xmath225-band tully fisher relationship .
prada et al .
@xcite have shown ( e.g. , their figure 12 ) that the dependence of the line of sight velocity dispersion on the virial mass of nfw halos scales as @xmath231 for the case that @xmath110 is computed as an average over scales @xmath232 , and that @xmath233 for the case that @xmath110 is computed at @xmath234 kpc . combining this with their results for the dependence of @xmath110 on @xmath180 at different scales leads to the conclusion that on scales
@xmath235 kpc , @xmath236 while on scales @xmath237 kpc , @xmath238 is a constant . similarly , if the halos of the 2dfgrs galaxies studied by brainerd @xcite are assumed to be nfw objects , the implication is that @xmath239 for the 2dfgrs hosts ( again , computed on scales @xmath235 kpc ) . while it certainly can not be said that there is a consensus from weak lensing and satellite dynamics as to the exact dependence of the galaxy mass to light ratio on @xmath180 , it does seem to be the case that all of these studies point towards a dependence of @xmath238 on @xmath180 that is , at most , rather weak . that is , with the notable exception of the brainerd & specian @xcite result for late type galaxies , all of the recent determinations of @xmath178 for @xmath240 find that , to within 2@xmath20 , @xmath178 is independent of @xmath180 .
in addition , when the weak lenses and host galaxies are each modeled as nfw objects , a fairly consistent value of the average virial mass of the halos of @xmath95 galaxies is found : @xmath241 .
further , it seems to be clear that both weak lensing and satellite dynamics indicate that the masses of the halos of early type galaxies are larger than that of late type galaxies , and that at fixed luminosity the mass to light ratios of early type galaxies are larger than those of late type galaxies .
although the simple isothermal sphere can reproduce the flatness of the rotation curves of the disks of spiral galaxies at large radii , there are both observational and theoretical arguments in favor of halos which are flattened , rather than spherical .
direct observational evidence for halo flattening that has come from studies of individual galaxies is somewhat scarce , however , owing to the fact that there are relatively few galaxies for which the shape of the halo potential can be probed directly . nevertheless , the evidence for flattened halos of individual galaxies is diverse and includes such observations as the dynamics of polar ring galaxies , the geometry of x - ray isophotes , the flaring of hi gas in spirals , the evolution of gaseous warps , and the kinematics of population ii stars in our own galaxy .
in particular , studies of disk systems which probe distances of order 15 kpc from the galactic planes suggest that the ratio of shortest to longest principle axes of the halos is @xmath242 ( see , e.g. , the comprehensive review by sackett @xcite and references therein ) .
studies of a number of strong lens galaxies have also suggested that the mass distributions of the lenses are not precisely spherical .
for example , maller et al .
@xcite found that , provided the disk mass is small compared to the halo mass , the halo of the spiral galaxy which lenses the quasar b1600 + 434 is consistent with @xmath243 .
in addition , the 17 strong lens systems studied by keeton , kochanek & falco @xcite showed some preference for flattened mass distributions , although extremely flattened ( i.e. , `` disky '' ) mass distributions were ruled out .
finally , a recent analysis of the luminous halos of 1047 edge on disk galaxies in the sdss suggests that the old stellar populations of these galaxies consist of moderately flattened spheroids with axis ratios of @xmath244 @xcite . on the theoretical side ,
high resolution simulations of dissipationless cdm models consistently produce markedly non spherical galaxy halos with a mean projected ellipticity of @xmath245 ( see , e.g. , @xcite , @xcite ) .
it is known , however , that the dark matter will react to the condensation of baryons during galaxy formation ( e.g. , @xcite ) and that the resulting increase in the central density leads to a more spherical shape than if dissipation were not considered ( e.g. , @xcite ) .
recent simulations performed by kazantzidis et al .
@xcite show that on scales @xmath246 , the effects of gas cooling cause a substantial circularization of the mass density profile , leading to a projected ellipticity of @xmath247 to 0.5 in the inner regions of the galaxy .
however , on scales @xmath248 kazantzidis et al .
@xcite find that the projected ellipticity is @xmath245 .
since both the weak lensing shear and satellite dynamics are determined primarily by the large scale mass distribution of the halos , the roundness of the mass distribution on small scales due to gas cooling should not have a dramatic effect . from a theoretical standpoint , therefore , it is not at all unreasonable to expect that galaxy
galaxy lensing and satellite dynamics should reflect a significant flattening of the halos . unlike a spherically symmetric lens for which the gravitational lensing shear is isotropic about the lens center , the shear due to an elliptical lens is anisotropic about the lens center .
specifically , at a given angular distance , @xmath249 , from an elliptical lens , source galaxies which are located closer to the major axis of the mass distribution of the lens will experience greater shear than sources which are located closer to the minor axis ( e.g. , @xcite ) . noting this ,
natarajan & refregier @xcite and brainerd & wright @xcite modeled the dark matter halos of field galaxies as infinite singular isothermal ellipsoids and made rough estimates of the sizes of observational data sets which would be required to detect `` anisotropic '' galaxy galaxy lensing and , hence , to constrain the net flattening of the halo population .
both studies concluded that , if the mean flattening of the halos is of order 0.3 , then only a relatively modest amount of imaging data would be necessary to observe the effects of halo flattening on the weak lensing signal . in estimating the amount of data that would be required to detect anisotropic galaxy galaxy lensing , both natarajan & refregier @xcite and brainerd & wright @xcite made the simplifying assumption that each distant source galaxy is lensed by only one foreground galaxy .
however , for a somewhat deep imaging survey ( @xmath250 ) , the simulations of galaxy galaxy lensing performed by brainerd , blandford & smail @xcite indicated that most of the galaxies with magnitudes in the range @xmath251 would have been lensed at a comparable level by two or more foreground galaxies . in a realistic data
set , these multiple weak deflections might significantly affect the signal to noise that could be achieved when attempting to detect anisotropic galaxy galaxy lensing . this motivated wright & brainerd @xcite to carry out detailed monte carlo simulations of galaxy galaxy lensing by flattened halos , including the effects of multiple weak deflections on the final images of distant galaxies .
wright & brainerd @xcite showed that multiple weak deflections create systematic effects which could hinder observational efforts to use weak lensing to constrain the projected shapes of the dark matter halos of field galaxies .
they modeled the dark matter halos of lens galaxies as truncated singular isothermal ellipsoids , and for an observational data set in which the galaxies had magnitudes in the range @xmath252 , they found that multiple deflections resulted in strong correlations between the post lensing image shapes of most foreground background pairs of galaxies . imposing a simple redshift cut during the analysis of the data set , @xmath253 and @xmath254 , was sufficient to reduce the correlation between the final images of lenses and sources to the point that the expected anisotropy in the weak lensing signal was detectable via a straightforward average .
wright & brainerd @xcite concluded that previous theoretical calculations of weak lensing due to flattened halos had considerably underestimated the sizes of the observational data sets which would be required to detect this effect .
in particular , for a multi color survey in which the galaxies had apparent magnitudes of @xmath252 and the imaging quality was modest , wright & brainerd @xcite found that a 4@xmath20 detection could be obtained with a survey area of order 22 sq .
deg . , provided photometric redshift estimates were made for the galaxies , the typical error in @xmath255 was @xmath256 , and only source galaxies with azimuthal coordinates that were within @xmath257 of the lens symmetry axes were used in the data analysis . to date
, only one intrepid team of investigators has claimed a detection of flattened halos from observations of galaxy galaxy lensing . in their analysis of the rcs galaxy galaxy lensing signal hoekstra , yee & gladders @xcite took the approach of modeling the lens galaxies as having halos with ellipticities that scaled linearly with the ellipticity of the image of the lens : @xmath258 .
further , they assumed that the major axis of the lens image was aligned with the major axis of the halo in projection on the sky .
this is a sensible assumption provided the majority of the lenses are relaxed systems , and it is justified at least partially by the observations of kochanek @xcite who found that the major axes of the mass and light of strong lens galaxies were aligned to within @xmath259 in projection on the sky .
hoekstra , yee & gladders @xcite performed a maximum likelihood analysis and concluded that spherical halos ( i.e. , @xmath260 ) could be ruled out at the 99.5% confidence level on the basis of their weak lensing signal ( see figure [ henk : flat ] ) .
formally , hoekstra , yee & gladders @xcite found @xmath261 .
since the mean ellipticity of the lens images in their study was @xmath262 , this implies a mean halo ellipticity of @xmath263 and a projected axis ratio of @xmath264 .
this is in excellent agreement with the expectations for cdm halos , as well as previous observational constraints on halo flattening obtained on large physical scales ( see , e.g. , @xcite ) .
while it may yet be a bit premature to call this result a `` definitive '' measurement of the flattening of field galaxy halos , it is certainly impressive and the statistics will only improve as weak lensing surveys become larger . in the case of substantially flattened halos of host galaxies
, one would naively expect that satellite galaxies would show a somewhat anisotropic distribution about the host .
that is , barring possible effects due to infall rates and orbital decay , one would expect the satellites to have some preference for being located near to the major axis of the host s halo . until very recently ,
however , such an observation had not been confidently made and , moreover , a preference for clustering of satellite galaxies along the _ minor _ axes of host galaxies has been reported at a statistically significant level by a handful of authors ( @xcite , @xcite , @xcite ) .
the apparent alignment of satellite galaxies with the minor axes of the host galaxies is often referred to as the holmberg effect and in the naive picture of satellite orbits in flattened potentials , observations of the holmberg effect lead to the uncomfortable conclusion that not only is the halo mass flattened , but it is also anti aligned with the luminous regions of the galaxy . while one is tempted to dismiss the minor axis clustering of satellites observed by zaritsky et al .
@xcite and holmberg @xcite as being due to some combination of selection biases and very small sample sizes , it is not easy to use this argument for the results of sales & lambas @xcite . in their study ,
sales & lambas @xcite selected hosts and satellites from the final data release of the 2dfgrs , with a resulting sample size of 1498 hosts and 3079 satellites .
the satellites were constrained to be within projected radii @xmath265 kpc of their host and to be within a velocity difference @xmath101 .
further , host images were required to have eccentricities of at least 0.1 in order that the orientation of their major axes be well determined . when sales & lambas @xcite searched their entire sample for anisotropies in the distribution of satellites about 2dfgrs hosts , their results were consistent with an isotropic distribution . however , when they restricted their sample to only hosts and satellites whose radial velocities differed by @xmath266 , an apparently strong detection of the holmberg effect ( i.e. , minor axis clustering of the satellites ) was found . more recently , brainerd @xcite investigated the distribution of satellites about hosts in the second data release of the sdss .
she selected her samples using three different criteria : ( 1 ) the criteria used by sales & lambas @xcite in their investigation of the holmberg effect for 2dfgrs galaxies , ( 2 ) the criteria used by mckay et al .
@xcite and brainerd & specian @xcite in their analyses of satellite dynamics in the sdss and 2dfgrs , respectively , and ( 3 ) the selection criteria used by zartisky et al .
@xcite in their investigation of the holmberg effect .
in addition , brainerd @xcite restricted the analyses to hosts with ellipticities @xmath267 and satellites that were found within a projected radius of 500 kpc .
the three selection criteria lead to samples of : ( 1 ) 1351 hosts and 2084 satellites , ( 2 ) 948 hosts and 1294 satellites , and ( 3 ) 400 hosts and 658 satellites respectively . in all three samples
, brainerd @xcite found that the distribution of satellites about their hosts was inconsistent with an isotropic distribution .
formally , when a kolmogorov
smirnov test was applied to the distribution of satellite locations , an isotropic distribution was rejected at a confidence level of @xmath268% for sample 1 , @xmath268% for sample 2 , and @xmath269% for sample 3 .
further , the mean angle between the major axis of the host and the direction vector on the sky that connected the centroids of the hosts and satellites was found to be @xmath270 for sample 1 , @xmath271 for sample 2 , and @xmath272 for sample 3 .
that is , a clear anisotropy in the distribution of satellites about the hosts was seen , and the satellites showed a preference for being aligned with the _ major _ axis of the host rather than the minor axis ( see figure [ holmberg1 ] ) .
in addition , brainerd @xcite investigated the dependence of @xmath273 with projected radius on the sky and found that the majority of the anisotropy arose on small scales ( @xmath274 kpc ) in all three samples ( see figure [ holmberg2 ] ) .
in other words , the anisotropy was detected on physical scales that are comparable to the expected virial radii of large , bright galaxies . on scales
much larger than the expected virial radii of galaxy
sized halos ( @xmath275 kpc to 500 kpc ) , the distribution of satellites about the sdss hosts was consistent with an isotropic distribution at the 1@xmath20 level . aside from the brainerd @xcite claim of `` planar '' ( rather than `` polar '' ) alignment of satellites with the symmetry axes of their hosts ,
there has been only one other similar claim .
valtonen , teerikorpi & argue @xcite found a tendency for compact satellites to be aligned with the major axes of highly inclined disk galaxies ; however , their sample consisted of only 7 host galaxies .
although it is extremely tempting to accept its veracity based upon an intuitive sense that planar alignment of satellites is more dynamically sensible than polar alignment , it is clear that the brainerd @xcite result is badly in need of independent confirmation .
sales & lambas @xcite used a data set of very similar size ( and in one case identical selection criteria ) to that of brainerd @xcite yet did not detect any anisotropy in the satellite distribution when they analysed their entire sample .
why this is the case remains a mystery at the moment , but it may be attributable to a combination of two things .
first , the velocity errors in the 2dfgrs are typically larger than those in the sdss ( @xmath276 versus @xmath37 to @xmath38 ) . at some level
, this would lead to a higher fraction of interlopers ( i.e. , false satellites ) in the sales & lambas @xcite sample than in the brainerd @xcite samples .
second , van den bosch et al .
@xcite found that when they combined mock redshift surveys with the 2dfgrs , there was a clear absence of satellites at small projected radii in the 2dfgrs .
since the majority of the anisotropy seen by brainerd @xcite appears to come primarily from small scales , it could be that sales & lambas @xcite simply had too few pairs of hosts and satellites at small separations to detect the anisotropy .
any lack of host satellite pairs in the 2dfgrs data , however , does not explain why a holmberg effect was detected by sales & lambas @xcite when they restricted their analysis to host
satellite pairs with @xmath266 .
when brainerd @xcite imposed the same restriction on her sample 1 ( i.e. , the sample selected using the sales & lambas @xcite selection criteria ) , she found that the satellites with @xmath266 displayed an anisotropy that was identical to that of the full sample : a clear alignment of the satellites with the host major axes .
the cause of this discrepancy is not at all obvious
. it may in part be attributable to the fact that a value of @xmath277 is comparable to the error in a typical measurement of @xmath109 for hosts and satellites in the 2dfgrs . also , work by van den bosch et al .
@xcite suggests that the interloper fraction is substantially higher for host satellite pairs with low values of @xmath109 than it is for host satellite pairs with high values of @xmath109 .
it could , therefore , be possible that the sales & lambas @xcite sample with @xmath266 is heavily contaminated with interlopers and some strange , unknown selection bias is giving rise to their signal .
finally , it is worth notating that not only are the observational conclusions about the distribution of satellite galaxies particularly muddy at the moment , so too are the theoretical conclusions .
zaritsky et al .
@xcite compared their observed holmberg effect with high resolution cdm simulations and were unable to recover their observations .
pearrubia et al .
@xcite investigated both polar and planar orbits of satellites inside a massive , flattened dark matter halo and found that the planar orbits decayed more quickly that the polar orbits .
they therefore suggest that such differences in orbital decay rates could be the origin of the holmberg effect .
abadi et al .
@xcite suggest that the holmberg effect could be caused by the cumulative effects of accretion of satellites by the primary .
however , knebe et al .
@xcite found that the orbits of satellites of primary galaxies in cluster environments were located preferentially within a cone of opening angle 40@xmath278 ( i.e. , planar alignment , not polar ) .
since the structure of cold dark matter halos is essentially independent of the mass scale of the halo ( e.g. , @xcite , @xcite ) , the implication of this result would be a preference for the satellites of isolated galaxies to be aligned with the major axis of the host .
all of this in mind , perhaps the only answer to the question `` are either the sales & lambas @xcite or brainerd @xcite observations of anisotropic satellite distributions consistent with galaxy halos in a cdm universe ? ''
is , for now , `` maybe '' .
there has been a long period of time over which it has been perfectly acceptable to write papers on investigations into the nature of the dark matter halos of field galaxies that begin with a statement along the lines of `` although modern theories of galaxy formation posit that all large galaxies reside within massive halos of dark matter , the characteristic properties of those halos ( e.g. , mass , radial extent , and shape ) are not well constrained by the current observations '' .
that time is now coming to an end .
the wealth of data that has been acquired in recent years is truly beginning to place strong , direct constraints on the dark matter halos of field galaxies .
weak lensing and satellite dynamics have proven themselves to be excellent probes of the gravitational potentials of large , bright galaxies on physical scales @xmath279 kpc . while one might be skeptical and discount the results that come from one technique or the other , the fact that both are yielding consistent constraints can not be ignored . both weak lensing and satellite dynamics
lead to statistical constraints on the halo population as a whole , rather than constraints on any one particular galaxy halo , and it is especially the acquisition of extremely large data sets that has allowed these techniques to begin to fulfill their promise of mapping out the gravitational potentials associated with large , massive halos .
weak lensing and satellite dynamics have inherent advantages and disadvantages , but since their systematic errors and selection biases are completely uncorrelated , they are extremely complementary to each other .
at least at the moment , when strong constraints are only just beginning to emerge from each technique , this complementarity is very reassuring . based upon my own critical , and hopefully unbiased , reading of the recent literature ,
i think it is fair to say that , both individually and in combination , weak lensing and satellite dynamics are pointing toward the following scenario for the nature of large , bright field galaxies and their halos : * the dark matter halos are well - characterized by nfw
type objects in terms of their gravitational properties .
the dynamics of satellite galaxies strongly prefer nfw halos to isothermal halos . *
the virial masses that are inferred for large field galaxies are in good agreement with the predictions for galaxy mass halos in the context of cold dark matter .
specifically , the virial mass of the halo of an `` average '' @xmath139 galaxy is in the range @xmath280 when nfw profiles are fit to the data .
* there are clear differences in the depths of the potential wells of the halos that surround galaxies of differing morphology and differing intrinsic luminosity .
specifically , the virial masses of the halos of @xmath139 ellipticals exceed those of @xmath139 spirals by a factor of at least 2 .
the actual value of the mass excess depends upon details of the data and its analysis .
in addition , the virial masses of the halos of high luminosity galaxies exceed those of low luminosity galaxies .
again , however , the amount by which they differ depends upon details of the data and its analysis . *
averaged over all galaxies with @xmath0 , the mass to light ratio computed on scales larger than the optical radii of the galaxies is , at most , weakly
dependent upon the luminosity of the galaxy . at the 2@xmath20 level ,
the mass to light ratio of the average galaxy with @xmath0 is consistent with a constant value . * the dark matter halos
are flattened , rather than spherical , and the degree of flattening on large scales ( @xmath281 kpc to @xmath282 kpc ) is consistent with the predictions of cold dark matter .
it is worth noting that the above list comes from quite a diverse set of data . in particular ,
the data are spread over a wide range in redshift . with the exception of preliminary data from deep2 ,
the satellite dynamics studies have median redshifts of @xmath283 .
the weak lenses in the sdss data have a median redshift of @xmath284 and the weak lenses in the rcs and combo17 data have considerably higher redshifts , @xmath285 .
since it is clear that the field galaxy population has evolved since @xmath286 , it is not entirely fair to lump the results from all of these studies together , and i think the big challenge to the weak lensing community in particular will be to eventually place constraints on the evolution of field galaxies and their halos from , say , @xmath287 to the present .
nevertheless , i think we have reached a particularly gratifying time in which we are really being able to measure some of the fundamental properties of dark mater halos on physical scales that extend well beyond the visible images of the galaxies at their centers .
a remarkably consistent picture of the large scale gravitational properties of the halos is emerging from the observations and , at least for now , that picture seems entirely in accord with a cold dark matter universe .
i am deeply indebted to henk hoekstra , martina kleinheinrich , and erin sheldon for their help with the preparation of numerous figures at a time when they all had much more important things do to , and to tom peterson , without whose indulgence and encouragement this article would probably never have been written .
support under nsf contracts ast0098572 and ast0406844 is also gratefully acknowledged .
kleinheinrich , m. , schneider , p. , erben , t. , schirmer , m. , rix , h.w . & meisenheimer , k. 2004 , in _ gravitational lensing : a unique tool for cosmology _ , asp conf .
series , eds . d. valls gabaud & j.p .
kneib ( in press ) , http://arxiv.org e print archive , astro
ph/0304208 colless , m. m. , dalton , g. b. , maddox , s. j. , sutherland , w. j. , norberg , p. , cole , s. m. , bland
hawthorn , j. , bridges , t. j. , cannon , r. d. , collins , c. a. , couch , w. j. , cross , n. , deeley , k. , de propris , r. , driver , s. p. , efstathiou , g. , ellis , r. s. , frenk , c. s. , glazebrook , k. , jackson , c. a. , lahav , o. , lewis , i. j. , lumsden , s. , madgwick , d. s. , peacock , j. a. , peterson , b. a. , price , i. a. , seaborne , m. & taylor , k. 2001 , mnras , 328,1039 colless , m. , peterson , b. a. , jackson , c. , peacock , j. a. , cole , s. , norberg , p. , baldry , i. k. , baugh , c. m. , bland
hawthorn , j. , bridges , t. , cannon , r. , collins , c. , couch , w. , cross , n. , dalton , g. , de propris , r. , driver , s. p. , efstathiou , g. , ellis , r. s. , frenk , c. s. , glazebrook , k. , lahav , o. , lewis , i. , lumsden , s. , maddox , s. , madgwick , d. , sutherland , w. & taylor , k. 2003 , http://arxiv.org e print archive , astro
ph/0306581 norberg , p. , cole , s. , baugh , c. m. , frenk , c. s. , baldry , i. , bland
hawthorn , j. , bridges , t. , cannon , r. , colless , m. , collins , c. a. , couch , w. j. , dalton , g. b. , de propris , r. , driver , s. p. , efstathiou , g. , ellis , r. s. , glazebrook , k. , jackson , c. a. , lahav , o. , lewis , i. j. , lumsden , s. , maddox , s. j. , madgwick , s. j. , peacock , j. a. , peterson , b. a. , sutherland , w. j. & taylor , k. 2002 , mnras , 336 , 907 mckay , t. a. , sheldon , e. s. , racusin , j. , fischer , p. , seljak , u. , stebbins , a. , johnston , d. , frieman , j. a. , bahcall , n. , brinkmann , j. , csabai , i. , fukugita , m. , hennessy , g. s. , ivezic , z. , lamb , d. q. , loveday , j. , lupton , r. h. , munn , j. a. , nichol , r. c. , pier , j. r. & york , d. g. 2001 , http://arxiv.org e print archive , astro
ph/0108013 kleinheinrich , m. , rix , h.w . ,
schneider , p. , erben , t. , meisenheimer , k. , wolf , c. & schirmer , m. 2005 , in _ the impact of gravitational lensing on cosmology _ , proceedings of iau symposium 225 , eds.y .
mellier & g. meylan ( in press ) , http://arxiv.org e print archive , astro
ph/0409320 seljak , u. , makarov , a. , mandelbaum , r. , hirata , c. m. , padmanabhan , n. , mcdonald , p. , blanton , m. r. , tegmark , m. , bahcall , n. a. & brinkmann , j. 2004 , submitted to prd , http://arxiv.org e print archive , astro
ph/0406594 thomas , p. a. , colberg , j. m. , couchman , h. m. p. , efstathiou , g. p. , frenk , c. s. , jenkins , a. r. , nelson , a. h. , hutchings , r. m. , peacock , j. a. , pearce , f. r. & white , s. d. m. 1998 , mnras , 296 , 1061 madgwick , d. s. , lahav , o. , baldry , i. k. , baugh , c. m. , bland
hawthorn , j. , bridges , t. , cannon , r. , cole , s. , colless , m. , collins , c. a. , couch , w. j. , dalton , g. b. , de propris , r. , driver , s. p. , efstathiou , g. , ellis , r. s. , frenk , c. s. , glazebrook , k. , jackson , c. a. , lewis , i. j. , lumsden , s. , maddox , s. j. , norberg , p. , peacock , j. a. , peterson , b. a. , sutherland , w. j. & taylor , k. 2002 , mnras , 333 , 133
|
here i summarize constraints on the nature of the dark matter halos of field galaxies that have been obtained from the most recent investigations of ( i ) weak galaxy
galaxy lensing and ( ii ) the dynamics of satellite galaxies in orbit about large host galaxies .
both of these techniques are statistical in their nature ( i.e. , large samples of galaxies are required to obtain a `` signal '' ) , but since they have inherently different selection biases and systematic errors , they are quite complementary to each other . results of work over the last several years on weak lensing and satellite dynamics is revealing a remarkably consistent picture regarding the dark matter halos of bright field galaxies ( @xmath0 ) .
the halos extend to large physical radii ( @xmath1 kpc ) and are flattened in projection on the sky , there is a marked difference in the depths of the potential wells of early type galaxies and late type galaxies , and the velocity dispersion profiles of the halos , @xmath2 , decrease at large projected radii .
all of these are expected to hold true in a cold dark matter universe and , while neither technique can address the possible small scale ( @xmath3 kpc ) conflicts between cold dark matter and observed galaxies , on scales @xmath4 kpc both techniques yield results that are consistent with each other , and with the predictions of cold dark matter .
address = boston university , institute for astrophysical research , 725 commonwealth ave . ,
boston , ma 02215
|
since the discovery of high - temperature superconductors ( htscs ) , their engineering applications at liquid nitrogen temperatures have drawn much attention .
the yba@xmath1cu@xmath2o@xmath3 ( ybco ) is one of the promising htscs materials for various technical applications .
however , in ceramic materials the critical current density ( @xmath10 ) is , due to the weak link effects , very low , rendering them unsuitable for applications . in order to enhance @xmath10 , the so - called melt - textured growth ( mg ) process
@xcite has been developed , which can significantly enhance @xmath10 .
one interesting phenomenon in the mg htscs is the fishtail effect ( fe ) in @xmath10 as well as in isothermal magnetic hysteresis loops ( @xmath11 curves ) @xcite . as for the fe origin
, some researchers propose the vortex ordered phase to disordered phase transition @xcite to explain this interesting phenomenon , while others attribute it to the bulk pinning @xcite . in the vortex phase transition explanation ,
the fe exists both on a single @xmath12 curve and on a single @xmath13 curve .
however , in the bulk pinning explanation , the fe appears on @xmath11 curves whereas on a single magnetization vs. temperature @xmath14 curve it is seldom reported . in the previous work , the critical state model is used to study the isothermal @xmath11 curves @xcite .
however , for htscs , due to their operation at higher temperatures and due to their small activation energy @xmath15 , the flux creep is significant .
hence , non - linear flux creep models were developed .
meanwhile , the surface barrier @xcite , which prevents vortex entering ( @xmath16 ) and exiting ( @xmath17 ) superconductors , has strong effects on the irreversibility of htscs . in @xcite
pointed out that in the case of a competition between bulk and surface pinning , the initial relaxation is determined by the smaller pinning energy between @xmath18 and @xmath19 .
thus , for example , if @xmath20 then the bulk relaxation should be observed first . in @xcite
it was explained that the pe appears due to both bulk pinning and surface barriers therewith bulk pinning is more important at low temperatures while surface barriers at high temperatures .
the goal of this work was to contribute to the understanding of the fe origin in terms of significant influence of both bulk pinning and surface barriers .
undoped and ag - doped ybco bulk single - grain samples were fabricated by a top seeded melt - growth process ( tsmg ) . oxide powders
yba@xmath1cu@xmath2o@xmath21 + 0.25y@xmath1o@xmath2 + 0.5 wt.@xmath22 ceo@xmath1 + addition of ag@xmath1o in concentration determined by @xmath23 = 0.02 in yba@xmath1(cu@xmath4ag@xmath5)@xmath2o@xmath3 were milled in a friction mill and pressed into cylindrical pellets of 20 mm diameter . a smba@xmath1cu@xmath2o@xmath21 type seed with geometrical dimensions 2@xmath242@xmath241 mm@xmath25 was placed in the middle of the top surface of the pellet .
the samples were heated in the chamber furnace to a sintering temperature , @xmath26 = 950@xmath27c , sintered for 24 h and than heated to melting temperature , @xmath28 = 1040@xmath27c with dwell 9 h , cooled to 1000@xmath27c and than ybco crystal was grown during slow cooling to 950@xmath27c . as grown samples were cut in two halves along the \{100 } plane .
the cut surface was grinded on sic papers and fine polished with alumina powder .
the microstructure of the sample was analyzed by an optical microscope in normal and polarized light @xcite .
the pureness of ag@xmath1o is 0.005 wt@xmath22 .
\{110 } type of twins with twin spacing about 200 nm .
twin boundaries are parallel to the @xmath29-axis and formed almost 45@xmath27 angle with @xmath30-axis . small samples for oxygenation and magnetization measurements were cut from the @xmath30 - growth sector of the bulks @xcite at a distance of 1 mm from the seed ( top surface of the pellets ) .
the samples dimensions are shown in table [ tab:1 ] .
two samples , pure crystal ( s1 ) and ag - doped crystal ( s2 ) , were used in the present work .
physical properties of the samples are presented in table [ tab:1 ] .
.[tab:1]physical characteristics for samples yba@xmath1cu@xmath2o@xmath3 ( s1 ) and yba@xmath1(cu@xmath4ag@xmath5)@xmath2o@xmath3 ( s2 ) [ cols="^,^,^,^,^",options="header " , ] field dependences of critical current density @xmath10 ( a ) and derivative @xmath31 ( b ) at temperature @xmath3270 k for sample s1 right after annealing . points a and b denotes positions of minimum and maximum respectively.,width=307 ] the measurements of isothermal hysteresis loops ( @xmath11 curves ) at various temperatures ( 20 k@xmath3385 k ) were carried out in fields ( -6 t@xmath346 t ) parallel to the c - axis by means of a commercial vibrating sample magnetometer ( vsm ) .
the field sweep rate was @xmath35 t / min .
the relaxation data were performed at each temperature at zero magnetic field ( remanent magnetization ) during 300 sec with @xmath36 and @xmath37 . to obtain an additional parameter @xmath38
the resistivity @xmath39 curves were measured on two samples ( the first one is pure and the second one is ag - doped ) which were obtained from the same single - domain bulks as s1 and s2 .
transport measurements were performed using a physical property measurement system ( ppms ) .
the experiment was divided into two steps .
@xmath11 curves were measured : a ) right after annealing of samples in oxygen ( during 240 hours at 400@xmath27c ) ; b ) after long - term aging at the room temperature ( six months ) , in which case almost all relaxation processes were finished .
the first value for @xmath40 in table [ tab:1 ] corresponds to a ) and the second value corresponds to b ) .
the magnetic relaxation was measured for each curve at magnetic field @xmath41 during 300 sec .
from the measured @xmath11 curves the critical current density was calculated using the critical state model @xcite that have been developed in @xcite : @xmath42 where @xmath43 ( @xmath440 , @xmath450 are the values of magnetic moment at field @xmath46 , @xmath47 is the weight of sample , a , b and c are geometric dimensions of sample ( @xmath48 ) .
dependences @xmath13 ( fig .
[ fig:1]a for @xmath49=70 k ) were calculated for the field range 0 t@xmath346 t. to obtain the exact position of maximum on @xmath13 curves the derivative @xmath31 was calculated for each temperature ( see example at fig . [
fig:1]b ) .
[ fig:1 ] shows positions of both minimum ( point a ) and maximum ( point b ) on j vs h curve , where derivative @xmath500 .
contour for @xmath51 for sample s1 right after annealing .
points a and b coincide with points a and b in the fig .
[ fig:1],width=307 ] then the 3d surface @xmath52 was obtained .
intermediate curves were defined by the 3rd degree polynomial ( spline - function ) .
the contour for @xmath51 is shown in fig . [ fig:2 ] . points a and b correspond to those in fig .
[ fig:1 ] .
this method is useful to obtain position of maximum @xmath53 ( or ensemble of points b ) not only at given temperatures but also for intermediate temperatures .
m - t curves with zfc process in fields from 0.5 t to 6 t for sample s1 6 month after annealing , width=307 ] fig .
[ fig:3 ] shows the zero - field - cooled ( zfc ) @xmath54-curves at certain values of magnetic field ( from 0.5 t to 6 t ) .
as one can see magnetization monotonically changes with temperature and no apparent peak is visible in each zfc @xmath54 curve .
this is characteristic that the fe in @xmath55 curves can not be simultaneously observed on a single zfc @xmath54 curve , contrasting with the fe caused by order - disorder vortex matter transition .
hence , this can be used as a criterion to distinguish the fe originated from bulk pinning from the fe induced by order - disorder vortex matter transition @xcite .
measured magnetic relaxation provides a linear dependence in @xmath56 coordinates .
the results for magnetic relaxation are shown in fig .
[ fig:5 ] .
temperature dependencies of relaxation rate r = dm / dlnt for samples s1 ( green ) and s2 ( blue ) 6 months after annealing , width=307 ] the total magnetic moment @xmath57 . if surface activation energy @xmath19 larger then the bulk one @xmath58 the rate @xmath59 @xcite , i.e. @xmath60 . otherwise , if @xmath61 or @xmath62 then @xmath63 .
the crossover between these two regimes manifests in changes of slope @xmath56 . as one can see ( fig .
[ fig:4 ] ) relaxation rate @xmath64@xcite non - monotonically changes with temperature .
temperature range can be divided into 3 parts : 1 . at low temperatures ( [email protected] )
bulk pinning8 dominates over surface barriers , i.e. activation energy @xmath66 and total relaxation tare @xmath67 decreases .
position of minimum of relaxation rate @xmath68 for ag - doped crystal shifts to a higher temperatures ( @xmath690.4 for doping - free sample , @xmath690.45 for ag - doped sample ) as compared to doping - free sample .
the bulk pinning for ag - doped sample is stronger than for a doping - free one due to an additional pinning centers - ag atoms @xcite .
that is why relaxation in ag - doped sample due to @xmath70 dominate to a higher temperatures then in undoped sample .
2 . at intermediate temperatures ( [email protected] )
bulk pinning strongly decreased compared to surface barriers .
it leads to a sufficient decreasing of activation energy @xmath18 and , hence , relaxation rate @xmath72 dominates over @xmath70 . as a result @xmath67 increases
it is significant to note that bulk pinning reduces with increasing temperature and at critical temperature disappears whereas surface barriers still finite . at the point
@xmath73 the relaxation rate @xmath74 .
3 . at high temperatures ( @xmath750.8 ) @xmath70 dominates over @xmath72 because of @xmath760 in the vicinity of @xmath40 . from the model in previous chapter
the @xmath77 dependences of @xmath53 were obtained ( see fig . [ fig:5 ] ) .
temperature - field dependences of @xmath53-position for samples s1 ( green lines ) and s2 ( blue lines ) , right after annealing ( curves 1,2 ) and after long - term aging ( curves 3,4),width=307 ] as one can see from table [ tab:1 ] the value of the critical temperature @xmath40 for sample s1 changed from 90.2 k to 89.8 k and for sample s2 changed from 91.3 k to 91.1 k. this indicates the optimal oxygen annealing process and aging effects are not caused by changing in @xmath78 ( especially for sample s2 ) .
[ fig:5 ] shows that the position of @xmath53 significantly evolves after long - term aging .
the maximum field @xmath9 for sample s1 decreases from 4.25 t to 3.8 t and the minimum temperature @xmath79 increases from 0.42 to 0.54 ( curves 1 and 3 in fig
. [ fig:5 ] ) .
such changes can be induced by the redistribution of pinning centers in the crystal or , in another words , bulk pinning changes during aging .
constancy of the maximum temperature @xmath80 for s1 corresponds to the influence of surface barriers at high temperatures .
the aging effect for sample s2 is not so significant as for s1 .
it is related to the presence of considerable mechanical tensions in the sample ( large value of an additional parameter @xmath811.5 k ) that prevent the redistribution of pinning centers .
larger value of @xmath9 ( for s2 as compared with s1 ) corresponds to the presence of additional pinning centers on silver atoms .
a possible reason of the decrease of @xmath9 and increase of @xmath79 for s1 is the decreasing number of dislocations .
in another words , the crystal s1 becomes more homogeneous after aging @xcite .
in summary , mg ybco samples were investigated by measuring @xmath0-curves . on these curves fe was observed in a wide temperature range .
the origin of the fe provides in competition between surface barrier and bulk pinning at intermediate temperatures [email protected] .
it is confirmed in non - monotonically behavior of relaxation rate @xmath68 , notably , increasing of @xmath68 at intermediate temperatures due to surface barriers .
bulk pinning influences parameters @xmath9 and @xmath79 while surface barrier influences @xmath80 .
undoped ybco is characterized by a significant aging effect which is manifested in a considerable change of @xmath9 and @xmath79 .
this corresponds to the change of bulk pinning . for ag - doped ybco
there are no significant changes in parameters @xmath9 , @xmath79 and @xmath80 .
such difference is caused by the redistribution of pinning centers in undoped ybco while ag - doped ybco is characterized by huge mechanical tensions which prevent redistribution .
the work was partly supported by the slovak research and development agency ( no .
apvv-0006 - 07 ) and the slovak grant agency vega ( no.1/0159/09 ) the financial support of u.s.steel - dz energetika koice is acknowledged .
one of author ( d.l . ) is very thankful to the national scholarship program of slovak republic ( saia ) for the financial support during his stay in slovakia .
|
@xmath0-curves were experimentally investigated .
fishtail - effect ( fe ) was observed in mg yba@xmath1cu@xmath2o@xmath3 and yba@xmath1(cu@xmath4ag@xmath5)@xmath2o@xmath3 ( at @xmath6 0.02 ) crystals in a wide temperature range 0.4 @xmath7 0.8 at the orientation of magnetic field @xmath8 .
it was obtained that the influence of bulk pinning on fe is more effective at low temperatures while surface barriers dominates at high temperatures .
the value @xmath9 for ag - doped crystals is larger than for a pure one that due to the presence of additional pinning centers , above all on silver atoms .
|
much attention has been focused on novel fe - based superconductors since the recent discovery of superconductivity at the high temperature 26k in lafeaso@xmath2f@xmath3.@xcite many theoretical and experimental studies on fe - based superconductors have been reported for the last year .
it is important to identify the superconducting order parameter to elucidate the mechanism of superconductivity in those high-@xmath4 materials .
a @xmath5-wave pairing symmetry has been theoretically proposed as one of the candidates for the pairing symmetry in fe - pnictide superconductors.@xcite the @xmath5-wave symmetry means that the symmetry of pair potentials on each fermi surface is @xmath6-wave and the relative phase between them is @xmath7 .
recently , we showed that a fully - gapped anisotropic @xmath5-wave superconductivity consistently explained experimental observations such as nuclear magnetic relaxation rate and superfluid density .
@xcite a key point to identify the @xmath5-wave symmetry is a detection of the sign change in the order parameters between fermi surfaces .
it is difficult to detect the relative phase of the order parameters in a bulk material .
however , as shown in studies of high-@xmath4 cuprates , andreev bound states are formed at a surface or a junction when the quasiparticles feel different signs of the order parameter before and after scattering.@xcite since one can extract the information on the relative phase through andreev bound states , several theoretical studies on junctions and surfaces have been reported recently .
@xcite andreev bound states at zero energy have been experimentally observed as a zero bias conductance peak ( zbcp ) in tunneling spectroscopy for fe - based superconductors.@xcite the fe - based superconductors are interesting also as novel unconventional multi - band superconductors since multi - band effects are essentially important there.@xcite fermi surfaces in these systems predominantly consist of the @xmath8 orbitals of fe atom .
et al._@xcite suggest that five orbitals are necessary to describe the properties of the superconductivity , and they elaborate an effective 5-band model . on the other hand , mgb@xmath9 is a 2-band system that is a conventional @xmath6-wave bcs - like superconductor .
the aim of this paper is to develop a method for analyzing surface bound states in multi - band superconductors .
matsumoto and shiba@xcite developed a method to analyze surface bound states in single - band systems such as high-@xmath4 cuprates .
we extend their method into multi - band systems . since the ratio of the superconducting gap @xmath10 to the fermi energy @xmath11 is small , @xmath12 , in fe - based superconductors
, we can adopt a quasiclassical approach . in this approach ,
all we need is only to consider quasiparticles at the fermi level .
thus , we can reduce computational machine - time and the physical picture becomes clear .
in addition , this approach enables us to easily calculate the surface - angular dependence of tunneling spectroscopy .
we find a general appearance condition of the zbcp for multi - band systems .
this general condition can be applied to various pairing symmetries including @xmath5-wave and @xmath8-wave . with our method
, we will discuss a two - band superconductor as a simple example .
this paper is organized as follows .
the formulation of our quasiclassical approach is shown in sec .
we apply a quasiclassical approximation to eliminate fast spatial oscillations with fermi wave length .
the appearance condition of the zbcp in multi - band systems is derived in sec .
the results for a two - band model are shown as a simple example of our approach in sec .
iv , where we will show both analytical and numerical results .
the discussions and conclusion are given in secs .
v and vi , respectively . in the appendix
, we describe the derivation of the appearance condition of the zbcp when a system can be treated without quasiclassical approximation .
let us consider the local density of states near a surface following a procedure by matsumoto and shiba.@xcite we assume a two - dimensional superconductor , and consider a specular surface , for which the component of the quasiparticle momentum along the surface is conserved as shown in fig .
[ fig : spec ] .
we treat the surface as a potential @xmath13 , where the time - reversal symmetry is conserved.@xcite here , @xmath14 ( @xmath15 ) denote pauli matrices in nambu space and @xmath16 is the position in the real space .
we consider a @xmath0-orbital system , which is a periodic crystal with @xmath0 atomic orbitals in unit cell . throughout the paper , _ hat _
@xmath17 denotes a @xmath18 matrix in the orbital space , and _ check _ @xmath19 denotes a @xmath20 matrix composed of the @xmath21 nambu space and the @xmath18 orbital space .
we calculate the green function under the influence of @xmath13 .
it is written as @xmath22 here , @xmath23 is an unperturbed green function in the absence of @xmath24 .
we take the @xmath25(@xmath26)-axis perpendicular ( parallel ) to the surface as shown in fig .
[ fig : spec ] .
considering the surface situated at @xmath27 and the scattering potential @xmath24 written as @xmath28 , eq .
( [ eq : green ] ) is reduced to @xmath29 where @xmath30 here , we have taken the fourier transformation with respect to @xmath26 .
we use units in which @xmath31 , and the coordinates @xmath16 and the momentum @xmath32 are dimensionless . the surface is actually represented in the limit @xmath33 . the green function
is then given by @xmath34 where @xmath35 the local density of states at the position @xmath25 for the momentum @xmath36 is written as @xmath37 , \label{eq : real - part - g}\ ] ] where @xmath38 here @xmath39 is the fermion matsubara frequency and @xmath40 is a positive infinitesimal quantity .
the unperturbed green function @xmath41 is given by @xmath42 where @xmath43 here , @xmath44 is the @xmath20 hamiltonian in nambu and orbital spaces written as @xmath45 in the `` orbital representation '' where the base functions are atomic orbitals in crystal unit cell . from now on
, the subscript `` @xmath46 '' indicates that matrices are represented with the orbital basis .
@xmath47 is the hamiltonian in the normal state represented as @xmath18 matrix in the orbital space .
remember that @xmath0 is the number of the orbitals .
@xmath48 is the superconducting order parameter .
let us introduce a @xmath18 hamiltonian in the `` band representation '' defined by @xmath49 here , @xmath50
( @xmath51 ) denote the eigenvalues where the relation @xmath52 is satisfied .
@xmath53 is a unitary matrix consist of the eigenvectors that diagonalizes the hamiltonian @xmath47 .
the @xmath20 hamiltonian in nambu and orbital spaces in the `` band representation '' is also defined by @xmath54 where @xmath55 in general , @xmath56 contains off - diagonal elements , which correspond to inter - band pairings . assuming that intra - band pairings are dominant , we neglect the off - diagonal ( inter - band ) elements in @xmath56 : @xmath57 that is , we consider that only single pair - potential is defined on each fermi surface . here
, @xmath58 is the pair - potential on the @xmath59-th band . substituting eq .
( [ eq : bhami ] ) into eq .
( [ eq : greeno ] ) , the green function @xmath60 is written as @xmath61 assuming eq .
( [ eq : dede ] ) and taking the inverse matrix of @xmath62 , one can obtain @xmath63 where @xmath64 we find that eq .
( [ eq : ugu ] ) can be rewritten as @xmath65 where @xmath59 is the band index and @xmath66}_{jk } & = & { [ } \hat{p } { ] } _ { ji } { [ } \hat{p } { ] } _ { ki}^{\ast } .
\end{aligned}\ ] ] equation ( [ eq:22 ] ) is divided into a sum of the green functions defined on each band .
substituting eq .
( [ eq:22 ] ) into eq .
( [ eq:8 ] ) , @xmath41 is expressed as @xmath67 hence , the @xmath68-integration is found to be performed on each band independently .
we assume @xmath69 .
this relation is satisfied in most of systems such as conventional superconductors and fe - based ones . in this case
, one can use a quasiclassical approach .
we consider a line with a fixed @xmath36 in the momentum space . on this line ,
we classify @xmath0-bands into two groups .
one group is composed of the bands on which the eigen energy @xmath70 crosses the fermi level ( for example , the bands @xmath71 and 2 in fig . [
fig : segments ] ) . the other group is composed of the bands on which the eigen energy does not cross the fermi level ( the band @xmath72 ) . for the former group
, we can analytically integrate @xmath73 over @xmath68 with the use of a quasiclassical approach since @xmath73 is a function localized near the fermi level . for the latter group ,
we need to integrate @xmath73 over @xmath68 numerically since @xmath73 is not a localized function . however , the integrand is a smooth function , so that it is easy to perform such a numerical integration .
we integrate @xmath73 on the bands of the first group with the use of the quasiclassical approach . to perform the @xmath68-integration
, we divide the @xmath68-line with a fixed @xmath36 into some segments as shown in fig .
[ fig : segments ] .
each segment has only single channel that is the point satisfying the relation @xmath74 . from now on
, @xmath75 denotes the channel index and @xmath76 denotes the maximum number of @xmath75 .
the integration for the @xmath59-th band is written as @xmath77 expanding @xmath78 in the first order of @xmath50 around @xmath74 as @xmath79 , one can carry out the integration by the residue theorem : @xmath80 where @xmath81 here , @xmath82 and @xmath83 are the fermi wave number and the fermi velocity on the @xmath75-th channel , respectively . using the above , eq .
( [ eq : g0wa ] ) can be written as @xmath84 where the elements in @xmath85 are the indices of the bands whose energy dispersions cross the fermi level for a fixed @xmath86 .
here , we assume @xmath87 , namely the superconducting order parameters are finite only around the fermi level .
it should be noted that the second term in the right - hand side of eq .
( [ eq : g0quasi ] ) can not be neglected since @xmath88 without this second term may have artificial divergences .
we assume the condition @xmath89 ( i.e. , @xmath69 ) , which is the quasiclassical condition . here
, @xmath90 is the coherence length of a superconductor . under this condition
, the short range spatial oscillations characterized by the fermi wave length @xmath91 can be eliminated .
we rewrite eq .
( [ eq : g0quasi ] ) as @xmath92 where @xmath93 the perturbed green function @xmath94 defined in eq .
( [ eq : pg ] ) can be written as @xmath95 setting @xmath96 \rightarrow 1 $ ] , we eliminate the short range oscillation while keeping the enveloping profile of the integrand . thus , the above equation is reduced to @xmath97 from this equation , it is concluded that the andreev bound states appear when @xmath98 diverges , i.e. , when @xmath99 .
let us consider the appearance condition of the zbcp in @xmath0-band system at a surface .
at the zero energy @xmath100 , @xmath101 defined in eq .
( [ eq : gf ] ) [ for @xmath102 is written as @xmath103 for @xmath104 , we have from eq .
( [ eq:23 ] ) with @xmath100 , @xmath105 where we have set @xmath106 because the superconducting order parameter is assumed to be finite only near the fermi level and the bands with the indices @xmath104 do not cross it . substituting the above equations into eq .
( [ eq : g0quasi ] ) , we can obtain the appearance condition of the zbcp from @xmath99 : @xmath107 where @xmath108 equation ( [ eq : l ] ) shows that the appearance condition _ does not _
depend on the anisotropy of the pair - potentials and it depends only on the signs of them because information on the pair potentials is included in the form , @xmath109 , in eq .
( [ eq : l ] ) .
this result shows that information on the normal state ( i.e. , the matrices @xmath110 , @xmath111 ) is important for the zbcp to appear .
we calculate the density of states in a two - band superconductor as a simple example .
we consider a two - band tight - binding model on a square lattice .
there are two orbitals on each lattice site .
the hamiltonian with a @xmath112 matrix form in the normal state is described as @xmath113 in the orbital representation ( @xmath114 ) . here ,
@xmath115 and @xmath116 are the axes fixed to the crystal axes in the momentum space , @xmath117 and @xmath118 are intra- and inter - orbital hopping amplitudes , respectively , and @xmath119 denotes the chemical potential .
we use the unit in which the lattice constant @xmath120 .
this hamiltonian can be diagonalized into the matrix in the band representation , @xmath121 , written as @xmath122 here , @xmath123 denotes the energy dispersion on the @xmath124-band . as shown in fig .
[ fig : fer ] , the fermi surfaces consist of two parts near the half filling . [ cols= "
< , < " , ]
the advantages of our method are that one can easily investigate the surface - angle dependence of the density of states with the use of the quasiclassical method and easily calculate the density of states in the @xmath0-band ( multi - band ) system with less computational machine - time .
therefore , we can take , for example , a realistic 5 band model in order to discuss the density of states for iron - based superconductors . we will report its results elsewhere near future .
we have assumed that the matrix of the pair potential in the band - representation does not have off - diagonal elements , which correspond to the inter - band pairings . when the inter - band pairing is dominant , the cooper pairs have center - of - mass momentum @xmath125 . usually such pairs are not energetically favorable since the pair potentials have spacial dependence even in bulk systems . starting with the same matsumoto - shiba method,@xcite onari _
et al._@xcite recently calculated the surface andreev bound states without the quasiclassical approximation .
their results show that the peak positions of the andreev bound states depend on the gap amplitudes on two bands in the same two - band model as considered in sec .
iv , and the zbcp does not always appear at the half - filling
. these results might seemingly be inconsistent with our quasiclassical results .
it is , however , not the case .
they obtained the perturbed green function by directly integrating the original unperturbed green function over @xmath68 and @xmath36 numerically .
the original unperturbed green function has sharp peaks on fermi surfaces in the momentum space and rapid fermi - wave - length oscillations in the real space .
we have integrated out those properties by the quasiclassical approximation .
it should be noted that the pair potentials are of the order @xmath126 in ref . .
this parameter is out of our quasiclassical approach ( @xmath127 ) . as shown in sec .
iv.b.3 , our analytical result , which depends on the details of the gap amplitudes and therefore is consistent with ref .
, is reduced to the quasiclassical result in the limit @xmath127 .
thus , the differences in the obtained results between onari _ et al._@xcite and the present paper would be due to the difference in applicable parameter regions .
the formulation derived in secs .
ii and iii can be applied to general multi - band superconductors including @xmath8-wave pairing superconductor .
the appearance condition for the zbcp is given as eq .
( [ eq : app ] ) in sec .
let us consider , for instance , the case of the two - band model discussed in sec .
iv at the half filling for [ 110 ] surface . from eq .
( [ eq : app ] ) , the appearance condition for the zbcp is given as @xmath128 here , @xmath129 and @xmath130 are the pair potentials on the inner fermi surface ( red ) in fig . [
fig : yokogiru ] , and @xmath131 and @xmath132 are the pair potentials on the outer fermi surface ( green ) there . for a two - band @xmath8-wave superconductor , @xmath133 and @xmath134 , so that the above condition is satisfied and the zbcp appears .
furthermore , in the case of a single - band @xmath8-wave superconductor , the appearance condition for the zbcp is obtained from eq .
( [ eq : app ] ) as @xmath135 this is consistent with previous results for @xmath8-wave pairing in refs . , where the zbcp appears when the quasiparticles feel a superconducting phase change @xmath7 in the surface scattering process a1@xmath136a2 on a fermi surface .
in conclusion , we extended the single - band method by matsumoto and shiba@xcite into general @xmath0-band case ( @xmath1 ) . with the use of the quasiclassical approximation
, we developed the way to integrate the unperturbed green function with respect to @xmath68 which is the momentum component perpendicular to a surface .
we showed that the appearance condition of the zbcp does not depend on any anisotropy in the pair - potential amplitude , but only on the relative phase , in the case of @xmath137 in @xmath0-band systems .
the properties of the normal state are influential for the zbcp to appear .
we also calculated the surface density of states in the two - band system as a simple example of our approach .
we suggested that our quasiclassical approach is appropriate when @xmath138 .
we showed that the peaks of the density of states due to the andreev bound states depend on the surface angle and the parameters in the normal state ( @xmath117,@xmath118 , @xmath119 ) , so that the sign - reversing @xmath6-wave ( @xmath5-wave ) superconductors exhibit complicate properties in the tunneling spectroscopy compared with single - band @xmath8-wave superconductors .
we thank y. kato , m. machida , n. nakai , h. nakamura , m. okumura , c. iniotakis , m. sigrist , y. tanaka and s. onari for helpful discussions and comments .
y.n . acknowledges support by grand - in - aid for jsps fellows ( 204840 ) .
is supported by jsps core - to - core program - strategic research networks , ` nanoscience and engineering in superconductivity ( nes ) ' .
we will show the zbcp appearance condition eqs .
( [ eq : dab ] ) and ( [ eq : iab ] ) for zero inter - orbital hopping amplitude ( @xmath139 ) , by integrating eq .
( [ eq : g0wa ] ) in the simple two - band model for [ 110 ] surface at the half filling.@xcite the hamiltonian in the normal state is described as @xmath140 with @xmath141 here , we have introduced @xmath142 and @xmath143 . considering the pair potentials @xmath144 which do not depend on @xmath32 and using the unitary matrix eq .
( [ eq : uni ] ) , the pair potential matrix in the orbital representation can be written as @xmath145 the unperturbed retarded green function @xmath146 is written as @xmath147 where @xmath148 to investigate the appearance condition of the zbcp , we set @xmath149 ( i.e. , zero energy ) , @xmath27 and @xmath150 ( i.e. , at the surface ) .
then , we calculate @xmath151 : @xmath152 each element in this matrix can be integrated analytically as @xmath153 integrating @xmath154 by using the above formulae , we finally obtain @xmath151 : @xmath155 where @xmath156 and @xmath157 are defined in eqs .
( [ eq : qui ] ) and ( [ eq : qud ] ) .
its inverse matrix is written as @xmath158^{-1 } & \propto & t \left(\begin{array}{cccc } \frac{-i_{1}}{(\delta_{ab})^2 + i_{1}^2 } & 0 & \frac { \delta_{ab}}{(\delta_{ab})^2 + i_{1}^2 } & 0 \\ 0 & \frac{-i_{2}}{(\delta_{ab})^{2 } + i_{2}^{2 } } & 0 & \frac{\delta_{ab}}{(\delta_{ab})^2 + i_{2}^2 }
\\ \frac { \delta_{ab}}{(\delta_{ab})^2 + i_{2}^2 } & 0 & \frac{i_{1}}{(\delta_{ab})^2 + i_{1}^2 } & 0 \\ 0 & \frac{\delta_{ab}}{(\delta_{ab})^2 + i_{2}^2 } & 0 & \frac{i_{2}}{(\delta_{ab})^{2 } + i_{1}^{2 } } \end{array}\right ) .
\nonumber \\\end{aligned}\ ] ] the zero energy bound states appear when @xmath159^{-1}$ ] diverges as noticed from eqs .
( [ eq:4 ] ) and ( [ eq : pg ] ) .
therefore , the appearance condition of the zbcp is expressed as @xmath160 y. kamihara , t. watanabe , m. hirano , and h. hosono , j. am .
* 130 * , 3296 ( 2008 ) . i. i. mazin , d. j. singh , m. d. johannes , and m. h. du , phys . rev
* 101 * , 057003 ( 2008 ) .
k. kuroki , s. onari , r. arita , h. usui , y. tanaka , h. kontani , and h. aoki , phys .
* 101 * , 087004 ( 2008 ) .
m. m. korshunov and i. eremin , phys .
b * 78 * , 140509(r ) ( 2008 ) .
k. seo , b. a. bernevig , and j. hu , phys .
* 101 * , 206404 ( 2008 ) .
t. nomura , j. phys .
. jpn . * 77 * , suppl .
c 123 ( 2008 ) . y. bang and h .- y .
choi , phys .
b * 78 * , 134523 ( 2008 )
. m. m. parish , j. hu , and b. a. bernevig , phys .
b * 78 * , 144514 ( 2008 ) .
r. arita , s. onari , h. usui , k. kuroki , y. tanaka , h. kontani , and h. aoki , proc . of the 25th international conference on low temperature physics ( lt2146 ) , to be published in j. phys .
v. stanev , j. kang , and z. tesanovic , phys .
b * 78 * , 184509 ( 2008 ) . y. senga and h. kontani , j. phys
. soc . jpn . *
77 * , 113710 ( 2008 ) .
y. nagai , n. hayashi , n. nakai , h. nakamura , m. okumura , and m. machida , new j. phys . * 10 * , 103026 ( 2008 ) .
|
we discuss the tunneling spectroscopy at a surface in multi - band systems such as fe - based superconductors with the use of the quasiclassical approach .
we extend the single - band method by matsumoto and shiba [ j. phys .
soc .
jpn . * 64 * , 1703 ( 1995 ) ] into @xmath0-band systems ( @xmath1 ) .
we show that the appearance condition of the zero - bias conductance peak does not depend on details of the pair - potential anisotropy , but it depends on details of the normal state properties in the case of fully - gapped superconductors .
the surface density of states in a two - band superconductor is presented as a simplest application .
the quasiclassical approach enables us to calculate readily the surface - angular dependence of the tunneling spectroscopy .
|
cellular automaton ( ca ) can be used as discrete mathematical models for a variety of phenomena@xcite .
one remarkable feature for ca being used as a model lies in the fact that the setup is mathematically simple . since both dependent and independent variables are discrete and a set of values which the dependent variable takes is finite , a rule table is more suitable for representing logical expression in a ca model .
thus , ca is free from mathematical conditions such as continuity and differentiability . due to the complete discreteness of ca ,
mathematical tools for analyzing its solutions of ca are very different from those for analyzing solutions of other mathematical objects such as differential equation , finite - difference equation and coupled map lattice .
pattern analysis , graph theory and statistical mathematics are often used to analyze ca @xcite . however , in seeking for a general solution starting from arbitrary initial data of a ca , this difference often becomes an obstacle .
compact expression of solution using special functions is rarely available for ca due to its discreteness .
one of the bridges between discrete and continuous mathematics to solve this problem was found in the area of integrable system . by applying the following simple formula for the transformation of variables ,
@xmath0 we can obtain ca from a finite - difference equation with continuous variables .
this procedure is called ` ultradiscretization'@xcite .
solution of the finite - difference equation can be ultradiscretized by the same procedure .
for example , we obtain a binary ca called ` box and ball system ' from the discrete lotka
volterra equation through the above transformation@xcite .
moreover , the elementary ca of rule number 184 ( eca184 ) is obtained from the discrete burgers equation@xcite .
eca184 is considered in the subsequent section .
note that we can obtain the differential equation and its corresponding solution by taking a continuum limit for a finite - difference equation and its solution .
the continuous analogue to the discrete lotka
volterra equation is the korteweg
de vries equation and that to the discrete burgers equation is the burgers equation .
therefore three different discrete levels of equations are directly related each other through two types of limiting procedure .
replacing basic operations is another way to obtain an ultradiscrete equation from a finite - difference one .
equation ( [ max - plus ] ) shows that the addition of the finite - difference equation corresponds to the @xmath1 operation of the ultradiscrete one .
the multiplication and division correspond to @xmath2 and @xmath3 respectively since we have @xmath4 though the subtraction is not well - defined for the ultradiscrete equation , we can automatically obtain the ultradiscrete equation and its solution through a procedure of replacing @xmath2 , @xmath5 and @xmath6 by @xmath1 , @xmath2 and @xmath3 respectively if the subtraction is not included explicitly in the finite - difference equation and its solution .
the algebra defined by operations of @xmath1 , @xmath2 and @xmath3 is called ` max - plus algebra'@xcite .
binary operations such as and , or and not can be easily converted to max - plus operations . since ca can be defined by binary operations , a rule table of ca
is usually transformed into a max - plus equation .
in contrast to a binary solution to the original ca , the solution to a max - plus equation is real - valued . in this sense , both the equation and the solution are extended to the real value range when we use max - plus algebra .
we should point out that it is important but difficult to express a rule table of ca by an appropriate max - plus equation in order to grasp the behavior of solution analytically . in the present paper
, we propose max - plus expression for rule table , general solution and asymptotic behavior of a class of evolutional binary ca .
the sum of values of dependent variables at all spatial sites is conserved through time evolution in this class .
eca184 is included in this class .
such ca can be used as an important digital model for transportation or flow phenomena . in regard to the asymptotic behavior of solution for this type of ca , a phase transition can often be observed @xcite . in order to clarify the asymptotic behavior mathematically , finding
a general solution for this type of ca is necessary .
it is found that the max - plus expression is more suitable in finding such a general solution .
the present paper is organized as follows . in section
[ sec : classification ] , we define the class of ca as mentioned above and show examples of solution and observed asymptotic properties . in section
[ sec : max - plus analysis ] , by giving the max - plus expressions of evolution rules , we present general solutions of the ca and , further , analyze the asymptotic behavior based on general solutions . in section [ sec :
remarks ] , concluding remarks are given . in [ sec : formulas ] , some convenient formulas about max and min operations are listed . in [ sec : proof solution ] , we prove the solution to a specific ca . in [ sec : proof asymptotic ] , we prove the asymptotic behavior of the solution .
let us consider the following evolution equation , @xmath7 where @xmath8 and @xmath9 are integers , representing the time step and the space site number , respectively . @xmath10 and @xmath11 are also integer constants .
assume that the value of @xmath12 takes either 0 or 1 , thus the above equation ( [ gen eq ] ) gives an evolution rule of binary ca .
the binary function @xmath13 has @xmath14 ( @xmath15 ) arguments and we call this ca ` @xmath14-neighbors ca ' . using a transformation of coordinates defined by @xmath16 , the above equation ( [ gen eq ] ) can be rewritten as @xmath17 thus , the solutions to these two equations ( [ gen eq ] ) and ( [ trans1 ] ) can be transformed from one to the other .
therefore , we can identify equation ( [ trans1 ] ) for any @xmath18 with ( [ gen eq ] ) . applying the transformation @xmath19 to ( [ gen eq ] )
, we obtain @xmath20 and can identify ( [ trans2 ] ) with ( [ gen eq ] ) by the same reasoning .
moreover , if we ` flip ' the dependent variable @xmath12 , that is , if we apply the transformation @xmath21 , it follows that @xmath22 although the function @xmath23 is generally different from @xmath13 , we can still identify ( [ trans3 ] ) with ( [ gen eq ] )
. moreover , there are cases where @xmath13 of ( [ gen eq ] ) does not depend on all of its arguments , for example , in the following equation @xmath24 the number of formal arguments of @xmath13 is three ( 3-neighbors ) but @xmath13 does not depend on the last argument .
therefore , this equation is reduced to @xmath25 , which is of 2-neighbors ca .
let us consider another equation , @xmath26 in which , the values of @xmath12 s at even ( odd ) sites at time @xmath27 depend only on those at even ( odd ) sites at time @xmath8 .
thus , this equation is reduced to @xmath25 , which becomes 2-neighbors ca by the separation of even and odd sites . in the present paper
, we exclude the equations which are reductive to those of less than @xmath14-neighbors from a class of @xmath14-neighbors .
let us consider the following condition for any @xmath8 , @xmath28 which means the sum of @xmath12 , that is , the number of 1 s at all space sites is conserved for arbitrary @xmath8 .
let @xmath29 denote the number of particle at @xmath9-th site and @xmath8-th time step , and each ` 1 ' in the solution represent a particle .
therefore , the particles move among the sites according to the evolution rule defined by ( [ gen eq ] ) without creation or annihilation .
we call the ca satisfying the condition ( [ cons ] ) ` particle ca ' ( pca ) in this sense .
classification and statistical analysis of pca was reported by fuk and boccara@xcite .
if a ca defined by ( [ gen eq ] ) is pca under appropriate boundary conditions , it can always be written in a conservation form , @xmath30 where @xmath31 is a ` flux ' for the ` density ' @xmath12@xcite .
the condition @xmath32 is necessary for a particle ca because a system with zero particle will produce zero flux .
this condition can be used to determine an indefinite constant in the expression of @xmath33 . in what follows ,
we list up all pca of 1-neighbor to 4-neighbors identified in subsection [ identification ] .
label ` @xmath8 ' or ` @xmath8-@xmath34 ' is assigned to each pca where @xmath8 denotes the number of neighbors , and @xmath34 denotes the subclass of the same neighbors ca if it exists .
rules of pca are defined in the form of ( [ flux ] ) together with a rule table of @xmath33 .
the upper row of the table shows all possible combinations of binary arguments and the lower row gives the values of @xmath33 .
* pca1 : @xmath35 , @xmath36 * pca2 : none * pca3 : @xmath37 @xmath38 * pca4 : @xmath39 @xmath40 for simplicity , let @xmath41 represent the initial time step of the evolution . for pca1 , we have always a trivial static solution @xmath42 .
pca3 is equivalent to eca184 @xcite , which is related to a simple traffic congestion model @xcite . in what follows
, we will discuss the dynamics of particles and the asymptotic behavior of solutions for pca3 and 4 .
although the max - plus analysis of pca3 ( eca184 ) was already given in @xcite , we here give a briefly review as a typical scenario of max - plus analysis . from the rule table of @xmath33
, we can determine the dynamics of a particle system .
in what follows , we show the motion representations of pca3 and pca4 .
note that ` 1 ' denotes a site occupied by a particle and ` 0 ' denotes an empty site .
the motion representation for each pca gives the configurations in which only neighborhoods resulting in a particle motion assuming that by default , in all other cases , particles do not move @xcite . the arrow indicates the initial and final positions for a moving particle in one time step .
@xmath43{pca3dyn.eps } \\ \mbox{pca4 - 1:\quad } \includegraphics[scale=1.2]{pca4 - 1dyn1.eps}\quad \includegraphics[scale=1.2]{pca4 - 1dyn2.eps } & \qquad \mbox{pca4 - 2:\quad } \includegraphics[scale=1.2]{pca4 - 2dyn1.eps}\quad \includegraphics[scale=1.2]{pca4 - 2dyn2.eps } \\
\mbox{pca4 - 3:\quad } \includegraphics[scale=1.2]{pca4 - 3dyn.eps}&\qquad \mbox{pca4 - 4:\quad } \includegraphics[scale=1.2]{pca4 - 4dyn.eps } \end{array}\ ] ] in this subsection , we discuss the asymptotic behavior of solution from arbitrary initial data as @xmath44 . the behavior is mainly dependent on the density of particles , or the average number of particles per site . to observe this dependence
, a periodic boundary condition for space sites is assumed .
let @xmath45 be the number of sites in one period , the density @xmath46 is defined by @xmath47 since the number of particles is conserved , @xmath46 is also a constant irrespective of time @xmath8 . for most of pca , the asymptotic behavior changes at certain critical values of @xmath46 .
the relation between the density and the average flux per site reflects this critical phenomenon well .
the average flux @xmath48 per site is defined by @xmath49 for large enough @xmath8 , the evolution of a pca often approaches a steady state , in which @xmath48 becomes a constant . in the case of convergence ,
the constant is defined by @xmath50 the value of @xmath51 is independent of the initial distribution of particles and depends solely on the density @xmath46 for pca3 , 4 - 1 , 4 - 2 and 4 - 3 , whereas for pca4 - 4 , the value of @xmath51 depends on both the density @xmath46 and the initial condition .
the graph of @xmath51 vs. @xmath46 is called ` fundamental diagram ' in the area of traffic flow analysis .
we use this terminology throughout the present paper and abbreviate it as fd .
the fd s for pca3 and pca4 , as well as examples of evolutions are shown in figures [ fig : pca3]-[fig : pca4 - 4 ] respectively .
c + [ cols="^ " , ] + the relations between the solutions of steady states at large enough @xmath8 and the critical values of @xmath46 are listed below .
@xmath52 @xmath53 @xmath54 @xmath55 @xmath56
the flux function of pca3 was found to be @xmath57 if @xmath58 and @xmath59 are binary . thus , the evolution rule can be rewritten in the following max - plus form , @xmath60 using a transformation of variable from @xmath12 to @xmath13 , @xmath61 we obtain the evolution rule for @xmath13 @xmath62 note that , for simplicity , an arbitrary constant occurring in the process of deriving ( [ potential pca3 ] ) is set to 0 . the initial value problem of ( [ potential pca3 ] ) can be easily solved as @xmath63 the inverse relation of ( [ trans pca3 ] ) is @xmath64 with an arbitrary constant @xmath18 , from which we have @xmath65 therefore , it is obvious that @xmath66 for @xmath67 and @xmath68 for @xmath69 . using above inequalities , we have @xmath70 for large enough @xmath8 .
furthermore , we conclude that @xmath71 for large enough @xmath8 .
these analytical relations explain the steady solution of pca3 shown in figure [ fig : pca3 ] . finally ,
substituting the relation @xmath72 under the case of @xmath73 into the evolution equation ( [ eq pca3 ] ) , we arrive at @xmath74 for any @xmath9 . in other words
, @xmath75 is a constant irrespective of @xmath9 . since @xmath73 , there exists at least one site @xmath9 satisfying @xmath76 or @xmath77 , which resulting in a zero value for the above constant .
thus , we have @xmath78 in the case of @xmath73 .
similarly we obtain @xmath79 in the case of @xmath80 .
this relation gives fd of pca3 as shown in figure [ fig : pca3 ] .
this pca is known as the fukui ishibashi model used for traffic congestion analysis@xcite and its euler and lagrange representation
are studied using max - plus expression@xcite .
asymptotic behavior for the solution of pca4 - 1 is similar to that of pca3 except a few slight differences : 1 ) critical value of @xmath46 is @xmath81 , 2 ) for the smaller region of @xmath46 ( @xmath82 ) , all particles move with a speed of 2 at large enough @xmath8 .
the evolution equation of pca4 - 1 can be described in the following max - plus form , @xmath83 introducing a transformation @xmath84 we obtain @xmath85 by choosing an appropriate value for the arbitrary constant .
consequently , the general solution for the initial value problem turns out to be @xmath86 on the other hand , the inverse relation of ( [ trans pca4 - 1 ] ) gives @xmath87 which leads to @xmath88 by similar calculations for pca3 , at large enough @xmath8 , we can derive steady state solutions : @xmath89 and further the following relations @xmath90 from above relations , it can be easily shown that @xmath91 in summary , the above analytical results agree with the solutions and fd shown in figure [ fig : pca4 - 1 ] .
it can be shown that the evolution equation of pca4 - 2 is @xmath92 through a transformation , @xmath93 it follows , from ( [ eq pca4 - 2 ] ) , that the evolution equation to @xmath13 is @xmath94 the general solution to equation ( [ potential pca4 - 2 ] ) turns out to be @xmath95 by expanding rhs of ( [ sol pca4 - 2 ] ) , specific examples of @xmath96 for @xmath97 , 2 , 3 are given as @xmath98 the proof for the general solution ( [ sol pca4 - 2 ] ) satisfying ( [ potential pca4 - 2 ] ) is given in [ sec : proof solution ] . moreover , at large enough @xmath8 , the asymptotic behavior of the solution , as well as the value of @xmath51 , can be shown to be : @xmath99 the proof of the above results is given in [ sec : proof asymptotic ] . in summary , the analytical results reveal the steady state solutions and fd as shown in figure [ fig : pca4 - 2 ] .
the evolution equation of pca4 - 3 can be expressed by @xmath100 by using a transformation , @xmath101 we arrive at an evolution equation for @xmath13 , @xmath102 from which , the general solution turns out to be @xmath103 by means of a similar proof as pca4 - 2 .
therefore , it is omitted here .
the specific expressions of @xmath96 for @xmath97 , 2 , 3 are as follows : @xmath104 at large enough @xmath8 , we can derive the asymptotic behavior of solution and , further , the value of @xmath51 , which both depend on the density @xmath46 as follows @xmath105 again , the above results reveal the solutions and fd as shown in figure [ fig : pca4 - 3 ] . since the proof is tedious and similar to that of pca4 - 2 , it is omitted here .
we are unable to find a general solution to this type of pca in the max - plus form .
however , it is worth pointing out the following observations from the binary rule table in the case of @xmath73 : if there exists a site @xmath9 satisfying @xmath106 in the initial data , the asymptotic solution will become static , that is , @xmath35 at large enough @xmath8 .
if there is no two consecutive 1 s in the initial data , the solution @xmath29 satisfies @xmath107 for any @xmath9 and @xmath8 .
it then follows that @xmath108 in the former case and @xmath109 in the latter case .
note that two different types of initial data are recognized , which correspond to two branches in fd as shown in figure [ fig : pca4 - 4 ] in the case of @xmath110 .
we propose the max - plus expressions to equations and solutions of pca4 - 1 , 4 - 2 and 4 - 3 .
we analyze the asymptotic behaviors of solutions and prove the fundamental diagrams mathematically . the max and min formulas
are effectively utilized in the proofs . in view of the role of basic operations appearing in max - plus algebra ,
the way of proofs is similar to that for the finite - difference or differential equation . in this sense ,
the mathematical technique used is analytic .
it is important to introduce such an analytic technique into the digital systems since it provides us with an alternative approach to study them , and sometimes can lead to rigorous results .
we restrict ourselves to the condition that the value of @xmath12 is binary in the proof of solutions for pca4 - 2 and 4 - 3 .
hence , the max - plus expressions for general solutions proposed only hold under this condition . on the other hand , solutions for pca3 and 4 - 1 are free from this condition .
their solutions and evaluation of asymptotic behaviors from more general data ( real valued data ) are valid .
it is a future problem to check the validity of the present expressions for pca4 - 2 and 4 - 3 from more general data .
finally , we have not yet obtained the max - plus expression for solution to pca4 - 4 and pca of more than four neighbors .
the general relation between pca and its corresponding continuous equation is not clear by using the ultradiscretization method .
it is worthy to solve these problems in the future .
there are useful formulas to derive or prove relations concerning @xmath1 and @xmath111 operations .
below we list up such formulas relevant to this paper .
the definition of @xmath1 and @xmath111 operations are as follows .
* @xmath112 , @xmath113 . from this definition , @xmath1 and @xmath111 are related each other .
* @xmath114,@xmath115 .
most basic formulas satisfy commutative , associative and distributive laws as follows , where @xmath1 or @xmath111 plays a role of the addition and @xmath2 , i.e. , the usual sum of real numbers serves as the multiplication .
* @xmath116,@xmath117 .
* @xmath118,@xmath119 . * @xmath120,@xmath121 . from the commutative and associative laws , we can easily extend @xmath1 and @xmath111 operations from binary to multinary .
* @xmath122 , + @xmath123 .
we have the following formulas about these multinary operations .
* @xmath124 , + @xmath125 . *
@xmath126 , + @xmath127 . * @xmath128 , + @xmath129 . *
@xmath130 , + @xmath131 .
* if there exist @xmath132 such that @xmath133 , @xmath134 .
+ if there exist @xmath132 such that @xmath135 , @xmath136 .
* @xmath137 + @xmath138 + the last set of formula regarding the exchange of max and min operations has been widely used in this paper .
we prove that ( [ sol pca4 - 2 ] ) satisfies ( [ potential pca4 - 2 ] ) by the mathematical induction . for @xmath97 ,
it is true since ( [ sol pca4 - 2 ] ) is equivalent to ( [ potential pca4 - 2 ] ) of @xmath41 .
let us assume that ( [ sol pca4 - 2 ] ) satisfies ( [ potential pca4 - 2 ] ) for @xmath139 ( @xmath140 )
. then @xmath141 is expressed by @xmath142 as @xmath143 using ( [ potential pca4 - 2 ] ) , we obtain @xmath144 using the max - min formula , we have @xmath145 expanding the terms , we have @xmath146 since the underlined terms are negligible , we obtain @xmath147 using the max - min formula again , we have @xmath148 expanding the terms , we have @xmath149 since the underlined terms are negligible , we obtain @xmath150 the above equation is nothing but ( [ sol pca4 - 2 ] ) for @xmath151 . thus the proof is complete by the mathematical induction .
the asymptotic behavior of solutions to pca4 - 2 is different for three ranges of @xmath46 , @xmath82 , @xmath152 and @xmath153 .
therefore we consider three cases separately .
first , we expand the solution ( [ sol pca4 - 2 ] ) for @xmath154 as follows to make evaluation easier ; @xmath155 in this case , we have @xmath157 for any @xmath9 and @xmath8 where @xmath158 is a non - negative integer and @xmath45 is the spatial period .
therefore we have @xmath159 let us define @xmath160 by @xmath161 for @xmath162 and @xmath163 , @xmath164 holds from ( [ ineq pca4 - 2 ] ) .
therefore the first @xmath111 term of ( [ expanded sol pca4 - 2 ] ) is equal to @xmath160 for @xmath154 .
we can make a similar evaluation from the second to the @xmath165-th @xmath111 terms of ( [ expanded sol pca4 - 2 ] ) as @xmath166 moreover , the relation @xmath167 holds since @xmath168 or 1 .
it implies that the difference between @xmath96 and @xmath169 is finite if @xmath158 is finite .
therefore we can evaluate from the @xmath170-th to the penultimate terms of ( [ expanded sol pca4 - 2 ] ) for @xmath154 as @xmath171 thus ( [ expanded sol pca4 - 2 ] ) becomes @xmath172 it coincides with ( [ asymp pca4 - 2 f ] ) in the case of @xmath82 . since we can easily show @xmath173 we have @xmath174 substituting @xmath175 into the evolution equation ( [ eq pca4 - 2 ] )
, we obtain @xmath176 therefore @xmath177 is a constant irrespective of @xmath9 .
since there exists at least a site @xmath9 such that @xmath178 or @xmath179 for @xmath82 , we obtain @xmath180 for any @xmath9 .
thus we have @xmath181 the above results coincide with ( [ asymp pca4 - 2 u ] ) , ( [ asymp pca4 - 2 q ] ) and ( [ asymp pca4 - 2 q ] ) in the case of @xmath82 . in this case , the inequality @xmath183 holds for any @xmath9 and @xmath8 where @xmath162 . under this condition , the first to the @xmath184-th @xmath111 terms of ( [ expanded sol pca4 - 2 ] ) for @xmath154 are evaluated as follows : @xmath185 regarding the rest @xmath111 terms of ( [ expanded sol pca4 - 2 ] ) , they are expressed generally by @xmath186 where @xmath187 .
we can evaluate them similarly as above and obtain @xmath188 then the inequality @xmath189 holds since @xmath190 and @xmath191 .
thus we obtain @xmath192 moreover , note that @xmath193 holds for @xmath162 by ( [ cond mid rho ] ) .
therefore we obtain the final results about the asymptotic evaluation of @xmath96 for @xmath154 in the case of @xmath152 as @xmath194 it coincides with ( [ asymp pca4 - 2 f ] ) in the case of @xmath152 .
it can be easily shown that the relation @xmath195 holds for @xmath96 in the above equation .
hence so does the relation @xmath196 .
substituting it into ( [ eq pca4 - 2 ] ) , we obtain @xmath197 and @xmath198 is a constant irrespective of @xmath9 . since there exists at least a site @xmath9 such that @xmath199 or @xmath200 for @xmath152 , we have @xmath201 . thus we derive @xmath202 the above results coincide with ( [ asymp pca4 - 2 u ] ) , ( [ asymp pca4 - 2 q ] ) and ( [ asymp pca4 - 2 q ] ) in the case of @xmath152 . substituting the transformation @xmath204 into ( [ eq pca4 - 2 ] )
, we obtain @xmath205 using the max - min formulas , we have @xmath206 since @xmath207 is the same as @xmath33 , the evolution equation ( [ eq pca4 - 2 ] ) is self - conjugate , that is , it is invariant under the above transformation .
therefore , the behaviors for solutions of @xmath82 and @xmath153 are symmetric to each other and there exist one - to - one correspondence between the solutions .
since @xmath175 holds for @xmath154 in the case of @xmath82 , @xmath208 , that is , @xmath175 again holds in the case of @xmath153 . substituting this relation to ( [ eq pca4 - 2 ] )
, @xmath177 is again a constant .
since there exists at least a site @xmath9 such that @xmath209 or @xmath210 for @xmath153 , we obtain @xmath211 for any @xmath9 .
thus we have @xmath212 these results coincide with ( [ asymp pca4 - 2 u ] ) , ( [ asymp pca4 - 2 q ] ) and ( [ asymp pca4 - 2 q ] ) in the case of @xmath153 .
moreover , introducing the transformation @xmath213 with @xmath214 , we obtain @xmath215 from ( [ trans pca4 - 2 ] ) , and @xmath216 from ( [ potential pca4 - 2 ] ) . these relations are in the same form of ( [ trans pca4 - 2 ] ) and ( [ potential pca4 - 2 ] ) .
thus the asymptotic solution @xmath96 for @xmath153 is derived from that for @xmath82 as @xmath217 it coincides with ( [ asymp pca4 - 2 f ] ) in the case of @xmath153 .
note that the above symmetry of bit inversion holds for any @xmath8 as well as for @xmath154 and it gives another expression of solution ( [ sol pca4 - 2 ] ) as follows : @xmath218
99 wolfram s 1986 _ theory and applications of cellular automata _
( singapore : world scientific ) wolfram s 2002 _ a new kind of science _ ( champaign : wolfram media ) tokihiro t , takahashi d , matsukidaira j and satsuma j 1996 _ phys .
lett . _ * 76 * 324750 matsukidaira j , satsuma j , takahashi d , tokihiro t and torii m 1997 _ phys .
a _ * 225 * 28795 takahashi d and satsuma j 1990 _ j. phys .
jpn . _ * 59 * 35149 takahashi d and hirota r 2007 _ j. phys . soc . jpn . _ * 76 * 104007 mada j , idzumi m and tokihiro t 2008 _ j. phys . a _ * 41 * 175207 nishinari k and takahashi d 1998 _ j. phys .
* 31 * 543950 gaubert s and plus m 1997 _ lecture notes in computer science _ * 1200 * 26182 fuk h and boccara n 2001 _ phys
* 64 * 016117 fuk h 2004 _ j. stat .
_ p07005 boccara n and fuk h 1998 _ j. phys . a _
* 31 * 600718 hattori t and takesue s 1991 _ physica d _ * 49 * 295322 fukui m and ishibashi y 1995 _ j. phys . soc .
* 64 * 186870 matsukidaira j and nishinari k 2004 _ int .
c _ * 14 * 50715
|
we concern with a special class of binary cellular automata , i.e. , the so - called particle cellular automata ( pca ) in the present paper .
we first propose max - plus expressions to pca of 4 neighbors .
then , by utilizing basic operations of the max - plus algebra and appropriate transformations , pca4 - 1 , 4 - 2 and 4 - 3 are solved exactly and their general solutions are found in terms of max - plus expressions .
finally , we analyze the asymptotic behaviors of general solutions and prove the fundamental diagrams exactly .
|
very cool white dwarfs ( wds ) with @xmath5 are among the oldest objects in our galaxy .
they are of great importance for cosmochronology and for understanding the formation and evolution of the milky way .
our main interest is in the atmospheres of these stars as the link between their physical characteristics and the observables . in this contribution
we concentrate on the atmospheres rich in helium . due to the high transparency of helium ( compared to hydrogen ) , our current models predicts for them densities that are typical of liquids , up to @xmath6 . under such extreme conditions ,
the input physics of atmosphere models needs to be reconsidered .
there are several physical effects that distinguish fluid from gaseous atmospheres .
they all arise from the strong interaction between particles .
helium atoms become strongly correlated , the refractive index of the fluid departs from unity , and free electrons strongly interact with the medium , which affects the opacity .
non - ideal effects in the equation of state are dominated by @xmath7 interactions . for a given density and temperature
, they result in increases in the pressure and ionization fraction and a decrease in the adiabatic gradient .
the refractive index inside helium - rich white dwarf atmospheres departs from unity and can be as large as 1.35 .
we have solved the equation of radiative transfer for refractive stellar atmospheres for the first time ( kowalski & saumon 2004 ) . the impact of the total internal reflection on the radiative equilibrium results in an increase of the temperature in the optically thin atmospheric layers .
this decreases the abundance of hydrogen molecules and the @xmath8 cia opacity in atmospheres of mixed composition .
due to strong refraction near the surface , the limb darkening almost disappears .
fluid atmospheres require a new description of the opacity that is different from that for diluted gases .
the calculation of the @xmath4 free - free absorption ( or inverse bremsstrahlung ) in wd atmospheres must be revised .
preliminary results obtained from quantum molecular dynamics simulations ( qmd ) suggest that fluid , helium rich , cool wd atmospheres are much more opaque than predicted by current models .
all these effects must be considered to properly model cool wd atmospheres .
realistic models of very cool wds are of fundamental importance for cosmochronology and to understand the spectra and physical properties of the coolest wds , especially the recently discovered , peculiar ultracool stars like lhs 3250 .
our equation of state ( eos ) for fluid @xmath9 mixtures includes the following species : @xmath10 and @xmath11 . currently , only the @xmath7 interactions are included as they dominate the non - ideal contributions to the eos .
these interactions are described with an effective pair potential ( ross & young 1986 ) that is calibrated to high - pressure eos data @xcite .
the eos is computed in a manner similar to that of bergeron , saumon & wesemael ( 1995 ) but explicitly includes non - ideal terms in the calculation of the chemical equilibrium of the h / he mixture .
we are therefore able to compute continuous sequences of models from pure h to pure he composition in a self - consistent manner with the non - ideal eos .
the importance of the non - ideal terms in the eos is illustrated in fig .
1 . as the he content increases , the overall atmospheric opacity decreases and the pressure rises . clearly , the non - ideal effects dominate the eos for @xmath12 . because of the repulsive @xmath7 interactions , it is energetically favorable for @xmath13 to break up into ionized species , and the degree of ionization increases .
the variation of the refractive index inside cool wd atmosphere models is displayed in figure 2 . applying geometric optics
we can obtain the radiative transfer equation modified for refraction @xcite , which has to be solved along curved , and eventually totally reflected ray - paths . in the atmospheres of cool wds , the refractive index returns to unity far above the photosphere .
therefore , the main effects on refraction are : 1 ) a significant weakening of limb darkening ( from @xmath14 to @xmath15 for pure he models of @xmath16 ) , 2 ) a reduction of the opacity , and 3 ) an increase of temperature in the radiative zone ( fig .
3 ) . this last phenomenon is a consequence of total internal reflection , which occurs in the upper atmospheric layers ( kowalski & saumon 2004 ) . in mixed h / he models , the larger temperature results in the dissociation of @xmath17 and a reduction in the @xmath17 cia absorption in the infrared ( fig .
the effect is strongest in the @xmath18 band where the flux is increased by @xmath19 in this particular model .
the third , and probably the most important density effect in he - rich wd atmospheres is its impact on the opacity . we suspect that the description of opacities presently used in modeling of he - rich wd atmospheres is inadequate because the widely used standard expressions for opacities of h and he species are valid for tenuous gases , not fluids ! the absorption cross section of the important @xmath4 free - free process is derived for an isolated atom - electron collision , while in a dense fluid , the absorbing electron interacts strongly with many surrounding helium atoms .
so far , a reduction of the @xmath4 ff absorption and the rayleigh scattering due to strong , collective interactions in dense helium , has been reported ( iglesias , rogers & saumon 2002 ; hereafter irs02 ) .
this calculation , done in the born approximation , is not quite justified for strong interactions .
however , irs02 show that these two sources of opacity in dense helium - rich wd atmospheres may be reduced by factors as large as 20 . due to its complexity ,
a fully analytical , quantum mechanical description of absorption processes in strongly correlated fluids does not exist .
however , we can get valuable insight into the opacities of fluid helium from quantum molecular dynamics simulations @xcite . the results are surprising .
the opacity obtained for pure helium at @xmath20 and @xmath21 is flatter and much larger than that of irs02 ( figure 4 ) .
moreover , kramer s @xmath22 behavior of the free - free absorption coefficient is strongly suppressed .
the frequency behavior of the absorption coefficient obtained from simulations is similar to the prediction of the classical model @xcite . in the classical picture
the opacity for small frequencies is asymptotically constant because in the presence of the slowly varying electric field of the electromagnetic wave ( photon ) , the mobility of electrons is determined by the frequency of electron - atoms collisions .
this effect is present also in quantum systems .
the average frequency of electron - atom collisions @xmath23 ( damping parameter ) for the conditions in figure 4 is @xmath24 .
the ionization fraction @xmath25 , extracted from the simulation , by a fit to the classical model is @xmath26 , compared to @xmath27 from our present eos ( 2 ) .
moreover , extending the @xmath4 ff opacity with the classical model for frequencies smaller than @xmath23 and using the ionization fraction extracted from the simulations , we get an opacity that is in good agreement with the simulations ! this apparently successfull description of the quantum system with a classical model is preliminary and needs to be better understood
. taken at face value , these results suggest that multiple electron - atom collisions are important , and that the ionization fraction predicted by our eos is incorrect .
based on the dense fluid physics described here ( non - ideal eos , refraction , and @xmath4 ff and rayleigh scattering of irs02 but not the qmd opacities ) , we have computed a sequence of cool wd atmosphere models ( fig . 5 ) .
the comparison with the observed sequence of cool wds ( bergeron , ruiz & leggett 1997 ; hereafter brl97 ) suggests that the coolest wds have mixed h / he atmospheres , rather than the pure helium composition assigned to them in brl97 .
however , our prediction for their atmospheric abundance @xmath28 seems too high for disk wds . assuming a hydrogen accretion rate of @xmath29 , and a maximal thickness of the convection zone of @xmath30 @xcite , the largest expected abundance of helium is @xmath31 .
since the qmd simulations predict that pure helium is much more opaque than in these models , we anticipate that the fit of the coolest wds with more complete models will result in a much smaller he / h ratio .
bergeron , p. , ruiz , m. t. , leggett , s. k. 1997 , apjs , 108 , 339 ( brl97 ) bergeron , p. , saumon , d. & wesemael , f. 1995 , apj , 443 , 764 cox , j.p . &
giuli , r. t. 1968 , principles of stellar structure , vol .
1 , ch . 2 - 5 ( new york : gordon & breach ) hansen , b. m. s. 2004 , physics reports , 399 , 1 iglesias , c. a. , rogers , f. j. & saumon , d. 2002 , apj , 569 , l111 ( irs02 ) jackson , j. d. 1975 , classical electrodynamics , ch .
7 ( new york : john willey & sons ) kowalski , p. m. & saumon , d. 2004 , apj 607 , 970 mazevet , s. , collins , l. a. , magee , n.h .
2003 , a&ap , 405 , l5 nellis , w. j. , holmes , n. c. , mitchell , a. c. et al .
1984 , phys .
lett , 53 , 1248 ross , m. & young , d. a. 1986 , physics letters a , 118 , 9 , 463
|
the atmospheres of cool , helium - rich white dwarfs constitute an exotic and poorly explored physical regime of stellar atmospheres . under physical conditions where the temperature varies from @xmath0 to @xmath1 ,
the density can reach values as large as @xmath2 , and the pressure is as high as @xmath3 , the atmosphere is no longer an ideal gas and must be treated as a dense fluid .
helium atoms become strongly correlated and refraction effects are present .
opacity sources such as @xmath4 free - free absorption must be calculated with a formalism that has never been applied to astrophysical opacities .
these effects have been ignored in previous models of cool white dwarf atmospheres .
|
log - normal distributions are frequently observed not only in natural phenomena but also in social ones . for representative example , the probability density function @xmath0 of personal income or firm size @xmath1 is considered to obey the log - normal distribution @xcite@xcite @xmath2 \label{ln } \end{aligned}\ ] ] in the middle scale region . here
@xmath3 is a mean value and @xmath4 is a variance . a large number of persons or firms are included in the middle scale region .
the study of the distributions is significant . the simplest model which describes the log - normal distribution is the pure multiplicative stochastic process defined by @xmath5 where @xmath6 is a positive random variable . by applying this process iteratively ,
we obtain @xmath7 the logarithm of this equation is @xmath8 if @xmath9 is negligible compared to @xmath10 in the limit @xmath11 and @xmath12 @xmath13 are independent probability variables , @xmath10 follows the normal distribution in the limit . as a result , this model leads the stationary log - normal distribution @xmath14 .
( [ msp ] ) means that the distribution of the growth rate @xmath6 does not depend on @xmath15 .
this is known as gibrat s law @xcite that the conditional probability density function @xmath16 of the growth rate is independent of the initial value @xmath17 : @xmath18 here @xmath17 and @xmath19 are two successive incomes , assets , sales , profits , the number of employees and so forth .
the growth rate @xmath20 is defined as @xmath21 and @xmath16 as @xmath22 by using the probability density function @xmath23 and the joint probability density function @xmath24 .
as far as firm sizes in the middle scale region , however , it is reported that the growth rate distributions do not follow the gibrat s law ( [ gibrat ] ) ( see refs .
@xcite@xcite for instance ) . the log - normal distribution in the middle scale region can not be explained by the pure multiplicative stochastic process model ( [ msp ] ) .
instead , we have shown that the log - normal distribution can be derived @xcite by using no model such as the pure multiplicative stochastic process . in the derivation ,
two laws are employed which are observed in profits data of japanese firms .
one is the law of detailed balance which represents symmetry in a stable economy @xcite .
the other is a non - gibrat s law which describes a statistical dependence in the growth rate of the past value @xcite . in ref .
@xcite , the non - gibrat s law and the static log - normal distribution in the middle scale region are uniquely derived from the detailed balance .
the derivation has been confirmed by the empirical data analysis . in this study , we extend the derivation by replacing the detailed balance with the detailed quasi - balance proposed in ref .
@xcite to derive a log - normal distribution in the quasistatic system . by this procedure , the log - normal distribution is described as quasistatic .
the derivation is consistently confirmed employing data on the assessed value of land in 19742007 japan .
in this section , we briefly review the study in ref . @xcite . as an equilibrium system
, we investigate profits data of japanese firms in 2003 , 2004 and 2005 which are available on the database cd eyes 50 " published in 2005 and 2006 by tokyo shoko research , ltd . @xcite .
figure [ logprofit0405db ] shows the joint probability density function @xmath25 of all firms in the database , the profits of which in 2004 ( @xmath17 ) and 2005 ( @xmath19 ) exceeded @xmath26 , @xmath27 and @xmath28 .
the number of firms is 232,497 " . from fig .
[ logprofit0405db ] , we approximately confirm the detailed balance which is time - reversal symmetry ( @xmath29 ) of @xmath25 @xcite : @xmath30 figure [ distributiondata ] shows probability density functions of profits in the database .
the distributions are almost stable and the following power - law is observed in the large scale region @xmath31 where @xmath32 is a certain threshold .
this power - law is called pareto s law @xcite and the exponent @xmath33 is named pareto index . notice
that the pareto s law does not hold below the threshold @xmath32 .
the purpose of this section is to exhibit the distribution in the middle scale region under the detailed balance ( [ detailed balance ] ) . in order to identify a statistical dependence in the growth rate of the past value
, we examine the probability density function of the profits growth rate in 20042005 . in the database ,
we divide the range of @xmath17 into logarithmically equal bins as @xmath34 $ ] thousand yen with @xmath35 .
figure [ profitgrowthrate0405 ] shows the conditional probability density functions for @xmath36 in the case of @xmath37 , @xmath38 , @xmath39 and @xmath40 . the number of firms in fig .
[ profitgrowthrate0405 ] is @xmath41 " ,
@xmath42 " , @xmath43 " and @xmath44 " , respectively . from fig .
[ profitgrowthrate0405 ] , we approximate @xmath45 by linear functions of @xmath46 : @xmath47 here @xmath48 is the conditional probability density function for @xmath46 , which is related to that for @xmath20 by @xmath49 . these approximations ( [ approximation1])([approximation2 ] ) are expressed as tent - shaped exponential forms as follows : @xmath50 where @xmath51 .
furthermore , fig .
[ profitgrowthrate0405 ] shows that the dependence of @xmath52 on @xmath17 is negligible for @xmath53 .
we assess the validity of these approximations against the results .
figure [ egibrat0405 ] represents the dependence of @xmath54 on the lower bound of each bin @xmath55 . for @xmath56 ,
@xmath54 hardly responds to @xmath17 .
this means that the gibrat s law ( [ gibrat ] ) holds only in the large scale region of profits ( @xmath57 ) . )
is derived from the gibrat s law ( [ gibrat ] ) valid only in the large scale region and the detailed balance ( [ detailed balance ] ) . in the derivation ,
linear approximations ( [ approximation1])([approximation2 ] ) need not be assumed . ] in contrast , @xmath58 linearly increases and @xmath59 linearly decreases symmetrically with @xmath60 for @xmath61 . from fig .
[ egibrat0405 ] , the slops are described as @xcite @xmath62 the parameters are estimated as follows : @xmath63 for @xmath64 , @xmath65 for @xmath66 , takes different values in two regions .
this is not an exact procedure .
however , for firms which are in the large scale region in both years ( @xmath64 and @xmath67 ) or in the middle scale one ( @xmath66 and @xmath68 ) , this procedure is exact . in the database ,
most firms stay in the same region .
this parameterization is , therefore , approximately valid for describing the probability density function @xmath23 .
this is confirmed in fig .
[ distribution0405 - 04 ] . ]
@xmath69 thousand yen and @xmath70 thousand yen .
notice that approximations ( [ approximation1])([approximation2 ] ) uniquely fix the expression of @xmath54 under the detailed balance @xcite .
this derivation is included in the proof in the next section .
we call eqs .
( [ tent - shaped1])([t0 ] ) a non - gibrat s law . for @xmath53 ,
the dependence of @xmath71 on @xmath17 is negligible . in this case
, the non - gibrat s law determines the probability density function of profits as follows : @xmath72 where @xmath73 @xcite .
this is the power - law in the large scale region ( @xmath64 ) and the log - normal distribution in the middle scale one ( @xmath66 ) . the relations between parameters @xmath4 , @xmath3 in eq .
( [ ln ] ) and @xmath74 , @xmath33 , @xmath32 are given by @xmath75 , @xmath76 .
figure [ distribution0405 - 04 ] shows that the distribution ( [ handm ] ) fits with the empirical data consistently .
notice that the distribution can not fit with the empirical data , if @xmath74 is different from the value estimated in fig .
[ egibrat0405 ] ( @xmath77 or @xmath78 for instance ) .
in the previous section , the log - normal distribution in the middle scale region is exhibited by a non - gibrat s law under the detailed balance .
the resultant profits distribution is empirically confirmed in data analyses of japanese firms in 20032004 , 20042005 and 20032005 .
the profits distribution ( [ handm ] ) is static , because the derivation is based on the detailed balance ( [ detailed balance ] ) which is static time - reversal symmetry observed in the system ( fig .
[ logprofit0405db ] ) . in the case
that an economy is not stable , the detailed balance should be extended to describe the state . in ref .
@xcite , we have derived pareto s law with annually varying pareto index under the detailed quasi - balance : @xmath79 it is assumed that , in an ideal quasistatic system , the joint probability density function has @xmath80 " symmetry where @xmath81 is a slope of a regression line : @xmath82 the detailed balance ( [ detailed balance ] ) has the special symmetry @xmath83 .
because the detailed quasi - balance ( [ detailed quasi - balance ] ) is imposed on the system , @xmath81 is related to @xmath84 as follows : @xmath85 here @xmath86 is a sufficient large value compared to the upper bound in which @xmath81 and @xmath84 are estimated . in ref .
@xcite , these results have been empirically confirmed by employing data on the assessed value of land @xcite@xcite in 19832005 . in the derivation , under the detailed quasi - balance
, we have used the gibrat s law ( [ gibrat ] ) valid only in the large scale region without linear approximations ( [ approximation1])([approximation2 ] ) .
the purpose of this section is to show that the approximations uniquely fix a non - gibrat s law even under the detailed quasi - balance ( [ detailed quasi - balance ] ) .
after that , we identify the quasistatic distribution not only in the large scale region but also in the middle scale one . by using the relation @xmath87 under the change of variables @xmath88 @xmath89 ,
these two joint probability density functions are related to each other @xmath90 where we use a modified growth rate @xmath91 . from this relation , the detailed quasi - balance ( [ detailed quasi - balance ] ) is rewritten as @xmath92 substituting @xmath24 for @xmath16 defined in eq .
( [ define ] ) , the detailed quasi - balance is reduced to be @xmath93 in the quasistatic system , we also assume that @xmath16 follows the tent - shaped exponential forms ( [ tent - shaped1])([tent - shaped2 ] ) . under the approximations ,
the detailed quasi - balance is expressed as @xmath94 for @xmath95 .
here we use the notation @xmath96 . by setting @xmath97 after differentiating eq .
( [ detailed quasi - balance 2 ] ) with respect to @xmath20 , the following differential equation is obtained : @xmath98 \tilde{p}(x ) + x \tilde{p}~{'}(x ) = 0~ , \label{de}\end{aligned}\ ] ] where @xmath1 denotes @xmath17 .
the same equation is obtained for @xmath99 .
similarly , from the second and third derivatives of eq .
( [ detailed quasi - balance 2 ] ) , the following differential equations are obtained : @xmath100 the solutions are uniquely fixed as @xmath101 this is the same expression under the detailed balance .
here we consider two log - normal distributions in the middle scale region : @xmath106~ , \label{ln1}\\ p_{\rm ln_2}(x_2 ) = \frac{1}{x_2 \sqrt{2 \pi { \sigma_2}^2 } } \exp \left[-\frac{\ln^2 \left ( x_2/\bar{x}_2 \right)}{2 { \sigma_2}^2 } \right]~. \label{ln2 } \end{aligned}\ ] ] by comparing eqs .
( [ handm1])([handm2 ] ) to ( [ ln1])([ln2 ] ) , we identify @xmath107 , @xmath108 , @xmath109 and @xmath110 . consequently , the relation between @xmath111 , @xmath112 and @xmath81 is expressed as @xmath113 this is the equation which quasistatic log - normal distributions satisfy .
in this section , we confirm the results in the previous section employing data on the assessed value of land in 19742007 japan . in japan , land is a very important asset which is distinguished from buildings .
the assessed value of land indicates the standard land prices evaluated by ministry of land , infrastructure and transport japan .
the investigation is undertaken on each piece of land assessed once a year according to the posted land price system from 1970 .
the probability distribution functions of land prices are shown in fig .
[ distpart ] .
the number of data points is 14,570 " , 15,010 " , 15,010 " , 15,010 " , 15,580 " , 16,480 " , 17,030 " , 17,380 " and 17,600 " in 19741982 , respectively . in 19831991 , 16,975 " , 16,975 " , 16,975 " , 16,635 " , 16,635 " , 16,820 " , 16,840 " , 16,865 " and 16,892 " , respectively . in 19921999 , 17,115 " , 20,555 " , 26,000 " , 30,000 " , 30,000 " , 30,300 " , 30,600 " and 30,800 " , respectively . in 20002007 ,
31,000 " , 31,000 " , 31,520 " , 31,866 " , 31,866 " , 31,230 " , 31,230 " and 30,000 " , respectively . in each figure , the power - law is observed in the large scale region .
in addition , pareto index @xmath33 varies annually and changes significantly before and after bubble years ( 19861991 ) .
this is represented in fig .
[ varyingparetoindex2 ] where each pareto index @xmath33 is estimated in the range of land prices from @xmath114 to @xmath115 @xmath116 . in fig .
[ distpart ] , the log - normal distribution is also observed in each middle scale region .
the standard deviation @xmath117 varies annually and changes significantly in bubble years .
this is also represented in fig .
[ varyingparetoindex2 ] where each standard deviation @xmath117 is estimated in the range from @xmath118 to @xmath119 @xmath116 .
it is well known that the pareto index for firm sizes hardly changes ( fig .
[ distributiondata ] ) .
in such a database , the detailed balance ( [ detailed balance ] ) is observed @xcite ( fig .
[ logprofit0405db ] ) . on the other hand ,
the pareto index for the assessed value of land varies annually .
this suggests that the system is not stable and the detailed balance does not hold .
actually most of the period , the detailed balance is not observed in the scatter plot of all pieces of land assessed in the database ( figs .
[ 95vs96 ] and [ 06vs07 ] for instance ) . for instance ) . ] in the previous section , we have proposed the detailed quasi - balance ( [ detailed quasi - balance ] ) in an ideal quasistatic system . in each scatter plot of all pieces of land assessed in the database , the parameter @xmath81 is measured and the result is shown in fig .
[ dqb ] . here
@xmath81 is estimated in the following two regions .
one is the large scale region between @xmath114 and @xmath115 @xmath116 where pareto index @xmath33 is estimated .
the other is the middle scale one between @xmath118 and @xmath119 @xmath116 where the standard deviation @xmath117 of the log - normal distribution is estimated . in fig .
[ dqb ] , we represent two parameters by @xmath120 and @xmath121 , respectively . by using these parameters , we confirm not only the relation between the ratio of @xmath122 and @xmath120 ( [ ratio1 ] ) in the large scale region but also the relation between the ratio of @xmath117 and @xmath121 ( [ ratio2 ] ) in the middle scale one ( figs . [ ratio ] and [ ratio2 ] ) .
this warrants approximations assumed in the previous section .
the derivation and the data analysis are consistent .
in this study , by using no model , we have derived a quasistatically varying log - normal distribution from a non - gibrat s law under the detailed quasi - balance . in the derivation ,
we have employed two approximations .
one is that the probability density function of the growth rate is described as tent - shaped exponential functions .
the other is that the value of the origin of the growth rate distribution is constant . even under the detailed quasi - balance ,
the first approximation uniquely fixes a non - gibrat s law to be the same expression under the detailed balance . together with the second approximation ,
the resultant distribution is described as power - law with varying pareto index in the large scale region . in the middle scale region ,
the distribution is reduced to the quasistatic log - normal distribution with the varying standard deviation .
notice that not only the change of pareto index @xmath33 but also the change of the standard deviation @xmath117 depends on a parameter @xmath81 of the detailed quasi - balance . employing empirical data on the assessed value of land in 19742007 japan ,
we have confirmed these analytic results . in the scatter plot of all pieces of land assessed in the database ,
the parameter @xmath81 of the detailed quasi - balance is measured in the following two regions .
one is the large scale region where pareto index @xmath33 is estimated .
the other is the middle scale region where the standard deviation @xmath117 is estimated .
we have observed that two parameters @xmath120 measured in the large scale region and @xmath121 measured in the middle scale one are in good agreement with the ratio of @xmath122 and @xmath117 , respectively .
intriguingly , it is observed that the change of pareto index influences the change of the standard deviation in the opposite direction ( fig .
[ varyingparetoindex2 ] ) . in this database ,
the change of the distribution in the large scale region propagates in the middle scale one .
this phenomenon is intelligible by two slopes @xmath120 and @xmath121 of the regression lines in the scatter plot ( fig .
[ dqb ] ) . in order to make these discussions
more precisely , we need to investigate wealth of data in unstable state for a long period . by analyzing the database ,
we directly observe the non - gibrat s law which defines the large and middle scale regions .
in addition , we should take a statistical test for the symmetry in the two arguments of @xmath123 to confirm the detailed quasi - balance directly .
these accurate data analyses are imperative to realize the applications of the study in this paper , such as credit risk management and so forth .
the author is grateful to the yukawa institute for theoretical physics at kyoto university , where this work was initiated during the yitp - w-05 - 07 on `` econophysics ii
physics - based approach to economic and social phenomena '' , and especially to professor h. aoyama for the critical question about my previous work .
thanks are also due to dr .
y. fujiwara , dr . w. souma and dr .
m. tomoyose for a lot of useful discussions and comments .
99 r. gibrat , les inegalites economiques , paris , sirey , 1932 .
badger , in : b.j .
west ( ed . ) , mathematical models as a tool for the social science , gordon and breach , new york , 1980 , p. 87 ; + e.w .
montrll , m.f .
shlesinger , j. stat .
32 ( 1983 ) 209 .
v. pareto , cours deconomique politique , macmillan , london , 1897 .
a. dr@xmath124gulescu , v.m .
yakovenko , physica a299 ( 2001 ) 213 ; + a.c .
silva , v.m .
yakovenko , europhys .
69 ( 2005 ) 304 .
m. anazawa , a. ishikawa , t. suzuki and m. tomoyose , physica a335 ( 2004 ) 616 ; + a. ishikawa and t. suzuki , physica a343 ( 2004 ) 376 .
stanley , l.a.n .
amaral , s.v .
buldyrev , s. havlin , h. leschhorn , p. maass , m.a .
salinger , h.e .
stanley , nature 379 ( 1996 ) 804 .
h. takayasu , m. takayasu , m.p .
okazaki , k. marumo , t. shimizu , cond - mat/0008057 , in : m.m .
novak ( ed . ) , paradigms of complexity , world scientific , 2000 , p. 243 .
h. aoyama , ninth annual workshop on economic heterogeneous interacting agents ( wehia 2004 ) ; + h. aoyama , y. fujiwara , w. souma , the physical society of japan 2004 autumn meeting .
a. ishikawa , physica a367 ( 2006 ) 425 .
a. ishikawa , physica a383/1 ( 2007 ) 79 .
y. fujiwara , w. souma , h. aoyama , t. kaizoji and m. aoki , physica a321 ( 2003 ) 598 ; + y. fujiwara , c.d .
guilmi , h. aoyama , m. gallegati and w. souma , physica a335 ( 2004 ) 197 .
a. ishikawa , physica a371 ( 2006 ) 525 .
tokyo shoko research , ltd .
, http://www.tsr - net.co.jp/. a. ishikawa , physica a363 ( 2006 ) 367 .
t. kaizoji , physica a326 ( 2003 ) 256 ; + t. kaizoji and m. kaizoji , physica a344 ( 2004 ) 138 .
the ministry of land , infrastructure and transport government of japan s world - wide web site , http://nlftp.mlit.go.jp / ksj/. w. souma , fractals 9 ( 2001 ) 463 .
|
employing data on the assessed value of land in 19742007 japan , we exhibit a quasistatically varying log - normal distribution in the middle scale region . in the derivation
, a non - gibrat s law under the detailed quasi - balance is adopted together with two approximations .
the resultant distribution is power - law with the varying exponent in the large scale region and the quasistatic log - normal distribution with the varying standard deviation in the middle scale region . in the distribution ,
not only the change of the exponent but also the change of the standard deviation depends on the parameter of the detailed quasi - balance .
these results are consistently confirmed by the empirical data .
pacs code : 89.65.gh +
|
there is a good reason to study wave propagation in turbid or random media : multiply scattered waves can _ probe _ temporal changes in physical systems @xcite .
thus , light transport through fog @xcite , milky liquids , nematic liquid crystals @xcite , granular media@xcite , foams@xcite , and human tissue @xcite ; propagation of elastic waves in the earth s crust @xcite ; acoustic waves in the fluidized or sedimenting suspensions @xcite ; etc . , have attracted much attention . in a turbid medium
, light undergoes many scattering events before leaving the sample , and the transport of light energy is diffusive @xcite .
therefore , the photon can be considered as a random walker .
the transport - mean - free path @xmath1 , over which the photon direction becomes randomized , depends on the structural details of the opaque medium .
experimental techniques like diffuse - transmission spectroscopy ( dts ) @xcite and diffusing - wave spectroscopy ( dws ) @xcite can be used to measure @xmath1 . in dts ,
the average fraction @xmath6 of incident light transmitted through a slab of thickness @xmath7 is measured .
the transport - mean - free path is then deduced from @xmath8 . utilizing the temporal intensity fluctuations in the speckle field of the multiply scattered light
, dws determines @xmath9 and the mean - squared displacement of the scattering sites due to time evolution , thermal motion , or flow .
a plethora of light - scattering experiments show that light transport reaches its diffusive limit in granular media@xcite and foams@xcite , which means that photons perform a random walk .
however , the mechanisms underlying this random walk are not elucidated .
a wet foam is composed of spherical gas bubbles dispersed in liquid .
a relatively dry foam consists of polyhedral cells separated by thin liquid films .
three of them meet in the so - called plateau borders which then define tetrahedral vertices @xcite . in their studies of foams with the liquid volume fraction @xmath10 in the range @xmath11 , vera , saint - jalmes , and
durian @xcite observed the empirical law @xmath12 where @xmath2 is the average bubble radius .
recent studies of scattering from plateau borders @xcite , vertices@xcite , and films @xcite ; or transport effects such as total internal reflection of photons inside the plateau borders@xcite , have not yet clarified the empirical law of vera et al .. for granular media , systematic measurements of the transport - mean - free path @xmath1 as a function of the refractive indices of grains and the host medium ( air , water , ... ) , grain size , and packing fraction , have not been performed .
menon et al .
@xcite determined @xmath13 for glass spheres of radius @xmath14 dispersed in air . for glass beads dispersed in water , leutz et al .
@xcite found @xmath15 for @xmath16 .
their samples had a packing fraction @xmath17 .
crassous@xcite performed numerical simulations to find @xmath1 as a function of refractive indices of the grain and host medium , but only for packing fraction @xmath17 .
it is instructive to consider _ simple _ or even toy models of granular media and wet foams , which allow an _ analytic _ access to the transport - mean - free path @xmath1 .
apparently , such models pave the way for deeper understanding of fascinating dws experiments . in this paper , we consider two - dimensional packing of monosize disks .
the disks are much larger than the wavelength of light , thus one can employ ray optics to follow a light beam or photon as it is reflected by the disks with a probability @xmath0 called the intensity reflectance .
we assume that the intensity reflectance is constant , and the velocity of light inside and outside the disks are @xmath18 and @xmath19 , respectively .
we show that the photon s random walk based on the above rules is a persistent random walk @xcite . writing a master equation to describe the photon transport , we find in sec . [ aa ] the transport - mean - free path as @xmath20 where @xmath21 and @xmath2 denote the packing fraction and radius of disks , respectively .
we further study our model by numerical simulation of the photon s random walk .
we observe the overall agreement between our numerical and analytical estimates of the transport - mean - free path . for glass beads immersed in air or water
, we find transport - mean - free paths about half the experimental ones@xcite . for the air bubbles immersed in the water , we use eq .
( [ mainresult ] ) to derive @xmath1 as a function of the liquid volume fraction @xmath22 .
@xmath23 is estimated as a weighted average of fresnel s intensity reflectance .
we find that in the range @xmath24 , our analytical result agrees well with the relation @xmath25 .
in other words , we find that @xmath3 is a _ linear _ function of @xmath26 .
using the hybrid lattice gas model for two - dimensional foams and fresnel s intensity reflectance , sun and hutzler performed numerical simulation of photon transport and found @xmath27@xcite .
quite remarkably , our _ analytic _ estimate of the transport - mean - free path throws new light on the empirical law of vera et al . and the numerical simulation of sun et al .. our article
is organized as follows . in section [ model ]
we introduce the two - dimensional packing of disks as a simple model for a granular medium or a wet foam .
photon transport in a random packing of disks using constant intensity reflectance is discussed in sec .
discussions , conclusions , and an outlook are presented in sec .
as a _ simple _ model for a two - dimensional disordered granular medium , wet foam , and bubbly liquid , we choose the random packing of circular disks .
all non - overlapping disks have the same radius @xmath2 , and cover a fraction @xmath21 of the plane .
to address the photon transport in such medium , we have made the following assumptions : ( i ) disks or grains , are much larger than the wavelength of light , thus one can employ ray optics to follow a light beam or photon as it is reflected by the disks with a probability @xmath0 called the intensity reflectance . ( ii ) @xmath0 is a constant , with no dependence on the incidence angle .
( iii ) although disks of refractive index @xmath28 are immersed in a medium of refractive index @xmath29 , the incident and the transmitted rays have the same direction .
in other words , we assume that the angle of refraction equals the angle of incidence .
( iv ) the velocity of light inside and outside the disks are @xmath18 and @xmath19 , respectively .
our first assumption is inspired by the experiments@xcite .
our second and third assumptions do not agree with fresnel s formulas and snell s law , respectively .
consequently , our model does not consider total internal reflection of rays .
however , we deliberately adopt a step - by - step approach to photon transport in granular media , and will consider more realistic models later . as already mentioned , we model single photon paths in a packing of disks as a random walk with rules motivated by ray optics , i.e. , an incoming light beam is reflected from a disk surface with a probability @xmath0 or it traverses the disk surface with a probability @xmath30 .
this naturally leads to a persistent random walk of the photons @xcite , where the walker remembers its direction from the previous step @xcite .
persistent random walks are employed in biological problems @xcite , turbulent diffusion @xcite , polymers @xcite , landauer diffusion coefficient for a one - dimensional solid @xcite , and in general transport mechanisms @xcite .
more recent applications are reviewed in @xcite . in the following section ,
we adopt the approach of @xcite to study persistent random walk of the photons in a granular medium .
( b ) path of a photon moving in a disk and hitting its surface with an incidence angle @xmath31 .
the step length inside the disk is @xmath32 , where @xmath2 is radius of the disk .
( c ) a photon impinging on a disk with an impact parameter @xmath33 .
note @xmath34 .
( d ) the distance @xmath35 is related to the incidence angle @xmath36 by @xmath37 . ] the photon random walk in a packing of disks consists of steps inside and outside the disks .
we denote the average length of steps inside and outside the grains by @xmath38 and @xmath39 , respectively .
we characterize each step by an angle relative to the @xmath40-axis . as fig .
[ f1](a ) demonstrates , on hitting a disk with an incidence angle @xmath41 , a photon moving in the host medium along the direction @xmath42 will be either reflected to the direction @xmath43 , or enters the disk .
the probability distribution of the random variable @xmath41 ( @xmath44 ) is @xmath45 , see appendix [ app1 ] .
similarly , a photon moving in a disk along the direction @xmath46 and hitting its surface with an angle @xmath31 , will be either reflected to the direction @xmath47 or enters the host medium , see fig .
[ f1](b ) .
quite remarkably , the probability distribution of the incidence angles @xmath41 and @xmath36 are the same , see appendix [ app1 ] . we can therefore conclude that diffusion of the photons inside and outside the grains are not inherently different .
as will be shown in the following , we write a master equation to describe the photon diffusion inside ( outside ) the grains , utilizing the step length @xmath38 ( @xmath39 ) and velocity @xmath18 ( @xmath19 ) , and extract the diffusion constant @xmath48 ( @xmath49 ) .
according to the two - state model of lennard - jones @xcite , the diffusion constant of photons in the granular medium is @xmath50 where @xmath51 ( @xmath52 ) is the fraction of time that the photons spend inside ( outside ) the disks .
we introduce the probability @xmath53 that the photon after its @xmath54th step of length @xmath55 along the direction @xmath56 , arrives in the area @xmath57 at position @xmath58 .
then the following master equation expresses the evolution of @xmath59 : @xmath60 the first term on the right - hand side describes the reflection of the photon with a probability @xmath0 .
the photon which has arrived at position @xmath61 along the direction @xmath62 , changes its direction by an angle @xmath63 according to the probability function @xmath64@xcite .
the second term describes the transmission with a probability @xmath30 .
the photon performs a ballistic motion with step length @xmath65 along direction @xmath56 from position @xmath61 to @xmath66 .
the diffusion constant follows from the evaluation of the second moment of @xmath59 with respect to the spatial coordinates @xmath40 and @xmath67 .
the probability distribution as an exact solution of the master equation ( [ master ] ) is hard to obtain .
however , there is a more direct method for the evaluation of the moments which employs the characteristic function @xmath68 associated with @xmath59@xcite : @xmath69 where @xmath70 and @xmath71 are either zero or positive integers , @xmath72 and @xmath73 .
we are interested in the first and second moments of @xmath59 , thus we focus on the taylor expansion @xmath74 where @xmath75 and @xmath76 are the polar representation of the vector @xmath77 . from eqs .
( [ mean ] ) and ( [ taylor ] ) it follows that @xmath78 fourier transforming eq .
( [ master ] ) , we obtain @xmath79 , \label{cagefourier}\end{aligned}\ ] ] where @xmath80 , and @xmath81 is the @xmath82th - order bessel function .
since we are only interested in the taylor coefficients @xmath83 and @xmath84 , we insert eq .
( [ taylor ] ) into eq .
( [ cagefourier ] ) . using the taylor expansion of the relevant bessel functions @xmath85 ( @xmath86 ) and @xmath87 @xcite and collecting all terms with the same power in @xmath75 , results in the following recursion relations for the @xmath88 : @xmath89 q_{0,n}(\alpha|m ) , \nonumber \\
\nonumber\\ & & q_{1,n+1}(\alpha|m)= \big[t+r ( -1)^m \mathbf{f}(2 m ) \big ] q_{1,n}(\alpha|m ) \nonumber \\ & & \hspace{0.7 cm } + \frac{e^{-i \alpha } } { 2}\big[t+r(-1)^{m+1}\mathbf{f}(2m+2)\big ] q_{0,n}(\alpha|m+1 ) \nonumber \\ & & \hspace{0.7cm}+ \frac{e^{i \alpha}}{2 } \big[t+r(-1)^{m-1}\mathbf{f}(2m-2)\big ] q_{0,n}(\alpha|m-1 ) , \nonumber\\ & & q_{2,n+1}(\alpha|m)=\big [ t+ r ( -1)^m \mathbf{f}(2m)\big ] q_{2,n}(\alpha|m ) \nonumber\\ & & \hspace{0.7 cm } + e^{-i \alpha } \big[t+r(-1)^{m+1}\mathbf{f}(2m+2)\big ] q_{1,n}(\alpha|m+1 ) \nonumber\\ & & \hspace{0.7 cm } + e^{i\alpha}\big[t+r(-1)^{m-1}\mathbf{f}(2m-2)\big ] q_{1,n}(\alpha|m-1)\nonumber \\ & & \hspace{0.7 cm } + \frac{1 } { 2 } \big[t+ r ( -1)^m \mathbf{f}(2 m ) \big ] q_{0,n}(\alpha|m ) \nonumber \\ & & \hspace{0.7 cm } + \frac{e^{-2i \alpha}}{4 } \big [ t+r(-1)^{m+2}\mathbf{f}(2m+4 ) \big ] q_{0,n}(\alpha|m+2 ) \nonumber \\ & & \hspace{0.7 cm } + \frac{e^{2i \alpha}}{4 } \big[t+r(-1)^{m-2}\mathbf{f}(2m-4 ) \big ] q_{0,n}(\alpha|m-2 ) .
\nonumber \\ \label{setqq}\end{aligned}\ ] ] we solve this set of coupled linear difference equations using the method of the @xmath90-transform @xcite .
the @xmath90-transform @xmath91 of a function @xmath92 of a discrete variable @xmath93 is defined by @xmath94 .
one then derives the @xmath90-transform of @xmath95 simply as @xmath96 .
note the similarities of this rule with the laplace transform of the time derivative of a continuous function@xcite .
the @xmath90-transform of equations ( [ setqq ] ) leads to a set of algebraic equations which immediately gives @xmath97 } , \nonumber \\
\nonumber\\ q_{1}(z,\alpha|m)&= & \frac{q_{1,n=0}(\alpha|m)}{1-z\big[t+r(-1)^m\mathbf{f}(2m)\big ] } + \frac{z}{2\big(1-z\big[t+r(-1)^m\mathbf{f}(2m)\big]\big ) } \nonumber\\ & \times & \bigg\{\frac{e^{-i \alpha}\big[t+r(-1)^{m+1}\mathbf{f}(2m+2)\big ] q_{0,n=0}(\alpha|m+1)}{1-z\big[t+r(-1)^{m+1}\mathbf{f}(2m+2)\big]}+ \frac{e^{i \alpha}\big[t+r(-1)^{m-1}\mathbf{f}(2m-2)\big ] q_{0,n=0}(\alpha|m-1)}{1-z\big[t+r(-1)^{m-1}\mathbf{f}(2m-2)\big]}\bigg\},\nonumber\\ \nonumber\\ q_{2}(z,\alpha|m)&= & \frac{q_{2,n=0}(\alpha|m)}{1-z\big[t+r(-1)^m\mathbf{f}(2m)\big]}+ \frac{zq_{0,n=0}(\alpha|m)}{2\big(1-z\big[t+r(-1)^m\mathbf{f}(2m)\big]\big)^2}+\frac{z}{1-z\big[t+r(-1)^m\mathbf{f}(2m)\big]}\nonumber \\ & \times & \bigg\{e^{-i \alpha}\big[t+r(-1)^{m+1}\mathbf{f}(2m+2)\big ] q_{1}(z,\alpha|m+1)+ e^{i \alpha}\big[t+r(-1)^{m-1}\mathbf{f}(2m-2)\big ] q_{1}(z,\alpha|m-1)\nonumber \\ & + & \frac{e^{-2i\alpha}\big[t+r(-1)^m\mathbf{f}(2m+4)\big]q_{0,n=0}(\alpha|m+2)}{4\big(1-z\big[t+r(-1)^{m+2}\mathbf{f}(2m+4)\big]\big)}+ \frac{e^{2i \alpha}\big[t+r(-1)^m\mathbf{f}(2m-4)\big]q_{0,n=0}(\alpha|m-2)}{4\big(1-z\big[t+r(-1)^{m-2}\mathbf{f}(2m-4)\big]\big)}\bigg\},\end{aligned}\ ] ] where @xmath98 , especially @xmath99 .
the above expressions contain the sum of several terms whose inverse @xmath90-transform are readily accessible : @xmath100 here @xmath101 is an arbitrary real number whose absolute magnitude is less than @xmath102 . for an arbitrary initial distribution @xmath103 , the relevant function @xmath104 contains terms which are either constant or behave as @xmath105 with @xmath106 .
they are associated with the randomization of the initial distribution of the random walkers but are not essential for large @xmath54 . according to eq .
( [ meanmean ] ) , @xmath107 the behavior of the mean - square displacements is associated with @xmath108 , see eq .
( [ meanmean ] ) .
we checked that for large @xmath54 or in the long - time limit it is purely diffusive , i.e. , @xmath109 where we introduced the time @xmath110 which passes when the random walker makes @xmath54 steps at a speed @xmath111 .
we extract the diffusion constants from @xmath112 and @xmath113 : @xmath114 as already mentioned , we write the master equation ( [ master ] ) to describe photon diffusion in the grains and in the host medium .
equation ( [ finald ] ) immediately leads to @xmath115 the task is now expressing @xmath38 , @xmath39 , @xmath51 and @xmath116 in terms of the model parameters @xmath2 , @xmath21 , @xmath0 , and using the two - state model of lennard - jones to derive @xmath117 .
first , we note that @xmath118 , where @xmath31 is the incidence angle of photons moving in the disk , and here @xmath119 denotes averaging with respect to the probability distribution @xmath120 , see appendix [ app1 ] and fig .
[ f1](b ) .
second , @xmath121 .
hence we find @xmath122 the evaluation of @xmath51 is more exacting .
each step length @xmath38 inside a disk takes a time @xmath123 .
the probability of @xmath124 internal steps before leaving the disk is @xmath125 .
hence the average time that a photon spends in the disk is @xmath126 .
the photon spends a time @xmath127 before reaching a disk .
hence @xmath128 now we utilize eq .
( [ lenard ] ) to derive the diffusion constant of photons in the granular medium as @xmath129 in two - dimensional space , the transport - mean - free path follows from @xmath130 where @xmath131 is the transport velocity of light in the medium . to a first approximation @xmath132 note that the velocity of light in the disks ( the host medium ) covering a fraction @xmath21 ( @xmath133 ) of the plane is @xmath134 ( @xmath135 ) . from eqs .
( [ lkj0]-[lkj2 ] ) , we find the transport - mean - free path @xmath1 mentioned already in eq .
( [ mainresult ] ) in the introduction .
disks , covering a fraction @xmath136 of the plane . ]
( in units of the disk radius @xmath2 times the velocity of light @xmath137 ) as a function of the intensity reflectance @xmath0 , for various packing fractions . here @xmath138 and @xmath139 are assumed .
monte carlo simulation results and @xmath140 are denoted , respectively , by points and the line . ]
we presented an _
analytic _ theory to calculate the diffusion constant of photons .
now we carefully examine this analytic result by performing numerical simulations . in order to generate random configurations of monodisperse disks with a desired packing fraction , we compress a dilute system of rigid disks into a smaller space .
simulation methods based on a confining box generate a packing whose properties in the vicinity of walls differ from those in the bulk .
hence we utilize the compaction method of ref .
@xcite which combines the contact dynamics algorithm @xcite with the concept of the andersen dynamics @xcite .
this combined simulation method involves variable area of the simulation box with periodic boundary conditions in all directions . due to the exclusion of side walls , the algorithm generates homogenous packings .
we let photons perform a random walk in our packing of disks by applying the rules introduced in sec .
[ model ] . for improving the speed of our ray tracing program , we adopt the cell index method commonly used in the molecular dynamics simulations@xcite .
the square simulation box is divided into a regular lattice of @xmath141 cells .
we maintain a list of disks in each of these cells .
a photon moving in the cell @xmath142 ( @xmath143 ) probably hits the disks in the cell @xmath142 or its neighbor cells .
thus it is not necessary to check collision between the photon and _ all _ disks of the medium .
our computer program shrinks an initial dilute sample of @xmath144 non - overlapping disks randomly distributed in a two - dimensional simulation box . in the course of shrinking the packing
, the program saves snapshots of the grain positions if the packing fraction @xmath145 $ ] , see fig .
[ f2a ] . for each medium
, the program takes @xmath144 photons at an initial position , and launches them in a direction specified by angle @xmath146 relative to the @xmath40-axis .
then it generates the trajectory of each photon following a standard monte carlo procedure and evaluates the statistics of the photon cloud at times @xmath147 $ ] ( in units of @xmath148 ) . as detailed in ref .
@xcite , we determine the diffusion constant @xmath117 from the temporal evolution of the average mean - square displacement of the photons : @xmath149 .
( in units of the disk radius @xmath2 times the velocity of light @xmath137 ) as a function of the packing fraction @xmath150 , for @xmath151 and @xmath152 . here @xmath138 and @xmath139 are assumed .
monte carlo simulation results and @xmath153 are denoted , respectively , by points and the line . ] for angles @xmath154 $ ] , the simulation is repeated for each intensity reflectance @xmath155 $ ] . as a reasonable result ,
no dependence on the starting point and the starting direction is observed . in fig .
[ f2 ] we plot the diffusion constant @xmath117 ( in units of the disk radius @xmath2 times the velocity of light @xmath137 ) as a function of the intensity reflectance @xmath0 , for the glass disks ( @xmath138 ) immersed in the air ( @xmath139 ) . for this medium ,
[ f3 ] shows the diffusion constant @xmath117 ( in units of the disk radius @xmath2 times the velocity of light @xmath137 ) as a function of the packing fraction @xmath21 , for the intensity reflectances @xmath151 and @xmath152 .
the errorbars reflect the the standard deviations when we average over all diffusion constants for different starting positions and angles .
we also performed simulations for the other examples ( @xmath156 ) , ( @xmath157 ) , @xmath158 ) , etc . , but the results are not shown here .
we observed the overall agreement between the numerical results and our theoretical value for @xmath117 .
quite remarkably , eq . ( [ lkj0 ] ) involves no free parameters , but reasonably agrees with the numerical results in a wide range of @xmath150 , @xmath0 , @xmath28 , and @xmath159 .
diffusing - wave spectroscopy provides invaluable information about the static and dynamic properties of granular media@xcite and foams@xcite .
the transport - mean - free path @xmath1 in terms of the microscopic structure , however , remains to be fully elucidated . in this paper
, we consider a _
simple _ model for photon diffusion in a two - dimensional packing of disks .
our _ analytical _ result for @xmath1 provides new insights into the light transport , and promotes more realistic models .
we have studied the photon s persistent random walk in a two - dimensional packing of monodisperse disks .
we employed ray optics to follow a light beam or photon as it is reflected by the disks .
we used a constant intensity reflectance @xmath0 .
moreover we assumed that on hitting a disk , the incident and transmitted rays have the same direction . to achieve a better understanding of photon diffusion in granular medium and wet foams
, we are extending our studies by considering fresnel s formulae for the intensity reflectance , snell s law of the refraction , and the three - dimensional packing of polydisperse spheres .
we are also improving our estimate of @xmath1 by taking into account the distribution of photons step length , and the transport velocity of photons @xcite .
we discuss these points in the following .
two_-dimensional systems offer a rich and unexpected behavior .
for an experimental observation of photon diffusion in a two - dimensional packing , a set of parallel fibres can be used .
photons injected in a plane perpendicular to the axis of fibres perform a planar diffusion .
one can also drill parallel cylinders in a host medium , and fill all the cylinders with a liquid : @xmath1 dependence on the refractive index @xmath28 can be measured .
note that a ray maintains its polarization state on hitting a disk , and fresnel s intensity reflectance depends on the polarization state : quite interesting , the transport - mean - free paths for the transverse electric and transverse magnetic polarizations , are different .
inspired by the rich optics of a two - dimensional packing and following a step - by - step approach to a real system , our attention is naturally directed to estimate @xmath1 of a _ three_-dimensional granular medium or wet foam .
we speculate that one can multiply eq .
( [ mainresult ] ) with a factor of about @xmath102 to estimate the transport - mean - free path for a three - dimensional packing .
apparently , only a detailed study will approve or disapprove our speculation which is based on the following observation : to understand the role of liquid films for light transport in dry foams , we studied the two- and three - dimensional voronoi foams @xcite .
the interesting result is that the transport - mean - free path for these dry foams are determined by the same factor @xmath160 for a constant intensity reflectance @xmath0 in spite of the difference in dimension : @xmath161 and @xmath162 , where @xmath2 denotes the average cell radius@xcite .
our `` mean - field '' theory presented in sec .
[ aa ] relies on the _ average _ step lengths @xmath38 and @xmath39 .
we note that @xmath163 and the probability distribution of the incidence angle @xmath31 is @xmath164 , see appendix [ app1 ] and fig .
[ f1](b ) .
it follows that the probability distribution of @xmath165 is @xmath166 , @xmath167 , and @xmath168 .
[ gg5 ] delineates the probability distribution @xmath169 as a function of @xmath170 for various packing fraction @xmath21 . after reaching its pronounced maximum
, the distribution @xmath169 decays exponentially . the pioneering work of heiderich et al .
@xcite suggests that the broadness of @xmath171 and @xmath169 affects the value of @xmath1 by a factor about @xmath102 . as a function of @xmath170 for various packing fraction @xmath21
inset : the same plot in the logarithmic scale .
note that @xmath169 decays exponentially . ] in two - dimensional space , @xmath1 and @xmath117 are related as @xmath172 , where @xmath131 is the transport velocity of light . in a medium composed of dielectric spheres comparable to the light wavelength @xmath173 ( mie scatterers ) ,
the transport velocity differs by an order of magnitude from the phase velocity@xcite .
when spheres are much larger than the light wavelength , the difference between the two velocities becomes unimportant .
however , our first approximation to @xmath131 [ eq . ( [ lkj2 ] ) ] can be improved by considering the real ( rather than infinite ) size parameter of disks , and the pair - correlation function of disks .
we assumed that on hitting a disk , the incident and transmitted rays have the same direction .
note that in the case @xmath174 , a photon moving in the host medium and hitting the disks , experiences an average scattering angle @xmath175 ( in radians ) due to the reflections .
taking into account inequality of the incidence and refraction angles , and the total internal reflection of rays with incidence angle greater than @xmath176 , the average scattering angle due to the transmissions is @xmath177 f(\gamma ) d \gamma = 0.132 $ ] ( in radians ) , where @xmath178 and @xmath179 are assumed .
reflections are more efficient than transmissions in randomizing the direction of photons .
thus the assumption that the incident and transmitted rays have the same direction , leads to a plausible estimate of @xmath1 .
fresnel s intensity reflectance depends on the incidence angle and the polarization state of the light .
however an average ( constant ) intensity reflectance @xmath0 may describe the photon diffusion . to estimate @xmath0 , we first consider the case @xmath180 and define the critical angle @xmath176 .
the probability distribution of the incidence angles @xmath41 and @xmath31 are @xmath181 and @xmath182 .
we define @xmath183 . here
@xmath184 indicates a ray incident from the host medium onto the disk .
the incident electric field can be decomposed into a component parallel ( @xmath185 ) to the plane of incidence , and a component perpendicular ( @xmath33 ) to this plane . to address the photon diffusion in a _
three_-dimensional random packing , as reflectance @xmath186 we have taken an average over the @xmath185 and @xmath33 polarizations : @xmath187 $ ] .
note that for @xmath188 , the fresnel s intensity reflectances are @xmath102 .
similarly , we define @xmath189 f'(\gamma ' ) d \gamma'$ ] , and then use @xmath190 in eq .
( [ mainresult ] ) to estimate @xmath1 . in the case
@xmath191 , the appropriate critical angle is @xmath192 . here
@xmath193 and @xmath194 .
the fact that @xmath195 ensures that a photon moving in the disk is able to enter the host medium , since the fresnel s intensity reflectance is @xmath102 for @xmath196 .
an extension of appendix [ app1 ] and numerical simulations indicate that the probability distribution of the incidence angles @xmath41 and @xmath31 are @xmath197 and @xmath198 , respectively .
we define @xmath199 f(\gamma ) d \gamma$ ] and @xmath200 f'(\gamma ' ) d \gamma'$ ] to estimate @xmath201 . in the limit @xmath202
we find @xmath203 . in this limit
the photon transport is ballistic and eq .
( [ mainresult ] ) correctly predicts @xmath204 . of our model wet foam , as a function of the liquid volume fraction @xmath10 .
@xmath1 is measured in units of the bubble radius @xmath2 .
@xmath205 and @xmath206 are assumed .
our _ analytical _ result agrees well with the relation @xmath207 .
numerical simulations of ref.@xcite are well described by @xmath208 ] for the glass disks ( @xmath138 ) immersed in the air ( @xmath139 ) , we estimate @xmath209 .
thus for a packing fraction @xmath210 , we find @xmath211 .
for the glass disks immersed in the water ( @xmath206 ) , we find @xmath212 and @xmath213 .
our transport - mean - free paths are smaller than the experimental values @xcite by a factor of about @xmath214 .
now we consider a simple model for wet foams .
for the air bubbles ( @xmath179 ) immersed in the water ( @xmath206 ) , we estimate @xmath215 .
figure [ f4 ] delineates @xmath1 ( in units of @xmath2 ) as a function of the liquid volume fraction @xmath216 . from fig .
[ f4 ] we find that in the range @xmath24 , our analytical result agrees well with the relation @xmath217 using the hybrid lattice gas model for two - dimensional foams and fresnel s intensity reflectance , sun and hutzler performed numerical simulation of photon transport @xcite . for @xmath218
, their numerical results can be fitted to @xmath219 .
[ f4 ] compares our analytic prediction with the numerical result of ref .
again our analytic estimate is smaller than the numerical simulations by a factor of about @xmath214 .
quite remarkably , our _ analytic _ estimate of the transport - mean - free path @xmath1 quoted in eq .
( [ mainresult ] ) , sheds some light on the empirical law of vera et al . and the numerical simulation of sun and hutzler : @xmath3 is a _ linear _ function of @xmath26 , see eqs .
( [ empi ] ) and ( [ ourempi ] ) . for a better understanding of the empirical law ( [ empi ] )
, we aim at a more realistic model which not only considers fresnel s intensity reflectance with its significant dependence on the incidence angle , but also the broad distribution of photons step length .
also an extension to the three - dimensional packing of polydisperse spheres is envisaged .
we thank iran s ministry of science , research and technology for support of the parallel computing facilities at iasbs under grant no .
1026b ( 503495 ) . mf.m . and z.s .
appreciate iranian telecommunication research center ( itrc ) for financial support .
to find the probability distribution of the random variable @xmath41 , we assume that the impact parameter @xmath33 in fig .
[ f1](c ) has a uniform distribution in the interval @xmath222 $ ] . in other words , we assume that the number of incident rays with an impact parameter less than @xmath33 is _ proportional _ to @xmath33 .
the cumulative distribution function @xmath223 is then @xmath224 .
it follows that @xmath225 now we consider path of photons inside the disk , see fig .
. each ray can be characterized by its distance @xmath35 from the center of the disk .
we assume that the random variable @xmath35 has a uniform distribution in the interval @xmath222 $ ] .
since @xmath37 , the cumulative distribution function is @xmath226 , and @xmath227 further numerical simulations of our model confirm these analytical results for @xmath64 and @xmath221 .
b. a. van tiggelen , r. maynard , and a. heiderich , phys .
lett . * 77 * , 639 ( 1996 ) ; h. stark and t. c. lubensky , _ ibid . _ * 77 * , 2229 ( 1996 ) ; b. a. van tiggelen and h. stark , rev .
* 72 * , 1017 ( 2000 ) .
l. margerin , m. campillo , and b. a. van tiggelen , geophys .
* 145 * , 593 ( 2001 ) ; r. hennino , n. trgours , n. m. shapiro , l. margerin , m. campillo , b. a. van tiggelen , and r. l. weaver , phys .
rev . lett.*86 * , 3447 ( 2001 ) ; r. snieder , a. grt , h. douma , and j. scales , science**295 * * , 2253 ( 2002 ) ; n. trgours and b. a. van tiggelen , phys . rev . e**66 * * , 036601 ( 2002 ) ; m. campillo and a. paul , science * 299 * , 547 ( 2003 ) ; e. larose , l. margerin , b. a. van tiggelen , and m. campillo , phys . rev .
lett . * 93 * , 48501 ( 2004 ) ; e. larose , j. de rosny , l. margerin , d. anache , p. gouedard , m. campillo , and b. a. van tiggelen , phys .
e * 73 * , 016609 ( 2006 ) .
note the other possibility : a photon moving in the host medium along the direction @xmath228 may be reflected into the direction @xmath43 .
however , @xmath229 .
the factor @xmath230 in the first term of eq .
( [ master ] ) accounts for the fact that movement in the directions @xmath231 have the same probability . in ref .
@xcite , two - dimensional voronoi structure is used to model _ dry _ foams .
on hitting a liquid film , a photon either reflects with a probability @xmath0 or continues its ballistic motion with a probability @xmath30 .
the average step length of a photon is about the average cell radius @xmath2 . in our model
_ wet _ foams , the photons reflect or transmit on hitting the bubbles .
note that the velocity of light inside and outside the bubbles are different .
the average step length of a photon not only depends on the bubble radius , but also on the liquid volume fraction , see eq .
( [ pppp ] ) .
moreover , @xmath0 , @xmath10 , @xmath28 and @xmath29 determine the time that a photon spends inside the bubbles , see eq .
( [ exacting ] ) .
these points explain the origin of the difference between the expressions for @xmath1 in a two - dimensional voronoi structure and a packing of disks .
|
we study photon diffusion in a two - dimensional random packing of monodisperse disks as a simple model of granular media and wet foams .
we assume that the intensity reflectance of disks is a constant @xmath0 .
we present an _
analytic _
expression for the transport - mean - free path @xmath1 in terms of the velocity of light in the disks and host medium , radius @xmath2 and packing fraction of the disks , and the intensity reflectance . for the glass beads immersed in the air or water
, we estimate transport - mean - free paths about half the experimental ones . for the air bubbles immersed in the water ,
@xmath3 is a _ linear _ function of @xmath4 , where @xmath5 is the liquid volume fraction of the model wet foam .
this throws new light on the empirical law of vera et .
al [ applied optics * 40 * , 4210 ( 2001 ) ] , and promotes more realistic models .
|
inventory management is of great economic importance to industry , but forecasting demand for spare parts is difficult because it is _ intermittent _ : in many time periods the demand is zero .
this type of demand occurs in several industries , for example in aerospace and military inventories from which spare parts such as wings or jet engines are infrequently required .
various methods have been proposed for forecasting , some simple and others statistically sophisticated , but relatively little work has been done on intermittent demand .
most work in this area is influenced by that of @xcite , who first separated the forecasting of demand size and inter - demand interval .
a survey of forecasting methods for spare parts is given in @xcite . to choose a good forecasting method we can test the alternatives empirically on demand series to see which gives the smallest error .
for this we require an _ error measure _ ( or _ accuracy measure _ ) .
unfortunately , there is no general agreement on which of the many existing error measures is best .
forecasting methods have been extensively compared on real and simulated data in the well - known m- , m2- and m3-competitions @xcite using several error measures .
however , these forecasting competitions did not deal specifically with intermittent demands , so the experience gained from these competitions can not be used as a guide . for intermittent demand series
some error measures are inapplicable because division by zero leads to infinities , but there are still several possibilities .
this is an important issue because if researchers are free to choose from a large set of measures then their results are likely to be incomparable .
moreover , there is a temptation to choose measures that give desired results @xcite , making experiments less objective . in 1992 the editor of the international journal of forecasting
wrote that the choice of error measure _ is not a matter of personal preference _ and urged researchers to follow contemporary recommendations @xcite . in this paper
we examine the suitability of known error measures for intermittent demand , and propose new improved measures .
section [ background ] provides background , demonstrates anomalous behaviour in existing error measures , and proposes new measures .
section [ experiments ] evaluates the new measures on simulated data .
section [ conclusion ] concludes the paper .
in this section we provide some necessary background and describe our contribution .
section [ methods ] describes the relevant forecasting methods , section [ measures ] surveys known error measures , section [ ranking ] argues that these measures can rank forecasters incorrectly , and section [ newmeasures ] proposes new measures .
first we describe the forecasting methods that will be used in the paper . _ single exponential smoothing _ ( ses ) computes a smoothed series @xmath0 via the formula @xmath1 where @xmath2 is a _
smoothing parameter_. the smaller the value of @xmath3 the less weight is attached to the most recent observations .
an up - to - date survey of exponential smoothing algorithms is given in @xcite .
they perform remarkably well , often beating more complex approaches @xcite .
however , ses is known to perform poorly on intermittent demand , at least under some error measures .
the standard method for handling intermittency is _
croston s method _
@xcite which applies ses to the non - zero demand sizes @xmath4 and inter - demand intervals @xmath5 independently , using smoothing factors @xmath3 and @xmath6 respectively . given
smoothed demand @xmath0 and smoothed interval @xmath7 at time @xmath8 , the forecast is @xmath9 . both @xmath0 and @xmath7 , and hence @xmath10 ,
are updated at each time @xmath8 for which @xmath11 .
alternative versions were proposed by @xcite and we shall use the variant of syntetos & boylan @xcite which is known to have low bias and variance on stochastic demand .
we also mention the _ random walk _
method ( rw ) , also known as the _ naive method _ : take the previous period s demand as a forecast . though rw is a rather trivial forecaster , it is often used as a baseline for evaluating other methods .
finally we mention the forecaster that always forecasts 0 , which following @xcite we call zf .
it was proposed by croston , mentioned by @xcite and studied by @xcite .
next we survey error measures , largely based on @xcite .
no one error measure is generally accepted as useful on intermittent demand , and opinion is highly divided @xcite .
a common compromise is to use more than one measure as in the forecasting competitions , and @xcite recommend using different measures for different types of demand .
the most common are : * mean [ signed ] error ( me ) : mean(@xmath12 ) * mean square error ( mse ) : mean(@xmath13 ) * root mean square error ( rmse ) : @xmath14 * mean absolute error ( mae ) : mean(@xmath15 ) * median absolute error ( mdae ) : median(@xmath15 ) where @xmath12 is the error @xmath16 .
these are useful for comparing methods on one series , but not for comparing over several series .
doubt has been cast on the suitability of mae for intermittent demand @xcite .
these are also popular : * mean absolute percentage error ( mape ) : mean(@xmath17 ) * median absolute percentage error ( mdape ) : median(@xmath17 ) * root mean square percentage error ( rmspe ) : @xmath18 * root median square percentage error ( rmdspe ) : @xmath19 * symmetric mean absolute percentage error ( smape ) : @xmath20 * symmetric median absolute percentage error ( smdape ) : @xmath21 where @xmath22 .
the last two measures are motivated by the fact that mape and mdape penalise positive errors more than negative ones . however , percentage errors are undefined if any @xmath23 ( and if @xmath24 in the last two cases ) and have very skewed distributions when @xmath25 .
it is also pointed out in @xcite that they assume a meaningful zero , which is not the case for some data such as temperatures . despite these drawbacks mape
is recommended by most textbooks and was the main error measure used in the m - competition , while mdape is recommended by @xcite , and smape and smdape were used in the m3-competition .
it is pointed out by @xcite that many commercial software packages report a mape even when a series contains zeros , although the mape is technically undefined in this case .
this is done by simply excluding periods with zero demands , which does not reflect the true errors of a forecast .
we shall denote this mape variant by imape .
we may also scale by using errors from other measures : * mean relative absolute error ( mrae ) : mean(@xmath26 ) * median relative absolute error ( mdrae ) : median(@xmath26 ) * geometric mean relative absolute error ( gmrae ) : gmean(@xmath26 ) where @xmath27 and @xmath28 is the error from a baseline method which is often rw .
gmrae is also known as relative geometric root mean square and has desirable statistical properties @xcite .
it is used by @xcite , and @xcite recommends the use of relative error - based measures .
it has been proposed for intermittent demand in particular @xcite . however , these measures have the drawback of infinite variance because @xmath28 can be arbitrarily small @xcite . in the particular case of intermittent demand with rw as baseline
, @xmath28 is often zero so these measures are undefined .
extreme values can be trimmed @xcite but this introduces some arbitrariness @xcite . instead of computing an absolute quantity to measure the accuracy of a method
, we may compare it with another method .
this can be done for many types of error measure , for example * relative mean absolute error ( relmae ) : mae / mae@xmath29 * relative mean squared error ( relmse ) : mse / mse@xmath29 * relative root mean squared error ( relrmse ) : rmse / rmse@xmath29 * etc where mae@xmath29 , mse@xmath29 and rmse@xmath29 are the mae , mse and rmse of a baseline measure .
the most popular baseline is rw , in which case relrmse is theil s u2 statistic @xcite and @xmath30 is thompson s lmr measure @xcite .
relmae was recommended for intermittent demand by @xcite and called cummae by @xcite .
it is unlikely for these measures to become infinite , because the denominator is only zero if the baseline forecaster gives perfect results .
another relative measure is percent better ( pb ) where a method is compared to another , usually rw , by how often its absolute error is smaller .
this is recommended by @xcite .
a related measure is percent best ( pbt ) which compares several methods and computes the percentage of times each is most accurate .
a drawback with pb and pbt is that they give no indication of the size of errors , so one large error is considered to be less serious than two tiny errors .
we mention two of these .
firstly , the mad / mean ratio @xcite , also called the weighted mape , which we shall abbreviate to mmr .
mad ( mean absolute deviation ) is another name for mae so @xmath31 .
secondly , the mase @xcite ( mean absolute scaled error ) defined by @xmath32 where @xmath33 is a scaled error defined by @xmath34 and @xmath35 is the set of sample periods used for forecasting .
mase effectively evaluates a forecasting method against rw .
the only situation in which it is unusable is when all in - sample demands are identical .
other scaled error measures defined analogously to mase include the root mean squared scaled error ( rmsse ) and the median absolute scaled error ( mdase ) .
mase has been argued to be superior to several other methods used in forecasting competitions .
an advantage of mase over mmr is that it is more reliable on demand with seasonality , trends or other forms of non - stationarity .
however , @xcite note that the mase of two series with identical forecasts and identical demands during the forecast horizon will differ if the two series differed in their historical demands .
this is counter - intuitive so mase is not always easy to interpret .
which forecaster is best for intermittent demand ?
there is no universally - agreed ranking but cr is often applied in practice to intermittent demand @xcite , and versions of cr are used in leading statistical forecasting software packages such as sap and forecast pro @xcite so we might expect it to be ranked first . however , there is some debate on this issue .
@xcite investigate whether mape , mse or u2 is the best error measure for intermittent demand , using more than one error measure . they find that zf does surprisingly well , beating ses and cr on lumpy demand under a modified mape , but losing under mse and u2 .
however , @xcite note that zf is of no practical use for inventory control .
@xcite found that ses beat cr on intermittent demand , using more than one error measure , though cr was better if issue point only were considered .
according to @xcite it is hard to conclude from the various studies that cr is best , because the results depend on the data and error measures used .
@xcite also note that there is no conclusive evidence pointing to a best method .
but @xcite conclude that the apparently poor performance of cr in some studies is caused by the use of inappropriate error measures , while @xcite found that cr beats ses on intermittent series .
our position is that it is both reasonable and consistent with current wisdom to rank cr above ses , and ses above zf , on intermittent demand series .
cr has maintained its popularity over several decades , and if practitioners prefer a method based on experience , then any error measure that disagrees with this preference is of little use to them .
we shall therefore take it as axiomatic that any error measure that fails this test , which we denote by cr @xmath36 ses @xmath36 zf , should not be applied to intermittent demand .
of course some researchers will disagree with this position , which is entirely reasonable , but we hope that our work will be of use to those who agree with our ranking axiom . in section [ experiments ]
we shall test known error measures against this axiom .
first we propose several new measures : in fact one for each existing measure , obtained by evaluating forecasts against the _ mean of the underlying stochastic process _ of the demand , which we denote @xmath37 , instead of the point demand @xmath38 .
so instead of the usual error @xmath39 we use @xmath40 , instead of @xmath22 we use @xmath41 , and similarly for baseline measures .
the error @xmath42 measures how well a forecaster ignores noise and estimates the underlying demand rate .
we shall call these _ mean - based _ measures , and denote them by adding the prefix `` m '' to the measure they are based on ( mmae , mmse , etc ) . for artificial data
it is easy to find @xmath37 . for stationary demand
we can use @xmath43 where the mean is either derived analytically or simply computed over the entire series . for non - stationary demand the mean of the stochastic process
is a function of time , but we can still use our knowledge of the data to obtain the dynamic underlying demand rate .
for example the obsolescence experiments of @xcite use artificial data whose non - zero demand probability drops either linearly or abruptly to 0 , whereas demand sizes follow a fixed distribution : in either case we can multiply the current probability by the fixed distribution mean to obtain @xmath37 . on real - world series
the stochastic process is of course unknown , though one can make assume a particular form ( for example a poisson process ) then estimate its parameters .
a simple approach is to estimate the current mean demand via standard techniques used to estimate seasonal components .
it is common to estimate a seasonal component by taking a moving average over a window , stretching forward and backward in time .
we can use the same moving window technique to obtain a smoothed version of the demand series , and use this as @xmath37 .
if the demand series is too short to use a moving window , we can estimate the changing mean by regression . or if we assume demand to be stationary , we can take the series mean as @xmath37 . as with seasonal component estimation
, there are several reasonable approaches .
mean - based error measures have wider applicability than their original counterparts .
percentage errors such as mape are undefined whenever @xmath23 , whereas mmape is only undefined when @xmath44 : for stationary demand @xmath37 is the series mean , which is only zero when @xmath23 for all @xmath8 .
relative error - based measures such as gmrae with rw as baseline are undefined on intermittent demand , because both the demand and rw s forecast are often zero so the rw error @xmath45 ( which is the denominator ) is also zero .
however , a measure such as mgmrae is only undefined when the denominator @xmath46 is zero : the rw forecast @xmath47 will often be zero , but again @xmath48 is only zero when @xmath23 for all @xmath8 .
we now have a large number of new error measures , none of which is likely to give infinite answers on intermittent demand .
this allows us to measure forecasting deviations using absolute or squared values , in scaled or unscaled ways , and on one or multiple demand series . in the next section
we shall evaluate them with respect to our forecaster ranking axiom .
the error measures we compare are selected from the various classes described in section [ measures ] . to represent the scale - dependent measures we use mae , mdae and mse and their mean - based equivalents mmae , mmdae and mmse ; from the percentage errors we use imape and mmape .
note that the mmape of zf is always @xmath49 and the imape of zf is always @xmath50 ( mape is undefined on intermittent demand ) . to represent the relative error - based measures we use only mgmrae with rw as baseline ( gmrae with rw as baseline is undefined for intermittent demand ) . to represent the relative measures we use pb and mpb with rw as baseline .
we start with data based on that used in the experiments of teunter _
demands occur with some probability in each period , hence inter - demand intervals are distributed geometrically , and we use a logarithmic distribution for demand sizes .
geometrically distributed intervals are a discrete version of poisson intervals , and the combination of poisson intervals and logarithmic demand sizes yields a negative binomial distribution , for which there is theoretical and empirical evidence : see for example the recent discussion in @xcite .
_ generate demand data that is nonzero with probability @xmath51 where @xmath51 is either 0.2 or 0.5 , and whose size is logarithmically distributed .
the logarithmic distribution is characterised by a parameter @xmath52 and is discrete with @xmath53= - \ell^k / k \log(1- \ell)$ ] for @xmath54 .
they use two values : @xmath55 to simulate low demand and @xmath56 to simulate lumpy demand .
tables [ results1][results4 ] show best results for ses , cr and zf using @xmath3 and @xmath6 values chosen from @xmath57 .
we initialise the forecasters by choosing arbitrary initial values @xmath58 then running them for @xmath59 periods using demand probability @xmath51 .
results are then computed over @xmath60 time periods . to estimate the stochastic process mean we simply compute the mean of all @xmath60 demands ( including zeros ) .
the results show that mae , mdae and imape are unreliable error measures for some types of intermittent demand because they incorrectly rank the three forecasters .
pb is more reliable but the differences are sometimes very small , and in one case pb ranked cr and ses equally . among existing measures
only mse behaves correctly .
however , all mean - based measures behave correctly , though mpb still scores cr and ses quite similarly .
we also tried geometrically distributed demand sizes as in @xcite , and regular intermittent demand as in @xcite , with similar results .
.results for artificial demand with @xmath61 and @xmath55 [ cols= " < , > , > , > , > , > , > , < , < ,
< , < , < " , ] collectively these results imply that almost all tested error measures are unreliable .
they also imply that other untested error measures are unreliable , because they are monotonic functions of mae or mse and hence rank forecasters in the same way .
these include rmse , relative measures such as relmae and relmse and as their special cases u2 and lmr , and scaled errors such as mmr and mase . hence most existing error measures are inapplicable to intermittent demand , if our ranking axiom is reasonable .
pb was almost correct , but in one case it was unable to distinguish between cr and ses ( see table [ results1 ] ) .
we conclude that all known error measures ( except me which measures bias , not deviation ) are unreliable on some types of intermittent demand , even when they do not incur infinities , so there is currently no reliable way of measuring deviation .
these results reinforce and complement those of @xcite , who show that mae and rmse rank zf above ses ( and above a moving average ) , and ses above cr , on a large data set of intermittent demand from an air force .
our results apply to more error measures and are more easily reproducible , being based on simple artificial data .
but their result shows that the inappropriateness of at least some current error measures extends to real data .
in contrast , our new measures gave correct results in all cases .
based on our experiments , and on inapplicabilities pointed out by other researchers , we make two proposals regarding measures of deviation for intermittent demand .
firstly , we do not recommend _ any _ existing error measures .
secondly , we recommend several new error measures : mmse , mrmse , mmae , mmdae , mmape , mrelmse ( including special cases mu2 and mlmr plus other relative measures such as mrelmae ) , mrelrmse , mmmr , mmase and mgmrae . which of these is best depends on user preference , and considerations such as whether errors are to be compared on one or across several series . in our experiments
we did not evaluate several other possible new measures : mmdape , mrmspe , mrmdspe , msmape , msmdape , mmrae , mmdrae , mrmsse and mmdase , all defined in the obvious way .
we leave their evaluation for future work .
we have shown that almost all known error measures rank forecasting methods incorrectly on some intermittent demand series . given this result , and the well - known fact that several error measures are inapplicable to intermittent demand because of infinities , there is currently no reliable way of measuring forecast deviation errors on such demands . to alleviate this problem , we described a simple way of modifying all known error measures so that they are more widely applicable and behave more correctly .
this yields many new _ mean - based _
error measures that can be used to compare forecasters on intermittent demand .
they are unlikely to be plagued by infinities , and in tests they consistently ranked forecasters correctly .
we have defined a large number of new error measures , and it might be argued that this only adds to the confusion .
but we believe that the improved behaviour and wider applicability of our measures make them worth considering when faced with intermittent demand . to simplify matters
we should perhaps recommend a small number of new measures . based on popularity and the recommendations of experts
, we choose the mean - based analogues of mape , gmrae and u2 .
in future work we shall evaluate other error measures that can be modified by our technique .
we shall experiment with artificial non - stationary data and real - world series .
the statistical properties of the new error measures should be investigated .
finally , the new error measures can also be applied to non - intermittent demand , and we shall evaluate their usefulness using series from the forecasting competitions .
s. makridakis , a. andersen , r. carbone , r. fildes , m. hibon , r. lewandowski , j. newton , e. parzen r. winkler .
the accuracy of extrapolation ( time series ) methods : results of a forecasting competition .
_ journal of forecasting _
* 1*:111153 , 1982 .
s. makridakis , c. chatfield , m. hibon , m. lawrence , t. mills , k. ord , l. f. simmons .
the m-2 competition : a real - time judgmentally based forecasting study . _ international journal of forecasting _ * 9*:523 , 1993 .
a. syntetos , z. babai , d. lengu , n. altay .
distributional assumptions for parametric forecasting of intermittent demand . in : n.
altay & a. litteral ( eds . ) , service parts management : demand forecasting and inventory control , springer verlag , ny , usa , 2011 , pp.3152 .
g. h. k. venkitachalam , d. b. pratt , c. f. deyoung , s. a. morris , m. l. goldstein .
forecasting and inventory planning for parts with intermittent demand a case study .
presented at the industrial engineering research conference , portland , or usa , 2003 .
t. r. willemain , c. n. smart , j. h. shockor , p. a. desautels .
forecasting intermittent demand in manufacturing : a comparative evaluation of croston s method . _
international journal of forecasting _
* 10*(4):529538 , 1994 .
|
to compare different forecasting methods on demand series we require an error measure .
many error measures have been proposed , but when demand is intermittent some become inapplicable , some give counter - intuitive results , and there is no agreement on which is best .
we argue that almost all known measures rank forecasters incorrectly on intermittent demand series .
we propose several new error measures with wider applicability , and correct forecaster ranking on several intermittent demand patterns .
we call these `` mean - based '' error measures because they evaluate forecasts against the ( possibly time - dependent ) mean of the underlying stochastic process instead of point demands .
|
the behavior of charged , fluctuating systems is an old problem in chemistry and physics and is of importance for very diverse disciplines . in this article
we will be concerned with the distribution of counter ions around charged objects , which is experimentally realized whenever an object with dissociable surface groups is brought in contact with water ( or some other high - dielectric solvent ) . in such a situation
, the counter ions will be attracted to the oppositely charged object , but also repelled from other counter ions .
the characteristic feature of charged systems is that the coulomb interaction is long - ranged , which gives these systems very special properties .
since , in general , one counter ion interacts with many different counter ions simultaneously , the mean - field approach is very successful and can be used to describe experiments or simulations quantitatively .
exact solutions of the so - called poisson boltzmann equation , which determines the electrostatic potential distribution on a mean - field level , are available for planar@xcite and cylindrical@xcite geometries . a very readable introduction into the poisson - boltzmann ( pb ) approach is given in ref.@xcite .
there are several factors which contribute to deviations from the pb equation , including additional , short - ranged interactions and solvent effects . in this article
we will consider deviations due to _ fluctuations and correlations_. these effects have been incorporated for the single charged wall by modified pb equations @xcite , by integral equation theories@xcite , and by numerical methods@xcite .
in general , one finds correlation effects to become important for ions of high valency , and packing effects dominate the ion distribution for very bulky ions or high ion concentration . for the case of counter ions or electrolyte solutions between two charged walls ,
there have been a number of studies using integral equations@xcite and monte - carlo methods@xcite .
an exhaustive review has been given by attard@xcite .
the main result of these studies is that correlations can lead to an effective attraction between similarly charged walls for high enough ion valency .
there have been a few attempts of using field theoretic methods to go beyond mean - field theory for charged systems .
here we only mention the works by podgornik and zeks@xcite and by attard and coworkers@xcite , who expanded around the mean - field solution and calculated within additional approximations fluctuation corrections to the interaction between two charged , planar walls .
in this article we consider a system of mobile , point - like counter ions in the presence of a fixed charge distribution , and formulate the field - theoretic framework for going beyond the poisson - boltzmann description .
as has been realized by many others before , the poisson - boltzmann equation constitutes the _ saddle point _ of the exact field theory .
we first reformulate this theory by a loop - wise expansion around the saddle point .
the poisson equation , which relates the electrostatic potential to the counter ion distribution , is obeyed at each level of this loop expansion .
clearly , the counter - ions are not distributed according to the boltzmann weight , and correlations and fluctuations lead to pronounced deviations from the poisson - boltzmann predictions .
the description is considerably facilitated by a legendre transformation , after which we obtain the gibbs potential as a function of a fixed electrostatic potential distribution .
this transformation restricts the number of diagrams to so - called one - particle - irreducible diagrams .
the electrostatic potential is given by the solution of the equation of state .
we then apply this general formalism to the case of counter ions confined to a half space and next to a charged wall .
we calculate without further approximations the one - loop correction to the electrostatic potential distribution and the counter - ion distribution .
due to correlations the counter ions are more densely packed close to the wall than predicted by the pb solution .
this is in accord with previous monte - carlo simulations and with the intuitive expectation , since mean field usually overestimates repulsive interactions due to the neglect of correlations .
the relative importance of fluctuation corrections to the electrostatic potential is measured by the single parameter @xmath1 , where @xmath2 is the surface charge density of the planar surface , @xmath3 is the distance at which two elementary charges interact with thermal energy , and @xmath4 denotes the counter - ion valency
. since the thickness of the counter ion distribution is , according to the mean - field prediction , given by @xmath5 , the average counter - ion concentration follows as @xmath6 and our effective fluctuation parameter can be rewritten as @xmath7 and thus measures the electrostatic interaction energy between two counter ions at their average separation ( in units of the thermal energy ) ; it is thus related to the plasma parameter . for vanishing values of @xmath1 mean field theory becomes exact and fluctuation corrections are unimportant . on the other hand , since the counter ion valency comes in as a cubic power , it is clear that by going from monovalent to divalent ions , fluctuation effects become much more pronounced , in accord with experiments and simulations .
our results , which are obtained only to first order in the loop expansion , become unreliable for large values of @xmath1 . a clear break down of our expansion is indicated by an unphysical negative density of counter ions , which occurs at @xmath8 . in section ii we formulate the general field - theoretic description for an ensemble of fluctuating counter ions in the presence of a fixed charge distribution .
the main steps consist of i ) formulating the initial partition function as a field theory using a hubbard - stratonovich transformation , ii ) going to the grand - canonical ensemble , iii ) expanding the action around the saddle - point , and iv ) performing another legendre transformation after which the gibbs potential depends on the electrostatic potential . in section
iii we show how the electrostatic potential follows from the equation of state , and in section iv we give explicit results for the case of a single charged wall . in section v we calculate the effective interaction between charged test particles , and section vi is devoted to a brief discussion .
the partition function for @xmath9 mobile counter ions of valency @xmath4 , interacting with an arbitrary fixed charge distribution , @xmath10 , reads @xmath11 \exp\left\ { -q^2 \sum_{j > k } v({\bf r}_j - { \bf r}_k ) + q \int { \rm d } { \bf r } \sigma({\bf r } ) \sum_j v({\bf r } - { \bf r}_j ) \right\}.\ ] ] we assume the counter ions to be point - like particles , and only include the coulomb interaction between ions , @xmath12 , where @xmath3 is the bjerrum length defined as @xmath13 .
we also neglect any image charge effects in the present formulation of our theory .
the charge distribution @xmath10 in general also contains test particles , the presence of which allows to calculate the effective interaction between charges , as we will demonstrate in section vi . introducing the particle density operator @xmath14 we can rewrite the partition function as @xmath15 = \frac{1}{n ! } \left [ \prod_{i=1}^n \int { \rm d } { \bf r}_i \right ] \exp\left\ { - \frac{1}{2 } \int { \rm d } { \bf r } { \rm d } { \bf r } ' [ q \hat{\rho}({\bf r } ) - \sigma({\bf r } ) ] v({\bf r}-{\bf r } ' ) [ q \hat{\rho}({\bf r } ' ) -\sigma({\bf r } ' ) ] + \int { \rm d } { \bf r } \hat{\rho}({\bf r } ) h({\bf r } ) \right\}.\ ] ] there is no need to subtract the diagonal terms involving self interactions at this point , because they will be canceled by a one - point renormalization at a later stage . using the generating field @xmath16
, the average counter ion density can be calculated as @xmath17 } { z_n[h ] \delta h({\bf r } ) } \right|_{h=0}.\ ] ] correlation functions can be calculated by taking multiple functional derivatives with respect to @xmath16 .
noting that the operator - inverse of the coulomb interaction is @xmath18 the partition function is after a hubbard - stratonovich transformation given by @xmath15 = \frac{1}{n ! } \int \frac{{\cal d}\phi}{z_0 } \ ; \exp \left\{- \int { \rm d}{\bf r } \left [ \frac{1}{2 \tilde{\ell}_b } ( \nabla \phi)^2 + \frac{\imath \sigma({\bf r } ) \phi({\bf r})}{q } \right ] \right\ } \left [ \int { \rm d}{\bf r } { \rm e}^{- \imath \phi({\bf r } ) + h({\bf r } ) } \right]^n,\ ] ] where we introduced the rescaled bjerrum length @xmath19 . in the above expression , @xmath20 denotes the partition function of the coulomb operator , @xmath21 . using the definition eq .
( [ density ] ) , the average particle density turns out to be @xmath22 and the normalization property of the density distribution , @xmath23 , is self - evident .
the two - point cumulant correlation function ( for @xmath24 ) follows as @xmath25 } { z_n[h ] \delta h({\bf r } ) \delta h({\bf r ' } ) } \right|_{h=0 } \nonumber \\ & = & n(n-1 ) \left\langle
\frac{{\rm e}^{- \imath \phi({\bf r } ) - \imath \phi({\bf r ' } ) } } { \left [ \int { \rm d}{\bf r } { \rm e}^{- \imath \phi({\bf r } ) } \right]^2 } \right\rangle , \end{aligned}\ ] ] and higher - order correlation function can be calculated in a similar manner .
the functional form of the particle distribution , eq.([density2 ] ) , shows that @xmath26 is the reduced electric potential .
the partition function is brought into a more manageable form by going to the grand - canonical ensemble , @xmath27 \equiv \sum_{n=0}^{\infty } \lambda^ { n } z_n[h ] = \int \frac{{\cal d}\phi}{z_0 } \ ; \exp \left\{-\ell { \cal h}[\phi , h ] + \imath \ell \int { \rm d}{\bf r } \phi({\bf r } ) \varrho ( { \bf r } ) \right\}.\ ] ] where @xmath28 is the fugacity of the counter - ions , and the hamiltonian is defined as @xmath29 \equiv \int { \rm d}{\bf r } \left [ \frac{1}{2 \tilde{\ell}_b } ( \nabla \phi)^2 + \frac{\imath \sigma({\bf r } ) \phi({\bf r})}{q } - \lambda { \rm e}^{- \imath \phi({\bf r})+ h({\bf r } ) } \right]\ ] ] and we added a source term to the partition function in such a way that one can calculate directly the reduced electrostatic potential @xmath30 } { \ell \delta \varrho({\bf r})}.\ ] ] in the grand - canonical partition function , eq.([zlambda ] ) , we arbitrarily multiplied the action by a constant @xmath31 which plays the role similar to the inverse planck s constant in front of the action integral in quantum field theory .
this constant serves as an expansion parameter in our systematic treatment of fluctuation effects and counts the number of diagrammatic loops .
the fugacity @xmath28 is related to the total number of ions by @xmath32 } { \ell \partial \lambda } = \langle n \rangle = \lambda \int { \rm d}{\bf r}\ ; \left\langle { \rm e}^{- \imath \phi({\bf r } ) } \right\rangle\ ] ] which establishes a useful relation between @xmath28 and @xmath9 .
the particle density is obtained from the grand - canonical partition function eq.([zlambda ] ) using the definition eq.([density ] ) and reads @xmath33 likewise , the two - point correlation function ( for @xmath24 ) reads @xmath34 which is simpler than the canonical form , eq.([corr1 ] ) ; higher - order correlation function can be calculated similarly . in the following
we will be interested in calculating the gibbs potential @xmath35 $ ] which depends on the reduced electric potential @xmath36 ; it is related to @xmath37 $ ] by a legendre transform @xmath38 + \ln z[\varrho , h]= \ell \int { \rm d}{\bf r}\ ; \psi({\bf r } ) \varrho ( { \bf r}).\ ] ] using the definition ( [ defpsi ] ) we obtain from the definition of the legendre transform ( [ legendre ] ) the inverse relation @xmath39}{\ell \delta \psi({\bf r})}.\ ] ] we will perform the legendre transformation using a loop expansion , using methods developed by schwinger@xcite and closely follow the notation of brzin , le guillou , and zinn - justin@xcite . to that end , we expand the hamiltonian around a fixed value of the potential , the so - called classical field , denoted as @xmath40 . at this point
, we do not specify what this classical field is .
we write @xmath41 and thus obtain the formal expansion @xmath42 - \imath \int { \rm d}{\bf r } \phi({\bf r } ) \varrho ( { \bf r } ) = { \cal h}[- \imath \psi_{\rm cl } ] -\int { \rm d}{\bf r } \varrho ( { \bf r})\left[\psi_{\rm cl}({\bf r } ) + \imath\chi({\bf r } ) \right ] + \sum_{j=1}\frac{1}{j ! } \int { \cal h}^{(j)}(\{{\bf r}_j\ } ; \psi_{\rm cl } ) \chi({\bf r}_1)\cdots \chi({\bf r}_j ) { \rm d}{\bf r}_1 \cdots { \rm d}{\bf r}_j.\ ] ] the vertex functions @xmath43 are defined by @xmath44 } { \delta \phi({\bf r}_1)\cdots \delta \phi({\bf r}_j ) } \right|_{\phi=-\imath \psi_{\rm cl}}\ ] ] and of course depend on the function @xmath40 around which one expands .
the function @xmath45 $ ] can now be evaluated by standard saddle - point methods , treating @xmath46 as the expansion parameter .
the saddle point @xmath47 is defined by @xmath48}{\delta \phi({\bf r } ) } - \imath \varrho({\bf r } ) \right|_{\phi=-\imath \psi_{\rm sp}[\varrho]}=0\ ] ] and depends on the source @xmath49 . using the definition of the two - point propagator @xmath50 ,
@xmath51 we can diagrammatically expand the expression for @xmath37 $ ] and obtain up to two loops @xmath52 & = & \ell \int { \rm d}{\bf r } \left\ { \frac{1}{2 \tilde{\ell}_b } ( \nabla \psi_{\rm sp})^2 -\frac{\sigma({\bf r } ) \psi_{\rm sp}({\bf r})}{q } + \lambda { \rm e}^{- \psi_{\rm sp}({\bf r})+h({\bf r } ) } + \varrho ( { \bf r})\psi_{\rm sp}({\bf r } ) \right\ } - \frac{1}{2 } \ln \det { \cal h}^{(2)}[\psi_{\rm sp } ] \nonumber \\ & + & \frac{1}{\ell } \left\ { -\frac{1}{8 } \int { \rm d}{\bf r}_1 \cdots { \rm d}{\bf r}_4 \ ; { \cal h}^{(4)}(\{{\bf r}_4\};\psi_{\rm sp } ) g({\bf r}_1,{\bf r}_2;\psi_{\rm sp } ) g({\bf r}_3,{\bf r}_4;\psi_{\rm sp } ) \right .
\nonumber \\ & & \;\;\;\;\;\;\left .
+ \int { \rm d}{\bf r}_1\cdots { \rm d}{\bf r}_3 { \rm d}{\bf r}'_1\cdots { \rm d}{\bf r}'_3 \ ; { \cal h}^{(3)}(\{{\bf r}_3\};\psi_{\rm sp } ) { \cal h}^{(3)}(\{{\bf r}'_3\};\psi_{\rm sp } ) \right . \nonumber \\ & & \left . \left [ \frac{1}{8 }
g({\bf r}_1,{\bf r}_2;\psi_{\rm sp } ) g({\bf r}'_1,{\bf r}'_2;\psi_{\rm sp } ) g({\bf r}_3,{\bf r}'_3;\psi_{\rm sp } ) + \frac{1}{12 } g({\bf r}_1,{\bf r}'_1;\psi_{\rm sp } ) g({\bf r}_2,{\bf r}'_2;\psi_{\rm sp } ) g({\bf r}_3,{\bf r}'_3;\psi_{\rm sp } ) \right ] \right\}.\end{aligned}\ ] ] the one - loop diagram and the two - loop diagrams are schematically represented in fig.1 . in order to perform the legendre transformation , we need the loop expansions for @xmath36 and @xmath53 , which are obtained from the definition ( [ defpsi ] ) and the saddle - point equation ( [ saddle ] ) , respectively .
these expressions depend on the saddle - point potential @xmath54 .
we then have to expand all functions and vertices around the saddle point and reexpress the potential dependence as a function of the expectation value @xmath55 . by inserting these expression into the legendre transform
, we obtain the loop expansion of the gibbs potential , @xmath56 & = & -\ell \int { \rm d}{\bf r } \left\ { \frac{1}{2 \tilde{\ell}_b } ( \nabla \psi)^2 - \frac{\sigma({\bf r } ) \psi({\bf r})}{q } + \lambda { \rm e}^{- \psi({\bf r})+h({\bf r } ) } \right\ } + \frac{1}{2 } \ln \det { \cal h}^{(2)}[\psi ] \nonumber \\ & - & \frac{1}{\ell } \left\ { -\frac{1}{8 } \int
{ \rm d}{\bf r}_1 \cdots { \rm d}{\bf r}_4 \ ; { \cal h}^{(4)}(\{{\bf r}_4\};\psi ) g({\bf r}_1,{\bf r}_2;\psi ) g({\bf r}_3,{\bf r}_4;\psi ) \right .
\nonumber \\ & & \;\;\;\;\ ; \left .
+ \frac{1}{12 } \int { \rm d}{\bf r}_1\cdots { \rm d}{\bf r}_3 { \rm d}{\bf r}'_1\cdots { \rm d}{\bf r}'_3 \ ; { \cal
h}^{(3)}(\{{\bf r}_3\};\psi ) { \cal h}^{(3)}(\{{\bf r}'_3\};\psi ) \right .
\nonumber \\ & & \left .
\ ; \;\;\;\;\;\ ; g({\bf r}_1,{\bf r}'_1;\psi ) g({\bf r}_2,{\bf r}'_2;\psi ) g({\bf r}_3,{\bf r}'_3;\psi ) \right\}\end{aligned}\ ] ] the effect of the legendre transformation is to cancel all one - particle - reducible diagrams , as expected . for the present case ,
the two - loop diagram to the right in fig.1 is removed from the expansion .
the equation of state is defined as @xmath57}{\delta \psi({\bf r})}=0\ ] ] and completely determines the electrostatic potential . in the following
, we will restrict ourselves to the one - loop order . using the explicit form of the two - point vertex function , which follows from the definitions eq.([ham ] ) and eq.([vertdef ] ) ,
@xmath58= \left ( -\frac{\nabla^2}{\tilde{\ell}_b } + \lambda { \rm e}^{-\psi({\bf r } ) } \right ) \delta({\bf r}-{\bf r'}),\ ] ] the equation of state can be explicitly written as @xmath59 note that from now one , we suppress the dependence of the two - point correlation function @xmath60 on the electrostatic potential distribution @xmath55 .
the correlation function is determined by the equation @xmath61 finally , the number of mobile ions can be calculated according to eq.([lambda / n ] ) and is given by @xmath62 the equation of state can be solved by a systematic expansion of all quantities which are to be determined in inverse powers of the loop parameter @xmath31 . at the one - loop order , only the leading term of @xmath63 contributes and therefore an expansion of @xmath63 itself is unnecessary . the electric potential and the fugacity have to be expanded to the next - leading order , @xmath64 @xmath65 the equation of state splits into two separate equations , the zero - loop equation @xmath66 which is the ordinary poisson boltzmann equation , and the next - leading correction , @xmath67 the correlation function is determined by @xmath68 the fugacity is at the zero - loop level determined by @xmath69 and at the next - leading level by @xmath70 combining the equations for @xmath71 and @xmath63 , eqs.([psi1def ] ) and ( [ gdef ] ) , we obtain for the correction to the electrostatic potential @xmath72,\ ] ] which contains an implicit dependence on @xmath71 on the right - hand side through the dependence of @xmath73 on @xmath71 , see eq.([lambda1a ] ) .
algebraically solving eqs.([lambda1a ] ) and ( [ psi1a ] ) we obtain @xmath74 } { 1-(\lambda_0/y ) \int { \rm d}{\bf r } { \rm d}{\bf r'}\ ; { \rm e}^{-\psi_0({\bf r})-\psi_0({\bf r ' } ) } g({\bf r}',{\bf r})},\ ] ] where we used the short - hand notations @xmath75 and @xmath76 the one - loop correction to the electrostatic potential is therefore determined by eq.([psi1a ] ) in conjunction with the equations determining @xmath63 and @xmath73 , eqs.([gdef ] ) and ( [ lambda1a ] ) .
in the following we consider the case of a charged wall which is impenetrable for the counterions .
the fixed charge distribution is given by @xmath77 where @xmath2 is the surface charge density at the charged wall .
the solution for the one - dimensional poisson - boltzmann equation , eq.([psi0def ] ) , is @xmath78 where @xmath79 is the gouy - chapman length .
enforcement of the normalization condition , eq.([lambda0 ] ) , leads to a relation between the surface charge density @xmath2 and the gouy - chapman length , @xmath80 , which can be further transformed to @xmath81 and @xmath82 , the standard results . due to lateral invariance
, the green s function @xmath83 can be fourier transformed in the directions perpendicular to the surface ; the equation for the green s function , eq.([gdef ] ) , becomes @xmath84 where @xmath85 are the fourier variables in the lateral directions , and @xmath86 .
the potential @xmath87 is given by eq.([psi0 ] ) for @xmath88 .
since the ions can not penetrate into the wall , the counter - ion density is zero for @xmath89 , or , equivalently , we put @xmath90 for @xmath89 .
the solution can be obtained in a straightforward manner and reads @xmath91\ ] ] for @xmath92 .
the solution for @xmath93 follows from eq.([gsol ] ) by an interchange of the arguments @xmath94 and @xmath95 . in the limit @xmath96
we obtain @xmath97 for @xmath92 ( and with the coordinates interchanged for @xmath98 ) .
the self interaction is defined as the equal - point green s function and follows from the partially fourier - transformed green s function as @xmath99 the integral to be solved is , subtracting the ( infinite ) coulomb self energy @xmath100 , @xmath101.\ ] ] the integral can be performed by some intermediate transformations , and the result can be written in a scaling form as @xmath102 where the scaling function @xmath103 is explicitly given by @xmath104(1+ix)^2 - i e^{(1+i)x } e_1[(1+i)x](1-ix)^2 - 4x \right]\ ] ] and plotted in fig.2 .
the function @xmath105 $ ] is the exponential - integral function@xcite .
the prefactor of the scaling function in eq.([gselfsol ] ) , @xmath106 is a measure of the importance of correlation effects and thus the departure from mean - field behavior .
we note that it is this single parameter combination which measures the relative importance of fluctuation corrections .
clearly , the cubic dependence on the counter - ion valency shows that fluctuation effects will become much more dominant for multivalent ions .
as can be seen from fig.2 , the self energy is negative , which reflects the fact that the screening due to counter ions lowers the local free energy of charged particles .
the scaling function @xmath107 shows a minimum at a finite distance from the wall , @xmath108 .
this maximum in screening is closely connected with a maximum in the counter - ion density at about this distance from the wall , as we will show in section iv.c .
the asymptotic behavior for small separations from the wall is @xmath109 and @xmath110 for large separations .
the functions eqs.([gasym1 ] ) and ( [ gasym2 ] ) are shown in figs.2a and 2b as broken lines .
the solution of the equation for @xmath71 is considerably simplified by the fact that , for the specific mean - field potential given in eq.([psi0 ] ) , we find @xmath111 independent of @xmath95 .
it immediately follows from eq.([psi1a ] ) that the integral @xmath112 is zero and , as a consequence , the equation for @xmath71 , eq.([psi1a ] ) , is independent of any constant appearing in the correlation function @xmath63 .
this holds in particular for the coulombic self energy and therefore justifies that we neglected this ( infinite ) constant in the calculation of the self interaction .
this is an example of a renormalization of a one - point correlation function , which otherwise is plagued by the ( infinite ) coulombic self - energy .
the normalization of the bare potential @xmath113 , eq.([psi0 ] ) is @xmath114 the renormalization constant appearing in eq.([psi1a ] ) thus is @xmath115 where the numerical constant is @xmath116 .
our final result for the potential is , according to eq.([psi1a ] ) , @xmath117,\ ] ] where the scaling function @xmath118 is defined by @xmath119 and graphically presented in fig.3 .
the asymptotic behavior for small separations is @xmath120 and for large separations we obtain @xmath121 the asymptotic formulas are shown in fig.3 as broken lines .
the counter ion density @xmath122 can be calculated from the gibbs potential eq.([gamma ] ) by taking a derivative with respect to the generating field @xmath16 .
the result is , including the one - loop correction , given by @xmath123 and leads via integration to the total ion number @xmath9 , eq.([expectn ] ) , as it should .
subtracting the equation of state , eq.([stateeq ] ) , we obtain the poisson equation , @xmath124 in fact , the equation eq.([poisson ] ) is exact to all orders in the loop expansion , since the equation of state contains exactly the same terms as does the particle density and thus leads to an exact cancelation of all higher - loop terms .
this is so because in the hamiltonian eq.([ham ] ) , the generating field @xmath16 enters in the same way as the fluctuating potential @xmath125 , except for the linear and quadratic terms .
expanding the counter ion density in powers of the loop parameter , @xmath126 and using the loop - wise expansions of the fugacity @xmath28 and the electrostatic potential @xmath36 introduced in section iii , the zero - loop ( or mean - field ) result for the counter ion density is @xmath127 and the one - loop correction reads @xmath128 using the explicit solution for the single charged wall , eq.([psi0 ] ) , and the result for the mean - field fugacity , @xmath129 , the zero - loop density can be written as @xmath130 and the one - loop correction reads @xmath131 where the scaling function @xmath132 can be written in terms of the previously defined scaling functions and reads @xmath133 in fig.4 we plot the scaling function @xmath132 .
the density correction vanishes at the charged wall , and for small separations the asymptotic behavior follows from eqs .
( [ gasym1 ] ) and ( [ fasym1 ] ) as @xmath134 the density corrections is positive for @xmath135 , i.e. for distances from the wall smaller than the gouy - chapman length @xmath136 fluctuations and correlations lead to an enhanced density , quite in accord with expectations : each counter ion is surrounded by a correlation hole , which is neglected on a mean - field level .
mean field theory therefore overestimates the repulsion between counter ions , and therefore underestimates the density close to the charged wall .
this is in qualitative agreement with monte - carlo simulations@xcite .
we note that since the slope of the density correction is infinite close to the wall , see eq.([hasym1 ] ) , our results predict always an initial increase of the counter - ion density as one moves away from the substrate .
the total integral over the density correction vanishes ; the increase in density close to the wall is therefore compensated by a decrease in density further away from the wall , as shown in fig.4b . for large separations from the wall ,
the asymptotic behavior follows from eqs .
( [ gasym1 ] ) and ( [ fasym1 ] ) as @xmath137 in fig .
5 we plot the rescaled density @xmath138 as a function of the rescaled distance from the wall @xmath139 , for four different values of the parameter @xmath140 .
the broken line , for @xmath141 , denotes the pb result , the other three solid lines are for @xmath142 , @xmath143 , and @xmath144 , with the difference to the broken line increasing as the parameter @xmath140 increases .
we see that already for rather small values of @xmath140 , the one - loop result for the counter ion distribution is quite different from the pb result .
the density depression far away from the wall is proportional to the parameter @xmath140 .
it is clear that for very large values of this parameter , the sum of the zero - loop and the one - loop densities will become negative .
this in fact happens at @xmath145 , which clearly sets an upper limit to the applicability of the present one - loop calculation .
it is likely that the one - loop calculation becomes quantitatively inaccurate even for smaller values of this parameter , but the precise value of @xmath140 when this happens can only be judged by a numerical calculation ( such as monte carlo ) or by a higher - loop calculation .
in this section we calculate the effective interaction between charged test particles . as mentioned before ,
test particles can be incorporated into the model by adding them to the charge distribution in eq.([model1 ] ) according to @xmath146 where we have considered a collection of @xmath147 test particles with charges @xmath148 and positions @xmath149 . inserting this shifted charge distribution into the grand - canonical partition function eq .
( [ zlambda ] ) , the reduced test - particle free energy turns out to be @xmath150 using the separation of the fluctuating field @xmath125 into the expectation value @xmath55 and the fluctuations around the expectation value , eq.([sep ] ) , the free energy can be reexpressed as @xmath151 using the loop - expansion of the electrostatic potential , @xmath55 , eq.([psiexp ] ) , and the definition of the self interaction , @xmath152 , eq.([gselfdef ] ) , the free energy becomes on the one - loop level @xmath153 + \frac{q_j^2}{2q^2 } g_{\rm self}(z_j ) \right\ } + \sum_{j < k}^m \frac{q_j q_k}{q^2 } g({\bf r}_j,{\bf r}_k)\ ] ] and thus separates into a single - particle part and a two - particle part .
the single - particle contribution contains the ordinary gouy - chapman potential , @xmath87 , given by eq.([psi0 ] ) , a contribution due to fluctuation - induced changes of the electrostatic potential , @xmath154 , given by eq.([psi1 ] ) , and a contribution due to correlations between counterions and a test particle , @xmath155 , given by eq.([gselfsol ] ) .
since the correlation contribution is attractive and proportional to the square of the test - particle charge , it will be dominant for large test - particle charge and lead to attraction even in the case of a similarly charged wall .
this somewhat surprising behavior has in fact been seen in monte - carlo simulations@xcite .
we note that on the two - loop level , in addition a three - point interaction appears .
finally we will present results for the interaction between two particles in the neighborhood of the charged wall and thus under the influence of the loosely bound counter - ion cloud . to make the results somewhat more transparent
, we will confine ourselves to particles which have the same vertical distance from the wall .
the effective interaction then follows by fourier transformation and reads @xmath156 where the green s function @xmath157 is determined by eq.([gsol ] ) and @xmath158 denotes the bessel function of the first kind@xcite . for particles which are very close to each other , @xmath159 , and/or
very far apart from the charged surface , @xmath160 , the interaction is basically unscreened and given by @xmath161 for particles that are far apart from each other , @xmath162 , but close to the surface , @xmath163 , the effective interaction is given by @xmath164 the main feature is that the effective interaction is screened and thus reduced compared to the bare coulomb interaction , but the screening is much weaker than for example in a salt solution and the interaction decays as an inverse cube of the distance .
this behavior has previously been predicted by liquid - state calculations and is interpreted as the effective dipole - dipole interaction between the test - charges and their associated counter - ion clouds@xcite .
the effective interaction between two induced dipoles is unscreened because the lower half - space is assumed to be impenetrable to ions and thus allows the unattenuated passage of electric field lines . from eq.([g1 ] ) it follows that the screening is maximal at a finite distance away from the wall .
this is connected with the maximum in the counter - ion distribution , see fig.4 . for particles that are far apart from each other , @xmath162 , and also relatively far apart from the surface , @xmath165 , the effective interaction is given by @xmath166 and thus increases gradually as one move away from the surface . at a separation from the surface which equals the interparticle distance , @xmath167
, the effective interaction has the same magnitude as the bare coulomb interaction , and we find the effective interaction to cross over smoothly to the bare interaction as given by eq.([g0 ] ) .
in this article we have formulated the non - linear field theory for mobile counter ions under the influence of fixed charge distributions .
we showed explicitly that the poisson - boltzmann equation corresponds to the saddle - point equation , and how correlations and fluctuations can be accounted for by a loop - wise expansion of the action around this saddle point .
we find the poisson equation to be satisfied at each order in this expansion . clearly , the boltzmann equation is not satisfied when going beyond the saddle point , and this is the reason for deviations from the poisson - boltzmann equation .
particularly useful is the introduction of the gibbs potential @xmath168 $ ] , which only contains one - particle irreducible diagrams ( and thus reduces the number of diagrams to be considered ) and which allows to directly calculate the electrostatic potential distribution via the equation of state .
we applied our formalism to the case of a charged wall which is impenetrable to counter ions and obtained , to one - loop order and without further approximations , the electrostatic potential , the counter - ion distribution , and the effective interactions between test particles .
we find that the counter - ion density is increased close to the wall as compared to the mean - field ( poisson - boltzmann , pb ) solution , however , right at the wall the density is unchanged as compared to pb .
the increase close to the wall is due to correlations between counter ions , which are not captured in the pb approach and which reduce the repulsion between counter ions .
the surface potential is increased at the surface .
the relative strength of the one - loop correction is proportional to the single parameter @xmath169 and thus is proportional to the cube of the counter ion valency .
our results become unphysical at @xmath170 .
the value of @xmath171 up to which our results are accurate could only be inferred from quantitative comparisons with numerical simulations , which we plan to do in the future .
our formalism can in principle applied to all situations where the pb equation can be solved in closed - form .
it would be particularly useful to obtain the one - loop correction for a charged wall in a salt solution , because this would represent the bridge between debye - hckel and poisson - boltzmann theory .
p. nielaba and f. forstmann , chem .
lett . * 117 * , 46 ( 1985 ) ; b. daguanno , p. nielaba , t. alts , and f. forstmann * 85 * , 3476 ( 1986 ) ; p. nielaba , t. alts , b. daguanno , and f. forstmann * 34 * , 1505 ( 1986 ) .
|
we formulate the non - linear field theory for a fluctuating counter - ion distribution in the presence of a fixed , arbitrary charge distribution .
the poisson - boltzmann equation is obtained as the saddle - point , and the effects of fluctuations and correlations are included by a loop - wise expansion around this saddle point .
we show that the poisson equation is obeyed at each order in the loop expansion and explicitly give the expansion of the gibbs potential up to two loops .
we then apply our formalism to the case of an impenetrable , charged wall , and obtain the fluctuation corrections to the electrostatic potential and counter - ion density to one - loop order without further approximations .
the relative importance of fluctuation corrections is controlled by a single parameter , which is proportional to the cube of the counter - ion valency and to the surface charge density .
we also calculate effective interactions between charged particles , which reflect counter - ion correlation effects .
pacs .
82.70.@xmath0y - disperse systems .
+ pacs .
61.20.qg - associated liquids .
+ pacs .
82.45.+z - electrochemistry .
|
here we investigate the dependence of the first plateau on the cavity loss @xmath3 to obtain data that support the essentiality of the cavity loss for the plateaus .
we denote the chemical potential at the beginning of the first plateau by @xmath93 , which is pointed with the up arrow in fig .
we plot @xmath93 ( circle symbols ) as a function of @xmath3 in fig .
[ fig : mub1st ] .
we observe that @xmath93 increases as @xmath3 becomes smaller .
this is consistent with the essentiality of the cavity loss for the plateaus .
we also evaluate @xmath93 from the equilibrium state as follows . as we mentioned in the main text
, we can identify @xmath93 with the point where @xmath44 becomes equal to @xmath43 . since the system is near to an equilibrium state for @xmath94
, we expect that we can calculate @xmath44 and @xmath43 for @xmath94 in good approximation by using the equilibrium state .
we thus evaluate @xmath93 from the approximate @xmath44 and @xmath43 . for the practical calculation
, we use the ground state @xmath95 as the equilibrium state because the temperature of the tls bath ( @xmath96 ) is sufficiently lower than the eigenenergy difference [ @xmath97 ; see fig .
in this case we have @xmath98 and @xmath99 .
we note that we set @xmath45 to @xmath100 to obtain the latter . in fig .
[ fig : mub1st ] , we plot @xmath5 ( triangle symbols ) where the thus - evaluated @xmath44 and @xmath43 get equal .
we see that the result provides a good approximation to @xmath93 .
we also observe that the approximate @xmath93 approaches to a finite value as @xmath3 decreases .
to see this more clearly , we use another evaluation of @xmath93 in the following analysis .
first we note that when we approximate the steady state by the ground state we have @xmath101 and @xmath102 .
next , from the steady - state equation for @xmath103 , we have a relationship @xmath104 . substituting this relationship into @xmath43 and equating @xmath43 with @xmath44 ( with @xmath105 )
, we have @xmath106 .
now we neglect @xmath107 term ( since we consider small @xmath3 ) and approximate @xmath108 ( since we consider the coherent phase ) .
then we obtain @xmath109 we plot this equation as the solid curve in fig .
[ fig : mub1st ] .
we observe that this equation provides a good approximation to @xmath93 . equation ( [ mub1st ] ) means that @xmath110 converges to @xmath111 in the limit of @xmath112 . at first glance
this may look inconsistent with the essentiality of the cavity loss for the plateaus .
however , in the equilibrium case , the region with @xmath113 is unstable @xcite .
therefore we conclude that the plateaus do not appear in the equilibrium situation ( @xmath38 ) .
we thus show that the results in this supplement support the essentiality . at the beginning of the first plateau ( circle symbols ) as a function of the cavity loss @xmath3 .
the cutoff for the maximum photon number in the cavity is set to 15 for @xmath114 , 25 for @xmath115 , and 30 for @xmath116 .
evaluation from the equilibrium state ( with the cutoff photon number being 500 ) is plotted with the triangle symbols .
the solid line is a plot of eq .
( [ mub1st ] ) .
the hopping constant @xmath117 is set to 0.6 .
the other parameter values are the same as those in fig . 2 .
|
nonequilibrium steady states are investigated in a coupled cavity qed array system which is pumped by a thermal bath and dissipated through cavity loss . in the coherent ( non - zero photon amplitude ) phase , plateau regions appear , where the steady states become unchanged against the variation of the chemical potential of the thermal bath .
the cavity loss plays a crucial role for the plateaus : the plateaus appear only if the cavity loss exists , and the photon leakage current , which is induced by the loss , is essential to the mechanism of the plateaus . _
introduction_.in recent years , much attention has been paid to coupled cavity qed arrays . in these systems , the effective photon - photon interaction in each cavity and photon hopping between cavities lead to rich quantum many - body phenomena of photons @xcite .
for example , in equilibrium states , these systems exhibit quantum phase transitions between the `` incoherent ( or mott insulator ) '' and `` coherent ( or superfluid ) '' phases @xcite .
nonequilibrium states in these systems have been also investigated @xcite .
most of the works have focused on the nonequilibrium effects of the pumping and dissipation on the properties observed in equilibrium states ; e.g. , signatures of the two phases in nonequilibrium states . however
, phenomena genuinely characteristic to nonequilibrium states in the systems remain unexplored .
apart from the coupled cavity qed array , examples of this kind of phenomena are seen in `` nonequilibrium phase transitions , '' such as the pattern formation outside equilibrium @xcite , dissipative structure @xcite , and lasing phenomena @xcite .
these are observed not in equilibrium states but only in nonequilibrium states .
it is important to find such phenomena from the viewpoint that they can be used as test models for nonequilibrium statistical mechanics and thermodynamics . in this paper
, we report that another genuine nonequilibrium phenomenon can occur in a coupled cavity qed array that is pumped by a thermal bath and dissipated through cavity loss .
we find that in the coherent phase there exist plateau regions where the states of the array system becomes unchanged to the variation of the chemical potential of the thermal bath .
we show that the plateau emerges only if the cavity loss ( dissipation ) exists , which implies that the nonequilibrium effect is crucial for this phenomenon . and inverse temperature @xmath0 ) , and a photon bath ( vacuum state ) .
each cavity photon couples with a two - level system ( tls ) ( coupling constant @xmath1 ) , with neighboring cavity photons ( hopping constant @xmath2 ) , and with the photon bath ( which induces the cavity loss @xmath3 ) .
each tls is incoherently pumped ( rate @xmath4 ) by the pumping bath . ]
_ setup_.we consider a cavity array that is connected to two baths .
an illustration of the total system ( array + baths ) is shown in fig .
[ fig : schematic ] .
each cavity contains a two - level system ( tls ) which is continuously excited through the coupling ( excitation exchange ) with the pumping bath ( tls bath ) .
the pumping bath is in an equilibrium state characterized by the chemical potential @xmath5 and inverse temperature @xmath0 .
thus the tlss in the cavity array are incoherently pumped ( rate @xmath4 ) .
the photon mode in each cavity interacts with the tls in the cavity ( coupling constant @xmath1 ) and is coupled with the photons in the neighboring cavities through mode overlap ( hopping constant @xmath2 ) .
it is also coupled with the photon bath ( vacuum ) , which induces the cavity loss @xmath3 .
the coupled cavity qed array is described by the jaynes - cummings - hubbard hamiltonian ( @xmath6 ) @xcite : @xmath7 where @xmath8 and @xmath9 represent the site indices of the cavities , @xmath10 and @xmath11 are the raising and lowering operators of the tls , and @xmath12 and @xmath13 are the creation and annihilation operators of the cavity photon in the @xmath8th cavity .
the first term in eq .
( [ h_s ] ) is the jaynes - cummings hamiltonian for single cavities given by eq .
( [ h_jc ] ) , and the second term is the hopping of the photons between the neighboring cavities ( @xmath14 means the nearest neighbor sites ) .
we assume that the transition frequency of the tls and the cavity frequency are resonant and the same for all of the cavities ( denoted by @xmath15 ) . to obtain insight over the steady - state properties of the pumped - dissipative system
, we employ a mean - field ( mf ) approximation and quantum master equation ( qme ) .
we introduce the steady - state average photon field @xmath16 as the mean field .
we assume that it oscillates with a single frequency @xmath17 in the rest frame : @xmath18 , where the amplitude @xmath19 is a real number . after using the mf approximation @xcite and transforming into the rotating frame with @xmath17 ,
we take the trace over the baths and use the born - markov approximation .
then we have the qme in the shrdinger picture ( within the rotating frame ) @xcite : @xmath20 , where @xmath21 is the density matrix of the array system in the rotating frame .
the qme superoperator @xmath22 is given by @xmath23 + \mathcal{l}_{\rm tls } + \mathcal{l}_{\rm ph}$ ] , where @xmath24 is the effective system hamiltonian in the rotating frame , and @xmath25 and @xmath26 are the superoperators associated with the tls and photon baths , respectively .
we will describe the details in the last part of this paper .
we determine the steady - state solution @xmath27 in the rotating frame by @xmath28 note that the solution @xmath29 depends on @xmath19 and @xmath17 .
then we have the self - consistent equation for the mean field : @xmath30 .
\label{selfconsistenteq}\end{aligned}\ ] ] by solving eqs .
( [ steadystateeq ] ) and ( [ selfconsistenteq ] ) numerically , we determine the values of @xmath19 and @xmath17 , and obtain the steady - state density matrix @xmath29 of the mf - qme for the pumped - dissipative cavity qed array system . for the numerical calculation we set the cutoff for the maximum photon number in the cavity to 10 .
_ phase diagram_.first we investigate a phase diagram of the nonequilibrium steady states in the array system by employing @xmath19 as the order parameter .
figure [ fig : phasediagram ] depicts the regions for @xmath31 and @xmath32 as a function of the hopping constant @xmath2 and the chemical potential @xmath5 of the tls bath .
we refer to the region with @xmath31 as incoherent phase and to the region with @xmath32 as coherent phase . for comparison
, we also show the phase boundary in the equilibrium states ( @xmath33 ) by the dotted curve . in small @xmath5 ( weak pumping ) regime ,
the phase boundary is almost the same as that of the equilibrium states .
this implies that the array system is nearly equilibrium in this regime . in larger @xmath5 ( strong pumping ) regime , by contrast
, the coherent phase region shrinks compared with the equilibrium case .
this results from a nonequilibrium effect ; in this regime , the number of the cavity photons increases , so that the effect of the cavity loss becomes larger .
since the loss destroys the coherence of the system , the area of the coherent phase decreases in this regime . of the tls bath and the hopping constant @xmath2 ( @xmath34 is the number of the nearest neighbor sites ) .
the boundary between the incoherent ( @xmath31 ) and coherent ( @xmath32 ) phases is plotted .
the parameter values are : @xmath35 , @xmath36 , and @xmath37 .
the dotted curve is the phase boundary for the equilibrium case ( @xmath38 ) .
the dashed line ( @xmath39 ) is the one along which the results in fig .
[ fig : frequency ] are plotted . ]
( solid line ; left vertical axis ) and photon number @xmath40 ( dashed line ; right vertical axis ) as functions of the tls bath chemical potential @xmath5 , which are calculated along the dashed line ( @xmath39 ) in fig .
[ fig : phasediagram ] .
the dotted line represents @xmath41 .
( b ) currents between the system and baths .
the solid , dashed , and dotted lines represent @xmath42 ( from the system to tls bath ) , @xmath43 ( from the tls bath to system ) , and @xmath44 ( from the system to photon bath ) , respectively .
the inset shows an illustration of the currents .
( c ) @xmath45 ( solid line ) , @xmath46 ( dashed line ) , @xmath47 ( dotted line ) , and @xmath48 ( dash - dotted line ) plotted against @xmath5 , where @xmath49 ( @xmath50 ) are the eigenenergies of @xmath24 .
the parameter values are the same as those in fig .
[ fig : phasediagram ] . ] _ main result_.in fig .
[ fig : frequency](a ) , we plot @xmath5-dependence of the frequency @xmath17 and the average photon number per cavity @xmath51 $ ] in the coherent phase along the dashed line in fig .
[ fig : phasediagram ] @xcite . in small @xmath5 ( weak pumping ) regime ,
we observe @xmath52 .
we note that in the equilibrium states @xmath41 holds .
therefore this result is consistent with the fact that the array system is near to equilibrium in this regime . in larger @xmath5 regime , we observe plateaus both in @xmath17 and @xmath40 ; i.e. , there are regions of @xmath5 where @xmath17 and @xmath40 take constant values .
we numerically confirm that the steady - state density matrix @xmath29 itself is unchanged in the plateau regions .
we note that the cavity loss is crucial for the emergence of the plateaus because @xmath41 holds for @xmath38 ( in the equilibrium states ) . furthermore , in the supplement @xcite , we will show that the chemical potential at the beginning of the first plateau [ pointed with the up arrow in fig .
[ fig : frequency](a ) ] increases as @xmath3 decreases and that it enters the unstable region @xcite in the limit of @xmath53 .
these cavity - loss induced plateaus , which are our main findings in this paper , seem counterintuitive since the loss usually smears out ( rather than generates ) sharp - cut phenomena .
we also note that the incoherent pumping is necessary to observe the plateaus because we see them in the dependence on @xmath5 .
the emergence of the plateaus implies that the properties of the cavity qed array system are completely stable to the uncertainty of @xmath5 ; for example , photoluminescence spectrum and photon statistics as well as @xmath17 and @xmath40 are unchanged with the variation of @xmath5 within the plateaus . _
mechanism_.we here discuss the mechanism of the plateaus . for this purpose
, we investigate the relationship of the plateaus with the currents and energy differences .
as we mentioned , the plateaus are induced by the cavity loss .
this leads us to expect that quantities directly related to the cavity loss are essential to the mechanism .
a typical one is the photon current that leaks to the photon bath .
therefore we first investigate the relationship with the currents . in fig .
[ fig : frequency](b ) , we plot the currents : @xmath54 $ ] ( from the system to tls bath ) , @xmath55 $ ] ( from the tls bath to system ) , and @xmath56 $ ] ( from the system to photon bath ) @xcite , where the superoperators @xmath57 and @xmath26 are respectively given by eqs . ( [ l_pump ] ) and ( [ l_diss ] ) ( see the last part ) .
a schematic illustration of the currents are shown in the inset of fig .
[ fig : frequency](b ) .
we observe that @xmath58 and @xmath59 in the plateaus ( note that @xmath60 holds in the steady states ) .
next we investigate the relationship with the energy differences .
as we mentioned , the plateaus are observed in the dependence on the chemical potential @xmath5 . in the qme , especially in the superoperator @xmath25 , @xmath45 ( rather than @xmath5 itself ) and the eigenenergy differences appear [ see the below of eq .
( [ l_pump ] ) ] .
therefore we investigate @xmath45 and the eigenenergy differences . in fig .
[ fig : frequency](c ) , we plot @xmath45 and @xmath61 , where @xmath49 ( @xmath50 ) are the eigenenergies for the ground and @xmath62th excited states of the effective hamiltonian @xmath24 .
we observe that @xmath46 and @xmath48 cross with @xmath45 respectively at the ends of the first and second plateaus , whereas @xmath47 crosses with @xmath45 in the transition region between the first and second plateaus . from these observations
, we have a following mechanism scenario of the plateaus .
first we discuss the first plateau .
when @xmath63 , i.e. , before the end of the first plateau , the ground state is the only channel available for pumping from the tls bath . before the first plateau starts , @xmath43 is a convex function of @xmath5 , whereas @xmath44 is almost proportional to @xmath5 .
these behaviors of the currents possibly result from the fact that , for each single pumping channel , @xmath64 is bounded above whereas @xmath65 is not . due to these behaviors , @xmath44 can catch up with @xmath43 at certain @xmath5 .
however , it can not go over @xmath43 since @xmath66 holds in the steady states .
therefore the first plateau starts from this value of @xmath5 .
as @xmath5 increases further , @xmath45 gets equal to @xmath46 , which results in the open of a new channel the first excited state for pumping from the tls bath .
therefore the system can go out of the first plateau when @xmath67 .
this scenario requires the existence of @xmath44 , which is consistent with the fact that the cavity loss is crucial for the plateaus .
similarly , the beginning of the second plateau is the point where @xmath44 catches up with @xmath43 again . in this case
, we should note that @xmath45 crosses with @xmath47 before the second plateau starts .
thus we do not observe the end of the plateau at this point although the pumping channel of the second excited state opens .
instead , we observe the effect of this channel open as the jumps in @xmath40 [ fig . [
fig : frequency](a ) ] and in the currents [ fig .
[ fig : frequency](b ) ] .
the end of the second plateau is the point where @xmath45 crosses with @xmath48 and the pumping channel of the third excited state opens .
finally we remark that in the plateau regions the input excitation of the tlss is almost perfectly converted to the output of the photons because @xmath58 holds there . in this sense ,
conversion efficiency of this system in the plateaus is nearly 100% .
_ details of the mf - qme_.the hamiltonian of the total system ( array + baths ) is given by @xmath68 , where @xmath69 is given by eq .
( [ h_s ] ) , and @xmath70 and @xmath71 are respectively the hamiltonians of the tls and photon baths including the system - bath interactions .
we assume that the baths are sufficiently larger than the array system , so that the tls bath remains in the equilibrium state with @xmath5 and @xmath0 and the photon bath remains in the vacuum .
we obtain the mf - qme for the array system as follows .
first , we explain the mf approximation . as we mentioned before , we introduce the oscillating photon field as the mean field : @xmath72 .
then we use the mf approximation @xcite : @xmath73 . substituting this into the second term in eq .
( [ h_s ] ) , we obtain the mf system hamiltonian as a sum over the single sites .
we can thus reduce the problem to the single - site one , and will omit the site index : @xmath74 , where @xmath34 is the number of the nearest neighbor sites .
furthermore , by transforming into the rotating frame with the frequency @xmath17 , we eliminate the time dependence in @xmath75 ; i.e. , @xmath76 , where @xmath77 , @xmath78 is the total number of photon and tls excitations in the cavity , and @xmath79 is the excitation number in the baths
. then we have the von neumann equation for the density matrix @xmath80 of the total system in the rotating frame as @xmath81 $ ] .
next , we explain the qme . starting from the above von neumann equation
, we take the trace over the baths and use the born - markov approximation to obtain the qme in the shrdinger picture ( within the rotating frame ) @xcite : @xmath20 , where @xmath82 and @xmath23 + \mathcal{l}_{\rm ph } + \mathcal{l}_{\rm tls}$ ] . here
@xmath83 is the effective system hamiltonian in the rotating frame .
the superoperator @xmath26 associated with the photon bath is given by @xmath84 where @xmath3 is the cavity loss rate .
to rewrite the superoperator @xmath25 associated with the tls bath into a tractable form , we introduce eigenoperators of @xmath24 as follows .
we denote by @xmath85 the projection operator onto the eigenspace belonging to @xmath49 ( eigenvalue of @xmath24 ) .
then we define eigenoperators : @xmath86 . by using these operators
we can write @xmath25 as @xmath87 @xcite , where @xmath88 , % \notag\\ & \mathcal{l}_{\rm tls}^\pm \tilde{\rho } \label{l_pump } \\ & = - \frac{\gamma}{2 } \sum_{\omega } f^\pm(\omega ) \bigl [ \hat{\sigma}^\mp \hat{\sigma}^\pm_\omega \tilde{\rho } + \tilde{\rho } \hat{\sigma}^\mp_\omega \hat{\sigma}^\pm - \hat{\sigma}^\pm \tilde{\rho } \hat{\sigma}^\mp_\omega - \hat{\sigma}^\pm_\omega \tilde{\rho } \hat{\sigma}^\mp \bigr ] . \notag\end{aligned}\ ] ] here @xmath4 is the pumping rate , the @xmath89-sum is over the eigenvalue differences , @xmath90\}$ ] is the fermi distribution function , and @xmath91 .
note that , if the cavity loss is absent ( @xmath38 ) , @xmath25 leads the array system to the equilibrium state with the chemical potential @xmath5 and inverse temperature @xmath0 . _
conclusion_.we have investigated the nonequilibrium steady states in a coupled cavity qed array system that is incoherently pumped by a thermal bath for the tls and dissipated by the vacuum of the photon .
we have determined a steady - state phase diagram by employing the photon amplitude @xmath19 as the order parameter . in the coherent phase ( region with @xmath92 ) , we have found that there exist cavity - loss induced plateaus where the state of the array system becomes unchanged as a function of the chemical potential @xmath5 of the tls bath .
we have proposed a mechanism of the plateaus .
there are two essential elements in the proposed mechanism : one is the existence of discrete energy levels ( or energy gap ) in the system , and the other is that the output current to one of the baths can catch up with the input current from the other bath ( the bounded excitation number of the tls and unbounded photon number would be the origin in our case ) .
therefore , if our scenario is correct , similar plateau phenomena should be observed in other systems which have these elements .
this work was supported by a jsps research fellowship for young scientists ( no .
24 - 1112 ) and by the jsps through the first program
. k. k. is also supported by the project for developing innovation systems of mext
. 99 m. j. hartmann , f. g. s. l. brando , and m. b. plenio , laser photonics reviews , * 2 * , 527 ( 2008 ) .
a. tomadin and r. fazio , j. opt .
soc .
am .
b * 27 * , a130 ( 2010 ) .
a. d. greentree , c. tahan , j. h. cole , and l. c. l. hollenberg , nat . phys . * 2 * , 856 ( 2006 ) .
m. j. hartmann , f. g. s. l. brando , and m. b. plenio , nat . phys . * 2 * , 849 ( 2006 ) .
d. g. angelakis , m. f. santos , and s. bose , phys .
rev .
a * 76 * , 031805 ( 2007 ) .
d. rossini and r. fazio , phys .
rev .
lett .
* 99 * , 186401 ( 2007 ) .
m. i. makin , j. h. cole , c. tahan , l. c. l. hollenberg , and a. d. greentree , phys .
rev .
a * 77 * , 053819 ( 2008 ) .
j. koch and k. le hur , phys .
rev .
a * 80 * , 023811 ( 2009 ) .
m. schir , m. bordyuh , b. ztop , and h. e. treci , phys .
rev .
lett . * 109 * , 053601 ( 2012 ) .
k. kamide , m. yamaguchi , t. kimura , and t. ogawa , phys .
rev .
a * 87 * , 053842 ( 2013 ) .
i. carusotto , d. gerace , h. e. treci , s. de liberato , c. ciuti , and a. imamoglu , phys .
rev .
lett . * 103 * , 033601 ( 2009 ) .
m. j. hartmann , phys .
rev .
lett . * 104 * , 113601 ( 2010 ) .
a. tomadin , v. giovannetti , r. fazio , d. gerace , i. carusotto , h. e. treci , and a. imamoglu , phys .
rev .
a * 81 * , 061801(r ) ( 2010 ) .
s. ferretti , l. c. andreani , h. e. treci , and d. gerace , phys .
rev .
a * 82 * , 013841 ( 2010 ) s. schmidt , d. gerace , a. a. houck , g. blatter , and h. e. treci , phys .
rev .
b * 82 * , 100507(r ) ( 2010 ) .
m. knap , e. arrigoni , w. von der linden , and j. h. cole , phys .
rev .
a * 83 * , 023821 ( 2011 ) .
f. nissen , s. schmidt , m. biondi , g. blatter , h. e. treci , and j. keeling , phys .
rev .
lett . * 108 * , 233603 ( 2012 ) .
t. grujic , s. r. clark , d. jaksch , and d. g. angelakis , new j. phys .
* 14 * , 103025 ( 2012 ) .
j. jin , d. rossini , r. fazio , m. leib , and m. j. hartmann , phys .
rev .
lett . * 110 * , 163605 ( 2013 ) .
m. c. cross and p. c. hohenberg , rev .
mod .
phys .
* 65 * , 851 ( 1993 ) .
g. nicolis , and i. prigogine , _ self - organization in nonequilibrium systems , from dissipative structures to order through fluctuations _
( wiley , new york , 1977 ) .
h. haken , _ light , volume 2 : laser light dynamics _
( elsevier north - holland , new york , 1985 ) .
h. p.
breuer and f. petruccione , _ the theory of open quantum systems _ ( oxford university press , london , 2002 ) . in the incoherent phase ,
the photon frequency @xmath17 is indefinite because the photon amplitude @xmath19 vanishes .
see supplemental material . there is no current from the photon bath to system because the photon bath is in the vacuum state .
|
it is important to determine the composition of unevolved stars in globular clusters in order to learn the basic composition of the cluster from stars that have not undergone any mixing with the interior layers .
we include as `` unevolved stars '' both main sequence and turn - off stars .
it is well known that there can be large spreads in the abundance of some elements in the giant stars of a given globular cluster ( e.g. kraft 1994 , sneden et al .
most previous high - resolution studies were limited to giant stars , whose atmospheres and interiors are less well modeled than those of dwarf stars .
more extensive studies of the unevolved stars , which have little or no advanced nuclear processing _ in situ _ , are expected to reveal the clusters original composition .
it now seems clear that mixing and evolutionary effects in cluster giants mean that these stars can not be used to probe the original abundance patterns in clusters . the the light elements ( e.g. c , n , o , na , al ) in the giants evince rich intra - cluster differences that makes it impossible to ascertain a universal pattern of light element alterations .
good evidence exists , e.g. , that cno cycling and nena / mgal burning in m13 has a considerable _ in situ _ component ( pilachowski et al .
1996 ; briley & cohen 2001 ) .
conventional wisdom now holds that such abundance modifications can be explained by a combination of @xmath5-capture reactions ( denisenkov & denisenkova 1990 ; langer , hoffman & sneden 1993 ) and some non - standard deep mixing mechanism ( cavallo , sweigart & bell 1996 ) perhaps operating particularly more efficiently in low metallicity clusters owing to smaller stellar interior mean molecular weight gradients .
the most direct means , however , to explore _ in situ _ versus primordial ( or self - enrichment ) mechanisms is to determine the composition of unevolved stars in globular clusters .
more extensive studies of the unevolved stars , which have little or no advanced nuclear processing _ in situ _ , are expected to provide a baseline of clusters red giants original composition(s ) onto which the additions - by other mechanisms - are made .
several groups have begun studying globular cluster stars over a range of evolutionary stages , including turn - off stars at high spectral - resolution ( e.g. gratton et al .
2001 ; cohen & melndez 2005 ) .
these initial efforts , and previous photometric and low - resolution spectroscopic results , have revealed strong evidence that some light element abundance alterations in some clusters are present prior to reaching the red giant branch , thus suggesting the possibilities of primordial variations , self - enrichment mechanisms , or totally unexpected mixing mechanisms .
briley et al . ( 1994 ) found anti - correlated cn and ch band strength variations at the turnoff of 47 tuc , and suggested that this cluster s cn distribution changes little with evolutionary status ; additionally striking is the existence of na abundance variations on the 47 tuc main - sequence reported by briley et al .
( 1996 ) . utilizing keck data ,
briley & cohen ( 2001 ) find that cno abundances in m71 show little evolution from the main - sequence to bright giants ; moreover , the bimodal and anti - correlated cn and ch distributions of m71 red giants are found to be in place on the main - sequence .
gratton et al .
( 2001 ) find that an o - na ( and mg - al and c - n ) anticorrelation , previously seen only in globular red giants , exists in turn - off stars in ngc 6752 .
ramrez & cohen ( 2002 ) ( hereafter rc02 ) utilized high - resolution keck / hires data to find that an o - na anticorrelation exists in m71 from its turnoff to giants above the red hb level . for an excellent review of globular cluster abundances and abundance variations , see gratton , sneden & carretta ( 2004 ) .
we have embarked on a high spectral - resolution study of turn - off stars in a number of globular clusters .
such stars are faint , so to observe them at high resolution requires long integration times on the largest telescopes .
we have used the keck i 10-m telescope with hires to make observations of @xmath6 turn - off stars in m92 , m71 , m13 , and m5 .
results on m92 have been published on li in boesgaard et al .
( 1998 ) and on fe , na , mg , ca , ti , cr and ba in king et al .
the latter paper found [ fe / h ] = log n(element / h)@xmath7/log n(element / h)@xmath8 = @xmath92.52 for unevolved stars , compared to @xmath92.25 for giants found by sneden et al .
the reason for the dwarf - giant difference remains unclear , but the homogeneous analyses of ngc 6397 and 6752 ( gratton et al .
2001 ) and m71 ( ramrez & cohen 2001 , rc02 ) indicate no difference in [ fe / h ] values ( or fe peak , @xmath10-capture , and @xmath3-element abundances for the m71 cluster ) between turnoff stars and more evolved cluster stars .
king et al . ( 1998 ) also found large mg depletions and na enhancements compared to hd 140283 , a mildly evolved halo subgiant with similar [ fe / h ] and evolutionary state ; [ ba / fe ] also appeared to be enhanced in the m92 stars compared to hd 140283 .
king et al .
( 1998 ) noted these qualitative patterns were reminiscent of those believed to be due to mgal and nena cycling in red giants .
( see the recent discussion by aikawa et al .
( 2004 ) on the reaction rates in these chains related to observations in red giants in globular clusters . )
additionally , they dedicated considerable discussion to wrestling with the difficulty in understanding how such processing ( whether _ in situ _ or from polluted material processed in a previous stellar generation ) could co - exist without resulting in remarkably depleted li abundances .
bonifacio et al .
( 2002 ) noted a similar `` paradox '' for ngc 6397 , which appears to be analogous to that seen in the metal - poor field ( spite & spite 1986 ) .
bonifacio et al .
( 2002 ) mention preliminary evidence of prominent n enhancements in ngc 6397 subgiants , indicative of cn processing or supernovae in massive stars .
if those enhancements are due to internal processing in their turnoff stars , then how does li remain `` normal '' ?
similarly , thvenin et al .
( 2001 ) note that one of their ngc 6397 turn - off stars demonstrates a relative 0.4 dex na deficiency ; however , this star also evinces unremarkable li .
this summary paints a complex and rich picture of globular cluster abundances one that will probably only be understood on a cluster - by - cluster ( if not star - by - star ) basis .
here we present observations , analysis , and results from a study of five turn - off stars in m71 .
we have determined abundances for three iron - peak elements ( fe , cr , and ni ) , four alpha - fusion elements ( mg , si , ca , and ti ) , two rare - earth elements ( y and ba ) and na .
the globular cluster m71 is especially interesting for this work because it is very old ( @xmath11 gyr ; grundahl , stetson & andersen 2002 ) , yet relatively metal - rich ( [ fe / h]@xmath12 ) for a globular cluster .
sneden et al .
( 1994 ) ( hereafter sklps ) find [ fe / h ] = @xmath90.79 from 10 red giants .
cohen , behr & briley ( 2001 ) , ramrez et al .
( 2001 ) and rc02 have reported on keck / hires observations of m71 of stars of a variety of evolutionary stages including three stars near the turn - off .
the stars we observed in m71 have the same age , mass , luminosity , and temperature as each other and as the stars in m 92 , but the two clusters differ in metallicity by a factor of @xmath050 .
we have selected m 71 because 1 ) it is a little brighter than m 92 at the turnoff ( @xmath13 versus 18.0 ) ; 2 ) despite considerable reddening ( @xmath14 ) , it has excellent photometry ( hodder et al .
1992 ) ; 3 ) it is a disk globular with relatively high metal content ; 4 ) its space motion indicates that it is probably a member of the thick disk population ( cudworth & hanson 1993 ) .
we report on our four general goals : first , to characterize the abundances of relatively unevolved stars in m71 .
second , to compare these with determinations from both other near - turnoff stars ( rc02 ) and red giants ( sklps , rc02 ) in m71 .
third , to compare our m71 turn - off abundances with those in field halo stars and field disk stars of the same metallicity .
fourth , to search for evidence of photospheric material having suffered @xmath5-capture and deep mixing .
keck / hires spectral observations with respectable signal - to - noise ( s / n ) ratios of stars near v = 18 require exposures of several hours .
however , individual intergration times did not exceed 45 - 60 minutes due to the contamination by `` cosmic ray '' events . on the other hand ,
the exposures needed to be long enough to obtain a high enough signal so that individual frames could be co - added reliably at the correct wavelength justification . in metal - poor stars where the spectral lines are weak
this can be problematic .
therefore , we could only make these observations on cloudless nights with good seeing .
the spectra for this project were obtained at the keck 10-m telescope with hires ( vogt et al .
1994 ) over three observing seasons on four observing runs comprising seven nights .
the same settings ( slit dimensions , echelle and cross - disperser angles , filters ) were used on each run .
approximately 2400 were covered from 4430 6880 with some gaps between orders longward of 5000 .
the positions in the color - magnitude diagram of our five stars can be seen in figure 1 .
the fiducial sequence is taken from hodder et al .
( 1992 ) .
the details of the observations for our five stars are shown in table 1 which gives the dates of the observations and the details of the multiple exposures on multiple nights .
the star identifications are those of hodder et al .
the photometric data for v and ( b@xmath9v ) are from hodder et al .
( 1992 ) as is the assumed reddening correction of e(b@xmath9v ) = 0.28 .
this value of the reddening agrees well with that of both zinn ( 1980 ) of 0.27 @xmath20.03 and that of geffert & maintz ( 2000 ) of 0.27 @xmath20.05 .
the final column shows the signal - to - noise ratio of the combined spectra . as can be seen in table 1 , the integration times were 135 - 285 minutes to reach the listed s / n ratios ( per pixel ) of 50 - 60 .
we have also made use of a spectrum of the moon as a solar proxy that we obtained with the same settings on hires as our stellar spectra .
the lunar spectrum was a 30 s exposure obtained on 29 july 1994 with a s / n ratio near 1400 .
standard data reduction procedures were carried out with iraf routines .
individual frames were trimmed , overscan - subtracted , and bias - subtracted ( the median of numerous overscan - subtracted bias frames from each night ) . typically 20 quartz flat - field exposures were obtained each night and these processed frames were combined to form a master nightly flat - field frame .
frames of the stars in m71 were pre - processed in the same fashion and then divided by a normalized flat field , which was produced by fitting a low order spline to the blaze+lamp function .
scattered light was removed by fitting a low - order polynomial to the inter - order regions across the dispersion and smoothing them along the dispersion .
cosmic ray hits were removed interactively .
typically , 29 orders were identified , traced and extracted .
the wavelength scale was found from the positions of many lines in each order of the th - ar comparison spectrum .
the measured spectral resolution is @xmath045,000 .
the dispersion - corrected , extracted 1-dimensional spectra were co - added order - by - order .
the final spectrum of each star was then continuum normalized . for most of the spectral analysis
the individual orders were smoothed by a 3-pixel boxcar in order to increase the effective s / n ratio with minimal impact upon the spectral resolution .
examples of two spectral regions in the spectra for three of the stars are shown in figures 2 and 3 .
spectral features from several ions are identified in these figures .
the three stars shown all have the same value of ( b@xmath9v ) and the line strengths are clearly similar in all three .
we have made measurements of the equivalent widths of some 200 lines in each of our five stars and 117 in the sun / moon spread throughout the 29 orders of our keck spectra .
the relevant features include those of the : fe - peak elements ( and , and , and ) ; alpha - fusion elements ( , and , , and and ) ; the alkali element na ( ) ; and the neutron - capture elements ( and ) . the splot package in iraf was used to measure the equivalent widths via gaussian profile fitting .
lines that were noticeably blended were not used ; thus , nearly all could be measured using both sides of the line profile .
the measured equivalent widths ranged from 7 to 200 m , but lines used for the parameter and abundance determinations were typically @xmath15 .
table 2 provides the species , wavelengths , lower excitation potentials , oscillator strengths , and equivalent widths for all five stars and the sun . the log gf values in table 2 are from the compilations presented in stephens ( 1999 ) and stephens & boesgaard ( 2002 ) .
we have used the @xmath16 values from hodder et al .
( 1992 ) to determine the stellar effective temperature of each star .
although the reddening of 0.28 is high , the good agreement with the values of zinn ( 1980 ) and geffert & maintz ( 2000 ) suggest that it is rather well - determined .
hodder et al .
give errors in b@xmath9v of 0.01 for stars in this v magnitude range .
we then use the ( b@xmath9v)@xmath17 values to find t@xmath18 for both the carney ( 1983 ) ( c83 ) and the king ( 1993 ) ( k93 ) temperature scales for metal - poor stars , as we have done in earlier work ( e.g. boesgaard et al .
in addition , we used ( b@xmath9v)@xmath17 and a nominal metallicity of [ fe / h]@xmath19 ) to find t@xmath18 values on the irfm scale of alonso et al .
1996 ) ; these temperatures agree almost exactly with those on the c83 scale .
the uncertainty in the temperatures comes from random errors in the photometry ( 0.01 mag ) , the large reddening , and the calibrations .
random errors of @xmath20 mag in both the photometry and reddening estimate both contribute an uncertainty of @xmath21 k. the two calibrations give @xmath22 values that systematically differ by @xmath23 k. we carry through the detailed abundance analysis for both sets of temperatures and their associated parameters : log g , [ fe / h ] , and microturbulence ( @xmath24 ) .
we have used the newest yale isochrones ( yi et al .
2001 ) to determine log g values using an age of 12 gyr , the median of the age values determined by grundahl , stetson & andersen ( 2002 ) , salaris & weiss ( 1998 ) , and hodder et al .
( 1992 ) , for both temperature scales for metallicities of both @xmath25 and @xmath26 ( and an interpolated value of @xmath27 ) .
armed with a set of initial temperatures , consistent theoretical gravities , and a range of overall metallicities , we constructed model atmospheres characterized by these parameters from the grids of kurucz ( 1993 ) . utilizing an updated ( 2000 ) version of the lte analysis code moog ( sneden 1973 )
, we computed absolute fe abundances from both our and equivalent width measurements for an array of trial microturbulence values .
for these determinations we used only lines with reduced equivalent width values log ( w/@xmath28)@xmath29 ( generally corresponding to equivalent widths @xmath30 m ) .
for a given log g and [ fe / h ] , we determined the combination of @xmath22 and @xmath24 that resulted in the minimal slopes between fe abundance and a ) lower excitation potential , and b ) reduced equivalent width .
we also tested the effect of varying log g on the agreement between fe abundance as derived from and from .
we adopted model parameters that are given in table 3 .
although we have carried out the abundance analysis throughout on both temperatures scales , we have a preference for the `` cooler '' scale .
( this preference does not imply a preference for the calibration used by c83 over k93 because the large reddening for m71 makes the @xmath31 values are uncertain by about @xmath20.03 . ) with the cooler scale : 1 ) there is better agreement with irfm temperatures , 2 ) the slopes for fe i abundance vs excitation potential are closer to zero with our spectroscopic temperatures , and 3 ) the match with the new evolutionary tracks of yi et al .
( 2001 ) seems better as the tracks do not reach the high turn - off temperatures of the hot stars on the hotter scale .
we employed the updated lte moog code to determine abundances from all our species .
moog s abfind routine was used with the measured equivalent widths in table 2 and the appropriate kurucz model atmosphere characterized by parameters in table 3 .
the final abundances for each species are given in tables 4 ( fe ) , 5 ( na and @xmath3 elements ) , and 6 ( fe - peak : ni and cr , and n - capture : y and ba ) and are discussed below in section 4 .
for a few lines of some species of particular interest we had to violate our practice of only using features with log ( w/@xmath28)@xmath32 ( e.g. @xmath286141 ) , but we stay within that limit for all the stellar and solar lines of , , , and . where abundances were determined from lines from two ionization stages for an element , an average was determined by weighting the results by the numbers of lines measured . also included in those tables are the results for the sun from our lunar spectrum from keck / hires . for these abundances
we used the kurucz solar atmosphere and the lines for each species that appear in the last column of table 2 .
because the metallicity of the sun is considerably higher than that of the m71 stars , we could not use exactly the same set of lines for the solar abundance determinations .
this means that accurate @xmath33 values might play a role in determining accurate normalized abundances [ x / h]an issue which we explore below .
however , there are a large number of well - determined solar equivalent widths in the set we used for the stars
. the solar abundances also appear in tables 4 , 5 , and 6 .
we have used the solar abundances that we determined from the solar / lunar keck spectrum for consistency in the analysis .
our results are in good agreement with the compilation of grevesse & sauval ( 1998 ) , usually within 0.05 dex . however , our ca and ni abundances are 0.18 and 0.20 dex lower than the respective solar photospheric values of grevesse & sauval . an abundance summary with abundances normalized to fe is given in table 7 .
the @xmath34 values here represent the error in the mean .
the internal uncertainty , @xmath35 , in the mean abundance for each species in the five individual stars and the sun are given in tables 4 , 5 , and 6 , along with the number of lines utilized in determining the mean abundance ; the former come from the line - to - line scatter for each species .
for those species where only 1 or 2 lines were used , @xmath35 could not be determined . in these cases ,
a conservative value had to be estimated for later use . in table 8 ,
internal uncertainties in the mean cluster abundance ratios ( column 3 ) are estimated from the star - to - star scatter of the prenormalized abundances . in column 4
we give the internal uncertainties in the mean abundance ratios that additionally include contributions from the solar values utilized in the abundance normalizations .
we note that , for elements with multiple species , all these listed uncertainties include contributions from each species weighted in the same fashion as the mean abundances .
table 8 also lists the sensitivities of the [ fe / h ] and [ x / fe ] ratios to the changes in parameter values of t@xmath18 , log g , and @xmath24 given above columns 5 - 7 .
typical errors are @xmath250 k in t@xmath18 , @xmath20.15 in log g , and @xmath20.15 in @xmath24 .
sensitivities include correlations of errors in element x and fe .
we note that all sensitivities take into account any differences arising from utilizing different ions for a given element .
sensitivity contributions from different species of a given element were weighted in the same fashion as for the mean abundance determinations , i.e. by the number of lines used . based upon the photometric errors and our parameter investigations , we believe that reasonably conservative estimates of random 1@xmath34-level parameter uncertainties are @xmath36 k , @xmath37 dex , and @xmath37 km / s in @xmath22 , log @xmath38 , and @xmath24 , respectively .
this means that total uncertainties in the mean abundance ratios are @xmath39 dex except for [ mg / fe ] and [ ba / fe ] ; the latter two ratios have total uncertainties of @xmath400.12 - 0.13 dex .
we investigated the effects of our choice of @xmath33 values by redetermining abundances for each species utilizing only those specific lines for which solar abundances abundances were also calculated .
the abundance differences ( `` new '' minus `` old '' ) are given for each star in the fourth row of each species in table 4 for those species which utilized different line samples for stellar and solar abundance determinations .
for example , had we used only those lines which were also measurable in the sun , our -based mean stellar log n(fe ) value would have been 0.060 dex lower .
given the lower weight assigned to the lines in determining [ fe / h ] , however , this difference effects our final results negligibly .
indeed , none of our abundances is significantly affected by line sample ( and , thus , presumably @xmath33 choice ) except na . table 5 indicates that a truly consistent solar normalization of our lines would have raised our final [ na / fe ] ratio by an additional @xmath41 dex .
whether this systematic change yields a more accurate abundance compared to a value derived from additional lines is , of course , unclear .
however , even the difference for this one element is modest .
the abundances for fe for each star are given in table 4 .
for each ion we give the 1@xmath34 uncertainty and the number of lines used .
the final abundance is the combined result obtained by weights equal to the number of spectral lines that were used for the determination .
for fe we have 79 - 86 fe i lines and 8 - 11 fe ii lines for each star that give the final fe / h .
there are no significant star - to - star variations among the five stars so the mean for the five - the cluster mean - is listed in the final column .
also in table 4 are the results for the sun from our lunar spectrum with the equivalent widths given in table 2 .
we used our solar values , obtained with the same spectrograph , to find [ fe / h ] .
our mean m 71 fe abundance is [ fe / h]@xmath42 dex , where the uncertainty includes internal uncertainties plus those in the parameters .
the final column of table 4 indicates that our -based metallicity ( [ fe / h]@xmath43 ) is in excellent accord with the -based value ( [ fe / h]@xmath44 ) within the internal uncertainties alone ( 0.03 and 0.07 dex respectively ) .
for these five turn - off stars we find @xmath45[fe / h]@xmath46 = @xmath90.80 @xmath20.02 , where the uncertainty is just the error in the mean .
this is in excellent agreement with [ fe / h ] = @xmath90.79 @xmath20.01 from ten red giant stars in m71 of sklps .
these comparisons suggest that any intra - cluster dwarf - giant fe discrepancy is small .
this is in contrast to our previous study of the metal - poor globular m92 ( king et al .
1998 ) , where we found fe abundances to be nearly a factor of two smaller in near - turnoff stars than deduced by others in m92 red giants .
one possible explanation may be the action of nlte effects in the near turnoff stars ( king et al .
thvenin & idiart ( 1999 ) have studied nlte effects on lines in metal - poor stars .
examination of their figure 9 suggests that , at the lte - based fe abundance of m92 determined by king et al .
( 1998 ) ( @xmath47 ) , nlte corrections are indeed 0.3 dex . since these corrections drop to @xmath48 dex at our lte - based m71 fe abundance , concordance with previous giant - based determinations is not remarkable . alternatively , as mentioned by king et al .
1998 , it is possible that the m92 dwarf - giant difference is not real or is due to analysis differences .
indeed , kraft & ivans ( 2003 ) suggest that their analysis of and lines in m92 giants indicates the thvenin & idiart ( 1999 ) nlte corrections are too large .
kraft & ivans ( 2003 ) conclude that the metallicity of m92 red giants is in the range @xmath49 to @xmath50 , consistent with the earlier estimate of @xmath47 by peterson , kurucz & carney ( 1990 ) and the king et al .
1998 turnoff results .
the agreement in [ fe / h ] between the turn - off stars and the red giants in m71 is a confirmation that high quality spectra and model atmospheres can provide the same benchmark abundance of fe for this cluster .
we can now examine the difference in other element ratios in giant vs. turn - off stars to explore whether nuclear processing and mixing have altered the abundances in the giants .
the abundances of the other elements are given in tables 5 and 6 and the mean cluster abundances and parameter sensitivities are given in tables 7 and 8 .
looking at the abundances of the various species in our five m71 stars , there is no firm evidence for genuine star - to - star differences ; in particular , no star stands out has having consistently deviant abundances given the uncertainties .
depending on the species , we infer that any genuine star - to - star differences must be at a level of @xmath51 dex for the particular elements in our sample of stars .
our first comparison is between our m71 stars and field stars of similar metallicity that have been analyzed by stephens & boesgaard ( 2002 ) ( hereafter sb02 ) because we have used the same line list and the same procedure .
there are eight stars in that sample with [ fe / h ] between @xmath90.6 and @xmath91.2 .
the abundances in these eight stars are shown in table 9 .
figure 4 shows the abundances of these fe - normalized abundances with our mean for m71 as a function of [ fe / h ] for four alpha - elements , two fe - peak elements and two n - capture elements .
compared to the field halo stars , m71 is high in ca ( but would be in the midst of the halo star points at + 0.14 had we used the grevesse & sauval ( 1998 ) solar ca of 6.36 ) . although the field stars appear to be low in ni , which is discussed by sb02 , the m71 point would fall among them at @xmath90.10 with the grevesse & sauval ( 1998 ) solar ni of 6.25 .
the field stars are also apparently low in na ( see below ) .
edvardsson et al . (
1993 ) ( hereafter eaglnt ) derived abundances of 13 elements in some 189 field f and g dwarf stars in the galactic disk .
several of these stars have [ fe / h ] values similar to that of m71 .
we have separated out those stars with [ fe / h ] within @xmath20.20 dex of @xmath90.80 to compare with our m71 turn - off mean abundances ; there are 17 such disk stars and their abundances are given in table 9 .
figure 5 is a similar plot to figure 4 with the disk stars as filled circles ; eaglnt did not determine [ cr / fe ] so the lower left panel in figure 5 is [ na / fe ] rather than [ cr / fe ] as in figure 4 .
we can compare figures 4 and 5 of the abundances in the halo field stars and in the disk dwarfs with our m71 results . note that although the samples of the disk stars of eaglnt and those of the halo stars of sb02 have similar metallicities , the orbital characteristics are very different in the two samples .
we have examined the u , v , w velocities , the perigalacticon distances , r@xmath52 , and the maximum distance above the galactic plane , z@xmath53 , for all the stars in each sample .
for example , the @xmath54u@xmath54 values are 4 - 123 km s@xmath55 in the disk stars , but 180 - 358 km s@xmath55 in the halo stars . with the exception of one disk star
, z@xmath53 is @xmath45 0.85 kpc for the disk stars , while the halo star sample has z@xmath53 values are @xmath461.2 kpc , except for one star with low z@xmath53 but with @xmath54u@xmath54 = 254 km s@xmath55 .
the m71 means for the @xmath3-elements appear to match the abundances of the disk dwarfs of comparable metallicity better than they match those of the halo field stars .
the value of [ mg / fe ] is at the lower part of the range for both the disk stars and the halo stars , while [ si / fe ] , [ ca / fe ] , and [ ti / fe ] are somewhat higher in m71 than in the halo stars . according to cudworth ( 1993 ) the metallicity and u , v , w velocities of m71 , @xmath976 , @xmath960 and @xmath93 km s@xmath55 ,
are characteristic of the thick disk population .
the abundance of the n - capture element , ba , in m71 is well - matched by the abundances in the halo stars , but the mean value is more than a factor of two higher than [ ba / fe ] in the disk stars .
figure 6 shows the elemental abundances normalized to fe as a function of atomic number and includes values for our m71 turn - off stars , the ten m71 red giants of sklps , and the halo field stars .
the giants and turn - off stars agree well in na and ni .
the alpha - element ca seems high in the turn - off stars , but would fit perfectly at + 0.14 with the grevesse & sauval ( 1998 ) solar ca .
the alpha - elements ti and si are somewhat lower in the turn - off stars than in the giants . here
our ni abundance is in excellent agreement with the values from the sklps giants .
there are three rc02 m71 stars that are similar to ours in temperature ( @xmath05800 k ) and log g ( @xmath04.05 ) values , which they class as turn - off stars . although the s / n ( per pixel ) at 46 , 44 , 32 for their three stars are somewhat lower than ours for our five stars , their abundances provide a reasonable comparison for our own results . if we determine the mean [ fe / h ] from those three stars the same way as we have done for ours , i.e. weighing the results from fe i and fe ii by the number of lines used , we determine @xmath45[fe / h]@xmath46 = @xmath90.76 @xmath20.05 , which is the same within the errors as our mean of @xmath90.80 @xmath20.02 .
figure 7 shows the mean abundances normalized to fe for our five turn - off stars and the three from rc02 along with the abundances of 19 red giants from rc02 .
one sigma error bars are shown which are the error in the mean .
generally , the abundances agree to within 1 - 2 @xmath34 for the turn - off stars .
an exception is [ si / fe ] , but although their two stars agree with each other , each has errors of @xmath20.13 dex .
they were only able to measure cr from one line in one star , so that discrepancy is not important ; our cr abundances come from 16 - 19 lines of cr i and cr ii in all five stars .
their ni abundance is from 2 - 7 lines in three stars , while ours is from 15 - 18 lines in five stars .
similarly , we have 21 - 25 lines of ti i and ti ii for our ti abundance in the five stars versus their 4 - 6 ti i lines in three stars .
inasmuch as we concentrated on five turn - off stars , we could measure 2 - 5 times as many lines as they did for most species , i.e. our turn - off abundances are are based on much more data . however , they examined a large range in stellar luminosity and evolutionary stage and determined abundances of 23 elements , many more than our sample of 10 elements .
it is interesting to compare their results for 19 red giants with those of sklps for ten red giants and for their red giants with our turn - off stars .
the agreement for the two sets of red giants is good for na , si , and ni .
the situation for ca and ti seems confusing ( see discussion below about alpha - elements ) .
their red giants seem to be enhanced in mg and , to a lesser extent , in na compared to our turn - off - star abundances .
the other two fe - peak elements , cr and ni , are very similar to fe in our m71 stars .
we derive abundances of [ cr / h ] of @xmath90.75 @xmath20.04 from 16 - 19 cr i and cr ii lines per star and [ ni / h ] @xmath90.71 @xmath20.04 from 15 - 18 ni i lines per star .
the fe normalized means for the cluster are [ cr / fe ] = + 0.05 @xmath20.04 and [ ni / fe ] = + 0.09 @xmath20.04 .
we can compare our results for ni with those of sklps which are based on 3 - 5 ni lines in ten giants in m71 ( see figure 6 ) .
( they did not measure cr . )
their value of @xmath45[ni / fe]@xmath46 = 0.07 @xmath20.04 ( with @xmath34=0.11 ) is the same as ours to within the errors .
the 19 red giants of rc02 have @xmath45[ni / fe]@xmath46 = + 0.02 @xmath20.01 ( with @xmath34=0.04 ) from 13 - 38 ni i lines .
the similarity of the abundances derived for our turn - off stars and the two sets of giant stars for ni indicates a constancy between the evolved and unevolved stars . in the case of cr rc02
find a lower @xmath45[cr / fe]@xmath46 = @xmath90.13 @xmath20.01 ( with @xmath34=0.06 ) from 3 - 10 cr i lines compared to our + 0.05 @xmath20.04 ( see figure 7 ) .
( their [ cr / fe ] for their one turn - off star is still lower at @xmath90.17 . )
our [ cr / fe ] ( + 0.05 @xmath20.04 ) is the same to within the errors of the halo stars of sb02 ( + 0.02 @xmath20.02 ) .
we have determined na abundances in each star from 3 - 4 na i lines .
these results appear in table 5 .
there appear to be no variations from one star to the next .
our mean ratio [ na / fe ] = + 0.14 @xmath20.04 agrees well with the @xmath56 value from rc02 from 3 na i lines for three turn - off stars just considering our internal uncertainty alone .
figure 8 shows our m71 results for [ na / fe ] compared to those of the halo field stars of sb02 ( @xmath45[na / fe]@xmath46 = @xmath90.23 @xmath20.13 ) and the disk dwarf stars of eaglnt ( @xmath45[na / fe]@xmath46 = 0.10 @xmath20.01 ) .
the na content of the m71 turn - off stars is remarkably similar to that of the disk dwarfs and more than a factor of two higher than that in the halo stars with similar [ fe / h ] .
our [ na / fe ] of + 0.14 for the turn - off stars has a spread of + 0.01 to + 0.23 , less than half that found in the red giants by both sklps and rc02 .
the spread in our turn - off stars is consistent with the uncertainties in the determinations of [ na / h ] for the individual stars and provides no indication of an intrinsic spread . in their study of 10 red giants in m71 , sklps derived na abundances from 4 na i lines which showed a spread in [ na / fe ] from -0.04 to + 0.53 and a mean of + 0.24 @xmath20.06 ( with @xmath34=0.18 ) .
similarly , rc02 find @xmath45[na / fe]@xmath46 = + 0.24 @xmath20.03 ( with @xmath34=0.14 ) and a range of + 0.04 to + 0.57 in their 19 red giant stars .
as can be seen in figures 6 and 7 there is little difference in the _ mean _ [ na / fe ] content between the m71 turn - off stars and the m71 red giant stars .
however , the sizeable spread in the abundances for individual giant stars indicates that some of the giant stars have na enhancements , as found in several globular clusters ( e.g. in m3 by kraft et al .
1992 , in m13 by kraft et al . 1993 , 1997 ) .
an increased na could result from the capture of a proton by @xmath57ne to form @xmath58na ; this could occur in the h - burning shell .
the resultant na is dredged up to the photosphere where it can be observed .
an _ in situ _ processing scenario may or may not be responsible for the cluster s mild o - na anti - correlation or the range in na enhancements in the red giants
. figures 9a and 9b show comparisons of na with mg for m71 and the two sets of field stars . whereas mg seems similar to the halo stars , the ratio of [ na / mg ] of + 0.05 is higher than that in any of the halo or disk field stars .
( one can see the m71 point as an extension of the disk - star points toward higher [ na / mg ] at lower [ mg / h ] ; this is analogous to the trend of [ o / fe ] vs. [ fe / h ] , but with mg as the chronometer rather than fe . ) for the mean of the 19 red giants , rc02 find [ na / mg ] = @xmath90.12 from an apparent enhancement in mg in the giants .
we have calculated abundances for four alpha - elements : mg , si , ca , and ti . for mg
there are only 2 - 3 lines that we can use for the abundance determination , only 4 for si , but 18 - 20 for ca and 21 - 25 for ti . for both ti and
si we have used lines from neutral and ionized states .
the abundances for these @xmath3-fusion elements , along with the 1@xmath34 uncertainty and the number of lines used in the abundance calculation are presented in table 5 .
the fe - normalized ratios are in table 7 along with the weighted mean of [ @xmath3/fe ] .
( the weights are the number of lines used in the determination of the abundance of each element . )
again we see no significant star - to - star variations so we have found a cluster average for turn - off stars for each element , both [ @xmath3/h ] and [ @xmath3/fe ] .
the average alpha - enhancement is [ @xmath59/fe ] = + 0.29 @xmath20.05 .
our @xmath3-element ratios [ si , ca , ti / fe ] show a clear overabundance given the mean values of + 0.22 , + 0.32 , and + 0.30 with typical total internal uncertainties of 0.05 dex .
rc02 find these same ratios from their 2 - 3 turn - off stars of + 0.01 , + 0.40 , and + 0.16 . for their 19 red giants they find + 0.25 , + 0.43 , and + 0.21 with uncertainties of typically 0.02 dex .
the red giants of sklps have those ratios as + 0.31 , + 0.14 , and + 0.48 with typical uncertainties of 0.04 dex .
the straight means for these three @xmath3 elements are + 0.30 ( us ) , + 0.19 ( rc02 t.o . ) , + 0.29 ( rc02 giants ) , and + 0.31 ( sklps ) ; this is a remarkably consistent alpha enhancement in this cluster in both turn - off and giant stars .
our mean [ mg / fe ] is + 0.09 .
this is lower than the rc02 value of + 0.28 ( based on only two turn - off stars they measured ) and lower than the rc02 value of + 0.37 ( based on 19 red giants ) .
it is also lower than the results of shetrone ( 1996 ) who found + 0.32 @xmath20.04 from eight m71 giants .
however , our internal uncertainties for mg are larger than for our other elements .
it is possible that our [ mg / fe ] is in accord with the delayed rise in [ mg / fe ] suggested by fuhrmann et al .
( 1995 ; see their figure 5 ) and implies a genuine dwarf - giant difference .
further , the nlte reanalysis of galactic mg ratios by idiart & thvenin ( 2000 ) suggests there is significant scatter ( @xmath00.6 dex ) in [ mg / fe ] at [ fe / h ] @xmath600.8 .
while we have not done the nlte calculations here , we surmise that our m71 turn - off stars with [ mg / fe ] @xmath0 + 0.1 simply lie in the midst of this scatter .
table 6 contains the abundance results for our five m71 turn - off stars for y and ba based on 3 lines of y ii and 2 of ba ii and table 7 gives the mean for each element normalized to fe .
we find and [ ba / fe ] = + 0.23 @xmath20.06 , [ y / fe ] = + 0.15 @xmath20.05 with [ ba / y ] = + 0.08 .
figures 4 and 5 show how our results compare to field stars of similar metallicity in the halo ( sb02 ) and in the disk ( eaglnt ) .
our m71 stars are quite similar to the halo stars in [ ba / fe ] where the halo mean is + 0.16 @xmath20.01 , but a factor of 2 larger than the disk stars where @xmath45[ba / fe]@xmath46 = @xmath90.08 @xmath20.01 . for [ y / fe ]
our mean of + 0.15 is higher than the value of @xmath90.03 found for both the halo stars and the disk stars .
these enhancement are modest , and only marginally significant given the uncertainties , but may simply be the consequence of exposure of pre - cluster material to slightly higher neutron flux than that received by most disk field stars .
the halo stars have [ ba / y ] = + 0.19 , while the disk stars [ ba / y ] = @xmath90.05 compared to our intermediate value of + 0.08 .
these heavy - to - light ratios , a measure of neutron exposure , are certainly not in conflict with this idea .
the three m71 turn - off stars studied by rc02 have no y determinations and show a large range in [ ba / fe ] : + 0.22 , + 0.33 and + 0.61 .
the quoted error for the third star is @xmath20.32 ; the mean for the other two is + 0.27 , similar to our [ ba / fe ] of + 0.23 @xmath20.06 .
their three evolved stars which do have determinations of [ y / fe ] yield @xmath90.22 while the 25 cluster stars of all evolutionary phases give [ ba / fe ] = + 0.34 . their [ ba / y ] of + 0.56 is a high value , comparable to those seen in some ba , s , ch , stars and @xmath61 cen giants . such large ratios can be produced by very large neutron exposure , which yields significant production of heavy-@xmath4 elements ( e.g. , ba ) at the expense of light-@xmath4 elements ( e.g. , y and zr ) as indicated by both early clayton et al .
( 1961 ) and more recent theoretical calculations ( figure 5 of busso et al .
the expectation from the @xmath4-process calculations is that the adjacent light-@xmath4 element zr should show an abundance comparable or just slightly larger than y , and this is seen in the results of rc02 whose cluster results indicate [ y / zr]@xmath62 .
this ratio is unremarkable compared to field stars of similar metallicity ( figure 29 of edvardsson et al .
1993 ; figure 5 of gratton & sneden 1994 ) ; thus , the rc02 zr results seem to confirm their y results .
these heavy elements have been studied in globular clusters and field halo stars by james et al .
( 2004 ) . for their five field stars with [ fe / h ] similar to @xmath90.80 for m71 ,
we find [ ba / fe ] = + 0.29 , [ y / fe ] = @xmath90.05 , and [ ba / y ] = + 0.34 . the difference is mainly in the low [ y / fe ] values .
for the three turn - off stars in the globular cluster 47 tuc , which has [ fe / h ] = @xmath90.69 , they find [ ba / fe ] = + 0.22 , [ y / fe ] = + 0.06 , and [ ba / y ] = + 0.16 , values very similar to ours .
we can conclude that the primordial material from which m71 formed underwent at least modestly larger neutron exposure than local field disk stars of similar [ fe / h ] .
this appears to be true in 47 tuc as well as deduced from their heavy element abundances in turn - off stars .
the 0.6 dex [ ba / y ] difference between our m71 turn - off stars and the cluster giants of rc02 is surprising , and has potentially significant implications for cluster self - enrichment as manifested by the effects of convective dilution .
the reality of this difference needs confirmation by additional observations of light s - peak elements ( y , rb , sr , zr ) in m71 . in this regard ,
we note the intriguing circumstance that the mean [ zr / fe ] ratio ( -0.23 + /-
0.09 ) of rc02 s 7 most highly evolved stars ( log g
< = 1.75 ) is lower than that ( + 0.10+/-0.13 ) of their 3 least - evolved stars ( log g>=2.70 ) with zr measurements at the 2.1 sigma level . in their study of 10 m71 _ giants
_ , sklps noted `` normal '' 0.2 - 0.3 dex enhancements of [ si / fe ] and [ ca / fe ] . these match our near - turnoff ratios well
. however , they note that their ti ratio is unusually large : [ ti / fe]@xmath64 , and call attention to the fact that a similar dichotomy between [ ti , mg / fe ] and [ si , ca / fe ] is seen in moderately metal - poor disk and bulge giants ( e.g. , mcwilliam & rich 1993 ) _ and _ in some disk dwarfs eaglnt .
however , both our [ ti / fe ] ratios and those of rc02 do not confirm the larger ti overabundances noted by sklps ; rather , they suggest unremarkable values of @xmath65 to @xmath66 indistinguishable from si and ca . in our previous keck / hires
work on m92 ( king et al .
1998 ) , we suggested that pollution of these stars atmospheres along with later _ in situ _ processing was needed to understand the global picture of this cluster s abundance patterns .
whether such pollution was in the form of abundance variations or previously processed material in the proto - cluster gas or whether it happened after ( at least the first generation of ? ) cluster stars formed ( self - enrichment ) was not clear . while the idea of such pollution and intra - cluster abundance variations from a _ non in situ _ source may not be easy to accept , we note additional support for this picture in m92 . in their nlte reanalysis of numerous
literature mg and ca data , idiart & thvenin ( 2000 ) include our m92 star equivalent widths .
their figure 3 contains the striking result that the nlte [ ca / fe ] abundances in six m92 stars fall nearly precisely along 3 bifurcated sequences defined by field star data . while significant scatter is present for mg field star data , they suggest a similar trichotomy for the nlte [ mg / fe ] ratios . as suggested by idiart & thvenin ( 2000 ) these intra - cluster differences
may indeed be similar to those now well - known in @xmath61 cen , and presumed to arise from self - enrichment from type ii supernovae .
moreover , the correspondence of the m92 star [ ca , mg / fe ] ratios and the field star sequences may provide renewed support for the idea of globular clusters as the original reservoirs of halo field stars .
regardless , investigating the correspondence of m 71 [ ca , mg / fe ] ratios and the sequences defined by field stars is clearly important . however , it is not yet clear that this important task is possible . at [ fe / h]@xmath12 , the field star [ mg , ca / fe ] sequences in figure 3 of idiart & thvenin ( 2000 )
are separated by a modest 0.05 - 0.1 dex , which is comparable to or smaller than the internal uncertainty in our star - to - star mg abundances .
moreover , our data need to be subjected to a nlte analysis in an uniform as possible manner compared to idiart & thvenin ( 2000 ) . at present
, we can only note that there is no strong evidence of intra - cluster [ ca / fe ] differences in m71 from our data ; the largest difference seen between our five stars is 0.16 dex , which is a 3@xmath67 difference based on the internal star - to - star uncertainties alone ; the [ ca / fe ] ratios for the three turn - off stars of rc02 show no evidence of statistically significant bifurcation either .
magnesium abundances of additional little - evolved m71 stars using even higher - quality data would be of great interest in pursuing unsettled questions regarding star - to - star differences perhaps indicative of cluster self - enrichment or primordial abundance variations . while it was unclear from the work of king et al .
( 1998 ) whether any _ non in situ _ abundance variations in m92 were primordial or the product of cluster self - enrichment , they noted the potential importance of turnoff - giant comparisons in addressing this issue .
if self - enrichment occurs after stars have formed , any products that are enhanced ( at levels which are easy to detect ) in the near turnoff stars may not appear to be enhanced in the red giants .
the turn - off stars have relatively thin surface convection zones and the enhancements there may subsequently become unrecognizably diluted as these stars ( or stars just slightly more massive ) develop deep surface convection zones as they ascend the red giant branch .
king et al . ( 1998 ) noted that the @xmath68 dex lower abundance for [ mg / fe ] in their m92 near turnoff stars compared to the m92 giants in shetrone ( 1996 ) might indicate the action of self - enrichment in the giants .
the difference between the mean m92 near turnoff nlte [ mg / fe ] ratios from idiart & thvenin ( 2000 ) and shetrone s ( 1996 ) giant mean remains large at @xmath69 dex .
comparison of intra - cluster dwarf - giant abundances of @xmath3-elements is an efficient manner to search for self - enrichment signatures since the production of these is canonically presumed to occur in explosive supernovae having massive ( i.e. , short - lived ) progenitors .
we note that , in principle , such stars could also be responsible for @xmath70-process contributions to a possible dwarf - giant @xmath10-capture difference noted above ; while we refer to our heavy elements as `` @xmath4-process '' ones , it must be remembered that such designations are based on _ solar system _ data . in m71 , the rc02 results for ti and ca do not provide any evidence for significant near turnoff - giant differences that might be taken as evidence of self - enrichment .
our near - turnoff [ mg / fe ] ratio of @xmath71 is considerably smaller than the value of the rc02 evolved stars ( @xmath72 ) , which itself should be raised ( or ours lowered ) by 0.04 dex to account for the modest fe difference between our studies . while interesting , the meaning of this difference is unclear since the two turn - off stars from rc02 give [ mg / fe]@xmath73 , in accord with their giant results .
our turn - off [ si / fe ] ratio is + 0.22 @xmath20.04 , essentially the same as that of rc02 for evolved stars of @xmath74 dex .
we conclude that no @xmath3 abundance difference is clear between the giants and the turn - off stars .
if the @xmath10-capture difference noted above is real and resulted from self - enrichment , then we surmise that polluting material arose in a traditional @xmath4-process environment though perhaps one providing an unusually large neutron fluence .
we reiterate that mapping any variation of light @xmath4-element variation with evolutionary status in m71 is critical to determining whether such pollution happened before cluster formation or via self - enrichment after star formation in the cluster .
we have derived abundances of numerous elements in five stars near the turnoff of the relatively metal - rich globular cluster m71 via high - resolution ( @xmath75 ) spectroscopy with s / n of 50 - 60 per pixel acquired with keck / hires .
our five stars are very similar to each other in temperature , luminosity , mass , and age . our derived fe abundance , [ fe / h]@xmath42 is in excellent accord with the values measured in both giants and near - turnoff stars by sneden ( 1994 ) and rc02 .
the lack of a dwarf - giant fe discrepancy like that we found in the metal - poor cluster m92 may reflect the reduced effects of nlte on the abundances in the metal - rich case as suggested by the work of thvenin & idiart ( 1999 ) . within our own limited data set
, we find no evidence of star - to - star abundance deviations in any element .
furthermore , there are no stars that are consistently higher or lower than the others in the abundances of the various species we studied ; among other things , this indicates that our relative temperatures are correct .
we have compared our abundances with two field star samples with metallicities similar to that in m71 : eight halo stars from sb02 and 17 disk stars from eaglnt .
our @xmath3-product abundances , mg , si , ca , ti , seem to fit with both the disk and the halo stars , but perhaps a somewhat better match with the disk stars .
( the space motion of m71 is similar to that of disk stars according to cudworth & hanson ( 1993 ) . ) for mg the m71 mean is near the lower edge of both samples .
our m71 mean is more similar in ni to the disk stars , but cr matches the halo sample well .
whereas the light s - process element y is enhanced in m71 relative to both field stars samples , the heavier s - process element ba is like the halo stars and enhanced more than a factor of 2 above the disk stars .
all the @xmath3 elements we observed ( mg , si , ca , ti ) are enhanced relative to fe as compared to the sun .
the mean enhancement of @xmath45[@xmath3/fe]@xmath46 is + 0.29 @xmath20.05 .
this is strikingly similar to the mean enhancements in those elements in the red giants in m71 studied by sklps and rc02 .
it is known that m71 evinces o - na anticorrelations down to the main - sequence turn - off ( rc02 ) that are presumed to arise from @xmath5-capture reactions .
the totality of data now available for m71 suggests to us the following two conclusions .
first , the similarity of our turn - off na abundances , as well as those of rc02 , with the values observed in m71 giants ( sklps ; rc02 ) indicates that this @xmath5-capture mechanism is not an _ in situ _ process unless there is a glaring fundamental failure in our basic understanding of stellar structure .
the atmospheres of m71 stars must be polluted by material having undergone such processing elsewhere .
second , the lack of clear evidence for intracluster @xmath3-element variations suggests that this processing was not the result of self - enrichment via explosive nucleosynthesis a process which could , in principle , also affect the heavy elements if the @xmath70-process came into play .
the m71 [ ba / fe ] ratio suggests that material in m71 stars once was present in an @xmath4-process environment of at least modestly higher neutron fluence compared to nearby field stars of similar [ fe / h ] .
it remains unclear whether the @xmath4-process polluting source acted prior to cluster formation or after ( via self - enrichment ) .
if there is a genuine dwarf - giant y and/or zr difference in m71 , then it would suggest that a source of modest neutron fluence self - polluted m71 ( to produce high light @xmath4 abundances perhaps seen near the turnoff ) in addition to a very high neutron fluence source acting prior to cluster formation ( to produce the high heavy - to - light @xmath4 ratios perhaps seen in the giants ) .
consistent determination of light @xmath4-elements ( rb , sr , y , zr ) in a considerably larger sample of m 71 turn - off , subgiant , and giant stars is needed to address this critical issue ; the possibility of intra - cluster mg variations also merits additional study . finally , we note the recent work on elemental abundances in the globular clusters m4 ( [ fe / h]@xmath76 ) and m5 ( [ fe / h]@xmath77 ) by ivans et al .
( 1999 , 2001 ) .
ivans et al .
( 2001 ) call attention to the differences in these clusters based upon the morphology of the o - na anticorrelation patterns and hb morphology . in particular , they note the similarity in the light element processing patterns between m4 and m71 , as well as the hb morphology .
indeed , it might be hoped that such an association is a meaningful classification perhaps related to the dominance of _ in situ _ processing versus pollution , or perhaps simply related to metallicity since this parameter affects the molecular weight gradient relevant for deep mixing in stellar interiors as well as serving as the `` first parameter '' controlling hb morphology .
the factor of @xmath78 difference in the m4 and m71 [ ba / fe ] ratios , however , reminds us that the detailed chemical evolution histories of globular clusters are individually unique . as is clear from this investigation ,
unraveling these histories will require continued study on a star - by - star and cluster - by - cluster basis involving determination of abundances in near turnoff stars via high - resolution spectroscopy on current and future generations of large aperture telescopes .
this work has been supported by an nsf award to amb , ast-0097955 , and the nsf reu program award to the institute for astronomy at the university of hawaii , ast-9987896 .
jrk gratefully acknowledges support for this work from nsf awards ast-0086576 and ast-0239518 as well as a generous grant from the charles f. curry foundation to clemson university .
lccclcc 239 ..... & 17.62 & 0.81 & 0.53 & d , d , f@xmath79 , g & 3 15 & 60 + 259 ..... & 17.66 & 0.82 & 0.54 & a@xmath79 , b , c , c , g&4 00 & 50 + 260 ..... & 17.66 & 0.79 & 0.51 & d , d , g@xmath79 & 2 30 & 50 + 264 ..... & 17.67 & 0.81 & 0.53 & d , d , d & 2 15 & 48 + 273 ..... & 17.68 & 0.81 & 0.53 & b , c , c , e , f@xmath79 , f & 4 45 & 60 + llrrrrrrr & + 5682.650 & 2.1000 & @xmath90.8200 & 40.2 & 45.6 & 48.1 & 38.6 & 40.1 & 106.5 + 5688.219 & 2.1000 & @xmath90.3700 & 49.5 & 69.9 & 63.7 & 65.4 & 62.2 & + 6154.227 & 2.1000 & @xmath91.6600 & 7.4 & 13.9 & 14.7 & & 8.6 & 37.8 + 6160.751 & 2.1000 & @xmath91.3500 & 12.5 & & 16.4 & 27.3 & 15.5 & 57.5 + & + 4571.099 & 0.000 & @xmath95.5850 & 71.5 & 71.5 & 73.3 & 63.0 & 71.3 & 111.0 + 4703.98 & 4.340 & @xmath90.5350 & 160.3 & 148.6 & 118.5 & 149.8 & 132.9 & + 4730.028 & 4.340 & @xmath92.4300 & 17.7 & 14.7 & & & 17.5 & 74.0 + 6318.72 & 5.110 & @xmath91.6840 & & & & & & 50.8 + & + 5772.148 & 5.0809 & @xmath91.7500 & 19.6 & 30.2 & 30.2 & 33.2 & 15.8 & 53.0 + 5948.545 & 5.0813 & @xmath91.2250 & 54.9 & 40.0 & 12.9 & 42.5 & 49.0 & 84.1 + 6155.141 & 5.6200 & @xmath90.8400 & 44.0 & 43.9 & 37.1 & 47.0 & 41.7 & 83.4 + & + 6347.100 & 8.1200 & 0.3199 & 28.3 & 23.0 & 23.0 & 29.9 & 30.7 & 45.5 + & + 4526.934 & 2.7092 & @xmath90.4890 & 50.0 & 53.1 & 42.8 & 56.8 & 44.5 & + 4578.551 & 2.5214 & @xmath90.6285 & 49.3 & 40.7 & & 51.6 & 49.6 & + 4685.268 & 2.9327 & @xmath90.8800 & 32.2 & 12.1 & 23.5 & & 14.8 & + 5188.844 & 2.9327 & @xmath90.0900 & & & & & & 91.1 + 5260.389 & 2.5210 & @xmath91.8095 & 13.9 & 11.1 & 9.1 & 10.9 & & + 5261.707 & 2.5214 & @xmath90.6545 & 62.6 & 58.0 & 45.5 & 60.9 & 62.1 & 98.2 + 5512.980 & 2.9327 & @xmath90.3685 & 49.7 & 46.5 & 47.3 & 34.9 & 51.8 & 86.5 + 5581.968 & 2.5232 & @xmath90.6325 & 73.2 & 55.2 & 50.3 & 66.3 & 66.3 & 94.4 + 5590.117 & 2.5214 & @xmath90.6405 & 56.1 & 52.1 & 63.1 & 63.3 & 50.8 & 91.1 + 5598.480 & 2.5214 & @xmath90.2200 & & & 98.6 & & & 138.4 + 5601.277 & 2.5259 & @xmath90.6900 & 69.3 & 55.8 & & 60.5 & 56.2 & + 5857.451 & 2.9327 & 0.2350 & 86.7 & 80.8 & 83.0 & 84.5 & 69.1 & 138.4 + 6163.755 & 2.5214 & @xmath91.2860 & 31.9 & 25.5 & 27.3 & 33.0 & 23.7 & + 6166.440 & 2.5214 & @xmath91.1400 & 34.6 & 37.6 & 25.3 & 36.6 & 23.9 & 70.3 + 6169.042 & 2.5232 & @xmath90.7970 & 59.3 & 48.0 & 49.7 & 58.9 & 49.0 & 93.7 + 6169.562 & 2.5259 & @xmath90.3740 & 60.6 & 56.4 & & 67.8 & 61.1 & 112.6 + 6449.810 & 2.5214 & @xmath90.5020 & 67.0 & 68.7 & 50.5 & 65.3 & 52.3 & 107.8 + 6455.600 & 2.5230 & @xmath91.3400 & 30.0 & 23.2 & 28.1 & 24.9 & 30.0 & 56.0 + 6471.661 & 2.5259 & @xmath90.6380 & 59.2 & & 41.7 & 56.4 & & 92.6 + 6493.781 & 2.5214 & 0.0155 & 87.4 & 87.2 & 75.5 & 98.5 & 76.8 & 126.7 + 6499.650 & 2.5232 & @xmath90.8180 & 44.0 & 38.4 & 33.0 & 57.6 & 36.1 & 86.3 + 6717.686 & 2.709 & @xmath90.5670 & 56.2 & 38.2 & 59.8 & 52.8 & 46.8 & + & + 4518.023 & 0.8259 & @xmath90.2690 & 48.0 & 39.7 & 38.3 & 40.5 & 31.7 & 69.8 + 4527.305 & 0.8130 & @xmath90.4700 & 42.2 & 40.4 & 40.9 & 46.4 & 38.2 & + 4534.778 & 0.8360 & 0.3360 & & & 71.1 & & & + 4535.570 & 0.8259 & 0.1200 & 67.7 & & & 69.0 & 63.0 & + 4544.688 & 0.8182 & @xmath90.5200 & & 35.2 & 39.9 & & & + 4617.254 & 1.7489 & 0.4450 & 25.6 & 28.1 & & 37.2 & 33.5 & 60.6 + 4681.908 & 0.0480 & @xmath91.0150 & 58.2 & 38.4 & & 51.7 & 33.0 & + 4840.874 & 0.8996 & @xmath90.4530 & 35.2 & 40.4 & 32.8 & 52.1 & 24.5 & 66.4 + 4999.504 & 0.8259 & 0.3060 & 74.5 & 66.7 & & & 74.1 & + 5016.162 & 0.8484 & @xmath90.5180 & 42.4 & 28.1 & 35.0 & 29.6 & 32.2 & 64.7 + 5020.028 & 0.8360 & @xmath90.3580 & 49.1 & 51.8 & 36.5 & 52.4 & 45.9 & 78.4 + 5022.871 & 0.8259 & @xmath90.3780 & 54.5 & 49.3 & 45.8 & 48.3 & 35.2 & 70.1 + 5035.907 & 1.4602 & 0.2600 & & & 59.0 & & & + 5036.468 & 1.4432 & 0.1860 & 43.7 & 43.2 & 53.4 & 51.5 & 47.8 & 74.3 + 5039.959 & 0.0211 & @xmath91.1300 & 49.6 & 49.0 & 34.2 & 50.5 & 42.4 & 73.2 + 5064.654 & 0.0480 & @xmath90.8550 & 58.8 & 46.1 & 41.3 & 62.9 & 56.0 & + 5192.969 & 0.0211 & @xmath90.9500 & 66.1 & 54.0 & 48.0 & 51.6 & & + & + 4501.272 & 1.1156 & @xmath90.7550 & & & & 112.7 & & + 4563.761 & 1.2214 & @xmath90.9600 & & & & 103.0 & & + 4571.968 & 1.5719 & @xmath90.5300 & & & & 108.2 & & + 4589.958 & 1.2369 & @xmath91.7900 & 61.6 & 70.5 & 47.1 & 73.9 & 54.0 & + 4657.203 & 1.2430 & @xmath92.2350 & & 36.9 & 27.1 & 43.9 & 43.7 & 51.4 + 4779.985 & 2.0478 & @xmath91.3700 & 50.0 & 53.3 & 48.0 & 56.5 & 47.4 & 62.2 + 4805.085 & 2.0614 & @xmath91.1000 & 63.7 & 66.4 & 56.6 & 80.0 & 58.1 & + 5129.152 & 1.8918 & @xmath91.3900 & & 48.5 & 69.3 & 73.6 & 45.8 & 79.4 + 5154.070 & 1.5659 & @xmath91.9200 & 53.7 & 44.8 & 43.2 & 75.3 & 47.2 & 72.4 + 5226.543 & 1.5659 & @xmath91.3000 & 78.6 & 81.9 & 60.9 & 82.9 & 73.8 & + 5336.781 & 1.5819 & @xmath91.6650 & 59.4 & 50.0 & 54.7 & 49.6 & 61.4 & 71.4 + 5381.018 & 1.5658 & @xmath92.0250 & 48.9 & 45.4 & 48.4 & 66.9 & 46.8 & 59.6 + & + 4496.842 & 0.9415 & @xmath91.1500 & & & 67.3 & & & + 4540.734 & 3.1046 & 0.0280 & & 16.6 & & & & 84.5 + 4545.945 & 0.9415 & @xmath91.3700 & 43.7 & & & & 41.5 & 84.5 + 4591.389 & 0.9685 & @xmath91.7400 & 26.1 & & & 44.9 & & 82.6 + 4600.741 & 1.0037 & @xmath91.2600 & 46.7 & 69.3 & & 61.6 & 42.7 & 92.1 + 4616.120 & 0.9829 & @xmath91.1900 & 42.7 & 55.4 & 44.1 & 55.5 & 45.8 & 87.6 + 4626.174 & 0.9685 & @xmath91.3200 & 48.0 & 44.4 & 49.3 & 44.0 & 38.9 & 78.7 + 4646.148 & 1.0301 & @xmath90.7000 & 68.9 & 66.5 & & 67.9 & & + 4651.282 & 0.9829 & @xmath91.4600 & 41.5 & 44.8 & 44.8 & 58.9 & 31.4 & 77.0 + 4652.152 & 1.0037 & @xmath91.0300 & 55.4 & 51.8 & 42.7 & 64.5 & 48.8 & 99.2 + 4718.426 & 3.1955 & 0.0900 & & 23.3 & & 21.6 & & 65.4 + 5247.566 & 0.9610 & @xmath91.6400 & 32.0 & 31.9 & 32.3 & 41.8 & 36.6 & 81.5 + 5296.691 & 0.9829 & @xmath91.4000 & 53.4 & 42.1 & 47.0 & 62.1 & 46.5 & 91.9 + 5298.277 & 0.9829 & @xmath91.1600 & & 53.7 & 63.7 & 79.7 & 57.3 & + 5345.801 & 1.0037 & @xmath90.9800 & 62.1 & 70.4 & 57.5 & 70.8 & 64.3 & 116.0 + 5348.312 & 1.0037 & @xmath91.2900 & 49.8 & 52.8 & 41.2 & 52.9 & 51.3 & 100.5 + 5409.772 & 1.0301 & @xmath90.7150 & 69.3 & 85.1 & 73.5 & & 70.4 & 141.4 + 6330.093 & 0.9415 & @xmath92.9200 & & & & & & 27.3 + & + 4558.650 & 4.0737 & @xmath90.6600 & 47.2 & 48.5 & 44.8 & 51.7 & 45.0 & 78.8 + 4588.199 & 4.0715 & @xmath90.6300 & 44.4 & 49.3 & 42.0 & 39.9 & 41.5 & 67.9 + 4634.070 & 4.0725 & @xmath91.2400 & & 21.9 & 26.2 & 33.3 & 36.9 & 61.5 + 4848.235 & 3.8647 & @xmath91.1400 & 30.5 & 40.7 & 20.1 & & 42.6 & 65.6 + 5237.329 & 4.0737 & @xmath91.1600 & 28.3 & 31.9 & 33.0 & 28.4 & 27.7 & 51.5 + & + 4489.739 & 0.1210 & @xmath93.9318 & 60.4 & 57.9 & 50.6 & 54.8 & 61.5 & + 4556.126 & 3.6030 & @xmath90.7870 & 61.9 & 67.8 & 47.0 & 61.2 & 53.3 & + 4592.651 & 1.5580 & @xmath92.4556 & 66.3 & & & & & + 4602.000 & 1.6080 & @xmath93.1439 & 24.2 & 27.7 & & & & + 4602.941 & 1.4850 & @xmath92.2080 & & & & & & 67.6 + 4619.287 & 3.6030 & @xmath91.1200 & 41.4 & 45.8 & 38.4 & 46.9 & 49.4 & + 4625.044 & 3.2410 & @xmath91.3400 & 45.7 & 55.6 & 50.7 & 55.6 & 48.2 & + 4630.120 & 2.2790 & @xmath92.5935 & 42.0 & & 35.2 & 44.3 & 34.1 & + 4637.503 & 3.2830 & @xmath91.3900 & 54.4 & 38.1 & 35.7 & 44.2 & 48.1 & + 4638.010 & 3.6030 & @xmath91.1195 & & 33.0 & 48.7 & 31.6 & 42.7 & + 4647.434 & 2.9490 & @xmath91.3305 & 60.7 & 62.7 & 46.9 & 50.6 & 56.9 & + 4669.171 & 3.6540 & @xmath91.3100 & & 33.5 & 22.1 & 24.1 & 36.6 & + 4678.846 & 3.6030 & @xmath90.7465 & 53.1 & 53.6 & 45.6 & 57.8 & 45.4 & + 4691.411 & 2.9910 & @xmath91.4865 & 55.7 & & 51.7 & 53.9 & 62.6 & + 4710.283 & 3.0180 & @xmath91.6120 & 56.6 & 53.9 & 47.0 & 60.4 & 50.0 & + 4728.546 & 3.6540 & @xmath91.1710 & 38.2 & 27.6 & 39.2 & 51.5 & 27.8 & + 4733.591 & 1.4850 & @xmath92.9872 & 47.3 & 46.1 & 59.8 & 54.8 & 50.1 & + 4736.773 & 3.2110 & @xmath90.7460 & 67.9 & & 62.0 & 70.2 & 69.9 & + 4741.529 & 2.8320 & @xmath91.8820 & 28.8 & 30.8 & 30.0 & 43.1 & 31.5 & 71.4 + 4882.144 & 3.4170 & @xmath91.6400 & 35.1 & 32.7 & & & 27.3 & 73.6 + 4924.770 & 2.2790 & @xmath92.2305 & 38.5 & 53.9 & 31.6 & 35.1 & 37.7 & + 4938.814 & 2.8760 & @xmath91.0770 & 73.8 & & 66.3 & & 64.6 & + 4939.687 & 0.8590 & @xmath93.3152 & & 55.4 & 50.9 & 72.4 & 60.3 & + 4946.385 & 3.3690 & @xmath91.1700 & 58.9 & 51.0 & 45.2 & 68.9 & 53.6 & + 4985.253 & 3.9290 & @xmath90.5590 & 42.8 & 44.8 & 41.1 & 43.6 & 59.1 & + 4985.547 & 2.8650 & @xmath91.3310 & 59.7 & 58.7 & 46.6 & 57.5 & 50.9 & + 4994.130 & 0.9150 & @xmath92.9690 & 68.0 & & 59.1 & & 68.3 & + 5001.862 & 3.8820 & 0.0100 & & & & & 70.5 & + 5022.236 & 3.9840 & @xmath90.5300 & & 52.7 & 50.6 & 70.6 & 57.1 & + 5028.127 & 3.5730 & @xmath91.1220 & 47.5 & 41.6 & 39.1 & & 48.5 & + 5044.212 & 2.8510 & @xmath92.0590 & 38.4 & 34.8 & 23.6 & 35.8 & 30.9 & 70.7 + 5049.819 & 2.2790 & @xmath91.3495 & & & 78.1 & & & + 5051.635 & 0.9150 & @xmath92.7640 & & & 76.6 & & & + 5068.766 & 2.9400 & @xmath91.1355 & 70.6 & & 64.1 & & 72.5 & + 5079.224 & 2.1980 & @xmath92.0860 & 70.6 & 63.2 & 52.2 & 57.8 & 61.8 & + 5079.740 & 0.9900 & @xmath93.2328 & 68.2 & 75.6 & 53.7 & & 50.2 & + 5083.339 & 0.9580 & @xmath92.8419 & 74.2 & 73.1 & 65.4 & & 65.0 & + 5110.413 & 0.0000 & @xmath93.7595 & & & 79.7 & & & + 5123.720 & 1.0110 & @xmath93.0630 & 70.1 & 75.6 & & 67.8 & 62.3 & + 5150.840 & 0.9900 & @xmath93.0200 & & 73.6 & 54.3 & 62.1 & 63.2 & + 5151.911 & 1.0110 & @xmath93.3215 & 61.3 & 56.6 & 42.7 & 60.2 & 51.5 & + 5198.711 & 2.2230 & @xmath92.1130 & 56.8 & 49.4 & 48.6 & 48.8 & 55.4 & + 5202.336 & 2.1760 & @xmath91.8545 & & & 78.3 & & 78.9 & + 5215.182 & 3.2660 & @xmath90.8710 & 59.8 & 69.2 & 56.4 & 56.7 & 60.6 & + 5216.274 & 1.6080 & @xmath92.1160 & & & 72.7 & & & + 5217.390 & 3.2110 & @xmath91.1160 & 56.6 & 71.3 & 46.2 & 55.6 & 41.9 & + 5225.525 & 0.1100 & @xmath94.7696 & 38.0 & 37.1 & 40.3 & 41.3 & 34.5 & 73.5 + 5242.491 & 3.6350 & @xmath90.9035 & 49.7 & 49.9 & 40.4 & 56.4 & 42.6 + 5250.210 & 0.1210 & @xmath94.9172 & 32.7 & 12.5 & 22.6 & 32.3 & 28.0 & 68.9 + 5250.646 & 2.1980 & @xmath92.1150 & 61.8 & 62.5 & 53.5 & & 62.0 & + 5263.305 & 3.2660 & @xmath90.9245 & 64.1 & 74.0 & 64.6 & 56.4 & 60.1 & + 5307.361 & 1.6080 & @xmath92.9496 & 52.3 & 53.4 & 45.9 & 47.1 & 50.2 & + 5332.900 & 1.5570 & @xmath92.8579 & 54.3 & 53.5 & 50.8 & 70.5 & 44.7 & + 5339.930 & 3.2660 & @xmath90.6800 & & & 80.8 & & 79.4 & + 5373.698 & 4.474 & @xmath90.8600 & & 21.2 & 15.1 & & & 61.0 + 5379.573 & 3.695 & @xmath91.4970 & 31.0 & 17.3 & & 33.6 & 17.8 & 59.3 + 5393.167 & 3.2410 & @xmath90.8125 & 75.1 & 68.5 & 69.4 & 79.8 & 72.5 & + 5497.516 & 1.0110 & @xmath92.8371 & 70.9 & 81.1 & 62.0 & 84.0 & 79.4 & + 5501.465 & 0.9580 & @xmath92.9978 & 77.9 & 69.6 & 62.7 & 75.4 & 72.1 & + 5506.779 & 0.9900 & @xmath92.7929 & 77.6 & 79.8 & 55.8 & 82.9 & 74.9 & + 5560.207 & 4.435 & @xmath91.1900 & & 13.9 & & & & + 5569.618 & 3.4170 & @xmath90.5130 & 71.3 & & & & & 50.8 + 5572.841 & 3.3970 & @xmath90.2925 & & & 82.2 & & & + 5576.090 & 3.4300 & @xmath91.0000 & 67.0 & 69.5 & 37.0 & 57.5 & 54.6 & + 5658.816 & 3.3970 & @xmath90.8360 & 76.3 & 71.1 & & & 74.9 & + 5717.835 & 4.285 & @xmath91.1300 & 17.8 & 17.8 & 14.7 & 17.8 & & 62.6 + 5775.080 & 4.220 & @xmath91.2970 & 16.4 & 16.9 & & 16.9 & & 55.5 + 6012.204 & 2.223 & @xmath94.0381 & & & & & & 24.2 + 6016.604 & 3.547 & @xmath91.8200 & 22.4 & 14.5 & & 15.1 & 21.4 & + 6027.048 & 4.076 & @xmath91.1495 & 27.6 & 23.9 & 15.9 & 27.8 & 26.0 & 63.4 + 6055.992 & 4.734 & @xmath90.4600 & 23.2 & 33.7 & 27.2 & 34.3 & 23.4 & 72.8 + 6065.481 & 2.6090 & @xmath91.4700 & 70.6 & 69.6 & 74.6 & 63.6 & 70.0 & + 6078.999 & 4.652 & @xmath91.1200 & & 9.6 & & 9.6 & & 43.3 + 6127.904 & 4.143 & @xmath91.3990 & 17.1 & 13.1 & & 17.6 & 6.5 & 49.7 + 6136.615 & 2.4530 & @xmath91.4050 & 78.7 & 78.4 & 71.0 & 85.8 & 82.0 & + 6137.691 & 2.5880 & @xmath91.3745 & 82.0 & 73.3 & 81.4 & 80.3 & 75.6 & + 6151.618 & 2.176 & @xmath93.2990 & 19.7 & 11.4 & 11.2 & 18.6 & & 49.0 + 6157.725 & 4.076 & @xmath91.2600 & 28.2 & 25.5 & 20.4 & 27.0 & & 62.6 + 6173.341 & 2.223 & @xmath92.8800 & 31.6 & 19.8 & 31.4 & 40.4 & 27.6 & 68.3 + 6180.203 & 2.728 & @xmath92.6225 & & & & & & 55.6 + 6219.280 & 2.1980 & @xmath92.4330 & 60.9 & 42.9 & 43.7 & 49.2 & 53.0 & 89.0 + 6230.723 & 2.5590 & @xmath91.2785 & 88.7 & 84.0 & 80.4 & 93.6 & 84.4 & + 6240.645 & 2.223 & @xmath93.2030 & & & & & & 48.0 + 6246.318 & 3.6030 & @xmath90.8770 & 64.0 & 60.6 & 46.4 & 58.1 & 58.7 & + 6252.555 & 2.4040 & @xmath91.7270 & 74.8 & 70.9 & 71.5 & 73.9 & 72.6 & + 6254.257 & 2.2790 & @xmath92.4430 & & 56.3 & & & & + 6322.684 & 2.588 & @xmath92.4475 & & & & & & 75.3 + 6335.330 & 2.1980 & @xmath92.2035 & 63.5 & 50.5 & 42.2 & 62.1 & 48.1 & 96.8 + 6344.148 & 2.433 & @xmath92.9000 & 17.7 & 22.4 & 18.9 & 22.4 & 18.6 & 66.4 + 6355.029 & 2.8450 & @xmath92.3575 & 28.3 & 20.2 & 25.6 & & 16.2 & 78.1 + 6380.742 & 4.187 & @xmath91.3875 & 18.1 & 13.7 & & 15.4 & & 51.3 + 6393.601 & 2.4330 & @xmath91.5760 & 71.8 & 68.0 & 63.4 & 72.8 & 76.0 & + 6481.869 & 2.279 & @xmath92.9840 & 24.2 & 19.4 & 22.2 & 26.9 & 18.6 & 64.1 + 6494.980 & 2.4040 & @xmath91.2560 & 96.1 & & 90.7 & 97.3 & 88.5 & + 6498.940 & 0.958 & @xmath94.6946 & & & & & & 47.1 + 6592.913 & 2.7280 & @xmath91.5365 & 70.9 & 73.6 & 65.4 & 66.7 & 59.4 & + 6593.868 & 2.433 & @xmath92.3940 & 52.4 & 42.3 & 47.1 & 51.3 & 46.2 & 84.7 + 6609.109 & 2.559 & @xmath92.6765 & 26.2 & 17.7 & 17.7 & 17.7 & 20.7 & 67.1 + 6750.152 & 2.4240 & @xmath92.6080 & 28.9 & 25.6 & & 29.5 & 30.9 & 73.7 + & + 4508.289 & 2.8580 & @xmath92.3210 & 63.0 & 60.3 & 41.4 & 65.4 & 55.2 & + 4515.339 & 2.8440 & @xmath92.4800 & 54.3 & 69.5 & & 74.5 & 55.2 & + 4522.634 & 2.8440 & @xmath92.1100 & 83.3 & & & 85.1 & 71.4 & + 4576.339 & 2.8410 & @xmath92.9550 & 51.7 & & 43.9 & & 42.2 & 61.4 + 4583.837 & 2.8070 & @xmath91.9200 & 85.6 & 76.3 & 80.2 & 75.7 & 83.3 & + 4629.339 & 2.8070 & @xmath92.3700 & 64.0 & 52.8 & 73.1 & 75.9 & 58.0 & + 4923.927 & 2.8910 & @xmath91.3200 & & & 84.3 & & & + 5197.576 & 3.2310 & @xmath92.1665 & 52.7 & 54.1 & 48.8 & 57.4 & 49.0 & 78.7 + 5234.630 & 3.2210 & @xmath92.2100 & 62.6 & 51.7 & 61.3 & 64.1 & 51.7 & + 5276.002 & 3.2000 & @xmath92.0350 & 68.0 & & 82.1 & & 74.8 & + 5316.615 & 3.1520 & @xmath92.0200 & & 86.5 & 87.9 & & & + 5325.560 & 3.221 & @xmath92.0750 & & & & & & 41.7 + 6149.249 & 3.890 & @xmath92.9300 & 23.7 & 25.4 & & & & 36.6 + 6247.562 & 3.890 & @xmath92.7240 & & 22.3 & & & & 53.5 + 6456.391 & 3.9000 & @xmath92.3290 & 51.6 & & 46.0 & 37.8 & 50.8 & + & + 4714.408 & 3.3801 & 0.2300 & 67.3 & & 63.9 & & 63.6 & + 4715.757 & 3.5435 & @xmath90.3400 & 34.4 & & & 49.3 & 32.7 & 79.3 + 4756.510 & 3.4802 & @xmath90.3400 & 43.8 & 34.9 & 40.8 & 49.8 & 35.1 & 80.7 + 4786.531 & 3.4198 & @xmath90.1700 & 51.8 & 41.7 & 57.7 & 44.9 & 43.9 & 95.4 + 4829.016 & 3.5424 & @xmath90.3300 & 30.5 & 21.9 & 19.1 & 30.5 & 29.2 & 78.5 + 4831.169 & 3.6063 & @xmath90.4200 & 35.8 & 33.7 & 36.3 & 34.3 & 40.3 & 72.9 + 4904.407 & 3.5424 & @xmath90.1700 & 44.0 & 54.9 & 44.8 & 35.5 & 42.4 & 85.5 + 4937.341 & 3.6063 & @xmath90.3900 & 30.0 & 36.5 & 29.1 & 40.8 & 22.4 & 85.3 + 5035.357 & 3.6356 & 0.2900 & 53.9 & 56.8 & 40.1 & 65.7 & 51.8 & + 5081.107 & 3.8476 & 0.3000 & 60.9 & 61.3 & 53.4 & 52.6 & 46.0 & 95.9 + 5084.089 & 3.6787 & 0.0300 & 54.8 & 36.4 & 47.9 & 46.5 & 47.7 & 89.5 + 5115.389 & 3.8342 & @xmath90.1100 & 41.6 & 32.1 & 37.5 & 40.6 & 34.9 & 74.6 + 5146.480 & 3.7060 & 0.1200 & & 44.6 & 37.2 & 51.8 & & 87.0 + 5155.762 & 3.8985 & @xmath90.0900 & 44.8 & 33.7 & 27.9 & 42.0 & 21.5 & 75.1 + 5476.900 & 1.8263 & @xmath90.8900 & & & 89.9 & & 96.7 & + 5587.853 & 1.9355 & @xmath92.1400 & 30.2 & 98.6 & & 25.0 & & 59.3 + 6108.107 & 1.6765 & @xmath92.4500 & & & 11.9 & & & 64.7 + 6482.796 & 1.9355 & @xmath92.6300 & 11.1 & 9.2 & & 10.8 & 10.2 & 41.5 + 6586.308 & 1.9509 & @xmath92.8100 & 13.9 & 11.6 & & & & 43.3 + 6643.629 & 1.6765 & @xmath92.3000 & 50.7 & 40.6 & & 51.6 & 38.6 & + 6767.768 & 1.8263 & @xmath92.1700 & 47.5 & 37.4 & & 41.2 & 33.8 & + & + 4883.684 & 1.0821 & @xmath90.0100 & 35.2 & 40.7 & 34.1 & 35.3 & 34.9 & 55.5 + 4900.120 & 1.0313 & @xmath90.1300 & 44.6 & 50.2 & 41.2 & 41.0 & 41.6 & 58.0 + 5087.416 & 1.0810 & @xmath90.3100 & 34.1 & 33.4 & 28.3 & 22.8 & 33.9 & 43.8 + & + 5853.700 & 0.000 & @xmath91.0000 & 48.3 & 46.7 & 53.6 & 52.7 & 41.5 & 63.3 + 6141.700 & 0.000 & @xmath90.0760 & 96.5 & 85.6 & 85.2 & 93.4 & 83.7 & 116.0 + 239 ............ & @xmath80 & 4.01 & 1.00 & @xmath90.70 + 259 ............ & @xmath81 & 3.98 & 1.00 & @xmath90.70 + 260 ............ & @xmath82 & 4.08 & 1.10 & @xmath90.70 + 264 ............ & @xmath80 & 4.01 & 1.00 & @xmath90.70 + 273 ............ & @xmath80 & 4.01 & 1.00 & @xmath90.70 + llrrrrrrrc & mean & 6.73 & 6.64 & 6.60 & 6.71 & 6.64 & 6.66 & 7.47 & + & [ fe i / h ] & @xmath90.74 & @xmath90.83 & @xmath90.87 & @xmath90.76 & @xmath90.83 & @xmath90.81 & & @xmath83 + & @xmath35 & 0.016 & 0.018 & 0.018 & 0.026 & 0.016 & 0.027 & 0.019 & 0.031 + & # lines & 76 & 76 & 77 & 71 & 76 & & 32 & + & @xmath84mean@xmath85 & @xmath86 & @xmath87 & @xmath88 & @xmath89 & @xmath90 & @xmath91 & & + & mean & 6.68 & 6.58 & 6.57 & 6.71 & 6.55 & 6.62 & 7.37 & + & [ fe ii / h ] & @xmath90.69 & @xmath90.79 & @xmath90.80 & @xmath90.66 & @xmath90.82 & @xmath90.75 & & @xmath92 + & @xmath35 & 0.053 & 0.096 & 0.118 & 0.078 & 0.062 & 0.037 & 0.053 & 0.065 + & # lines & 11 & 9 & 10 & 8 & 10 & & 6 & + & @xmath84mean@xmath85 & @xmath93 & @xmath94 & @xmath95 & @xmath96 & @xmath97 & @xmath98 & & + fe i & ii & wted mean & 6.72 & 6.63 & 6.59 & 6.71 & 6.63 & 6.66 & 7.45 & + & [ fe / h ] & @xmath90.73 & @xmath90.82 & @xmath90.86 & @xmath90.75 & @xmath90.83 & @xmath90.80 & & @xmath90.80 llrrrrrrrc & mean & 5.53 & 5.69 & 5.75 & 5.68 & 5.61 & 5.66 & 6.32 & @xmath90.66 + & @xmath35 & 0.08 & 0.14 & 0.10 & 0.17 & 0.06 & 0.05 & 0.02 & 0.05 + & # lines & 4 & 3 & 4 & 3 & 4 & & 3 & + & @xmath84mean@xmath85 & @xmath99 & @xmath100 & @xmath101 & @xmath102 & @xmath103 & @xmath104 & & + & [ na / h ] & @xmath90.79 & @xmath90.63 & @xmath90.57 & @xmath90.64 & @xmath90.71 & @xmath90.66 & & @xmath90.66 + & mean & 6.90 & 6.80 & 6.79 & 6.78 & 6.90 & 6.83 & 7.54 & @xmath105 + & @xmath35 & 0.06 & 0.08 & 0.42 & 0.05 & 0.18 & 0.03 & 0.14 & 0.14 + & # lines & 3 & 3 & 2 & 2 & 3 & & 3 & + & @xmath84mean@xmath85 & @xmath106 & @xmath107 & @xmath108 & @xmath109 & @xmath110 & @xmath111 & & + & [ mg / h ] & @xmath90.64 & @xmath90.74 & @xmath90.75 & @xmath90.76 & @xmath90.64 & @xmath90.71 & & @xmath90.71 + & mean & 7.08 & 7.06 & 6.93 & 7.13 & 7.00 & 7.04 & 7.61 & @xmath112 + & @xmath35 & 0.06 & 0.12 & 0.13 & 0.13 & 0.07 & 0.04 & 0.03 & 0.05 + & # lines & 3 & 3 & 3 & 3 & 3 & & 3 & + & mean & 6.95 & 6.82 & 6.76 & 6.99 & 7.01 & 6.91 & 7.55 & @xmath113 + & @xmath35 & & & & & & 0.06 & & 0.07 + & # lines & 1 & 1 & 1 & 1 & 1 & & 1 & + si i & ii & wted mean & 7.05 & 7.00 & 6.89 & 7.10 & 7.00 & 7.01 & 7.59 & + & [ si / h ] & @xmath90.54 & @xmath90.59 & @xmath90.70 & @xmath90.49 & @xmath90.59 & @xmath90.58 & & @xmath90.58 + & mean & 5.78 & 5.62 & 5.68 & 5.77 & 5.63 & 5.70 & 6.18 & @xmath90.48 + & @xmath35 & 0.03 & 0.03 & 0.04 & 0.03 & 0.04 & 0.04 & 0.03 & 0.04 + & # lines & 20 & 19 & 18 & 19 & 18 & & 21 & + & @xmath84mean@xmath85 & @xmath89 & @xmath114 & @xmath115 & @xmath116 & @xmath117 & @xmath118 & & + & [ ca / h ] & @xmath90.40 & @xmath90.56 & @xmath90.50 & @xmath90.41 & @xmath90.55 & @xmath90.48 & & @xmath90.48 + & mean & 4.47 & 4.28 & 4.41 & 4.58 & 4.28 & 4.40 & 4.83 & @xmath119 + & @xmath35 & 0.03 & 0.04 & 0.05 & 0.04 & 0.04 & 0.05 & 0.03 & @xmath120 + & # lines & 14 & 14 & 13 & 13 & 13 & & 8 & + & @xmath84mean@xmath85 & @xmath121 & @xmath122 & @xmath123 & @xmath124 & @xmath125 & @xmath126 & & + & mean & 4.52 & 4.44 & 4.36 & 4.68 & 4.40 & 4.48 & 5.09 & @xmath127 + & @xmath35 & 0.028 & 0.058 & 0.082&0.079 & 0.041 & 0.064&0.067 & 0.086 + & # lines & 7 & 9 & 9 & 12 & 9 & & 6 & + & @xmath84mean@xmath85 & @xmath128 & @xmath129 & @xmath130 & @xmath131 & @xmath132 & @xmath114 & & + ti i & ii & wted mean & 4.49 & 4.34 & 4.39 & 4.62 & 4.33 & 4.43 & 4.94 & + & [ ti / h ] & @xmath90.45 & @xmath90.55 & @xmath90.55 & @xmath90.32 & @xmath90.61 & @xmath90.50 & & @xmath90.50 + & mean & 5.44 & 5.30 & 5.29 & 5.37 & 5.29 & 5.34 & 6.05 & @xmath105 + & @xmath35 & 0.04 & 0.04 & 0.05 & 0.04 & 0.03 & 0.03 & 0.03 & @xmath133 + & # lines & 18 & 17 & 15 & 17 & 17 & & 16 & + & @xmath84mean@xmath85 & @xmath134 & @xmath135 & @xmath136 & @xmath125 & @xmath137 & @xmath134 & & + & [ ni / h ] & @xmath90.61 & @xmath90.75 & @xmath90.76 & @xmath90.68 & @xmath90.76 & @xmath90.71 & & @xmath90.71 + & mean & 4.76 & 4.79 & 4.80 & 5.01 & 4.72 & 4.81 & 5.53 & @xmath138 + & @xmath35 & 0.03 & 0.04 & 0.06 & 0.05 & 0.03 & 0.06 & 0.04 & @xmath139 + & # lines & 13 & 14 & 12 & 12 & 12 & & 15 & + & @xmath84mean@xmath85 & @xmath140 & @xmath141 & @xmath142 & @xmath143 & @xmath118 & @xmath90 & & + & mean & 4.79 & 4.88 & 4.76 & 4.89 & 4.90 & 4.84 & 5.73 & @xmath144 + & @xmath35 & 0.04 & 0.04 & 0.11 & 0.11 & 0.10 & 0.03 & 0.09 & @xmath145 + & # lines & 4 & 5 & 5 & 4 & 5 & & 5 & + cr i & ii & wted mean & 4.77 & 4.81 & 4.79 & 4.98 & 4.78 & 4.83 & 5.58 & + & [ cr / h ] & @xmath90.81 & @xmath90.77 & @xmath90.79 & @xmath90.60 & @xmath90.80 & @xmath90.75 & & @xmath90.75 + & mean & 1.53 & 1.60 & 1.48 & 1.40 & 1.51 & 1.50 & 2.15 & @xmath146 + & @xmath35 & 0.11 & 0.11 & 0.09 & 0.10 & 0.10 & 0.04 & 0.05 & 0.06 + & # lines & 3 & 3 & 3 & 3 & 3 & & & + & [ y / h ] & @xmath90.62 & @xmath90.55 & @xmath90.67 & @xmath90.75 & @xmath90.64 & @xmath90.65 & & @xmath90.65 + & mean & 1.63 & 1.49 & 1.61 & 1.67 & 1.47 & 1.57 & 2.14 & @xmath112 + & @xmath35 & & & & & & 0.045 & & 0.05 + & # lines & 2 & 2 & 2 & 2 & 2 & & 2 & + & [ ba / h ] & @xmath90.51 & @xmath90.65 & @xmath90.53 & @xmath90.47 & @xmath90.67 & @xmath90.57 & & @xmath90.57 + lcccl @xmath147}$ ] & @xmath90.80 & @xmath20.02 & 79 - 86 & and + @xmath148}$ ] & + 0.14 & @xmath20.04 & 3 - 4 & + @xmath149}$ ] & + 0.09 & @xmath20.03 & 2 - 3 & + @xmath150}$ ] & + 0.22 & @xmath20.04 & 4 & and + @xmath151}$ ] & + 0.32 & @xmath20.04 & 18 - 20 & + @xmath152}$ ] & + 0.30 & @xmath20.06 & 21 - 25 & and + @xmath153}$ ] & + 0.29 & @xmath20.05 & 45 - 52 & mg , si , ca , ti + @xmath154}$ ] & + 0.05 & @xmath20.04 & 16 - 19 & and + @xmath155}$ ] & + 0.09 & @xmath20.03 & 15 - 18 & + @xmath156}$ ] & + 0.15 & @xmath20.04 & 3 & + @xmath157}$ ] & + 0.23 & @xmath20.04 & 2 & + lrrcccc @xmath147}$ ] & @xmath158 & 0.025 & 0.035 & @xmath159 & @xmath160 & @xmath161 + @xmath148}$ ] & @xmath162 & 0.064 & 0.085 & @xmath163 & @xmath164 & @xmath165 + @xmath149}$ ] & @xmath166 & 0.028 & 0.12 ? & @xmath167 & @xmath168 & @xmath169 + @xmath150}$ ] & @xmath170 & 0.015 & 0.046 & @xmath171 & @xmath172 & @xmath173 + @xmath151}$ ] & @xmath174 & 0.023 & 0.043 & @xmath175 & @xmath176 & @xmath177 + @xmath152}$ ] & @xmath178 & 0.032 & 0.056 & @xmath179 & @xmath180 & @xmath181 + @xmath154}$ ] & @xmath182 & 0.044 & 0.068 & @xmath183 & @xmath184 & @xmath185 + @xmath155}$ ] & @xmath166 & 0.013 & 0.040 & @xmath160 & @xmath163 & @xmath186 + @xmath156}$ ] & @xmath187 & 0.051 & 0.076 & @xmath188 & @xmath189 & @xmath172 + @xmath157}$ ] & @xmath190 & 0.037 & 0.11 ? & @xmath191 & @xmath192 & @xmath193 + lcccccccccc & + g33 - 31 & @xmath91.12 & @xmath90.16 & + 0.29 & @xmath1940.17 & + 0.27 & + 0.27 & @xmath1940.03 & @xmath90.04 & @xmath1940.06 & + 0.12 + g5 - 19 & @xmath91.16 & @xmath90.21 & + 0.20 & @xmath1940.12 & + 0.19 & + 0.22 & @xmath1940.00 & @xmath90.10 & @xmath90.05 & + 0.15 + g82 - 5 & @xmath90.71 & @xmath90.48 & + 0.07 & @xmath90.06 & + 0.01 & + 0.08 & @xmath1940.00 & @xmath90.13 & @xmath90.03 & + 0.17 + g9 - 36 & @xmath91.12 & @xmath90.06 & + 0.32 & @xmath1940.18 & + 0.23 & + 0.22 & @xmath1940.04 & @xmath90.07 & @xmath1940.04 & + 0.17 + g114 - 42 & @xmath91.11 & @xmath90.20 & + 0.37 & @xmath1940.20 & + 0.21 & + 0.19 & @xmath1940.01 & @xmath90.06 & @xmath90.07 & + 0.14 + g116 - 53 & @xmath91.03 & @xmath90.27 & + 0.16 & @xmath1940.12 & + 0.16 & + 0.19 & @xmath1940.00 & @xmath90.09 & @xmath90.10 & + 0.21 + g121 - 12 & @xmath90.80 & @xmath90.29 & + 0.06 & @xmath90.02 & + 0.10 & + 0.13 & @xmath90.05 & @xmath90.22 & @xmath90.10 & + 0.15 + g168 - 42 & @xmath90.84 & @xmath90.14 & + 0.27 & @xmath1940.08 & + 0.11 & + 0.26 & @xmath1940.11 & @xmath90.11 & @xmath1940.02 & + 0.13 + mean & @xmath90.99 & @xmath90.23 & + 0.22 & @xmath1940.10 & + 0.16 & + 0.20 & @xmath1940.02 & @xmath90.10 & @xmath90.03 & + 0.16 + & + hr 2883 & @xmath90.75 & + 0.03 & + 0.29 & + 0.15 & + 0.17 & + 0.21 & & + 0.03 & @xmath1940.01 & @xmath90.04 + hr 8181 & @xmath90.67 & + 0.14 & + 0.14 & + 0.07 & + 0.05 & + 0.09 & & + 0.00 & @xmath90.03 & @xmath90.03 + hr 3018 & @xmath90.78 & + 0.18 & + 0.38 & + 0.21 & + 0.26 & + 0.14 & & @xmath90.03 & @xmath1940.01 & @xmath90.05 + hr 4657 & @xmath90.70 & + 0.12 & + 0.34 & + 0.23 & + 0.19 & + 0.32 & & + 0.06 & @xmath1940.01 & @xmath90.13 + hd 22879 & @xmath90.84 & + 0.03 & + 0.44 & + 0.19 & + 0.19 & + 0.19 & & + 0.01 & @xmath1940.04 & @xmath90.09 + hd 25704 & @xmath90.85 & + 0.10 & + 0.35 & + 0.15 & + 0.24 & + 0.27 & & @xmath90.02 & @xmath1940.00 & @xmath90.03 + hd 51929 & @xmath90.64 & + 0.07 & + 0.32 & + 0.16 & + 0.14 & + 0.27 & & + 0.03 & @xmath90.05 & @xmath90.11 + hd 62301 & @xmath90.69 & + 0.03 & + 0.30 & + 0.18 & + 0.07 & + 0.23 & & + 0.08 & @xmath90.08 & @xmath90.09 + hd 78747 & @xmath90.64 & + 0.12 & + 0.34 & + 0.19 & + 0.20 & + 0.30 & & + 0.03 & @xmath1940.07 & @xmath1940.00 + hd 130551 & @xmath90.62 & + 0.17 & + 0.18 & + 0.10 & + 0.10 & + 0.20 & & + 0.01 & @xmath90.01 & @xmath1940.02 + hd 134169 & @xmath90.83 & + 0.15 & + 0.31 & + 0.19 & + 0.15 & + 0.29 & & + 0.04 & @xmath90.07 & @xmath90.16 + hd 148211 & @xmath90.65 & + 0.05 & + 0.30 & + 0.15 & + 0.15 & + 0.21 & & + 0.02 & @xmath90.10 & @xmath90.13 + hd 148816 & @xmath90.74 & + 0.13 & + 0.32 & + 0.18 & + 0.15 & + 0.27 & & + 0.02 & @xmath90.03 & @xmath90.15 + hd 159307 & @xmath90.71 & + 0.17 & + 0.24 & + 0.14 & + 0.12 & + 0.09 & & + 0.03 & @xmath1940.00 & @xmath90.03 + hd 210752 & @xmath90.64 & + 0.03 & + 0.16 & + 0.04 & + 0.07 & + 0.06 & & @xmath90.03 & @xmath90.19 & @xmath90.14 + hd 215257 & @xmath90.65 & + 0.09 & + 0.10 & + 0.02 & + 0.10 & + 0.17 & & @xmath90.02 & @xmath90.06 & @xmath90.05 + hd 218504 & @xmath90.62 & + 0.04 & + 0.29 & + 0.14 & + 0.13 & + 0.17 & & + 0.02 & @xmath90.08 & @xmath90.10 + mean & @xmath90.71 & + 0.10 & + 0.28 & + 0.15 & + 0.12 & + 0.20 & & + 0.02 & @xmath90.03 & @xmath90.08 +
|
we have made observations with the keck i telescope and hires at a resolution of @xmath045,000 of five nearly identical stars at the turn - off of the metal - rich globular cluster m 71 .
we derive stellar parameters and abundances of several elements . our mean fe abundance , [ fe / h]@xmath1 , is in excellent agreement with previous cluster determinations from both giants and near - turnoff stars .
there is no clear evidence for any star - to - star abundance differences or correlations in our sample .
abundance ratios of the fe - peak elements ( cr , ni ) are similar to fe .
the turn - off stars in m71 have remarkably consistent enhancements of 0.2 - 0.3 dex in [ si / fe ] , [ ca / fe ] and [ ti / fe ] like the red giants .
our [ mg / fe ] ratio is somewhat lower than that suggested by other studies .
we compare our mean abundances for the five m 71 stars with field stars of similar metallicity [ fe / h ] 8 with halo kinematics and 17 with disk kinematics .
the abundances of the alpha - fusion products ( mg , si , ca , ti ) agree with both samples , but seem a closer match to the disk stars .
the mg abundance in m71 is at the lower edge of the disk and halo samples .
the neutron - capture elements , y and ba , are enhanced relative to solar in the m71 turn - off stars . our ratio [ ba / fe ] is similar to that of the halo field stars but a factor of two above that for the disk field stars .
the important [ ba / y ] ratio is significantly lower than m71 giant values ; the pre - cluster material may have been exposed to a higher neutron flux than the disk stars or self - enrichment has occured subsequent to cluster star formation .
the na content of the m71 turn - off stars is remarkably similar to that in the disk field stars , but more than a factor of two higher than the halo field star sample .
we find [ na / fe ] = + 0.14 @xmath20.04 with a spread less than half of that found in the red giants in m71 .
excluding mg , the lack of intracluster @xmath3-element variations ( turn - off vis - a - vis giants ) suggests the polluting material needed to explain the abundance patterns in m71 did not arise from explosive nucleosynthesis , but arose in a more traditional @xmath4-process environment such as agb stars . the determination of light @xmath4-peak abundances should reveal whether this pollution occurred before or after cluster formation
|
characterizing couplings between interacting systems plays an important role in numerous scientific fields , ranging from physics to the neurosciences @xcite . over the last years
, a large number of linear and nonlinear analysis techniques has been proposed to reveal couplings from passive observations of the systems behavior , e.g. , from time series of certain observables , and thus allow a data - driven quantification of the strength and direction of an interaction @xcite .
knowing interaction properties is important for the construction of functional network structures in diverse scientific fields @xcite . among these techniques ,
the information - theoretic concept of transfer entropy @xcite provides a model - free approach to characterizing directed interactions , because it can be viewed as transfers of information .
transfer entropy is related to the concept of granger causality @xcite and to conditional mutual information @xcite , and has widely been used to distinguish the driving and responding elements and to detect asymmetry in the interaction of subsystems in various scientific fields . since its invention , techniques that allow a data - driven estimation of transfer entropy are being steadily improved @xcite . among these improvements
are methods that allow one to characterize information transfers at various time scales by incorporating delays @xcite . knowing coupling delays is of importance as it allows for improved physical interpretations @xcite . in ref .
@xcite , symbolic transfer entropy has been proposed as a permutation analogue of transfer entropy and constitutes an efficient and conceptually simple way of robustly quantifying the dominating direction of flow of information between time series from observed data . using this approach ,
transfer entropy is estimated from the probabilities of ordinal patterns that are derived from the amplitude values of the time series via symbolization @xcite .
symbolic transfer entropy has been used to study interactions in various disciplines ranging from quantum @xcite and laser physics @xcite via neurology @xcite , cardiology @xcite and anesthesiology @xcite to the neurosciences @xcite .
recently , an ordinal time series analysis technique has been introduced that detects the direction and the coupling delays of information exchange in coupled systems @xcite . here
we follow this line of approach and propose a straightforward extension of symbolic transfer entropy , which we refer to as delayed symbolic transfer entropy .
this paper is organized as follows . in sec .
[ sec_methods ] we briefly recall the definition of symbolic transfer entropy before we present our extension to detect coupling delays and to quantify the dominating direction of flow of information . in sec .
[ sec_logistic_map ] we present our numerical simulation studies that aim at demonstrating the applicability of our method and at exploring its limitations . in sec .
[ sec_eeg ] we present our findings obtained from inferring delayed directed interactions in the human epileptic brain before we draw our conclusions in sec .
[ sec_conclusion ] .
let @xmath0 and @xmath1 with @xmath2 denote time series of observables of systems @xmath3and @xmath4 .
relating previous samples @xmath5 and @xmath6 in order to predict @xmath1 allows for a quantification of the deviation from the generalized markov property , @xmath7 , where @xmath8 denotes the conditional transition probability density . if system @xmath3has no influence on system @xmath4 , there is no deviation from the markov property .
transfer entropy quantifies the incorrectness of this assumption and is formulated as kullback
leibler entropy between @xmath9 and @xmath10 .
transfer entropy is non - symmetric under the exchange of @xmath3and @xmath4 . in order to estimate the transition probabilities , the authors of ref .
@xcite proposed to use a symbolization technique with symbols that are derived from reordering the amplitude values of time series @xcite .
let @xmath11 and @xmath12denote the lag and embedding dimension , which have to be chosen appropriately for symbolization @xcite , e.g. , by making use of embedding theorems @xcite .
then @xmath12amplitude values @xmath13 for a given , but arbitrary time @xmath14 are arranged in ascending order @xmath15 with rank @xmath16 and @xmath17 .
equal amplitude values are arranged by their time index , i.e. , such that @xmath18 if @xmath19 .
this ensures that every @xmath20 is uniquely mapped onto one of the @xmath21 possible permutations , and a permutation symbol is defined as @xmath22 relative frequencies of symbols provide an estimator for joint and conditional probabilities of the sequences of permutation indices . with given symbol sequences @xmath23 and @xmath24 , symbolic transfer entropy is defined as @xcite : @xmath25 where the sum runs over all symbols .
@xmath26 is defined in complete analogy .
@xmath27 is positive and explicitly non - symmetric under exchange of @xmath3and @xmath4since it measures the flow of information from @xmath3to @xmath4and not vice versa .
the difference @xmath28 provides an estimate for the dominating flow of information and thus for the dominating direction of interaction .
when analyzing empirical data , one often needs to account for delayed interactions ( cf .
[ fig_coupling_scheme ] ) , where the flow of information from system @xmath3to system @xmath4needs some finite time @xmath29 ( and/or @xmath30from @xmath4to @xmath3 ) @xcite . addressing this issue , we here extend eq .
[ eq_ste ] and allow for symbols in transition probabilities that are @xmath31(@xmath32 ) time steps past the actual symbol : @xmath33 @xmath31denotes the number of time steps into the systems own past and @xmath32the number of time steps into the past of the influencing system , i.e. , the system from which we expect the flow of information .
we therefore did not interchange the parameters @xmath31and @xmath32 in the definition of @xmath34 .
if there is a delayed flow of information from @xmath3to @xmath4(from @xmath4to @xmath3 ) and if @xmath35 ( @xmath36 ) , we expect _ delayed symbolic transfer entropy _
@xmath37 ( @xmath34 ) to attain highest values for all @xmath31 .
the use of the parameter @xmath31may seem somewhat arbitrary , but we will see in the next section that @xmath34 ( @xmath37 ) carries additional information for specific pairs @xmath38 , which can assist in detecting delayed directed interactions in empirical data . in the aforementioned definitions of entropies
, we use a logarithm to base 2 , thus entropies are given in bit .
in the following , we investigate the conditions under which delayed symbolic transfer entropy allows one to infer the coupling delays @xmath39and @xmath30and the direction of interaction . mimicking a typical experimental situation with a priori unknown coupling delays ,
we perform a parameter scan with ( @xmath40 in a range where we expect our maximum coupling delays .
we consider two delay - coupled logistic maps @xcite @xmath41 with @xmath42 where @xmath43denotes the strength of coupling between systems @xmath3and @xmath4 , and @xmath44the respective strength between @xmath4and @xmath3 . for a slight mismatch of control parameter ( @xmath45 ) as well as for given coupling strengths ( @xmath43 , @xmath44 ) and coupling delays ( @xmath39 , @xmath30 ) , we generate 20 realizations of the system by randomly choosing the initial conditions ( @xmath46 , @xmath47 ) from the unit interval .
these time series consist of @xmath48data points each after @xmath49 transients .
if not stated otherwise , we will report mean values of the delayed symbolic transfer entropies obtained from the 20 realizations of the coupled systems .
( top ) and @xmath34 ( middle ) and of the directionality index @xmath50(bottom ) for unidirectionally delay - coupled logistic maps ( @xmath51 , @xmath52 ; embedding parameters @xmath53 , @xmath54 ; @xmath55 data points ) .
amplitude values are scaled linearly in @xmath56 $ ] and logarithmically otherwise . ] in fig .
[ fig_logistic_te_single ] we show , as an example , @xmath37 , @xmath34 , and the _ directionality index _
@xmath57 for unidirectionally delay - coupled maps ( @xmath58 and @xmath52 ) obtained with embedding parameters @xmath53 and @xmath59 and @xmath55 data points .
when comparing findings for @xmath37 with those for @xmath34 there are two prominent effects : first , if @xmath35 and for all @xmath31 , @xmath37 attains highest values ( up to three orders of magnitude larger than for other pairs ( @xmath60 ) , upper part of fig . [ fig_logistic_te_single ] ) , as expected and given our definition of delayed symbolic transfer entropy . in the following , we will refer to this structure as _ resonance - like pattern_. second , we observe @xmath37 to attain lowest values , if @xmath61 and @xmath62 .
the same holds for @xmath34 , if @xmath63 and @xmath64 .
interestingly , this strongly diminished ( or even absent ) flow of information for these _ secondary diagonals _ in the plots shown in fig . [ fig_logistic_te_single ] ( upper part : upper diagonal ; middle part lower diagonal ) also provide information about delay and direction of interaction . for exactly these pairs ( @xmath60 ) , the delayed flow of information has just taken place and thus these past states of system @xmath3and @xmath4provide almost the same amount of information for the current state of system @xmath4(cf .
[ fig_coupling_scheme ] ) .
this leads to only a few combinations of symbols contributing to the transition probabilities .
consequently , the ratio of conditional probabilities in eq . [ eq_dste_single ] approaches 1 and thus @xmath37 approaches 0 ( the same holds for @xmath34 ) .
for all other pairs ( @xmath60 ) not considered yet ( referred to as _ background _ in the following ) , the delayed flow of information @xmath37 ( @xmath34 ) only approaches 0 for an increasing number of data points and for appropriately chosen embedding parameters ( see above ) . for a wide range of coupling strengths differentiability of the secondary diagonals from the background ( i.e.
, the difference to the background ) is thus best for small numbers of data points accompanied by non - optimally chosen embedding parameters as is often the case when analyzing empirical data . as an example , for embedding parameters @xmath65 and @xmath54 , which are optimal for the system considered here , differentiability is almost 0 for @xmath66 but increases almost exponentially with decreasing the number of data points to @xmath67 .
the directionality index @xmath50 , as defined here , provides information about delay and direction of interaction .
if we exchange system @xmath4for @xmath3 , this leads to a change of sign of values of the directionality index @xmath50 , since the resonance - like pattern and the upper secondary diagonal can now be observed with @xmath34 and the lower secondary diagonal with @xmath37 . note that for @xmath68 , delayed symbolic transfer entropies correspond to the non - delayed ones and fail to correctly detect the delayed coupling , as expected ( cf .
[ fig_coupling_scheme ] ) . since @xmath69
, this also applies for the direction of interaction , independent on coupling strength , number of data points , and embedding parameters ( at least for the cases considered in this section ) . for unidirectionally coupled maps ( @xmath70 ) with coupling delays @xmath71 and coupling strengths @xmath72 $ ] , we generate time series consisting of @xmath73 data points and estimate @xmath50for embedding dimensions @xmath74 and lags @xmath75 ( cf .
@xcite ) . in fig .
[ fig_logistic_numdp ] we demonstrate exemplarily , how inference depends on the number of data points @xmath48and on the embedding dimension @xmath12 .
when decreasing @xmath48 , the amplitude of the resonance - like pattern decreases and , dependent on the chosen embedding dimension @xmath12 , even vanishes . instead , for smaller @xmath48 , the secondary diagonals can be observed . as a rule of thumb ( and at least for the systems investigated here ) , @xmath76 marks the border above which delay and direction of interactions can be inferred from the resonance - like pattern . below this border but above a lower
bound which depends on system properties , the same information can be inferred from the secondary diagonals .
the width of the patterns increases linearly with the embedding dimension @xmath12 .
this broadening can be attributed to the applied symbolization technique @xcite , since the overlap of symbols grows linearly with the embedding dimension @xmath12(cf .
[ fig_coupling_scheme ] ) .
the influence of the embedding lag @xmath11is demonstrated exemplarily in fig .
[ fig_logistic_embedding ] for the resonance - like pattern .
for given @xmath48 and @xmath12and with @xmath77 , highest values of @xmath50can still be observed for @xmath78 and for all @xmath31 , but we observe additional resonance - like patterns , if @xmath79 , for @xmath80 , however with lower values of @xmath50 . within these patterns , @xmath50attains lower values if @xmath81 , which again is linked to the applied symbolization technique . for these conditions ,
permutation symbols share up to @xmath82 amplitude values and are therefore not independent .
analogous observations hold for the secondary diagonals , and we obtained similar findings for other coupling delays . in the following
, we fix the embedding parameters ( @xmath65 and @xmath54 ) and investigate the impact of type and strength of coupling on the inference of delayed directed interactions . for unidirectional couplings with delay @xmath52
we show , in fig .
[ fig_logistic_coupling_comparison ] , the dependence of delayed symbolic transfer entropies on the coupling strength and for two numbers of data points .
we make use of a priori knowledge for which pairs ( @xmath60 ) we can expect the resonance - like pattern ( @xmath83 , upper row ) and the secondary diagonals ( @xmath84 , middle row , and @xmath85 , lower row ) ; the assignments for the opposite direction @xmath86 are analogous .
in addition , we show the mean of both directions for a pair ( @xmath60 ) , for which there is no pattern , i.e. , from the background : @xmath87 . for a larger number of data points ( here @xmath88 ) the flow of information can be inferred even for small coupling strengths ( @xmath89 ) and differentiability of @xmath90 from the background increases with increasing coupling strength up to a maximum at @xmath91 , for which the systems are lag - synchronized ( fig . [ fig_logistic_coupling_comparison ] , upper row ) .
for larger coupling strengths differentiability remains at its maximum value . for the opposite direction
, there is no flow of information and we obtain @xmath92 for all coupling strengths , as expected . for smaller number of data points ( here @xmath93 ) , standard deviations of estimates
are generally enlarged , as expected .
in addition , mean values of estimates are increased , and the increase is stronger for @xmath94 ( and @xmath95 ) than for @xmath90 .
inference of flow of information is thus diminished and restricted to coupling strengths @xmath96 .
making use of information gained from the upper secondary diagonal ( fig .
[ fig_logistic_coupling_comparison ] , middle row ) , the deviation of @xmath97 from @xmath94 for @xmath98 and @xmath88 also indicates the inference of flow of information . for @xmath93
, inference can already be achieved for @xmath99 .
again , @xmath100 for all coupling strengths and number of data points .
note , however , that both means and standard deviations of estimates are increased by one order of magnitude when decreasing @xmath48from @xmath101 to @xmath102 .
an even better inference of flow of information can be achieved from information gained from the lower secondary diagonal ( fig .
[ fig_logistic_coupling_comparison ] , bottom row ) .
although similar observations can here be made for means and standard deviations of estimators , @xmath103 ( and not @xmath104 , given our definitions ; see eq .
[ eq_dste ] ) deviates clearly from @xmath94 for coupling strengths @xmath105 for both numbers of data points considered here . summarizing these findings , in the case of smaller number of data points ,
directed interactions can be inferred for a larger range of coupling strengths with information from the lower secondary diagonal ( @xmath103 ) than from the resonance - like pattern ( @xmath90 ) . for bidirectionally delay - coupled maps ,
similar observations can be made ( data not shown here ) , as long as the coupling delays @xmath39 and @xmath30as well as the coupling strengths @xmath43 and @xmath44are not identical .
even for the case @xmath106 the dominating delayed flow of information can be inferred , if the coupling strengths are sufficiently different ( cf . fig .
[ fig_logistic_coupling_comparison ] ) . as before ,
inference is influenced by alterations of the patterns ( the resonance - like pattern and the secondary diagonals ) related to the choice of embedding parameters necessary for the applied symbolization technique .
next we estimate the performance of our method , particularly with respect to the analysis of empirical data , by investigating the influence of noise on the inference of delayed directional couplings . for unidirectional couplings with delay @xmath52
, we add noise to the time series @xmath0 of the driver and to the time series @xmath1 of the responder , and estimate @xmath50for signal - to - noise ratios @xmath107 $ ] ( @xmath108 , where @xmath109 and @xmath110 denote the standard deviations of the noise - free and the noise - contaminated time series ) .
we use different types of noise as well as different noise - contamination schemes .
with gaussian @xmath111-correlated noise , we simulate measurement errors , and with the concept of surrogates @xcite , we generate in - band noise from the original time series , thus mimicking observational noise .
the surrogate time series have a power spectrum and a distribution of amplitude values that are identical to those of the original time series . with each of these types of noise
we contaminate both time series , @xmath0 and @xmath1 , using either the same @xmath112__(symmetric noise contamination ) _ _ or different @xmath112for the driver and responder _ ( asymmetric noise contamination)_. the latter contamination scheme is more likely in field applications and is known to affect various time series analysis techniques aiming at an inference of the direction of interactions @xcite . in fig .
[ fig_logistic_noiselevel ] , we show exemplary findings for a symmetric contamination with in - band noise . for various @xmath112we plot the dependence of delayed symbolic transfer entropies on the coupling strength and for different number of data points .
as in the previous subsection , we restrict ourselves to the pairs ( @xmath60 ) for which we can expect the correct direction of flow of information from the resonance - like pattern ( @xmath90 ) and from the secondary diagonals ( @xmath97 and @xmath103 ) .
as expected , differentiability of all estimators of flow of information from the background @xmath94 decreases with an decreasing signal - to - noise ratio .
likewise , the range of coupling strengths for which directed interactions can be inferred shrinks with decreasing the signal - to - noise ratio and is shifted towards higher coupling strengths . for a smaller number of data points , the inference of flow of information and with this the direction of interaction gained from the secondary diagonals ( @xmath97 and
@xmath103 ) is more robust to noise contaminations than for a larger number of data points .
as expected , the opposite is true for the inference gained from the resonance - like pattern ( @xmath90 ) .
we obtained similar findings for the other types of noise and contamination schemes .
taking advantage of the conceptual simplicity , efficiency , and robustness of symbolic transfer entropy , we demonstrated that our extension allows to infer of delayed directed interactions .
our method provides information about delay and direction of couplings even for smaller number of data points and , moreover , for the case of a non - optimal choice of embedding parameters used for the symbolization .
this renders delayed symbolic transfer entropy attractive for the analysis of empirical data . [ fig_fielddata_implscheme ] in this section , we apply our method to check whether consistent delayed directed interactions between brain regions can be inferred from long - lasting , multichannel electroencephalographic ( eeg ) recordings .
the eeg was recorded from an epilepsy patient using electrodes implanted under the skull , hence with high signal - to - noise ratio , prior to surgical treatment of a focal epilepsy .
the patient had signed informed consent that her / his clinical data might be used and published for research purposes .
the study protocol had previously been approved by the ethics committee of the university of bonn .
we here consider eeg recordings from strip electrodes ( 8 or 16 contacts ) placed onto the cortex and from a pair of needle - shaped depth electrodes with 10 contacts each , implanted into deeper structures of the brain ( see upper left part of fig .
[ fig_fielddata_implscheme ] ) .
data were sampled at ( sampling interval @xmath113 ) using a 16 bit analog - to - digital converter and filtered within the frequency band . for our analyses ,
we consider a continuous recording of duration during the seizure - free interval , which covered different physiologic and pathophysiologic states of the patient . here
we restrict ourselves to eeg data from six recording sites ( see upper part of fig .
[ fig_fielddata_implscheme ] ) : two from within the epileptic focus ( tr01 and tr05 ) , one remote site on the same brain hemisphere ( tlr04 ) , and three from homologous positions on the other brain hemisphere ( tl01 , tl05 , and tll04 ) .
a widely used approach to analyze the dynamics of non - stationary systems is to perform the analysis in sliding windows with a duration , for which the dynamics can be regarded as approximately stationary . for the eeg , the duration of such a window typically amounts to duration @xcite . using this approach
, we perform for each combination of pairs of recording sites a time resolved estimation of delayed symbolic transfer entropies from non - overlapping eeg segments of duration ( corresponding to 4096 data points ) . following ref .
@xcite , we set embedding parameters to @xmath114 and @xmath115 .
since time delays in the human brain can vary considerably , depending on brain regions and functions , and may reach up to @xcite , we estimate @xmath50with @xmath116{ms } , \ldots , \unit[250]{ms}\}$ ] . moreover , by time - averaging separately over all windows for data recorded during day and during night times
, we check whether major delays as well as preferred directed interactions can be identified and whether delay and direction depend on the state of consciousness ( awake vs. asleep ) . in fig .
[ fig_fielddata_results ] , we show the mean directionality indices @xmath50separately for data from day and night times for three exemplary pairs of recording sites . in general , we do not observe the resonance - like patterns , which is to be expected given the number of data points and embedding parameters .
for some cases , however , we observe secondary diagonals , from which we can extract information about delay and direction of an interaction . in particular , we observe a consistent driving with an average delay of ( ) from posterior ( position tl05 ) to anterior sites ( position tl01 ) within the non - epileptic ( left ) mesial temporal brain structures both during day and night times ( upper right part of figure ) . for homologous recording sites within the epileptic ( right ) mesial temporal brain structures a similar directed driving with an average delay of ( ) can be observed for data recorded during night times ( lower left part of figure ) .
this delay is comparable to findings gained from analyses of propagation of specific patterns during seizures @xcite . identifying a delay for data recorded during day times , however , is more demanding , possibly due to multiple delays ( which may be associated with the epileptic process ) . interestingly , for @xmath68 , for which @xmath50corresponds to the non - delayed directionality index , we observe the direction of driving to be reversed , i.e. , from anterior to posterior sites
. our findings for long - ranged interactions between regions in the left and right temporal lateral neocortex ( lower right part of figure ) also point to multiple delays , and it remains to be shown whether they differ from those obtained for the short - ranged interactions within the epileptic focus . for data recorded during day times , the brain region in the right temporal lateral neocortex constantly drives the homologous brain region in the left hemisphere .
this unidirectional driving vanishes for data recorded during night times , and we can only speculate whether this is due to , e.g. , a bidirectional interhemispheric driving or a diminished interhemispheric interaction during sleep ( cf .
we have proposed a straightforward extension of symbolic transfer entropy @xcite that enables the inference of delayed directional relationships between coupled dynamical systems from time series . with numerical examples , which are representative of interacting chaotic systems contaminated with noise
, we have exemplified the applicability of our approach and have shown that delay and direction of an interaction can be inferred with delayed symbolic transfer entropy even for smaller number of data points and , moreover , with non - optimally chosen parameters for the applied symbolization technique @xcite . applying our method to infer delayed directed interactions in the human epileptic brain , we could show that major interaction delays can be identified , particularly from short - ranged interactions , and that these delays are influenced by the pathophysiology and by physiologic states of the brain .
moreover , we could also show , that not taking into account possible delays in interactions can lead to a possibly erroneous inference of the direction of interactions .
our approach can thus help to avoid misinterpretations and to further improve the construction of functional network structures from data @xcite . at present
, our approach requires estimating the directionality index @xmath50with parameters ( @xmath117 ) in a range where we expect maximum coupling delays .
although a more direct detection of coupling delays would be preferable , we note that the identification of delayed directed interactions from time series ( 4096 data points , embedding dimension @xmath114 ) for all @xmath118 can be performed in about on a cpu core due to the underlying conceptual simplicity , efficiency , and robustness of symbolic transfer entropy .
we are grateful to gerrit ansmann , christian geier , stephan porz , and alexander rothkegel for critical comments on earlier versions of the manuscript .
this work was supported by the deutsche forschungsgemeinschaft ( grant no : le 660/5 - 2 ) .
|
we propose a straightforward extension of symbolic transfer entropy to enable the investigation of delayed directional relationships between coupled dynamical systems from time series .
analyzing time series from chaotic model systems , we demonstrate the applicability and limitations of our approach .
our findings obtained from applying our method to infer delayed directed interactions in the human epileptic brain underline the importance of our approach for improving the construction of functional network structures from data .
20cm(3cm,27 cm ) published as phys .
rev .
e * 90 * , 062706 ( 2014 ) .
copyright 2014 by the american physical society .
|
in recent years people have become interested in higher - dimensional spacetimes with motivation , among others , coming from string theories .
one of the important classes of exact solutions of the einstein equations are the spacetimes representing black holes . in the last years
various generalizations of black hole solutions to the higher - dimensional gravity have been discovered . starting with nonrotating and rotating black holes @xcite and following with the metric describing the rotating black hole with the cosmological constant in five dimensions @xcite ,
the generally rotating black hole in arbitrary dimension with the cosmological constant was found in @xcite . in 2006 this solution
was rewritten @xcite in much more convenient coordinates which allowed to add nut parameters and compute the curvature tensors explicitly @xcite .
recently , the properties of this solution have been discussed in various papers .
it has been shown that the spacetime possesses hidden symmetries which can be described by the principal killing - yano tensor @xcite , the properties of which have been thoroughly discussed in @xcite .
one of the main consequences of these hidden symmetries is a possibility to find a complete set of integration constants for a geodesic motion which are in involution @xcite .
this result is related to the the separability of the hamilton - jacobi equation which , together with the separability of the klein - gordon equation , was demonstrated in @xcite .
it seems that a generalization of the black hole solution to include an electromagnetic field is not straightforward .
some partial results are know @xcite , however a solution representing a rotating black hole with electromagnetic field parameterized by full number of black hole parameters and electric / magnetic charges is not known . in this work we present the test electromagnetic field on the background described by the metric with the structure of black hole solution @xcite
, however , we do not enforce the specific form of the metric function @xmath0 which is needed to satisfy the vacuum einstein equations .
the electromagnetic field is described by @xmath1 electric and magnetic charges and it is adjusted to the explicit as well as hidden symmetries of the spacetime . in even dimensions
our field is equivalent to the harmonic forms independently found in @xcite , our results supply the proof that these forms are harmonic in all even dimensions .
we also show that our field is a generalization of the electromagnetic field known from the carter @xcite and plebaski
demiaski @xcite form of the black hole solution in @xmath2 . in this special case , however , the metric function can be modified in such a way that that the metric and the electromagnetic field satisfy the full coupled einstein
maxwell equations .
unfortunately , the same is not true in the higher - dimensional case .
the plan of the paper is following .
we discuss separately even and odd spacetime dimensions
although our procedure is very similar in both cases , the resulting expressions are slightly different and it would be artificial to try to them into the same equations .
we discuss the case of even dimensions in detail and than sketch the corresponding results for the odd dimensions .
first we review the metric and the symmetries of the black hole spacetime , next we find the test electromagnetic field to the symmetries of the spacetime .
for even dimensions we also discuss the case of the physical dimension @xmath2 , in the odd dimension case we study the chern
simons modification of the electromagnetic field .
finally , we discuss a possibility to generalize our fields to a solution of the coupled einstein
maxwell equations .
the results are recapitulated in the summary and the paper is concluded by the technical appendix .
we start with a simpler case of even spacetime dimension @xmath3 .
the spacetime of generally rotating black hole with nut charges can be described by the metric @xmath4 \period\ ] ] here @xmath5 , @xmath6 , correspond to radial in order to put the metric to a more symmetric and compact form cf .
, e.g. , @xcite . ] and latitudinal directions , while @xmath7 , @xmath8 , to temporal and longitudinal directions . the functions @xmath9 and @xmath10 ( together with @xmath11 and @xmath12 used below ) are defined as @xmath13 we call the remaining functions @xmath0 .
in the following we consider a broader class of spacetimes than merely the black hole spacetime , namely we just assume that the metric functions @xmath0 depend arbitrarily on a _ single _ coordinate , @xmath14 .
for the black hole spacetime they acquire a specific from ( see below ) determined by the vacuum einstein equations .
the metric can be diagonalized @xmath15 introducing the unnormalized frame @xmath16 and normalized frame @xmath17 of 1-forms @xmath18^{\!1\!/2}}\,\ep^{\mu}\commae\\ \ep^{\hat\mu } & = \sum_{k=0}^{n-1}\a{k}_{\mu}\grad\psi_k\commae & \eb^{\hat\mu } & = { \textstyle\bigl[\frac{x_\mu}{u_\mu}\bigr]^{\!1\!/2}}\,\ep^{\hat\mu}\period \end{aligned}\ ] ] the dual frames of vectors are given by @xmath19^{\!1\!/2}}\,\ep_{\mu}\commae\\ \ep_{\hat\mu}&= \sum_{k=0}^{n-1}\frac{(-x_{\mu}^2)^{n\!-\!1\!-\!k}}{u_{\mu}}\,\cv{\psi_{k}}\commae & \eb_{\hat\mu } & = { \textstyle\bigl[\frac{u_\mu}{x_\mu}\bigr]^{\!1\!/2}}\,\ep_{\hat\mu}\period \end{aligned}\ ] ] here we use the convention @xmath20 and greek indieces run from @xmath21 to @xmath22 .
the inverse relations can be easily obtained with help of the relations in the appendix .
let us note that the definition of unnormalized 1-forms @xmath16 and vectors @xmath23 does not depend on the metric functions @xmath0 .
it was shown in @xcite that the ricci curvature for such a metric is @xmath24 where @xmath25\period \end{split}\raisetag{5ex}\ ] ] the scalar curvature simplifies to @xmath26 here , the primes denote the differentiation with respect to the single argument of the metric function , @xmath27 .
the requirement that the metric satisfies the einstein equations with the cosmological constant implies that the metric functions have the form @xmath28 see @xcite or in the appendix .
the constants @xmath29 and @xmath30 are then related to the cosmological constants , angular momenta , mass , and nut charges , see , e.g. , @xcite for details . for @xmath31
we obtain @xcite the constant curvature spacetime with the scalar curvature @xmath32 .
the spacetime with the metric possesses the explicit symmetries given by the killing vectors @xmath33 and hidden symmetries which are related to the principal killing - yano tensor discovered in @xcite and discussed in detail in @xcite .
the principal killing - yano tensor is dual to the rank-2 closed conformal killing - yano tensor which has a very simple form in the frames @xmath16 and @xmath17 : @xmath34 it has been demonstrated in @xcite that it is possible to generate a series of higher - rank killing
yano tensors and a series of rank-2 killing tensors from @xmath35 . the conformal killing
yano tensor also identifies eigenspaces spanned on the pairs @xmath36 and coordinates @xmath5 as corresponding eigenvalues .
now we turn to the task to find an algebraically special test electromagnetic fields on the background given by the metric . by
_ algebraically special _ we mean that the maxwell tensor @xmath37 shares the explicit symmetry of the metric ( it is independent of @xmath7 ) and it is aligned with the hidden symmetry of the spacetime , namely it has the same eigenspaces as the principal conformal killing - yano tensor @xmath35 .
we thus require @xmath38 the maxwell tensor is generated by the vector potential , @xmath39 .
we assume the vector potential @xmath40 with the components @xmath41 and @xmath42 independent of @xmath7 . comparing @xmath43
with we find that @xmath41 terms are gauge - trivial and @xmath42 must satisfy @xmath44 , from which follows that the vector potential can be written as @xmath45 where @xmath46 are functions of a _ single variable _ only , @xmath47 . evaluating the maxwell tensor we get the components @xmath48 : @xmath49 alternatively , we could apply directly the first maxwell equation @xmath50 to the maxwell tensor .
with help of identity we find that @xmath48 are generated by an auxiliary potential @xmath51 , @xmath52 which satisfies the equation @xmath53 the field found above is generated by the potential @xmath54 next we proceed to calculate the source @xmath55 of the electromagnetic field using the second maxwell equation @xmath56 .
expressing the maxwell tensor in coordinates @xmath57 , using @xmath58 , the fact that the determinant @xmath59 of the metric in these coordinates is @xmath60 , identities , , and relations , we obtain @xmath61 with @xmath62\period\ ] ] substituting we finally obtain @xmath63\period\ ] ] it is worth to mention that the expression for the source @xmath55 does not contain any reference to the metric functions @xmath0 .
we are interested in the electromagnetic field without sources , so we require @xmath64 .
integrating we find that the sum in the square brackets has to be a constant .
however , this sum has a special form discussed in the appendix .
using we find that @xmath65 are given by a single polynomial of the @xmath66-th order in variable @xmath67 .
integrating once more we find @xmath68 substituting into the vector potential or the scalar potential we find with help of the relations that the terms containing the constants @xmath69 are gauge trivial ( they contribute by @xmath70 into @xmath71 or by just the constant @xmath72 into @xmath51 ) and they can be ignored .
we thus have found that the algebraically special electromagnetic field ( i.e. , the field of the form ) satisfies the maxwell equations on the background described by the metric if and only if it is generated by the vector potential @xmath73 the components @xmath48 of the maxwell tensor are easily determined by from the auxiliary potential @xmath74 and they are @xmath75 here , @xmath76 are constants which can be related using the gauss and stokes theorems to the electric and magnetic charges of the field .
if we set all charges except one , say @xmath77 , to zero , the maxwell tensor @xmath37 corresponds to the harmonic form @xmath78 recently found and verified for particular cases in @xcite .
the surprising property of our field is that it satisfies the maxwell equations independently of a specific form of the metric functions @xmath0 .
moreover , as we will see in , the stress - energy tensor corresponding to the field has the form consistent with the structure of the ricci ( and the einstein ) tensor .
these facts open a possibility that we could solve the full einstein
maxwell equations : modifying the metric functions @xmath0 we could construct the spacetime in which the stress - energy tensor @xmath79 would be a source for the einstein equations , and the electromagnetic field would remain the solution of the maxwell equations . first , we will show that this goal can be achieved in @xmath2 dimensions . unfortunately , next we will demonstrate that this procedure does not work in higher dimensions .
the metric is a generalization of the @xmath2 black hole solution in the form found by carter @xcite and elaborated by plebaski and demiaski @xcite .
the full class of plebaski demiaski solutions includes accelerated rotating nut and electromagnetically charged black holes .
the meaning of all parameters of the solutions was recently discussed in the series of papers @xcite .
we use the charged plebaski
demiaski metric with the acceleration set to zero : .
here we also used the gauge freedom to set @xmath80 and redefined @xmath81 . ]
@xmath82 where @xmath83 the electromagnetic field is given by @xmath84 here @xmath85 are temporal , radial , latitudinal , and longitudinal coordinates , @xmath86 are parameters related to mass , nut charge , angular momentum , and cosmological constant , and @xmath87 are electric and magnetic charges .
this metric and the electromagnetic field satisfy coupled einstein maxwell equations . for @xmath88
the metric satisfies the vacuum einstein equations , however , even in this case the electromagnetic field ( with non - zero charges ) satisfies the maxwell equations and it is thus a valid test electromagnetic field on the given vacuum background .
now , we can easily identify the plebaski
demiaski @xmath89 and @xmath71 with our general metric and field using the dictionary @xmath90 clearly , if the charges @xmath87 are missing in the metric functions @xmath91 and @xmath92 , these functions corresponds exactly to @xmath93 and @xmath94 given by and the electromagnetic field corresponds to the test field given by .
however , it is possible to modify @xmath92 and @xmath91 by adding @xmath95 and @xmath96 respectively ( i.e. , changing the vacuum @xmath0 s by adding @xmath97 ) , and we obtain the metric and the electromagnetic field satisfying the coupled einstein maxwell equations . in a generic dimension
we first evaluate the stress - energy tensor @xmath79 of the electromagnetic field .
a straightforward calculation leads to @xmath99 with the trace @xmath100 where the function @xmath101 is defined as @xmath102 we explicitly see that the trace of the stress - energy is non - vanishing for @xmath98 which is related to the fact that the electromagnetic field is not conformally invariant in a general dimension .
now we would like to solve the einstein equations @xmath103 .
the trace gives the condition @xmath104 however , the scalar curvature has the form and it immediately follows that @xmath105}\commae\ ] ] which is a function of @xmath5 only . applying this to the right - hand - side of we obtain the condition @xmath106 it was checked by _ mathematica _ that this conditions does not hold for the electromagnetic field given by , at least for the lowest non - trivial values of @xmath22 .
it seems that the main problem is that @xmath107 behaves as @xmath108 while @xmath101 as a square of such sums .
we thus may conclude that in a generic even dimension the electromagnetic field of the form , can not couple to the metric given by .
let us briefly review modifications which appear in the odd dimensional case .
we have an additional coordinate @xmath110 which labels an angular direction .
the metric contains an additional term , @xmath111\\ & \mspace{190mu}+\frac{c}{\a{n}}\bigl(\sum_{k=0}^n \a{k}\grad\psi_k\!\bigr)^{\!2 } \period\end{split}\ ] ] here @xmath112 is a conventional constant .
could be eliminated by a appropriate rescaling of the coordinates and other parameters . ]
we can again introduce the frames of 1-forms given by completed with 1-forms which we label by the index @xmath113 @xmath114^{\!1\!/2}}\,\ep^{\hat0}\period\ ] ] the dual frames then become @xmath115^{\!1\!/2}}\,\ep_{\hat\mu}\commae\\ \ep_{\hat0 } & = \frac1{\a{n}}\cv{\psi_n}\commae & \eb_{\hat0 } & = { \textstyle\bigl[\frac{\a{n}}{c}\bigr]^{\!1\!/2}}\,\ep_{\hat0}\period \end{aligned}\ ] ] the metric diagonalizes @xmath116 as well as the the ricci tensor @xcite @xmath117 here @xmath118\commae\\ r_0 & = \sum_{\nu=1}^n\frac{\bx_\nu'}{x_\nu u_\nu}\commae \end{split}\ ] ] and we used shifted metric functions @xmath119 the scalar curvature becomes @xmath120 finally , the vacuum einstein equations require @xmath121 the spacetime has @xmath122 killing vectors @xmath33 and there exists the principal killing yano tensor dual to the conformal killing
yano tensor which is given again by . as in the even dimensions we look for the electromagnetic field with the structure given by .
on could ask if in the odd dimension could nt be this ansatz extended by an additional term related to the direction , i.e. , by the term of the form @xmath123 with functions @xmath124 and @xmath125 independent of the coordinates @xmath7 . however , we do not consider such terms since they lead to some unwanted consequences ( e.g. , they lead to a non - diagonal stress - energy tensor ) .
it follows from the first maxwell equations that the vector potential can be written as @xmath126 where @xmath46 is a function of @xmath5 only and @xmath127 is a constant .
the components of the maxwell tensor can be again generated by from the auxiliary potential @xmath128 and in terms of @xmath129 s they take the form @xmath130 the electromagnetic current becomes @xmath131\ep_{\hat\mu } + 2\sum_{\nu=1}^n\frac1{x_\mu}\,\phi_{,\nu}\ep_{\hat0}\period\ ] ] solving the second maxwell equation without sources ( i.e. , @xmath64 ) we obtain @xmath132 the last two terms are gauge trivial and the vector potential can thus be written as @xmath133 and the auxiliary potential @xmath51 as @xmath134 finally , for the components of the maxwell tensor we find @xmath135 in an odd number of dimensions we can consider chern
simons modification of the electromagnetic field .
since the chern - simons term in the action does not refer to the metric , it does not change the einstein equation and the stress - energy tensor of the electromagnetic field .
however , it modifies the divergence @xmath136 in the second maxwell equation by adding a non - linear term @xmath137 proportional ( with a constant coefficient ) to the hodge dual of the @xmath22-th wedge power of @xmath37 , @xmath138 since the levi - civita tensor @xmath139 used in the hodge dual is given by the product @xmath140 we find that the chern
simons term @xmath137 for the field is aligned with the direction @xmath141\ ; \eb_{\hat0}\period\ ] ] we thus have to solve the maxwell equation @xmath142 with @xmath55 given by .
the condition that @xmath55 vanishes in @xmath143 , @xmath6 , directions is only slightly weaker than that we solved in the previous section .
we find that the vector potential @xmath71 , the auxiliary potential @xmath51 , and the components of the electromagnetic field are @xmath144 \commae\end{aligned}\ ] ] where @xmath145 is a constant parameter .
moreover , the source term @xmath55 must be @xmath146 although we did not prove it rigorously it seems evident that this component can not be equal to with @xmath48 given by .
we thus conclude that the chern
simons modification of the algebraically special electromagnetic field is not possible .
the stress - energy tensor for the field in odd dimensions is also diagonal ( with the additional component @xmath148 ) .
the condition for the scalar curvature following from the einstein equations is again given by , and by the same argument as in the even dimensions ( now with help of ) we conclude that the algebraically special electromagnetic field , can not couple to the metric given by .
we have found explicitly the test electromagnetic field on the background of the high - dimensional rotating nut charged black hole @xcite .
the field solves the maxwell equations even in a broader class of spacetimes since the specific form of the metric functions @xmath0 is not needed .
the common feature of these spacetimes is the presence of @xmath149 killing vectors and the existence of the principal killing yano tensor @xcite .
the constructed electromagnetic field is adjusted to this structure it shares the explicit symmetries and it has the same eigenspaces as the principal conformal killing yano tensor .
the electromagnetic field depends on @xmath1 constants @xmath76 related to the global electric and magnetic charges .
it generalizes the field known on the background of the plebaski
demiaski spacetime in @xmath2 dimensions . in this case
the metric functions can be modified in such a way that the field and the metric solve the full einstein
maxwell equations .
unfortunately , an analogous modification is not possible in a generic dimension .
finally , we have shown that the chern
simons generalization in an odd dimmension is also not permitted for the electromagnetic field of this form .
this work was supported by the grant gar 202/06/0041 and by the czech ministry of education under the project lc06014 .
the author would like to thank don n. page , valeri p. frolov , david kubizk and ji podolsk for numerous discussions , and ji podolsk and david kubizk for reading the manuscript .
the functions @xmath10 and @xmath9 defined in satisfy the following useful identities @xmath150\label{aurel } \\[-1ex]\notag \sum_{\mu=1}^n \a{i}_\mu \frac{(-x_\mu^2)^{n}}{u_\mu } = -\a{j+1}\comma \sum_{\mu=1}^n \frac{\a{j}_\mu}{x_\mu^2u_\mu } = \frac{\a{j}}{\a{n}}\commae\end{gathered}\ ] ] @xmath151 and @xmath152 the fact that the polynomial functions solve equation follows from .
the opposite implication is less trivial first one has to show that @xmath159 must be polynomials of the order @xmath162 ( by differentiating repeatedly @xmath163 ) .
next , a more intricate task is to prove that the polynomials for different @xmath164 are the same .
it can be achieved by an induction in @xmath22 . the solution of the functional equation with a non - trivial right - hand - side is then given by the sum of a particular solution with the homogeneous solution .
we mention three important particular solutions for simple right - hand - side terms : @xmath165 the property has been used , for example , to derive the specific forms and of the metric functions @xmath0 starting from the condition @xmath166 with the scalar curvature given by or .
it has been used also to find the components and of the vector potential .
p. krtou , d. kubizk , d. n. page , and v. p. frolov , killing - yano tensors , rank-2 killing tensors , and conserved quantities in higher dimensions , j. high energy phys .
jhep02(2007)004 , arxiv : hep - th/0612029 .
|
a special test electromagnetic field in the spacetime of the higher - dimensional generally rotating nut(a)ds black hole is found .
it is adjusted to the hidden symmetries of the background represented by the principal killing - yano tensor .
such electromagnetic field generalizes the field of charged black hole in four dimensions . in higher dimensions , however , the gravitational back reaction of such a field can not be consistently solved .
|
the molecular setup and grain types in the circumstellar envelopes ( cses ) of agb stars are to a large extent determined by the c / o - ratio in the photosphere of the central star .
there exists three chemical types : the carbon stars , with c / o@xmath01 , the m - stars , with c / o@xmath11 , and the s - stars , with c / o@xmath21 . the s - stars are believed to be a short transitional phase , as dredge - up of carbon from he - shell burning , changes the spectral type from m - stars to carbon stars .
the agb stars contribute to the chemical evolution of galaxies through their extensive mass loss and an understanding of their circumstellar properties is important in the study of galactic chemistry .
the mass - loss characteristics and molecular setup of m- and carbon stars are relatively well - studied .
however , comparatively little attention has been given to the s - stars . in order to improve this situation
we have started a project aimed at obtaining an extensive database of molecular line emission towards a sample of 40 s - stars from common molecules such as co , sio , hcn , cs and sis .
the properties of the gas present in the cses have been determined from co data using detailed non - lte radiative transfer modelling , which self - consistently also calculates the kinetic temperature of the gas @xcite .
the properties of the dust present in the cses were also determined and included in the co excitation analysis .
it was found that the s - stars resembles the mass - loss rate distributions of m - type agb stars and carbon stars with a median mass - loss rate of 2@xmath310@xmath4m@xmath5yr@xmath6 , possibly with a scarcity of objects with high mass - loss rates ( @xmath710@xmath8m@xmath5yr@xmath6 ) .
the expansion velocities of the envelopes are on average similar to the m - type agb stars but lower than the carbon stars .
once the physical properties of the dust and gas present in the cses are determined , accurate molecular abundances can be estimated and compared with predictions from chemical modelling , as well as with results for m - stars and carbon stars .
in addition to the already published co data we have detected sio @xmath9=2@xmath101 in 20 stars and sio @xmath9=5@xmath104 in 18 stars using the iram 30 m telescope . also , sio @xmath9=8@xmath107 has been detected in 17 of the sample stars with apex .
this will be further complimented by observations at jcmt in november 2006 .
we have also obtained interferometric sio data for @xmath11 gru and w aql using atca .
moreover , we have gathered data of radio line emission from other molecules , mainly sis and hcn .
we have started modelling the sio radio line emission ( see fig .
1 ) towards the sample , to determine accurate abundances using a detailed excitation analysis ( ramstedt et al . , in prep ) . of interest is to see if the s - stars shows the same decline of the average sio abundance in the cse with mass - loss rate as is found for m - type agb - stars @xcite and carbon stars @xcite , indicative of freeze - out on to dust grains .
also , the effect of non - equilibrium chemistry can be further tested .
|
the s - stars have been suggested to be a brief transitional phase as stars evolve from oxygen - rich m - type stars into carbon stars , through the dredge up of carbon from he - shell burning . as possible transition objects , s - stars might help achieve a deeper understanding of the chemical evolution as a star ascends the agb , as well as shed more light on the mass - loss mechanism .
we have initiated a large survey of 40 s - stars to observe line emission in common molecules such as co , sio , hcn , cs and sis .
detailed radiative transfer modelling of multi - transition co radio line observations towards a sample of 40 s - stars shows that the mass - loss rate distribution of s - stars is consistent with those found for m - type agb stars and carbon stars .
initial results from modelling of the circumstellar sio emission are also presented .
|
terrorist and insurgent networks are known for their ability to regenerate leadership after targeted attacks .
for example , the infamous al qaeda in iraq terrorist leader abu musab al - zarqawi was killed on june 8th , 2006 only to be replaced with abu ayyub al - masri about a week later . here ,
we introduce the notion of a _ shaping _ operation in which the terrorist network is first targeted for the purpose of reducing its leadership re - generation ability .
such shaping operations would then be followed by normal attacks against high value targets however the network would be less likely to recover due to the initial shaping operations . in this paper , we look to shape such networks by increasing network - wide centrality , first introduced in @xcite . intuitively , this measure provides insight into the criticality of high - degree nodes .
hence , a network with a low network - wide centrality is a more decentralized organization and likely to regenerate leadership . in the shaping operations introduced in this paper
, we seek to target nodes that will maximize this measure - making follow - on attacks against leadership more effective .
previous work has primarily dealt with the problem of leadership regeneration by focusing on individuals likely to emerge as new leaders @xcite . however ,
targeting or obtaining information about certain individuals may not always be possible .
hence , in this paper , we target nodes that affect the reduce the _
network s _ ability regenerate leadership as a whole .
the main contributions of this paper is the introduction of a formal problem we call _ fragility _ ( section [ prelim - sec ] ) which seeks to find a set of nodes whose removal would maximize the network - wide centrality .
we also included in the problem a `` no strike list '' - nodes in the network that can not be targeted for various reasons .
this is because real - world targeting of terrorist or insurgent networks often includes restrictions against certain individuals .
we also prove that this problem is np - complete ( and the associated optimization problem is np - hard ) which means that an efficient algorithm to solve it optimally is currently unknown .
we then provide two algorithms for solving this problem ( section [ algsec ] ) .
our first algorithm is an integer program that ensures an exact solution and , though intractable by our complexity result , may be amenable to an integer program solver .
then we introduce a greedy heuristic that we show experimentally ( in section [ expsec ] ) to provide good results in practice ( as we demonstrate on six different real - world terrorist networks ) and scales to networks of @xmath3 nodes and @xmath4 edges . in examining five real - world terrorist networks
, we found that successful targetting operations against only @xmath0 ( or less ) of nodes can increase the network - wide centrality between @xmath1 and @xmath2 .
additionally , we discuss related work further in section [ rwsec ] .
we would like to note that the targeting of individuals in a terrorist or insurgent network does not necessarily mean to that they should be killed .
in fact , for `` shaping operations '' as the ones described in this paper , the killing of certain individuals in the network may be counter - productive .
this is due to the fact that the capture of individuals who are likely emergent leaders may provide further intelligence on the organization in question .
we assume that an undirected social network is represented by the graph @xmath5 .
additionally , we assume a `` no strike '' set , @xmath6 .
intuitively , these are nodes in a terrorist / insurgent network that can not be targeted .
this set is a key part of our framework , as real - world targeting of terrorist and/or insurgents in a terrorist / insurgent network is often accompanied by real - world constraints .
for example , consider the following : * we may know an individual s relationships in the terrorist / insurgent network , but may not have enough information ( i.e. where he or she may reside , enough evidence , etc . ) to actually target him or her .
* the potential target may be politically sensitive .
* the potential target may have fled the country or area of operations but still maintains his or her role in the terrorist / insurgent network through electronic communication . * the potential `` target '' may actually be a source of intelligence and/or part of an ongoing counter - intelligence operation ( i.e. as described in @xcite ) . throughout this paper we will also use the following notation .
the symbols @xmath7 will denote the sizes of @xmath8 respectively . for each @xmath9 , we will use @xmath10 to denote the degree of that node ( the number of individuals he / she is connected to ) and @xmath11 to denote the set of neighbors and we extend this notation for subsets of @xmath12 ( for @xmath13 ) .
we will use the notation @xmath14 to denote all edges in @xmath15 that are adjacent to node @xmath16 and the notation @xmath17 to denote the maximum degree of the network . given some subset @xmath18
, we will use the notation @xmath19 to denote the subgraph of @xmath20 induced by @xmath21 .
we describe an example network in example [ zeroex ] .
[ zeroex ] consider network @xmath22 in figure [ ex1fig ] .
nodes @xmath23 and @xmath24 may be leaders of a strategic cell that provides guidance to attack cells ( nodes @xmath25 and @xmath26 ) .
note that no members in the attack cells are linked to each other .
also note that if node @xmath23 is the leader , and targeted , he could easily be replaced by @xmath24 . ) for example [ zeroex ] . ]
we now introduce the notion of network - wide degree centrality as per @xcite . the key intuition of this paper is to use this centrality as a measure of the network s ability to re - generate leadership .
[ firstdefine ] the degree centrality of a network @xmath20 , denoted @xmath27 is defined as : @xmath28 we note that there are other types of network - wide centrality ( i.e. network - wide betweenness , closeness , etc . ) .
we leave the consideration of these alternate definitions of network - wide centrality to future work .
freeman @xcite shows that for a star network , the quantity @xmath29 equals @xmath30 - and this is the maximum possible value for this quantity .
hence , the value for @xmath27 can be at most @xmath31 .
as this equation is clearly always positive , network - wide degree centrality is a scalar in @xmath32 $ ] . turning back to example [ zeroex ]
, we can compute @xmath33 - which seems to indicate that in this particular terrorist / insurgent network that , after leadership is targeted , there is a cadre of second - tier individuals who can eventually take control of the organization . throughout this paper , we find it useful to manipulate equation [ firstdef ] as follows .
@xmath34 we notice that the centrality of a network really depends on three things : number of nodes , number of edges , and the highest degree of any node in the network .
we leverage this re - arranged equation in many of our proofs .
further , we will use the function @xmath35 to denote the level of network - wide of the graph after some set of nodes is removed .
hence , @xmath36 .
we note that this function has some interesting characteristics .
for example , for some subset @xmath37 and element @xmath38 , it is possible that @xmath39 or @xmath40 , hence @xmath41 is not necessarily monotonic or anti - monotonic in this sense .
further , given some additional element @xmath42 , it is possible that @xmath43 or @xmath44 . hence , @xmath41 is not necessarily sub- or super- modular either .
consider example [ ex1 ] .
[ ex1 ] consider the network @xmath22 in figure [ ex1fig ] .
here , @xmath45 , @xmath46 , @xmath47 , and @xmath48 .
the fact that @xmath49 and @xmath50 illustrate that @xmath51 is not necessarily monotonic or anti - monotonic .
now let us consider the incremental increase of adding an additional element .
adding @xmath52 to @xmath53 causes @xmath51 to increase by @xmath54 while adding @xmath52 to @xmath55 causes @xmath51 to decrease by @xmath56 - implying sub - modularity .
however , adding @xmath57 to @xmath53 causes @xmath51 to decrease by @xmath58 while adding @xmath57 to set @xmath59 causes @xmath51 to increase by @xmath60 ( as @xmath61 ) - implying super - modularity .
hence , @xmath51 is not necessarily sub- or super- modular .
we now have all the pieces to introduce our problems of interest .
we include decision and optimization versions .
+ * @xmath62 : * + input : natural number @xmath63 , real number @xmath64 , network @xmath5 , and no - strike set @xmath65 + output : `` yes '' if there exists set @xmath66 s.t .
@xmath67 and @xmath68 `` no '' otherwise .
+ * @xmath69 : * + input : natural number @xmath63 , network @xmath5 , and no - strike set @xmath65 + output : set @xmath66 s.t
. @xmath67 s.t .
@xmath70 s.t .
@xmath71 and @xmath72 .
+ as our problems seek to find sets of nodes , rather than individual ones , it raises the question of `` how difficult are these problems . ''
we prove that @xmath73 is np - complete - meaning an efficient algorithm to solve it optimally is currently unknown .
following directly from this result is the np - hardness of @xmath74 .
below we state and prove this result .
[ fnpc ] @xmath73 is np - complete .
membership in np is trivial , consider a set @xmath21 of size @xmath63 , clearly we can calculate @xmath75 in polynomial time .
+ next we consider the vertex - cover ( @xmath76 ) problem and show that it can be embedded into an instance of @xmath73 . in the @xmath76 problem ,
the input consists of undirected graph @xmath77 and natural number @xmath63 .
the output is `` yes '' iff there is a set @xmath78 of size at most @xmath63 s.t . for all @xmath79 , either @xmath16 or @xmath80 ( or both )
are in @xmath81 .
this problem is well - known to be np - hard .
first we create a new network @xmath5 which consists of graph @xmath82 but with @xmath83 additional nodes which form a star that is disconnected from the rest of the network .
all of the new nodes are put in the no - strike set @xmath65 ( part of the input of @xmath73 ) .
clearly , the center of this star is always the most central node in the graph , no matter what is removed from set @xmath84 .
this allows us to treat @xmath17 as a constant equal to @xmath85 .
also note that with this construction , for both problems , if a solution exists of less than size @xmath63 , there also exists a solution of exactly size @xmath63 .
further , we note that for any subset of @xmath12 whose removal does not affect the overall maximal degree of the network ( which is any node outside the set @xmath65 - hence in some corresponding subset of @xmath81 in the graph of the dominating set problem ) , when some set @xmath21 ( of size @xmath63 ) is removed from @xmath12 , the network - wide degree centrality for the resulting graph can be expressed as follows : @xmath86 .
+ the proof of correctness of the embedding rests on proving that a `` yes '' answer is returned for the vertex cover problem iff + @xmath87 .
+ first , suppose by way of contradiction ( bwoc ) there is a `` yes '' answer to the @xmath76 problem and a `` no '' answer to the corresponding @xmath73 problem .
let @xmath88 be the set of nodes that cause a `` yes '' answer to @xmath76 .
if we remove the corresponding nodes from @xmath20 , there are @xmath89 edges left in that network .
hence , as this is a set of size @xmath63 ( thus , meeting the cardinality requirement of @xmath73 then @xmath90 which would cause a `` yes '' answer for @xmath73 hence a contradiction .
+ going the other direction , suppose bwoc there is a `` yes '' answer to the @xmath73 problem and a `` no '' answer to the corresponding @xmath76 problem .
let @xmath21 be the nodes in the solution to @xmath73 .
clearly , this set is of size @xmath63 and by how we set up the no - strike list ( @xmath65 ) , there are corresponding nodes in @xmath91 .
as these nodes cause a `` yes '' answer to @xmath73 , they result in the removal of @xmath92 number of edges in @xmath20 . by the construction ,
none of these edges are adjacent to nodes in @xmath65 . hence , there are corresponding edges in @xmath82 . as
this is also the number of edges in @xmath82 , then this set is also a vertex cover - hence a contradiction .
hence , as we have shown membership in np and that this problem is at least as hard as the dominating set problem ( resulting in np hardness ) , the statement of the theorem follows .
@xmath74 is np - hard follows directly from theorem [ fnpc ] .
now with the problems and their complexity identified , we proceed to develop algorithms to solve them .
first , we develop an integer program that , if solved exactly , will produce an optimal solution .
we note that solving a general integer program is also np - hard .
hence , an exact solution will likely take exponential time .
however , good approximation techniques such as branch - and - bound exist and mature tools such as qsopt and cplex can readily take and approximate solutions to integer programs .
we follow our integer program formulation with a greedy heuristic .
though we can not guarantee that the greedy heuristic provides an optimal solution , it often provides a natural approach to approximating many np - hard optimization problems .
our first algorithm is presented in the form of an integer program .
the idea is that certain variables in the integer program correspond with the nodes in the original network that can be set to either @xmath93 or @xmath31 .
an objective function , which mirrors the @xmath94 function is then maximized .
when this function is maximized , all nodes associated with a @xmath31 variable are picked as the solution . for each @xmath9 ,
create variables @xmath95 . for each undirected edge @xmath96 , create three variables : @xmath97 .
note that the edge is considered in only `` one direction '' for the @xmath98 variables and both directions for the @xmath99 variables .
we define the @xmath100 integer program as follows : @xmath101 subject to : @xmath102 next we prove how many variables and constraints @xmath100 requires as well as prove that it provides a correct solution to @xmath74
. @xmath100 has @xmath103 variables and @xmath104 constraints . _
_ given the vector @xmath105 returned by @xmath100 , the set @xmath106 is a solution to @xmath74 .
+ _ _ ( 2.)__given a solution @xmath21 to @xmath74 , @xmath107 and @xmath108 will maximize @xmath100 . _
_ suppose , bwoc , @xmath106 is not an optimal solution to @xmath74 .
then there is some @xmath109 that is .
suppose @xmath107 and @xmath108 . clearly , by the definition of a solution to @xmath74 , constraints [ kconst],[nosconst ] and [ zerooneconst ] are all met .
constraints [ adjedge1 ] and [ adjedge2 ] set variables associated with edges adjacent to nodes not in @xmath21 to @xmath31 .
hence , the quantity @xmath110 is equal to the number of edges in the network .
the @xmath98 edge variables ( both of them for each edge ) are also set in a similar manner .
constraints [ justonemax],[zeroonemax ] ensures that only one set of such edge variables are set to @xmath31 .
hence , the quantity @xmath111 is the degree of one node in the network .
as this quantity is present in the objective function and non - negative , it corresponds to the @xmath17 . as we note that @xmath112 is equal to the number of nodes in @xmath20 when @xmath21 is removed , we see that this function is @xmath41 . as this quantity
is maximized , we have a contradiction .
+ _ ( 2 . ) _ suppose , bwoc , @xmath107 and @xmath108 is not an optimal solution to @xmath100 .
using the same line of reasoning as above , we see that the objective function of @xmath100 is the same as @xmath41 , which also gives us a contradiction .
note that this integer program does not have a linear objective function .
however , this can be accommodated for by instead solving @xmath63 different integer programs and taking the solution from whichever one returns the greatest value for the objective function ( that is greater than the initial network - wide degree centrality , of course ) . in this case , each integer program is identified with a natural number @xmath113 and the @xmath16th integer program has the following objective function : @xmath114 as well as constraint [ kconst ] as follows : @xmath115 notice that now the quantities @xmath116 and @xmath117 can be treated as constants , making the objective function linear . however , for networks with a heterogeneous degree distribution where @xmath118 , it is likely that only the integer program for the case where @xmath119 is needed as removing any node with edges that is unconnected to a maximal degree node will result in an increase in network - wide degree centrality .
again , we stress that @xmath100 provides an exact solution .
as integer - programming is also np - hard , solving these constraints is likely intractable unless @xmath120 .
however , techniques such as branch - and - bound and mature solvers such as qsopt and cplex can provide good approximate solutions to such constraints .
even if the integer program must be linear , we can use the techniques described above to solve @xmath63 smaller integer programs or obtaining an approximation by treating the terms involving the total number of nodes in the resulting graph ( in the objective function ) as constants .
additionally , a relaxation of the above constraints where @xmath121 and @xmath122 variables lie in the interval @xmath32 $ ] is solvable in polynomial time and would provide a lower - bound on the solution to the problem ( although this would likely be a loose bound in many cases ) .
the integer program introduced in the last section can be leveraged by an integer - program solver for an approximate solution to @xmath74 .
however , it likely will not scale well to extremely large networks .
therefore , we introduce a greedy heuristic to find an approximate solution .
the ideas is to iteratively pick the node in the network that provides the greatest increase in @xmath94 - and does not cause a decrease .
network @xmath5 , no - strike set @xmath6 , cardinality constraint @xmath63 subset @xmath123 .
the following two propositions describe characteristics of the output and run - time of @xmath124 , respectively . if @xmath124 returns a non - empty solution ( @xmath21 ) , then @xmath67 and @xmath125 .
as the algorithm terminates its main loop once the cardinality of the solution reaches @xmath63 and as in each iteration , the variable @xmath126 is initialized as zero , the statement follows .
@xmath124 runs in @xmath127 time .
we note that @xmath94 is computed in @xmath128 time as it must update the node with the maximum degree .
as the outer loop of the algorithm iterates at most @xmath63 times and the inner loop iterates @xmath129 times , the statement follows .
though our guarantees on @xmath124 are limited , we show that it performs well experimentally in the next section .
[ greedy - ex ] following from examples [ zeroex]-[ex1 ] using the terrorist / insurgent network @xmath22 from figure [ ex1fig ] , suppose a user wants to identify @xmath130 nodes that will cause the network to become `` as fragile as possible '' and is able to target any node .
hence , he would like to solve @xmath131 and decides to do so using @xmath124 .
initially , @xmath45 . in the first iteration , it selects and removes node * a * , increasing the fragility ( @xmath132 ) . in the next iteration , it selects node * j * , giving us @xmath133 .
finally , in the third iteration , it picks node * c*. this results in @xmath134 .
the algorithm then terminates .
all experiments were run on a computer equipped with an intel core 2 duo cpu t9550 processor operating at @xmath135 ghz ( only one core was used ) .
the machine was running microsoft windows 7 ( 32 bit ) and equipped with @xmath136 gb of physical memory .
we implemented the @xmath137 algorithm using python 2.6 in under @xmath138 lines of code that leveraged the networkx library available from http://networkx.lanl.gov/. we compared the results of the @xmath137 to three other more traditional approaches to targeting that rely on centrality measures from the literature .
specifically , we look at the top closeness and betweenness nodes in the network . given node @xmath16 , its closeness is the inverse of the average shortest path length from node @xmath16 to all other nodes in the graph .
_ betweenness _ , on the other hand , is defined as the number of shortest paths between node pairs that pass through @xmath16 .
formal definitions of both of these measures can be found in @xcite .
we studied the effects of our algorithm on five different datasets .
the network * tanzania * @xcite is a social network of the individuals involved with the al qaeda bombing of the u.s .
embassy in dar es salaam in 1998 .
it was collected from newspaper accounts by subject matter experts in the field .
the remainder networks , * genterrornw1 * -*genterrornw4 * are terrorist networks generated from real - world classified datasets@xcite .
the * tanzania * and the * genterrornw1 * -*genterrornw4 * datasets used in our analysis were multi - modal networks , meaning they contain multiple node classes such as agents , resources , locations , etc . the presence of the different node classes generate multiple or meta networks , which , in their original state , do not provide the single - mode agent by agent network needed to test our algorithms .
johnson and mcculloh @xcite demonstrated a mathematical technique to convert meta networks into single - mode networks without losing critical information . using this methodology
, we were able to derive distant relationships between nodes as a series of basic matrix algebra operations on all five networks .
the result is an agent based social network of potential terrorist .
characteristics of the transformed networks of agent node class only can be found in table [ datasettable ] .
.network datasets [ cols="<,<,<,<,<",options="header " , ] in our experiments , we showed that our algorithm was able to significantly increase the network - wide degree centrality by removing nodes - hence increasing the @xmath94 function with respect to a given network . in each of the five real - world terrorist networks that we examined , removal of only @xmath0 of nodes
can increase the network - wide centrality between @xmath1 and @xmath2 ( see figures [ fragile1]-[fragile5 ] ) . in figure [ fragile0 ]
we show a visualization of how the * tanzania * network becomes more `` star - like '' with subsequent removal of nodes by the greedy algorithm . .
panel * a * shows the original network .
panel * b * shows the network after @xmath130 nodes are removed , panel * c * shows the network after @xmath139 nodes are removed , and panel * d * shows the network after @xmath140 nodes are removed .
notice that the network becomes more `` star - like '' after subsequent node removals . in our experiment ,
after @xmath137 removed @xmath141 of the nodes in the network , it took the topology of a star . ] for comparison , we also looked at the removal of high degree , closeness , and betweenness nodes .
removal of high - degree , closeness , or betweenness nodes tended to increase the network - wide centrality .
in other words , traditional efforts of targeting leadership _ without _ first conducting shaping operations may actually _ increase _ the organization s ability to regenerate leadership - as such targeting operations effectively cause an organization to de - centralize .
we display these results graphically in figures [ fragile1]-[fragile5 ] .
notice that @xmath137 consistently causes an increase in the network - wide degree centrality .
an analysis of variance ( anova ) reveals that there is a significant difference in the performance among our algorithm and the centrality measures with respect to increase or decrease in network - wide degree centrality ( @xmath142-value less than @xmath143 , calculated with * r * version 2.13 ) .
additionally , pairwise analysis conducted using tukey s honest significant difference ( hsd ) test indicates that the results of our algorithm differ significantly from any of the three centrality measures with a probability approaching @xmath144 ( @xmath145 confidence , calculated with * r * version 2.13 ) .
typically , the ratio of percent increase in fragility to the percent of removed nodes is typically @xmath146 or greater .
, top degree , top closeness , and top betweenness .
the scale of the x - axis is positioned at @xmath147 . ] , top degree , top closeness , and top betweenness .
the scale of the x - axis is positioned at @xmath147 . ] , top degree , top closeness , and top betweenness .
the scale of the x - axis is positioned at @xmath147 . ] , top degree , top closeness , and top betweenness .
the scale of the x - axis is positioned at @xmath147 . ] , top degree , top closeness , and top betweenness .
the scale of the x - axis is positioned at @xmath147 . ]
we also evaluated the run - time of the @xmath137 algorithm . with the largest terror network considered ( genterrornw4 ) , we achieved short runtime ( under @xmath148 seconds ) on standard commodity hardware ( see figure [ runtime ] ) .
hence , in terms of runtime , our algorithm is practical for use by a real - world analyst .
as predicted in our time complexity result , we found that the runtime of @xmath137 increases with the number of nodes removed .
we note that the implementations of top degree , closeness , and betweenness calculate those measures for the entire network at once - hence increasing the number of nodes to remove does not affect their runtime . , top degree , top closeness , and top betweenness . ] to study the scalability of @xmath137 , we also employed it on two large social networks .
note that these datasets are not terrorist or insurgent networks .
however , the larger size of these datasets is meant to illustrate how well our approach scales . for these experiments , we used an e - mail network from university rovira i virgili ( urv e - mail ) @xcite and a network science collaboration network ( ca - netsci ) from @xcite ( see table [ datasettable ] ) . in figure [ scale1 ]
we show the percentage of nodes removed vs. the percent increase in fragility .
we note that @xmath146 ratio of percent increase in fragility to the percent of removed nodes appears to be maintained even in these large datasets . in figure [ scale2 ]
we show the runtime for @xmath137 on the two large networks .
we note that the behavior of runtime vs. number of nodes removed resembles that of the genterrornw4 network from the previous section . also of interest
is that the algorithm was able to handle networks of over a thousand nodes in about @xmath149 minutes on commodity hardware . .
various aspects of the resiliency of terrorist networks have been previously explored in the literature .
for instance , @xcite studies the ability such network to facilitate communication while maintaining secrecy while @xcite studies how such networks are resilient to cascades .
however , to our knowledge , the network - wide degree centrality in such networks - and how to increase this property - has not been previously studied . there has been much work dealing with the removal of nodes from a network to maximize fragmentation @xcite where the nodes removed are mean to either increase fragmentation of the network or reduce the size of the largest connected component . while this work has many applications , it is important to note that there are special considerations of terrorist and insurgent networks that we must account for in a targeting strategy .
for instance , if conducting a counter - intelligence operation while targeting , as in the case of @xcite , it may be desirable to preserve some amount of connectivity in the network .
additionally , fragmentation of a network may result in the splintering of an organization into smaller , but more radical and deadly organizations .
this happens because in some cases , it may be desirable to keep certain terrorist or insurgent leaders in place to restrain certain , more radical elements of their organization . such splinter was observed for the insurgent organization jaysh al - mahdi in iraq @xcite .
further , these techniques do not specifically address the issue of emerging leaders .
hence , if they were to be used for counter - terrorism or counter - insurgency , they would likely still benefit from a shaping operation to reduce organization s ability to regenerate leadership .
there has been some previous work on identifying emerging leaders in terrorist networks .
although such an approach could be useful in identifying certain leaders , it does not account the organizations ability as a whole to regenerate leadership . in @xcite , the topic of _ cognitive demand _
is studied .
the cognitive load of an individual deals with their ability to handle multiple demands on their time and work on complex tasks .
typically , this can be obtained by studying networks where the nodes may represent more than individual people - but tasks , events , and responsibilities .
however , it may often be the case that this type of information is often limited or non - existent in many situations . additionally ,
as discussed throughout this paper , the targeting of individual nodes may often not be possible for various reasons .
hence , our framework , that focuses on the _ network s ability to regenerate leadership _ as opposed to finding individual emerging leaders may be more useful as we can restrict the available nodes in our search using the `` no strike list . '' by removing these nodes from targeting consideration - but by still considering their structural role - our framework allows a security force to reduce the regenerative ability of a terror network by `` working around '' individuals that may not be targeted . in more recent work
@xcite looks at the problem of removing leadership nodes from a terrorist or criminal network in a manner that accounts for new links created in the aftermath of an operation .
additionally , @xcite look at identifying leaders in covert terrorist network who attempt to minimize their communication due to the clandestine nature of their operations .
they do this by introducing a new centrality measure called `` covertness centrality . ''
both of these approaches are complementary to ours as they focus on the leadership of the terrorist or insurgent group - as this approach focuses on the networks ability to re - generate leadership . a more complete integration of this approach leadership targeting method such as these ( i.e. using a network - wide version of covertness centrality ) is an obvious direction for future work .
in this paper we described how to target nodes in a terrorist or insurgent network as part of a _ shaping _ operation designed to reduce the organization s ability to regenerate leadership .
our key intuition was to increase the network - wide degree centrality which would likely have the effect of eliminating emerging leaders as maximizing this quantity would intuitively increase the organization s reliance on a single leader . in this paper , we found that though identifying a set of nodes to maximize this network - wide degree centrality is np - hard , our greedy approach proved to be a viable heuristic for this problem , increasing this quantity between @xmath150 in our experiments .
future work could include an examination of other types of network - wide centrality for instance network - wide closeness centrality instead of network - wide degree centrality .
another aspect that we are considering in ongoing research is determining the effectiveness of the shaping strategy when we have observed only part of the terrorist or insurgent organization as is often the case as such networks are created from intelligence data .
we would like to thank jon bentley and charles weko for their feedback on an earlier version of this paper .
+ some of the authors are supported under by the army research office ( project 2gdatxr042 ) .
the opinions in this paper are those of the authors and do not necessarily reflect the opinions of the funders , the u.s .
military academy , or the u.s .
army .
s. wasserman and k. faust , _ social network analysis : methods and applications _
, 1st ed .
structural analysis in the social sciences.1em plus 0.5em minus 0.4emcambridge university press , 1994 , no
. 8 .
m. ovelgonne , c. kang , a. sawant , and v. subrahmanian , `` covertness centrality in networks , '' in _ proc .
2012 intl .
symposium on foundations of open source intelligence and security informatics _ , aug .
|
security organizations often attempt to disrupt terror or insurgent networks by targeting `` high value targets '' ( hvt s ) .
however , there have been numerous examples that illustrate how such networks are able to quickly re - generate leadership after such an operation . here ,
we introduce the notion of a _ shaping _ operation in which the terrorist network is first targeted for the purpose of reducing its leadership re - generation ability before targeting hvt s .
we look to conduct shaping by maximizing the network - wide degree centrality through node removal .
we formally define this problem and prove solving it is np - complete .
we introduce a mixed integer - linear program that solves this problem exactly as well as a greedy heuristic for more practical use .
we implement the greedy heuristic and found in examining five real - world terrorist networks that removing only @xmath0 of nodes can increase the network - wide centrality between @xmath1 and @xmath2 .
we also show our algorithm can scale to large social networks of @xmath3 nodes and @xmath4 edges on commodity hardware .
|
in this paper we describe a new approach to perturbation theory for systems with local moments or , in principle , any local order parameter .
we begin by summarizing the present status of weak coupling perturbation theories for such systems , with the hubbard model as a primary example .
over the course of the last decade weak coupling perturbation theories have been used extensively to investigate the physics of the hubbard model and its extensions . the simplest perturbative approach to the hubbard model
is just hartree fock theory for one body correlation functions .
the conserving approximation implied for two body correlations is then the random phase approximation , or rpa . in the large@xmath1 limit at half filling the static spin response is unstable within hartree fock , and the system exhibits a broken symmetry ground state with infinite
range antiferromagnetic order @xcite .
furthermore a gap of order @xmath1 appears in the one body density of states for the ordered phase .
if the hartree fock equations are extended to finite temperature , long range antiferromagnetic order onsets at a temperature of order @xmath1 in the large@xmath1 limit .
the zero temperature behavior of the hartree
fock approximation away from half filling has also been investigated , and incommensurate states , as well as vertical and diagonal domain walls , have been studied @xcite .
however , the appropriate treatment for finite temperatures has remained unclear , because the energetic competition of the various states is delicate .
in fact the hartree
fock approximation at finite temperatures provides only a crude description of the physics in a two ( or one ) dimensional system . in accordance with the mermin
wagner theorem @xcite , infinite
range magnetic order is forbidden for @xmath0 in two dimensions .
the magnetic correlation length @xmath2 instead diverges exponentially in @xmath3 for @xmath4 at half filling , and the quantitative behavior of @xmath5 , as well as the zero temperature ordered moment , is renormalized by quantum fluctuations .
nevertheless the half filled model exhibits a gap ( or _ pseudogap _ , since we do not mean to imply a rigorously zero density of states ) at finite temperature in the spin disordered phase . with the intention of modeling strong spin fluctuations in a disordered system , the present authors proposed the fluctuation exchange , or flex , approximation @xcite as a plausible starting point . the self consistent treatment of spin fluctuations in the one body self
energy suppresses the magnetic ordering transition to zero temperature .
nevertheless the new approximation is itself flawed : it does not produce a sharp gap at half filling and instead smears out the density of states , leaving a finite weight at zero energy for all temperatures @xcite .
furthermore the spin response function diverges with decreasing temperature in a way which only roughly mirrors the exact behavior of the model .
the failure at half filling calls into question results obtained for large @xmath1 at small doping , and in the end the flex approximation fails to describe the pseudogap regime .
the inadequacy of the flex approximation has motivated various attempts to improve the theory .
kampf and schrieffer have calculated the electron spectral density near an antiferromagnetic transition to lowest order in an effective spin fluctuation interaction , and their results exhibit a pseudogap when the spin correlation length is large @xcite .
vilk and tremblay @xcite have introduced a factorization approximation for the electron four point function which is constrained to be exact when all space time coordinates coincide .
others @xcite have argued that perturbative calculations based on an unrenormalized electron propagator are justified by a cancellation between vertex corrections and self energies , leading to the appearance of a pseudogap in some parameter regimes .
in addition efforts have been made to extend the flex approximation within a scheme based on infinite order ( or `` parquet '' ) vertex renormalization @xcite .
quantitative studies of parquet renormalization have had the goal of establishing the validity or inadequacy of perturbation theory resummations for systems with a gap , but without infinite range order .
why would one assume that such resummations have any chance of succeeding ?
for at least one nontrivial model , the anderson impurity in a metallic host , it has been known since the mid1980 s that zero temperature perturbation theory for thermodynamic properties has an infinite radius of convergence in @xmath1 @xcite .
this fact , as well as some suggestive results in the low density limit , initially encouraged hope that an infinite order parquet summation for the doped mott insulator might succeed if the technical difficulties impeding its implementation could be overcome .
in fact this hope is not justified .
full numerical solutions of the minimal parquet equations for the two dimensional hubbard model in the large@xmath1 limit @xcite yield the following results : ( i ) no gap appears in the converged solution at half filling .
( ii ) the spectral density at low energies is actually larger than that in a flex calculation .
we interpret these results to mean that conventional perturbation theory @xcite fails to describe the physics of a mott
hubbard gap or pseudogap in a spin disordered system .
furthermore the parquet results , which consistently incorporate vertex renormalization to infinite order , bring into question whether the basic problem can be circumvented by the constrained factorization approximation or a cancellation between parts of perturbative diagrams .
rather we believe the failure of perturbation theory around a uniform ( i.e. , fermi liquid ) starting point is due to a qualitative change in the saddle point structure of the action for the hubbard model at temperatures of order @xmath1 in the large@xmath1 limit .
this change is induced by the appearance of local moments ( or , more generally speaking , a local order parameter field ) in the theory .
conventional perturbation theory becomes unreliable below a temperature scale which depends on both the interaction @xmath1 and the chemical potential @xmath6 . below this scale
one needs a revised perturbation theory which incorporates `` anomalous '' feynman graphs , but does so without inducing infinite
range order .
these seemingly contradictory objectives may be accomplished by introducing a spatially varying static field with the full symmetry of the associated classical order parameter .
one way to introduce such a field is via a stratonovich - hubbard ( s h ) transformation @xcite of the interaction .
in particular we employ below a spin rotation
invariant s
h field to decouple the static @xmath7 component of the hubbard interaction .
about thirty years ago wang et al .
@xcite showed for the magnetic alloy problem that the effective static free energy induced by a time varying s
h field provides a physical picture of the crossover from band like to local
moment behavior .
evenson et al .
@xcite analyzed the coupling between isolated moments which arises within this approach and noted that the simplest scheme generates an ising , rather than heisenberg , interaction .
below we show that when a vector s
h field is introduced , the effective action reduces to a classical heisenberg form at half filling .
this means no infinite range order at @xmath0 , an exponentially diverging correlation length @xmath2 , _ and _ a mott
hubbard gap in the spin disordered phase .
furthermore the approximation scheme should provide a systematic way to study doping of the mott insulator . in the doped case both orientation and amplitude fluctuations of the field become important .
in particular the effective action for the static field is strongly non
gaussian @xcite .
since the action arises directly from integrating out the fermion degrees of freedom , its form reflects both the underlying spin and charge structure of the system . in this way
domain wall stripe fluctuations can occur naturally if they are present in the model .
we describe this new approximation scheme for the hubbard model in section 2 , then derive expressions for specific electron correlation functions in section 3 . in section 4
we provide a brief description of our monte carlo sampling scheme for integration over the fluctuating local fields .
in section 5 we present some illustrative results obtained on a work station for a @xmath8 system .
we outline in section 6 the form of generalized approximations which extend the present approach in the same sense that flex extends hartree fock .
finally we close in section 7 with some speculations on possible implications for high temperature superconductivity and with an outline of the next steps in this work .
in this section we show how to decouple the the static spin interaction in the repulsive hubbard model by introducing a spin rotation
invariant s h field .
this field may be interpreted as a classical heisenberg order parameter which generalizes the saddle point in conventional temperature dependent hartree fock calculations .
it is conventional to decouple the interaction term by introducing independent fields at each point in imaginary time .
when this is done , some freedom remains in choosing the form of the interaction , due to the equal time fermion anticommutation relations .
one may write equivalently un_in_i & = & u(n_i^2 - m_iz^2 ) [ intone1 ] + & = & u(n_i - m_iz^2 ) [ intone2 ] + & = & un_i -u(m_ix^2 + m_iy^2 ) , [ intone3 ] where n_i & = & n_i + n_i + _ i & = & _ c^_i_
c^_i . [ opsone ] there is in fact considerably more freedom allowed , since the interaction is rotationally invariant : the z component of the moment operator in eqns .
[ intone1 ] and [ intone2 ] may be replaced by the x or y component , and likewise different pairwise combinations of moment operators may be substituted in eqn .
[ intone3 ] .
one is free to consider in addition arbitrary linear combinations of these expressions , weighted properly to reproduce the original interaction .
this was realized by several authors around 1970 during early studies of the single impurity anderson model and the hubbard model . in particular
wang et al . employed eqn .
[ intone2 ] @xcite and hamann employed eqn .
[ intone1 ] @xcite as the starting points for equal time stratonovich hubbard decoupling schemes .
note that both these expressions imply an ising like s h field governing the spin degrees of freedom , while the third expression implies an xy like field .
all three expressions result in actions which break the spin rotational symmetry of the theory , and the symmetry is only restored by the integration over field configurations .
as long as all configurations are included there is no difficulty , but problems arise if approximations are introduced for any system beyond a single impurity .
this was realized by evenson et al .
@xcite in large@xmath1 studies of the two site hubbard model within a static approximation : the effective spin
spin coupling assumes an ising , rather than heisenberg , form when eqn .
[ intone2 ] is used as a starting point .
it would clearly be desirable to introduce a symmetrical heisenberg decoupling , but this is difficult for the following reason : no matter how the spin quantization axis is chosen , the three equivalent components of the spin spin interaction are divided up between the two particle hole channels ( longitudinal and transverse ) of the electron vertex , which `` overlap '' completely at equal time .
a heisenberg decoupling can be introduced artificially if , for example , the interaction is written as an equal weight linear combination of eqn .
[ intone1 ] for the three choices of moment operator , i.e. , un_in_i = un_i^2 -u _
[ intone4 ] this procedure introduces spurious factors of @xmath9 in the saddle
point equations when a direction is chosen for the broken symmetry spin axis , however , and so offers an unsatisfactory solution .
we may restate the problem as follows : a conventional equal
time s h transformation , followed by a static approximation within the resulting theory , violates the crossing symmetry of the hubbard interaction vertex and prevents a spin rotation invariant treatment of fluctuations . to insure an equivalent static treatment of all three spin components , we propose the introduction of a zero frequency , rather than equal time , s
such a decoupling scheme is possible because the static components of the direct and exchange channels in the hubbard model have essentially zero overlap at low temperatures .
this allows the simultaneous treatment of interactions in both crossed channels and leads to a unique prescription for the decoupling scheme independent of the expression employed for the interaction in eqns .
[ intone1]-[intone4 ] . the same procedure can be applied in more general situations when it is necessary to decouple multiple interactions in crossed channels .
it is possible to carry out this transformation directly within an operator based approach , but the manipulations are more transparent with anticommuting c numbers , and we follow the latter approach .
we consider a discretized representation of the partition function z = tre^-h , [ partition ] where h = -t_ic^_i+,c^ _ i + u_ic^_ic^_ic^_i c^_i - _ ic^_ic^ _ i .
[ hamone ] let _
& = & + & = & /l , [ tau ] with @xmath10 the number of time slices .
we write the anticommuting c
numbers for time @xmath11 , site @xmath12 , and spin @xmath13 as @xmath14 , @xmath15 .
it is convenient to use the frequency transformed variables @xmath16 , @xmath17 , where ^i _ & = & _
n =- l/2^l/2 - 1e^i_n _ ^i_n + c^i _ & = & _ n =- l/2^l/2 - 1e^-i_n _ c^i_n [ fourier ] with _
n = ( 2n+1)t [ matsubara ] for integer @xmath18 . in order to keep track of discretization effects , we define i_n^ 1-e^-i_n , [ discrete ] which approaches @xmath19 for @xmath20 .
the discretized fermion action is s & = & -_in(i_n^+)^i_n c^i_n +
_ in(-t)^i+,_n c^i_n + & & + u_imnn^i_n+m ^i_-n-1 c^i_-n-1c^i_n+m .
[ actionone ] the interaction term is represented in figure 1 .
( note that the integer label @xmath21 corresponds to frequency @xmath22 . )
equivalently one may write s = s_0 + s_int , [ actiontwo ] with s_0 = _ ijn ^i_n_ijc^j_n , [ actionthree ] where the matrix in brackets has spatial indices , and _
ij & = & _ ij + ( * h_0*)_ij & = & [ mats ] now note that the interaction can be rewritten in two equivalent ways to emphasize the particle hole channels ( see figures 2a b ) : s_int & = & -u_imnn ^i_n+mc^i_n ^i_nc^i_n+m + & = & u_imnn ^i_n+m c^i_n^i_nc^i_n+m .
[ crossing ] at this stage we restrict attention to terms with @xmath23 , i.e. , the static component of the particle hole interactions . for @xmath23 the two sums in eqn .
[ crossing ] are essentially independent : the only terms which appear in both sums are restricted to @xmath24 . therefore the static particle hole interaction may be written s^m=0_int
= -u_i + u_i[actionfour ] where we have dropped from the double sums an @xmath24 correction term , which becomes negligible at low temperatures @xcite .
now all terms in eqn .
[ actionfour ] may be reduced to sums of perfect squares in preparation for the introduction of s h fields .
first note that _ n^i_nc^i_n & = & _
n ^i_n(_x+i_y)_c^i_n + _
n^i_nc^i_n & = & _
n ^i_n(_x - i_y)_c^i_n .
[ sumsone ] using these identifications one has = , [ sumstwo ] where m^i_x0 & = & _ n^i_n(_x)_ c^i_n + m^i_y0 & = & _ n^i_n(_y)_ c^i_n . [ sumsthree ] likewise , _ n^i_nc^i_n & = & ( n^i_0+m^i_z0 ) +
n^i_nc^i_n & = & ( n^i_0-m^i_z0 ) , [ sumsfour ] where n^i_0 & = & _
n(^i_nc^i_n+^i_nc^i_n ) + m^i_z0 & = & _
n^i_n(_z)_ c^i_n . [ sumsfive ] in this case one identifies = .
[ sumssix ] thus s^m=0_int = u_i(n^i_0)^2 - u_i(^i_0^i_0)^2 .
[ actionsix ] the repulsive charge interaction is retained for treatment by perturbation theory ( i.e. , saddle point plus fluctuations ) .
the attractive spin interaction in eqn .
[ actionsix ] may be decoupled using an independent static s
h field at each site @xmath12 : & & = + & & = _ i ( ) ^3/2 d^i_xd^i_yd^i_z .
+ & & [ stratonovich ] the s
h fields @xmath25 may be interpreted as classical magnetic fields which couple to the local electron spin polarization .
the hubbard action may be rewritten as s = _ i|^i|^2 + s_0 + s^m=0_charge + s^m 0_int , [ actionseven ] where s_0 = _ ijn^i_n\{-i^ _ n__ij + } c^j_n [ actioneight ] and s^m=0_charge = u_i(n^i_0)^2 .
[ actionnine ] the interaction contributions to the action for @xmath26 take precisely the form they would in a conventional perturbative treatment , _ except _ that all vertices related to the zero frequency spin vertex by crossing are absent . note that because we have decoupled only the @xmath23 spin components , we have left open the possibility of higher order perturbative expansions about a static approximation .
we consider now the static charge interaction @xmath27 in eqn .
[ actionseven ] . this term could be decoupled in parallel with @xmath28 by introducing a real valued s
h field @xmath29 .
since the charge interaction is repulsive , however , the fermion bilinear coupling to @xmath29 would be pure imaginary and unsuitable for monte carlo simulation .
it is straightforward to show that a saddle point approximation for the @xmath29 fields in the presence of a fixed @xmath25 configuration yields the same result as a spatially non uniform hartree fock .
for this reason we do not introduce @xmath29 fields , but reserve @xmath27 for treatment by perturbation theory ( i.e. , hartree fock at lowest order ) .
within such a hartree
fock approximation for fixed @xmath25 , s^m=0_charge u_inn^i _
^i_nc^i_n - u_in^i_^2 , [ actionten ] where @xmath30 is the mean occupancy at site @xmath12 .
it is essential to retain the `` constant '' term , since it depends on the local field configuration @xmath31 .
from this point we use the notation @xmath32 to denote the expectation value of the fermion observable @xmath33 in the presence of a fixed field configuration .
the static @xmath34dependent action assumes the form s ( ) = s_0 ( ) + s_f(,,c ) , [ actioneleven ] where s_0 ( ) & = & _ i|^i|^2 - u_i n^i_^2 + s_f(,,c ) & = & _ ijn^i_n\ { -i^_n_ij_ + } c^j_n .
[ actiontwelve ] the fermion contribution to the action may now be integrated out exactly , since it takes a one body form ( albeit a form non diagonal in spin ) : ddc = _ r , [ fermion ] where the @xmath35 are eigenvalues of the one body hamiltonian h ( ) = -t_ic^_i+,c^_i - _
ic^_i_ c^_i
+ _ i c^_ic^_i .
[ hamtwo ] thus the final expression for the effective action after the fermion integration is s_eff ( ) = s_0 ( ) - _ r .
[ newaction ] note that for fixed @xmath6 and @xmath25 a hartree
fock iteration is required to adjust the non uniform charge field @xmath36 to satisfy the condition n^i & = & _
rf(_r)_|r|i|^2 + f(_r ) & = & 11+e^_r .
[ scfone ] in this equation @xmath37 is the one body eigenvector corresponding to eigenvalue @xmath38 ; and the @xmath39 constitute a complete set of basis vectors corresponding to spin @xmath40 on site @xmath12 .
note that for an @xmath41 real space lattice , the size of the eigenvector space is @xmath42 .
the form of the charge
interaction hartree
fock employed here provides a template for the family of baym
kadanoff @xcite ( or flex @xcite ) generalizations of the present approach discussed in section 6 .
the charge field is here viewed as a nonsingular `` slave '' to the spin field : it is fixed to its saddle point , which changes when the @xmath34 configuration changes .
no further approximation to the action @xmath43 is now possible without explicitly breaking spin rotational and translational invariance . in conventional hartree
fock treatments it is assumed that the static field @xmath25 assumes a rigid configuration . at half filling the mean field ansatz is just & = & e^iqr_iq= ( , ) + n^i & = & 1 .
[ ansatz ] a saddle point evaluation of @xmath44 within this restricted field space reproduces the conventional hartree
fock equations for a commensurate antiferromagnet .
furthermore inhomogeneous hartree
fock solutions @xcite for @xmath45 follow by assuming a spin
space form ( collinear , spiral , etc . ) for the order parameter and solving the resulting saddle point equations .
we propose instead to evaluate electron correlation functions using @xmath44 as the action in a classical field monte carlo .
the monte carlo is classical since all finite frequency contributions to the action have been set aside for treatment by perturbation theory : such contributions , when included , alter the form of the static field action , but are not themselves statistically sampled . by this
means it is possible to insure that higher order approximations have a positive definite weighting factor @xmath46 , avoiding the sign problem inherent in a full quantum monte carlo .
it is important to note that this approach has been explicitly designed to treat the low temperature regime in which local moments exist , i.e. , in which the most probable configurations of @xmath25 have non zero amplitude ( at least for some sites @xmath12 ) .
we do not address in this paper the problem of local moment formation per se .
it is clear that at high temperatures the `` free '' @xmath25 fluctuations have variance @xmath47 , and it is not appropriate to use @xmath43 as a starting point for perturbation theory in this regime .
instead our purpose is to construct an approximation scheme which builds in the presence of disordered local moments at temperatures satisfying 4t^2/u t u [ condition1 ] in three dimensions and t u [ condition2 ] in two dimensions .
it is also important to note that this approximation omits quantum effects associated with `` tunneling '' of the order parameter fields , i.e. , possible @xmath48dependent instantons . as a result
the kondo effect is outside the present description .
we restrict our attention here to the implications of a classical ( i.e. , @xmath23 ) order parameter description and do not discuss additional quantum effects .
we next examine the behavior of @xmath44 in the large@xmath1 limit , then comment on the action s relationship to other work in the literature . in the large@xmath1 limit the hubbard model may be mapped to a quantum spin@xmath49 heisenberg model with near neighbor antiferromagnetic exchange integral @xmath50 .
we show below that in this limit the approximate action @xmath44 describes a classical heisenberg model with exchange integral @xmath50 .
this insures that in a two dimensional system the approximation obeys the mermin wagner theorem ,
i.e. , infinite
range spin ordering is absent down to @xmath51 .
however this approach does not include the quantum fluctuations which renormalize the zero temperature moment and reduce the finite temperature correlation length .
it is essential to include finite frequency fluctuations in order to address these points .
nevertheless the static action @xmath44 achieves the goal of reproducing the exact classical critical behavior within a well defined zeroth order approximation . to demonstrate the mapping to the classical heisenberg model ,
consider first the local one site action ( @xmath52 ) in the large@xmath1 limit for @xmath53 : s_eff^one - site = ||^2 - _ r .
[ onesite ] in this case there are only two eigenvalues @xmath35 . to find @xmath35 , choose the spin quantization axis along the direction of @xmath34 .
then the field
dependent term in @xmath54 is just -(c^_ic^_i - c^_i c^_i ) , [ intern ] with eigenvalues _ = -,=1 .
[ eigs ] minimization of the action with respect to @xmath5 gives = _ e^1+e^ , [ minimiz ] i.e. , = m_z , [ solution ] since m_z = _ = 111+e^- .
[ mzz ] for large @xmath1 the solution has m_z & & 1 + & & /2 . [ limits ] as expected , the one site gap is just _ - _ = u-(-u ) = u .
[ gapp ] note that this result has been obtained without breaking the rotational invariance of the one site hamiltonian , since the direction of @xmath34 is arbitrary , and the boltzmann weight must be integrated over @xmath34 . thus to zeroth order in @xmath55 , the half filled system consists of fluctuating fields @xmath25 of fixed length @xmath56 and uncorrelated directions .
the first non zero terms in perturbation theory enter at @xmath57 , and describe the process of one electron hopping to a neighboring site , then back to its starting point .
physically the hopping terms should induce correlations in the field directions at neighboring sites to lower the kinetic energy .
let the field directions at neighboring sites @xmath12 and @xmath58 be @xmath59 and @xmath60 .
suppose also that @xmath59 may be obtained by rotating @xmath61 in a right handed sense by angle @xmath62 about axis @xmath63 .
then the eigenstates with spin projection @xmath64 along @xmath59 can be written |n^i , & = & |z , + ( -i_x+_y)|z , + [ sumsquares ] the one site solution gives ||
= = n^i^i , [ meanfield ] so that @xmath65 will be occupied and @xmath66 will be empty ( and higher in energy by @xmath1 ) .
now apply second order perturbation theory to the electron ground state for the frozen field configuration with direction vectors @xmath67 .
the matrix element for hopping from |n^i,|n^j , |n^j,|n^j , [ hoppingsites ] is ( -t)n^j,|n^i , = ( -t ) ( -i_x-_y)^ * , [ hoppingmatrix ] where @xmath60 is obtained from @xmath59 by a right handed rotation by @xmath68 about axis @xmath69 .
the energy lowering of the ground state due to the virtual hop from @xmath70 and back is then = -t^2u^22 , [ lowering ] since _
x^2+_y^2 = 1 .
[ sumup ] the hopping from @xmath71 supplies the same energy , so the total perturbation energy lowering in this particular field configuration is -2t^2u_ij^2(_ij/2 ) & = & t^2u_ij ( _ ij-1 ) + & = & 4t^2u_ij + & = & 4t^2u_ij , [ heisen ] where @xmath72 is the electron spin expectation value in configuration @xmath31 for @xmath52 .
note finally that the effective action at half filling reduces to s_eff ( ) = _ i|^i|^2 - _ r .
[ actionthirteen ] in the large@xmath1 limit _
r , [ prodd ] with @xmath73 the perturbed ground state energy , since all other occupancies result in an energy higher by at least @xmath1 . since @xmath74 is a constant in this limit , the part of the action dependent on the fluctuating fields is precisely @xmath75 , with h_eff = 4t^2u_ij(n^in^j - ) , [ hamthree ] establishing the mapping to the classical heisenberg model .
the behavior of the action away from half filling in the large@xmath1 limit is not so easy to deduce , since the hartree
fock correction from the static charge interaction is crucial .
if one assumes @xmath76 sites are singly occupied and @xmath77 sites are empty , the one site saddle point equation implies @xmath78 on the singly occupied sites and @xmath79 on the empty sites .
a degeneracy then arises , since the low
lying one electron states on the singly occupied sites pick up a hartree
fock density contribution @xmath80 , while the states on the empty sites have no hartree
fock correction .
the two sorts of sites then have degenerate low lying states in the absence of @xmath81 : the occupied state on each singly occupied site has energy @xmath82 , and the two states on each unoccupied site also have energy 0 . as expected , the hopping becomes crucial for determining @xmath36 and the favored @xmath25 configurations .
thus in the doped case there are important amplitude , as well as orientation , fluctuations in the s
h field configurations arising from @xmath44 .
we next comment briefly on the relationship of our approximation to other work in the literature .
we have argued in section 1 that perturbation theory in @xmath1 fails to resum properly at low temperatures in the doped mott insulator phase due to a fundamental change in the system s saddle point structure .
nevertheless a number of authors @xcite@xcite have observed that gaps or pseudogaps in the one electron density of states for the positive@xmath1 and negative@xmath1 hubbard models may be obtained within approximations which resemble perturbation theory about a non interacting ground state .
such approximations have in common the feature that the one electron self energy @xmath83 may be written in the form ( k , i_n ) = t_qm v(q , i_m)g_0(k - q , i(_n- _ m ) ) , [ perturb ] where @xmath84 is a momentum and frequency dependent effective potential and @xmath85 is a bare electron propagator .
crucial points to note are that ( i ) the electron propagator is left bare ; and ( ii ) higher order perturbation theory in @xmath84 leads to a deterioration , rather than an improvement , in the approximation for the one electron density of states .
we argue that the reasons for this behavior are as follows : the local moments which underlie the mott
hubbard gap generate anomalous diagrams in perturbation theory , i.e. , terms in the one electron self energy off diagonal in momentum and spin . at zero temperature and half filling the moments have true long range staggered order , so that a momentum
space description becomes possible . if the static off diagonal self
energy is denoted @xmath86 , with @xmath87 , then the effective diagonal self energy becomes ( k , i_n ) = ^2_qg_0(k- ) , [ afmhartree ] generating a hartree
fock antiferromagnetic gap of @xmath88 in the density of states . in this case , as is well known , @xmath85 enters eqn .
[ afmhartree ] rather than @xmath89 to avoid double counting .
thus approximations of the form in eqn .
[ perturb ] may be viewed as generalizations of eqn .
[ afmhartree ] which `` smear '' the off diagonal self
energy @xmath90 over a range in momentum and frequency
space @xcite . in this sense
these approximations are quite different from perturbation theory in @xmath84 , and they do not provide support for the convergence of perturbation theory in the doped mott insulator .
we would argue that such approximations model only roughly the physics of this system @xcite . in the doped phase the order parameter field has strong non
gaussian fluctuations in both amplitude and orientation , and these are coupled to the charge degrees of freedom .
furthermore , as discussed in reference @xcite , this general approach has difficulties addressing rare large amplitude fluctuations that play an essential role .
we comment also on the relationship of the current approach to the phenomenological spin fermion ( sf ) model . in recent years a number of perturbative studies @xcite have been carried out for this model , which describes a system of electrons coupled to a gaussian distributed vector spin field @xmath91 .
the microscopic action @xmath44 differs from @xmath92 in two crucial ways : ( i ) @xmath92 omits coupling to charge degrees of freedom , while @xmath44 describes a highly nonlinear coupling between spin and charge .
the form of this coupling is dictated by the hubbard interaction itself , and its omission prevents a proper description of the doped mott insulator phase .
( ii ) the finite frequency corrections to the actions differ significantly , since we have introduced only a static s
h field to circumvent the problems associated with an equal time decoupling .
finally , our point on the non gaussian nature of fluctuations in @xmath44 applies to @xmath92 as well , i.e. , in the interesting parameter regime for this model we expect a conventional perturbative treatment to be impossible due to the formation of local moments .
electron correlation functions for the action @xmath93 in eqn .
[ actioneleven ] have been calculated using the metropolis algorithm for statistical sampling of @xmath25 configurations.the monte carlo algorithm adopted relies on single
site updating and resembles a hybrid of glauber and kawasaki dynamics @xcite for the ising or classical heisenberg models .
the algorithm is described more precisely in section 4 .
electron correlation functions have the same general form as in full quantum monte carlo calculations with continuous valued s h fields .
let @xmath94 be an electron observable written in anticommuting c number notation , and let d _ i ( ) ^3/2d^i_xd^i_yd^i_z .
[ measure ] then a & = & dddc a(,c ) dddc + & = & 1zdz_f ( ) , + & & [ expect ] where z_f ( ) & = & ddc + & = & _ r , [ expect2 ] and the last line in eqn .
[ expect ] follows by multiplying and dividing by @xmath95 within the integral .
the expression in square brackets is just an electron thermal average in the presence of the background @xmath25 configuration .
thus finally a & = & 1zd a(,c ) _ + & & a(,c ) _ , [ expect3 ] where s_eff ( ) & = & s_0 ( ) - z_f ( ) + & = & s_0 ( ) - _ r .
[ expect4 ] the simple form for @xmath44 means that approximations which incorporate finite frequency fluctuations , i.e. , which generalize @xmath96 , may in principle be developed to extend the approach worked out in detail here ( see section 6 ) .
we derive below the form of several one and two body electron correlation functions @xmath97 in the presence of the background fields . as stated previously , n^i _ & = & _
g_(ii;=0 ^ - ) _ + & = & _
rf(_r)|r|i|^2 , [ density ] where g_(ij ; ) _
= -c^i()^j(0 ) _ [ gg ] in terms of anticommuting c numbers ; and the @xmath37 are eigenstates of the spin
dependent hamiltonian @xmath54 in eqn .
[ hamtwo ] .
likewise , the spin polarization at site @xmath12 is ^i _ & = & _ ^i _ c^i _ + & = & _
_g_(ii;=0 ^ - ) _ + & = & _ rf(_r)_r|i i|r , [ polarization1 ] in terms of the @xmath98basis states @xmath99 . writing out the vector components explicitly gives m^i_x _ & = & _
rf(_r ) + m^i_y _ & = & _
rf(_r ) + m^i_z _ & = & _
rf(_r ) . [ polarization2 ]
note that these last results generalize to give the one body propagator g_i , j(i_n ) _ & = & _ 0^de^i_n + & = & -c^i_n^j_n _ + & = & _ r i|rr|ji_n-_r .
[ propagator ] the real axis spectral density may be calculated directly by analytic continuation before the field averaging is performed .
thus the one particle spectral density for spin @xmath62 at site @xmath12 after field averaging is n_i ( ) & = & -1img_(ii;+i0^+ ) _ + & = & _
r|i|r|^2(-_r ) . [ rhoa ] the total one particle spectral density for spin
@xmath62 is n _
( ) = _ in_i ( ) .
[ rhotot ] ( note that in a properly equilibrated calculation translational and spin rotational invariance are restored by field averaging , and
the quantity @xmath100 is site and spin independent . )
the momentum resolved spectral density @xmath101 is only slightly more difficult to calculate .
note again that translational invariance is only restored by field
averaging , so that the propagator in a single field configuration is not diagonal in momentum . after averaging we have a_k ( ) = _ r(-_r)_ije^-ik(r_i - r_j)i|rr|j all the spectral densities may be calculated conveniently by histogram binning , i.e. , by incrementing a bin centered on frequency @xmath102 whenever an eigenvalue @xmath35 falls in the range @xmath103 . in this way
spectral densities may be obtained which integrate exactly to unity within machine precision .
in some cases it may be desirable to compare imaginary time propagators @xmath104 with the results from a full quantum monte carlo simulation .
the necessary expressions within the present approach are g_k ( ) = [ gtau ] where f^r_k = 1_ije^-ik(r_i - r_j)i|r r|j .
[ ffactor ] a quantity of particular interest in the local moment regime is the average double occupancy d = 1_id^i _ , [ double1 ] where d^i _
= n^in^i _ .
[ double2 ] in a general @xmath25 configuration with broken spin rotational invariance the appropriate expression for @xmath105 is not @xmath106 . instead d^i _ & = & _ ^i ( ) c^i()^i()c^i( ) _ + & = & n^i_n^i
_ + c^i^i_^ic^i _ .
[ double3 ] note ^ic^i _ & = & + c^i^i _ & = & - , [ double4 ] while n^i _ & = & + n^i _ & = & . [ double5 ] thus d^i _ = .
[ double6 ] the double occupancy assumes an explicitly spin rotation
invariant form as required .
the expression in eqn .
[ double6 ] should in principle be supplemented by zero and finite frequency rpa corrections to insure that eqn .
[ double3 ] reproduces the @xmath107 limit of the general two body correlation function .
( as noted below in the discussion of response functions , rpa corrections appear at zero frequency because the charge interaction is being treated within hartree fock .
there is no zero frequency rpa in the spin channel , since that portion of the interaction is being treated exactly . )
rpa corrections in equal
time expectation values @xmath108 are actually problematic in any perturbation theory for the following reason : the pauli principle implies an exact identity n^i()n^i ( ) = n^i .
[ double7 ] in perturbation theory , this identity can only be preserved exactly in approximations with a crossing symmetric two body vertex .
such a vertex appears in order by order perturbation theory ( and in parquet theory ) , but not in a standard rpa , whether about the unperturbed vacuum or about broken symmetry states as in the present case .
the rpa can be augmented to make it crossing
symmetric by incorporating the missing contributions in the crossed channel ( see figure 3 ) , but the result is no longer conserving in the baym kadanoff sense .
the rpa corrections to the double occupancy are in any event expected to be small in the present case . since they are also problematic , we omit them from the calculation of @xmath109 reported here .
we next consider the static spin and charge susceptibilities @xmath110 and @xmath111 . as mentioned above
, one expects the appearance of rpa corrections in the static charge response , since the charge interaction is being treated by `` slave '' hartree
fock , but naively one might expect no corrections in the spin response .
this is not correct : an average must be performed over the symmetry breaking field @xmath25 , and in the presence of @xmath25 spin and charge excitations are mixed . for this reason the repulsive charge interaction makes its presence
felt weakly in the spin channel , as well as in the charge channel .
the static spin response function takes the form ^ij _
m^i _ _ /h^j_|_h=0 , [ chi1 ] where the assumed local coupling to the external field @xmath112 is h = -h^j_c^_j_ c^_j .
[ chi2 ] temporarily neglecting rpa corrections , the field derivative yields ^ij _ = -1tm^i _ _ m^j _ _ + 1 t
_ n^a_n^_c^i_n _ n^i_n^_c^j_n _ . [ chi3 ] the first term on the right hand
side vanishes for zero applied field ( since the system has no spontaneous magnetization at finite temperature ) , and the second term ( see figure 4 ) reduces to ^ij _ = 1tm^i _ _ m^j _ _ + -t_n_ g_(ji;i_n)_g_(ij;i_n ) _
^_^_ , [ chi4 ] i.e. , formally an `` anomalous '' and a `` normal '' contribution .
the anomalous contribution may be evaluated using previous expressions for the site spin polarizations , while the normal term may be written as the configurational average of _
m^ij_,(i=0 ) _ ^_^_ , [ chi5 ] where m^ij__1_2,_1_2(i ) _ -_rs j_1|rr|i_1i_2|s s|j_2 f(_r)-f(_s ) _ r + i-_s .
[ chi6 ] ( note that for @xmath113 , the ratio involving the fermi functions must be replaced by @xmath114 . ) the static charge response function is ^ij
n^i _ /^j|_=0 , [ chi7 ] where in this case the assumed coupling to the external field is h = -^j_c^_jc^_j .
[ chi8 ] in complete analogy with the spin response calculation the result before the inclusion of rpa corrections is ^ij & = & 1 t + & & + _ m^ij_ , (i=0)___ , [ chi9 ] with the normal `` bubble '' @xmath115 defined as in eqn . [ chi6 ] .
as emphasized above , rpa corrections necessarily appear when an interaction is treated perturbatively by hartree fock . in the present case , where spin and charge excitations mix before the configurational average is performed , rpa charge vertex corrections appear in both the static charge and spin response functions . the calculation of rpa corrections for _ finite frequency _ response functions is trickier than in conventional perturbation theory because the zero frequency spin vertices have been explicitly removed from the theory .
this affects response functions at @xmath116 by partially removing vertices which carry zero frequency in a crossed particle hole channel . the analytic continuation to real energies
is also complicated by this feature of the theory .
we defer discussion of spin and charge response at finite frequency and concentrate in the present paper on rpa corrections to the static susceptibilities .
the repulsive rpa charge vertex produces a geometric series of corrections as in conventional perturbation theory . due to the lack of symmetry in individual configurations ,
the series can only be summed in matrix form . writing v^ij__1_2,_1_2
= u_ij__1_2 _ _
2_1 , [ chi10 ] the complete rpa corrected expressions for the spin and charge susceptibilities are given by ^ij _ & = & 1tm^i__m^j _ _
+ _ ^ij_,(i=0)_^ _ ^_ + & & [ chi11 ] and ^ij & = & 1 t + & & + _ ^ij_,(i=0)__ _ .
[ chi12 ] finally the momentum
resolved spin susceptibility takes the form _ ( q )
= 1_ije^iq(r_i - r_j)^ij _ , [ chi13 ] and similarly for the charge susceptibility .
( measurements of the rpa corrected susceptibilities require the inversion of small matrices , but the effect on overall calculational efficiency is negligible . )
the effective action in eqn . [ expect4 ] has been simulated using the metropolis algorithm . in comparison with a classical heisenberg spin model ,
the present theory has several complications : ( i ) the theory has three degrees of freedom ( orientation on the unit sphere and amplitude ) at each site , rather than two .
( ii ) local field updates induce global changes in the action through the fermi partition function @xmath96 .
( iii ) a hartree
fock convergence step is necessary after each spin field update to keep the `` slave '' charge fields @xmath30 on the saddle point .
( iv ) single
site updating ( i.e. , glauber dynamics ) is insufficient to achieve efficient equilibration for @xmath117 .
an analog of kawasaki dynamics , or spin swapping , is required , since the doping @xmath118 plays a role similar to a conserved order parameter .
we consider each complication in turn .
( i ) the natural physical interpretation of field fluctuations in terms of amplitude and orientation suggests an optimal representation for the functional integral over field configurations . at half filling it is expected that the only important fluctuations for @xmath119 correspond to magnons , i.e. , slow modulations of the field orientation for fixed amplitude .
a cartesian representation of the fields @xmath120 with independent component updates leads to a small acceptance ratio in the low temperature limit .
therefore we write d^i_xd^i_yd^i_z = _ 0^(r^i)^2dr^i_-1 ^ 1 d(^i)_0 ^ 2d^i , [ measure1 ] choosing as independent variables @xmath121 , @xmath122 , and @xmath123 .
variations in the measure @xmath124 at each site are incorporated into the metropolis updating factor , and moves are restricted , in a way consistent with detailed balance , to lead to @xmath125 . proposed changes to @xmath121 are restricted to a maximum scale @xmath126 , while large scale changes are permitted in @xmath127 and @xmath123 .
\(ii ) for the initial monte carlo calculations reported here we have simply diagonalized the effective hamiltonian matrix @xmath54 , which has row dimension @xmath42 , for each proposed @xmath34 configuration . the time for a single site update thus grows as the cube of the system size , and the time for a lattice sweep as the fourth power of the system size .
we believe that a hybrid molecular dynamics / monte carlo algorithm can be developed to significantly reduce the time needed for the simulation of large lattices @xcite .
\(iii ) the hartree fock treatment of the static charge interaction is an integral component of the present approach .
more sophisticated treatments of fluctuations would also in principle involve a self consistent calculation in the presence of symmetry breaking background fields . for the parameter sets studied in the present paper , the hartree
fock iteration loop generally converges in three to four steps . since each step requires a matrix rediagonalization , overall calculational time scales with the average number of hartree
fock iterations .
\(iv ) at half filling the hartree fock charge
field convergence becomes trivial , and @xmath36 remains equal to unity at all sites in all configurations .
this just reflects the particle hole symmetry of the model for this special parameter set .
away from half filling the situation is quite different .
typical configurations contain a few sites with small fields , while the majority of sites have fields with amplitudes close to the half filled value .
conventional single
spin updating tends to leave `` holes '' trapped at isolated locations , since moving a hole requires passing through high energy intermediate configurations . in this sense
the number of hole @xmath118 functions like a conserved order parameter , preventing the equilibration of the system by glauber dynamics . to combat this effect
we have introduced a second type of updating move , viz .
, a swap of the spin fields @xmath34 at two sites chosen at random .
when the spin fields are swapped , the `` slave '' charge fields are dragged along , and the system equilibrates to a uniform set of site occupancies @xmath128 , even though the hole distribution in individual configurations is highly non uniform .
the results for @xmath8 periodic lattices reported below were obtained on an ibm powerpc work station and a sun3000 server .
for all runs reported here , a total of 100 monte carlo warmup sweeps and 5000 sampling sweeps were carried out . runs on @xmath129 and @xmath130 lattices are now in progress on a sun10000 work station array at the san diego supercomputer center .
the total density of states for spin @xmath62 , @xmath131 , is plotted in figure 5 for @xmath132 and in figure 6 for @xmath133 , both at temperature @xmath134 .
the chemical potential @xmath135 is lowered from @xmath136 ( half filling ) as holes are doped into the insulating state .
the mean site occupancy @xmath137 and double occupancy @xmath109 are summarized in table 1 .
note that the double occupancy at half filling is reduced by factors of order two and six from the value 0.25 for a non interacting system . a prominent mott
hubbard gap , or pseudogap , appears in the density of states for both parameter sets at half filling ( figures 5a and 6a ) .
the remnants of the discrete quantum energy levels in the non interacting system remain visible .
note again that these systems are not magnetically ordered : the spin up and spin down densities of states are identical within statistical sampling errors , and the spin susceptibility is isotropic , as discussed below . as the chemical potential is pulled below @xmath138 , the site occupancy changes slowly at first , particularly for the large charge gap which appears when @xmath133 ( figures 6b c ) .
this is just what one would expect for rigid band doping in a semiconductor , and such behavior has been observed in previous lanczos @xcite and quantum monte carlo @xcite studies of the hubbard model on small lattices @xcite .
note that the rigid band picture begins to break down as soon as the site occupancy drops appreciably below unity ( figure 5b ) . in the doped system with @xmath132 and @xmath139 ,
spectral weight has been drawn from the lower edge of the upper hubbard band into the gap . due to this transfer of weight ,
the gap feature rapidly collapses with doping , though remnants remain for a wide range of fillings ( figures 5c d ) . by the time @xmath135 reaches 0.4 for
@xmath132 , the shape of the non interacting band has begun to re emerge , though levels are still significantly broadened by the fluctuating spin background .
this broadening is especially evident in the evolution of the k
resolved spectral density for points near the fermi surface . recall that a special symmetry of the @xmath8 lattice induces an equivalence between the generally distinct k
points @xmath140 and @xmath141 .
this means all points on the half filled fermi surface have the same @xmath101 for the small systems discussed here . to contrast the behavior of points near and far from the fermi surface , we plot @xmath101 for @xmath142 and @xmath143 at @xmath132 in figures 7 and 8 .
while the @xmath144 point at the top of the band ( figure 8) has a sharp quasiparticle like spectral density whose profile changes only slightly with doping , the @xmath140 point behaves quite differently . at half filling @xmath145 for this point
is symmetrically split between the upper and lower hubbard band edges , just as one would expect in an antiferromagnetic hartree
fock solution @xcite .
away from half filling , the spectral density remains exceptionally broad , reflecting the continued presence of strong short range antiferromagnetic correlations between the fluctuating spins in the doped system .
as noted above in the discussion of @xmath146 , remnants of the half filled gap remain for a wide range of fillings .
the spectral broadening decreases for fillings far from unity , consistent with the restoration of the non interacting band profile ( and presumably the recovery of fermi liquid behavior ) .
a broadening of the spectral density is also present in perturbative studies , including self consistent field methods such as flex @xcite ; however , the details of the broadening mechanism are quite different in the present case . flex and phenomenological spin fluctuation models @xcite assume a small amplitude scattering mechanism which may be treated by perturbation theory about a non interacting ( i.e. , fermi liquid ) background .
in contrast the present approach treats scattering by large
amplitude background distortions , with highly non gaussian static fluctuations in both amplitude and orientation ; these distortions evolve smoothly into the ordered antiferromagnetic background at half filling and zero temperature .
the behavior of the spectral density at the @xmath140 point for @xmath133 is illustrated in figure 9 .
as expected , the spectral weight is split symmetrically at half filling , and the two peaks are separated by an energy of order @xmath1 .
in contrast with the @xmath132 result , the weight is distributed broadly throughout the upper and lower hubbard bands .
signs of a more narrow coherent feature in the doped lower band are present for @xmath147 ( figure 9c ) . the momentum dependence of the static charge and spin susceptibilities is plotted in figures 10 and 11 for @xmath132 and 8 at temperature @xmath134 .
note that the uniform charge susceptibility ( or compressibility ) is exponentially small at half filling , as expected for a system with a charge gap .
in addition the compressibility increases very rapidly with doping : note particularly the increase for @xmath133 ( figure 11a ) . for general fillings
@xmath111 is relatively small ( note the difference in scales for the charge and spin susceptibility ) and structureless at this temperature .
this is expected in such a small system .
possible manifestations of charge ordering within the fluctuating spin background may become apparent , however , in larger systems .
in contrast the spin susceptibility is large and strongly peaked at the antiferromagnetic wave vector @xmath144 .
the peak is expected to move off @xmath144 for doped systems on larger lattices . for the parameter sets illustrated here ,
the spin susceptibility is dominated by the contributions of the `` anomalous '' term _
ije^iq(r_i - r_j ) m^i _ _
[ anom ] the @xmath148 component of @xmath149 is actually plotted in figures 10 and 11 . for 5000
lattice sweeps the agreement between the @xmath150 , @xmath151 , and @xmath152 components of the @xmath144 spin susceptibility is at the level of the statistical error bars , i.e. , a few percent .
( the agreement can be arbitrarily improved by increasing the monte carlo sampling time . )
the present approach consists of simulation of the action for a set of local static fields coupled to electron spin degrees of freedom .
when the action is constrained to be at its temperature dependent global saddle point ( if such a point indeed exists in the most general case ) , these static fields are determined by inhomogeneous hartree
fock equations and reduce to the anomalous fields of conventional diagrammatic perturbation theory .
just as hartree fock theory with an anomalous self energy may be extended using the baym
kadanoff prescription @xcite , there is no formal barrier preventing such an extension in the present case .
the computational cost of carrying out self consistent field ( scf ) perturbation theory in a broken symmetry background is quite large , however , since total momentum and spin are no longer good quantum numbers . in this section
we sketch out the steps necessary for a generalized approach and note several of the technical pitfalls which must be studied and overcome .
the form for the scf action must be determined separately for each configuration of the @xmath25 fields . in full analogy with baym
kadanoff theory for a uniform system , the action takes the form s^scf ( ) = _ i|^i|^2 - tr(-g^scf _ ) ^-1 + , [ actionfourteen ] where ( ) = -tr(g^scf ) + [ freeenergy ] and = _ (xx ) .
[ phi ] in these formal expressions @xmath153 is the discrete time propagator ( retained to insure a proper high frequency regularization of the @xmath154 term ) ; x ( r^i , ) , [ xx ] and @xmath155 is the scf generating functional ( e.g. , a sum of ring and ladder diagrams in flex ) .
since the scf action may be interpreted as @xmath156 , with @xmath157 a well behaved free energy , the statistical weighting factor @xmath158 is positive definite .
this is in sharp contrast to the behavior in imaginary
time quantum monte carlo simulations , where the weighting factor can be positive or negative with nearly equal probability .
what is not so clear is the nature of approximations which insure the stability of @xmath159 , particularly for exponentially rare field configurations .
the interaction which enters @xmath160 is not the original hubbard interaction , since the zero frequency spin components have been removed for special treatment .
this means that there are no obvious zero frequency instabilities left to appear in @xmath159 .
behavior at finite frequencies must also be addressed , however , and we are not prepared to comment on the stability of general baym kadanoff approximations ( and in particular flex ) in the present paper . as an example of the baym
kadanoff prescription in practice , note that when the static charge interaction is treated within hartree
(xx ) = un^i__ii_ _ .
[ sighf ] the appropriate generating functional is = u_in^i_^2 .
[ phii ] furthermore ( g^scf ) = 2[trace ] and so & = & - + & = & -u_in^i_^2 . [ calf ] note that this form for the hartree
fock `` constant '' in the action is exactly that which has appeared previously in eqn .
[ actiontwelve ] .
we believe the present hubbard model analysis , based on monte carlo simulation of a classically disordered local spin field , incorporates several desirable features which have eluded previous perturbative studies .
most important the present theory has classical critical behavior at half filling and describes the formation of a mott hubbard gap ( or pseudogap ) in a system with a finite spin correlation length .
in addition the spectral densities for the doped systems have the features suggested by exact methods .
the present static field analysis can be extended to significantly larger systems .
since the computations involve monte carlo field sampling and have reasonably small memory requirements , it would be natural to carry out studies on multiple processors with different random seeds , then to merge the equilibrated results from numerous short runs .
it is almost unavoidable to speculate on the implications for charge density wave or stripe formation @xcite and d wave singlet pairing in the cuprate superconductors . since the present approach reduces to standard inhomogeneous hartree
fock for @xmath4 , hartree fock results previously obtained for stripe formation should re
emerge @xcite . by allowing sampling of configurations removed from the zero temperature saddle point in a well defined way
, the present approach provides a framework to describe the destruction of striped order by thermal fluctuations . to study
potential d wave superconductivity it seems essential to incorporate the exchange of low frequency spin fluctuations about the disordered background configurations .
the low energy fluctuations which become magnons in the directions transverse to an ordered spin state ( i.e. , the goldstone modes of the order parameter field ) must evolve into a broadened spectrum of fluctuations in the important disordered configurations . if low
frequency spin fluctuations are important , the pair order parameter should exhibit retardation effects , in contrast with the spin order parameter .
in addition the disordered background has important implications for the onset of pairing .
presumably the pair field appears initially with a highly non uniform distribution in a small subset of @xmath34 configurations .
such behavior would imply superconducting precursor effects in the density of states well above the transition temperature for global phase coherence .
the effects would be more akin to the behavior of a system of superconducting grains with variable grain size than to conventional kosterlitz
thouless phase fluctuations in a clean two dimensional superconductor . in any case
the scenario described above is at this stage purely speculative .
the next steps in the development of the present approach consist of ( i ) studies of larger systems ; ( ii ) analysis of the spin and charge response at finite frequencies , with particular care given to the treatment of potential instabilities in rare field configurations ( see section 6 ) ; and ( iii ) analysis of pairing eigenvalues and eigenvectors from exchange of fluctuations in sets of non uniform configurations .
we gratefully acknowledge conversations with s. haas , w. hanke , s.a .
kivelson , h. monien , s. moukouri , r.t .
scalettar , r.l .
sugar and s.c .
zhang during the course of this work .
neb acknowledges support from the national science foundation under grant no .
dmr9811381 , and djs acknowledges support from the u.s .
department of energy under grant no .
de fg0385er45197 .
the behavior of high order perturbation theory for the anderson impurity model seems intimately connected with the re emergence of a fermi liquid description for @xmath174 .
see , e.g. , a.c .
hewson , phys .
@xmath175 , 4007 ( 1993 ) .
we suggest that perturbation theory in @xmath1 fails to resum properly for the anderson impurity model at temperatures @xmath176 , where a stiff local moment exists .
the @xmath24 correction is present because equivalent components in the order parameter field appear in crossed scattering channels .
in contrast no such correction appears when one decouples a general instantaneous pairing interaction or a holstein phonon interaction .
while the @xmath24 term must be retained to properly regain the high temperature limit , it seems clear that it has no special physical implications at low temperatures .
further the correction may at least formally be treated exactly by introducing a `` strong coupling '' ( but still @xmath23 ) generalization of the @xmath25 fields .
it is possible to provide a heuristic derivation of a one electron self
energy with the general form in eqn .
[ perturb ] starting from the full theory for @xmath44 in section 2 .
the steps are as follows : ( i ) write down the expression for the one electron propagator dictated by eqn .
[ expect ] .
( ii ) expand the expression for @xmath184 within the @xmath25 integral as a power series in @xmath185 , noting that the odd order terms integrate to zero .
( iii ) resum the series within the integral as @xmath186^{-1}$ ] , where @xmath187 takes the form of a self energy second order in @xmath25 .
( iv ) approximate the resulting field integral as @xmath188^{-1}$ ] , i.e. , average the second order self energy , rather than the propagator . the field averaged self
energy assumes exactly the form of eqn .
[ perturb ] , with @xmath84 proportional to the full static spin susceptibility of the system .
the electron propagator which appears in the self
energy is bare for the reason discussed in the text .
step ( iv ) is of course not justifiable , since the action is strongly non
gaussian , but it provides the phenomenological link between some previous theories and the current work .
|
we describe a new microscopic approach for analyzing interacting electron systems with local moments or , in principle , any local order parameter .
we specialize attention to the doped mott insulator phase of the hubbard model , where standard weak coupling perturbation methods fail . a rotationally invariant stratonovich
hubbard field is introduced to decouple the static spin components of the interaction at each site .
charge degrees of freedom are then treated by `` slave '' hartree fock in the presence of a spatially varying random spin field .
the effective action reduces to a classical heisenberg model at half filling , insuring that the system has ( i ) finite range order at @xmath0 with an exponentially diverging correlation length and ( ii ) a one electron mott
hubbard gap in the presence of disorder .
away from half filling properties are determined by strongly non
gaussian fluctuations in the amplitude and orientation of the local spin fields .
the saddle point equations of the theory at zero temperature reduce to inhomogeneous hartree
fock , so that disordering of domain walls at finite temperature is in principle included .
we present preliminary small system results for the intermediate and large interaction regimes obtained by monte carlo simulation of random spin field configurations .
# 1 0= -.025em0 - 0 .05em0 - 0 -.025em.0433em0
|
let @xmath0 be an @xmath7-dimensional compact complex khler manifold with an ample line bundle @xmath8 over @xmath0 .
let @xmath9 be the khler metric on @xmath0 corresponding to the khler form @xmath10 for some positive hermitian @xmath11 metric on @xmath8 .
such a khler metric @xmath9 is called a polarized khler metric .
the metric @xmath11 induces a hermitian metric @xmath12 on @xmath13 for all positive integers @xmath6 .
let @xmath4 be an orthonormal basis of the space @xmath14 with respect to the inner product @xmath15 where @xmath16 and @xmath17 is the volume form of @xmath9 .
the quantity @xmath18 is related to the existence of khler - einstein metrics and stability of complex manifolds .
a lot of work has been done for on compact complex khler manifolds .
tian @xcite applied hmander s @xmath19-estimate to produce peak sections and proved the @xmath20 convergence of the bergman metrics . later , ruan @xcite proved the @xmath21 convergence . about the same time , zelditch @xcite and catlin @xcite separately generalized the theorem of tian by showing there is an asymptotic expansion @xmath22 for certain smooth coefficients @xmath23 with @xmath24 . in @xcite , lu proved that each coefficient @xmath23 is a polynomial of the curvature and its covariant derivatives .
in particular , @xmath25 , where @xmath1 is the scalar curvature of @xmath0 . these polynomials can be found by finitely many steps of algebraic operations . recently , song @xcite generalized zelditch s theorem on orbifolds of finite isolated singularities .
the information on the singularities can be found in the expansion . on the riemann surfaces with bounded curvature ,
lu @xcite proved that there is a lower bound for .
later , the result of lu and tian @xcite implies that on the riemann surfaces with constant scalar curvature @xmath1 , the asymptotic expansion is given by @xmath26 for any @xmath27 . in the current paper
, we obtain a more precise result for .
[ liu ] let @xmath0 be a regular compact riemann surface and @xmath28 be the canonical line bundle endowed with a hermitian metric @xmath11 such that the curvature @xmath29 of @xmath11 defines a khler metric @xmath9 on @xmath0 .
suppose that this metric @xmath9 has constant scalar curvature @xmath1 .
then there is a complete asymptotic expansion : @xmath30 where @xmath4 is an orthonormal basis for @xmath5 for some @xmath31 , where @xmath32 is the injective radius at @xmath33 .
our result indicates that the asymptotic expansion is in exponential decay . + englis @xcite has an asymptotically expansion for the berezin transformation on any planar domain of hyperbolic type .
he also showed that berezin kernel @xcite has @xmath34 where @xmath35 is a positive constant .
[ [ acknowledgements ] ] * acknowledgements * + + + + + + + + + + + + + + + + + +
the author thanks her advisor z. lu for his suggestions and help during the preparation of this paper .
she also thanks r. chiang for her encouragement .
let @xmath0 be an @xmath7-dimensional compact complex khler manifold with a polarized line bundle @xmath36 .
choose the @xmath37-coordinates @xmath38 on an open neighborhood @xmath39 of a fixed point @xmath40 .
then the khler form @xmath41 satisfies @xmath42 for @xmath43 and any nonnegative integers @xmath44 with @xmath45 .
we also choose a local holomorphic frame @xmath46 of the line bundle @xmath8 at @xmath47 such that @xmath48 is the local representation function of the hermitian metric @xmath11 .
that is , @xmath49 under the @xmath37-coordinate , the function @xmath48 has the properties @xmath50 for any nonnegative integers @xmath44 with @xmath45 .
let @xmath4 be a basis of @xmath14 .
assume that at the point @xmath40 , @xmath51 if the set @xmath4 is not an orthonormal basis , we may do the following : + let the metric matrix @xmath52 with respect to the inner product . by definition , @xmath53 is a positive definite hermitian matrix .
we can find a @xmath54 matrix @xmath55 such that @xmath56 let @xmath57 be the inverse of @xmath58 .
then @xmath59 forms an orthonormal basis of @xmath14 .
the left hand side of is equal to @xmath60 let @xmath61 be the inverse matrix of @xmath53 .
denote that @xmath62 in order to compute , we need a suitable choice of the basis @xmath63 .
we select some of tian s peak sections in our basis .
the following lemma an improved version of tian s result ( * ? ? ? * lemma 1.2 ) , which is done by lu and tian .
let @xmath64 be the set of @xmath7-tuple integers @xmath65 such that each @xmath66 is a nonnegative integer for @xmath67 . for @xmath68
, we denote that @xmath69 and @xmath70 .
[ section ] [ tian ] suppose @xmath71 , where @xmath72 is a constant .
for @xmath68 and an integer @xmath73 , let @xmath6 be an integer such that @xmath74 then there is a holomorphic section @xmath75 , satisfying @xmath76 moreover , @xmath77 can be decomposed as @xmath78 such that @xmath79 and @xmath80 where @xmath81 is a smoothly cut - off function @xmath82 satisfying @xmath83 and @xmath84 and @xmath85 define the weight function @xmath86 a straightforward computation gives @xmath87 by using , we can verify that @xmath88 for @xmath89 , consider the @xmath90-form @xmath91 since @xmath92 in a neighborhood of @xmath47 , we have @xmath93 by ( * ? ? ?
2.1 ) , there exists a smooth @xmath13-valued section @xmath94 such that @xmath95 and @xmath96 by direct computation , we get @xmath97 under the @xmath37-coordinate , we have @xmath98 hence we get @xmath99 for some constant @xmath100 .
let @xmath101 and @xmath102 .
use a result in @xcite @xmath103 for some constant @xmath104
. then we have @xmath105 choosing @xmath106 , we obtain @xmath107
let @xmath0 be a smooth compact riemann surface with a metric @xmath9 that has constant scalar curvature .
let @xmath33 be a fixed point .
let @xmath108 where @xmath32 is the injective radius at @xmath33 .
it is well known that on a riemann surface there is an isothermal coordinate at each point on @xmath39 .
we may assume that there is a holomorphic function @xmath109 on @xmath39 and it defines the conformal structure on @xmath39 .
that is , @xmath110 and @xmath111 .
the metric @xmath9 satisfies @xmath112 where @xmath113 is the complex laplace of @xmath0 .
since the metric @xmath9 has conformal structure , it is rotationally symmetric .
we can write in polar coordinates @xmath114 : @xmath115 where @xmath116 , and @xmath117 .
there exists a solution @xmath118 to for @xmath119 if @xmath120 .
suppose that there exists another solution @xmath121 to .
we have @xmath122 for @xmath120 , let @xmath123 .
since @xmath9 and @xmath124 are rotationally symmetric , they remain constant on @xmath125 .
the harmonic function @xmath126 is a constant on @xmath127 by maximum principle . by definition
, we have @xmath128 .
therefore , the solution in is unique around @xmath47 . by the same reason , @xmath129 on @xmath130 for some @xmath131 for @xmath132 .
let @xmath48 be the local representation of the metric @xmath11 on @xmath28 such that @xmath133 if we normalize @xmath48 and @xmath48 satisfies @xmath134 since @xmath135 @xmath136 is also rotationally symmetric . since @xmath137 satisfies ,
the local uniqueness is due to the same reason .
we need to choose sufficient large @xmath6 such that @xmath138 . with these particular solutions of @xmath9 and @xmath48 ,
we further compute @xmath139 for @xmath140 , we have @xmath141 . for @xmath142 ,
this gives @xmath143 for @xmath144 , we have @xmath145 hence we obtain @xmath146 from the properties of @xmath9 and @xmath48 , the isothermal coordinate @xmath147 is a @xmath37-coordinate . according to lemma [ tian ] , we may choose two peak sections @xmath148 in @xmath149 for some @xmath150 .
obviously , we have @xmath151 and @xmath152 .
let the subspace @xmath153 where @xmath154 is a covariant derivative on @xmath2 .
let @xmath155 be an orthonormal basis of @xmath156 .
let @xmath157 then @xmath158 forms a basis for @xmath149 .
locally , each @xmath159 has the form @xmath160 for some holomorphic function @xmath161 defined in @xmath39 .
the holomorphic function @xmath161 has taylor expansion as @xmath162 in @xmath39 , since @xmath163 and @xmath164 for @xmath165 by definition , we have @xmath171 for @xmath172 .
the inner product of @xmath173 for @xmath174 is directly from lemma [ tian ] .
since @xmath175 is rotationally symmetric , we have @xmath176 then we get @xmath177 consider @xmath178 thus we have @xmath179 similarly , consider @xmath180since @xmath175 is rotationally symmetric , @xmath181 for @xmath182 .
hence we obtain @xmath183 according to ( * ? ? ?
* definition 3.1 ) , the metric matrix @xmath53 can be represented by the block matrices @xmath184 where @xmath185 , @xmath186 , @xmath187 , @xmath188 , and @xmath189 is a @xmath190 identity matrix . by using ( * ? ? ?
* lemma 3.1 ) , we obtain @xmath191 where @xmath192 applying lemma [ matrix ] in , we get @xmath193 in order to evaluate the expansion of , we are left to find @xmath194 .
from , we have @xmath195 therefore , the tian - yau - zelditch expansion according to on a riemann surface with constant scalar curvature @xmath1 is @xmath196 for @xmath31 .
|
let @xmath0 be a regular riemann surface with a metric which has constant scalar curvature @xmath1 .
we give the asymptotic expansion of the sum of the square norm of the sections of the pluricanonical bundles @xmath2 .
that is , @xmath3 where @xmath4 is an orthonormal basis for @xmath5 for sufficiently large @xmath6 .
[ section ]
|
dilute iii - n - v alloys with the group - v cations partially substituted by nitrogen , e.g. , gaasn and ingaasn , have recently attracted particular attention owing to a number of their unusual properties .
thus , an increase of nitrogen content in the alloy is accompanied by an anomalous reduction of the fundamental band gap , by more than 0.1ev per 1% of n as the nitrogen concentration increases from zero up to 4% @xcite , as well as by a drastic increase of the electron effective mass and enhanced nonparabolicity of the @xmath7 conduction band @xcite ( for reviews , see also @xcite ) .
these specific properties are related to the n - substitution - induced resonant electronic states within the conduction band continuum and an anticrossing of the localized nitrogen states with the extended conduction states of the semiconductor matrix @xcite .
qualitatively , the available experimental data are consistent with the simple phenomenological band anticrossing ( bac ) model which takes into account the anticrossing hybridization only with the levels of isolated ( single ) nitrogen atoms @xcite . as a result ,
the two hybridized conduction bands , @xmath8 and @xmath9 , are formed .
as the nitrogen concentration increases , the lower band , @xmath9 , shifts down in energy reducing the fundamental energy - band gap @xmath10 .
simultaneously , the band - bottom effective mass becomes heavier and the nonparabolicity coefficient rises considerably .
the band - structure modification involves a remarkable change in the gyromagnetic factor , @xmath0 , of electrons in the conduction band @xmath9 @xcite .
the change is caused by two reasons .
firstly , as in the case of conventional alloys in@xmath6ga@xmath5as , the band - gap shrinkage leads to a monotonous downward shift of the effective @xmath0-factor below the value of _ g_-factor in gaas matrix ( @xmath11 in gaas at helium temperature ) @xcite .
secondly , the coupling with isolated nitrogen levels , with the land factor being positive and close to the electron @xmath0-factor in vacuum @xmath12 , tends to shift the _
g _ value upwards , as it has been first found experimentally in ingaasn when studying the dependence of the absorption coefficient on the magnetic field @xcite .
the two effects have opposite signs and partially compensate each other .
the bac model predicts an initial growth in the dependence @xmath13 @xcite . the dependence is smooth and ,
as calculated for gaas@xmath5n@xmath6 in @xcite , it has a maximum value @xmath14 at @xmath15 . at the same time , detailed measurement and comparison of the left- and right - circularly polarized photoluminescence spectra performed in gaas@xmath5n@xmath6 in a longitudinal magnetic field at helium temperature for @xmath16 have shown that _
g _ grows sharply with increasing @xmath17 , reverses its sign from negative to positive at @xmath18 , reaches a value of 0.7 in the range @xmath19 and retains that value with small deviations as @xmath17 increases up to @xmath20 @xcite .
such an abrupt dependence of @xmath0 on @xmath17 at small @xmath17 has been interpreted in terms of a modified @xmath21 model @xcite .
this model takes into account the hybridization of the host conduction band not only with the localized levels due to single n atoms but also with the levels which are formed by clusters of n atoms , pairs and triplets , and lie closer to the conduction - band bottom .
the data on the @xmath0-factor in gaas@xmath5n@xmath6 with high nitrogen concentration are highly desirable for the purpose of checking various theoretical models . however , to the best of our knowledge
, such experimental data for @xmath22 are lacking . in the present work
the sign of the electron @xmath0-factor in the gaas@xmath5n@xmath6 conduction band has been measured at room temperature for @xmath23 . to this end
, we used the dependence of the sign of the asymmetry of hanle effect of optically spin - polarized electrons , measured under oblique excitation , on the @xmath0-factor sign .
this method is based on the relationship between the direction of larmor precession of electron spin in a magnetic field and the @xmath0-factor sign . as compared to other methods ,
this method is advantageous for studying crystals with high nitrogen content , particularly at high ( up to room ) temperatures when the photoluminescence occurs over a broad spectrum .
the fundamentals of the method are presented in section 3 , together with the discussion of peculiarities of the electron spin polarization and the oblique hanle effect caused by the splitting of the @xmath24 valence band due to a uniaxial strain of the gaasn films grown on gaas substrates .
section 4 provides the theory extended in order to calculate the magnitude and orientation of photoelectron spin polarization , the circular polarization of photoluminescence and the hanle - curve shape as a function of the energy of absorbed and emitted photons in a uniaxially strained semiconductor with the top of the light - hole subband to be higher in energy than the top of the heavy - hole subband . in the gaasn alloys under study
, the optical spin polarization of conduction electrons at room temperature has been found to be anomalously high @xcite .
this is the result of spin - dependent recombination of polarized conduction electrons on deep paramagnetic centers which apparently appear during the introduction of nitrogen atoms into gaas .
the capture of free electrons is accompanied by efficient dynamic spin polarization of electrons bound on the centers . in its turn
, the polarized centers kinetically affect the polarization of free electrons so that a strongly coupled spin system of free and localized electrons is formed .
a change in the polarization of localized electrons manifests itself in a remarkable change of the free - electron polarization .
thus , from the analysis of the curve of magnetic depolarization of the interband photoluminescence contributed by free particles , one can determine the @xmath0-factor signs of both free and bound electrons .
in section 5 we present experimental results together with the results of numerical calculation taking into account the influence of spin - dependent recombination on the electron spin polarization .
we studied the undoped @xmath25m - thick gaas@xmath1n@xmath2 layer grown by rf - plasma - assisted solid - source molecular - beam epitaxy at 350450@xmath26c on semi - insulating ( 001 ) gaas substrate @xcite .
the nitrogen content and crystallinity of the grown layer were examined by x - ray diffraction technique .
the as - grown structure was annealed for 5min at 700@xmath26c in a flow of arsenic in the growth chamber .
continuous - wave tunable ti : sapphire laser was used for photoluminescence ( pl ) excitation .
spin polarization of electrons was created upon the interband absorption of circularly polarized light @xcite .
it was monitored by measuring the degree of circular polarization of pl , defined as @xmath27 , where @xmath28 and @xmath29 are the right ( @xmath30 ) and left ( @xmath31 ) circularly polarized pl components .
the value @xmath32 and pl intensity in a wavelength range up to 1.4@xmath33 m were measured using a high - sensitive polarization analyzer @xcite comprising a quartz polarization modulator @xcite , a lock - in two - channel photon counter , and a photomultiplier with an ingaasp photocathode .
the measurements were carried out at 300k .
spectral dependencies of the pl intensity ( solid curve ) and of the pl circular polarization degree ( circles ) for the gaas@xmath1n@xmath2 layer under normal incidence ( @xmath34 ) of the pump beam and detection of the pl
. excitation energy @xmath35ev .
arrows indicate the pl detection energies @xmath36 and @xmath37 at which the hanle curves presented in fig . 2 are measured . ]
figure[pl^ro(det ) ] shows the spectra of pl intensity ( solid curve ) and of pl circular polarization degree ( circles ) measured in gaas@xmath1n@xmath2 layer under normal incidence of the pump beam and the detection of luminescence along the growth axis .
the pl spectrum consists of two strongly overlapping inhomogeneously broadened bands where the low - energy pl band is negatively polarized ( relative to the polarization of the exciting beam ) , whereas the high - energy band is polarized positively .
as we have shown earlier @xcite , the two pl bands appear due to splitting of the light- and heavy - hole valence subbands induced by uniaxial compression along the growth axes of gaas@xmath5n@xmath6 film grown on gaas substrate . in its turn
, this compression arises from large lattice mismatch between the film and substrate .
the crystal splitting of the pl bands , @xmath38 , increases with increase of the nitrogen content and at @xmath23 equals 29mev .
an interrelation of the sign of the pl circular polarization and the type of recombination transition in gaas@xmath5n@xmath6 is considered in detail in @xcite . under simultaneous excitation of electrons from both valence subbands , realized in fig.[pl^ro(det ) ] , the negative and positive pl polarizations are due to recombination of conduction electrons with the light and heavy holes , respectively . since under an uniaxial compression the top of the light - hole subband is situated above the top of the heavy - hole subband @xcite , the low - energy pl band is polarized negatively . in bulk
semiconductor , the spin relaxation rate of conduction electrons grows dramatically with temperature and decreases electron polarization down to units and fractions of @xmath39 at room temperature @xcite . at the same time , the absolute values both of negative and positive polarization in fig.[pl^ro(det ) ] reach @xmath40 , which is near to the maximal magnitude of @xmath41 determined by selection rules .
such anomalous enhancement of free electron polarization in gaasn is due to dynamic polarization of electrons bound on deep paramagnetic centers @xcite .
the paramagnetic centers arise with an incorporation of nitrogen in gaas and are polarized as a result of spin - dependent capture on them of the polarized conduction electrons .
the polarized centers increase the free electron polarization which can reach @xmath42 at strong pumping .
the hanle effect is a depolarization of photoluminescence by a magnetic field @xmath43 directed perpendicular to continuous - wave pump beam @xcite .
the effect originates from larmor precession of electron spins , which destroys their polarization . in the simplest case ,
hanle effect is described by lorentzian @xmath44 with half - width at half maximum , @xmath45 , where @xmath0 is the electron land @xmath0-factor , @xmath46 is the bohr magneton , and @xmath47 is the electron spin - polarization lifetime .
figure[0teta ] shows the hanle curves measured in gaas@xmath1n@xmath2 for the positive ( a ) and negative ( b ) pl polarization under the normal incidence of pump light onto the sample and the detection of the pl in the opposite direction ( backscattering configuration ) .
the measurements were carried out with the energy of the excitation quantum @xmath35ev , at which the conduction electrons are excited from both valence subbands .
one can see that for both positive and negative pl polarization the hanle curve is superimposition of the narrow and wide curves with the half - widths being different by two orders of magnitude .
a qualitative explanation of such a complex shape of the @xmath48 dependency is the following .
spin lifetime of electrons bound on paramagnetic centers , @xmath49 , can exceed spin lifetime of free electrons , @xmath47 , by orders of magnitude at room temperature .
therefore , the half - width of hanle curve of bound electrons , @xmath50 ( here @xmath51 is the gyromagnetic _
g_-factor of bound electrons ) , should be far less than the half - width of hanle curve of free electrons @xmath52 . as noted above , the spin - dependent recombination results in formation of a strongly coupled spin - system of free and bound electrons , where a variation of the polarization of the centers is accompanied by change in the free - electron polarization .
due to this , the @xmath48 curve is the sum of two curves with strongly different half - widths , where the narrow curve describes the hanle effect of bound electrons and the wide curve presents depolarization of free electrons .
solid curves in fig.[0teta ] are calculated using the equation @xmath53 + \rho_{0c}/[1+(b / b_{1/2}^{c})^2]+\rho_{\rm res}$ ] at @xmath54 g and @xmath55 g . it is seen that the calculated curves describe reasonably the measured ones in fig.[0teta ] for both positive and negative polarization . during the calculation
the fitting parameters were @xmath56 , @xmath57 and @xmath58 . at present , the origin of the constant polarization @xmath59 is not clear and additional studies should be done for its elucidation . experimental hanle curves ( open circles ) measured in gaas@xmath1n@xmath2 at excitation and detection along the normal to the sample .
excitation energy @xmath35ev , detection energy @xmath60 is equal to @xmath61ev ( _ a _ ) and @xmath62ev ( _ b _ ) .
solid curves are superpositions of the calculated hanle curves for free ( dotted curves ) and bound electrons ( see text for details ) .
inserts show the initial parts of the hanle curves in the range of small magnetic fields . ]
the hanle effect can provide measuring the sign of the land @xmath0-factor of electrons @xcite since a rotation direction of mean electron spin in a magnetic field depends on the @xmath0-factor sign . the most suitable way to find
the @xmath0-factor sign is to use a specular geometry of experiment . in this geometry ,
the exciting light falls on the sample obliquely , the luminescence is detected at an angle to the exciting beam while a magnetic field is perpendicular to both the excitation and detection directions , and lies in the crystal surface plane @xcite , as shown in fig.[mirrorconfig ] . in specular geometry
, the dependence @xmath48 in bulk zinc - blende semiconductor has the form @xcite : @xmath63 where @xmath64 is the angle between the directions of excitation and detection , @xmath65 is the degree of polarization for @xmath66 and @xmath67 , @xmath68 is the angle through which the mean spin @xmath69 of electrons with the _ g_-factor equal to _ g _ turns during the electron spin lifetime @xmath70 , @xmath71 and @xmath72 are the electron lifetime and spin relaxation time in the conduction band , @xmath73 is the larmor frequency , and the born magneton @xmath74 .
equation([hanlebulk ] ) was derived taking into account the fact that in a bulk unstrained gaas - type crystal the mean electron spin in the moment of creation , @xmath75 , is parallel to the exciting beam , and the degree of pl polarization is equal to the projection of @xmath69 on the direction of detection : @xmath76 , where @xmath77 is the unit vector along the direction of registration @xcite . at @xmath67 ,
the hanle curve is a symmetrical function on magnetic field as , for example , one can see in fig.[0teta ] . for @xmath78 ,
the @xmath48 plot is not symmetrical with respect to the @xmath66 point . the value of @xmath32 is larger for the magnetic field direction at which the electron spin precesses toward the luminescence observation axis .
note , that for measuring the _ g_-sign it is sufficient to record the asymmetry in the dependence of the absolute value of @xmath32 on @xmath79 . experimental configuration for the @xmath0-factor sign measurements .
the angle @xmath64 is the angle between the excitation beam and the direction of the detected photoluminescence . ]
the above method acquires distinctive features @xcite if the heavy and light hole subbands are split due to the uniaxial deformation in a bulk crystal or due to quantum - size effect in a nanostructure .
first , in this case the direction of the mean spin of photogenerated electrons @xmath80 may not coincide with the direction of the exciting beam .
theory @xcite ( see also @xcite ) developed for uniaxially strained crystals yields the following expressions for the mean spins of the electrons created with @xmath81 from states close to the tops of light - hole ( @xmath82 transition ) and the heavy - hole ( @xmath83 transition ) subbands : @xmath84 where @xmath85 and @xmath86 are unit vectors along the deformation ( growth ) axis and pump beam direction .
as seen from eqs.([k0 ] ) , the spin @xmath87 is parallel to the deformation axis for any light incidence angle , whereas the angle between @xmath88 and @xmath85 is twice the angle @xmath89 between the excitation direction and the growth axis ( fig.[ezhik ] ) @xcite .
the light - hole and heavy - hole splitting can be neglected if the energy of the photogenerated holes exceeds substantially the splitting energy @xmath90 @xcite .
such situation can be realized when @xmath91 , where @xmath10 is the forbidden gap of the deformed crystal . in this case , the electrons are excited from both valence subbands , and , as in a bulk unstrained crystal , their mean spin @xmath92 is directed along the exciting beam : @xmath93 ( fig.[ezhik ] ) .
thus , with the increase of excitation energy from @xmath94 towards @xmath95 , the spin @xmath96 changes from @xmath88 to @xmath97 rotating counterclockwise through angle ( @xmath98 ) .
scheme of orientation of spins @xmath99 , @xmath88 , @xmath97 and vectors @xmath100 , @xmath101 with respect to the directions @xmath102 of excitation , @xmath103 of detection and @xmath85 of deformation ( @xmath85 is a perpendicular to the sample surface ) in the case of specular configuration . ]
second , in the case of valence - subband splitting , the pl polarization is determined by projection of @xmath104 not on the direction of observation @xmath103 , as this should be for a bulk unstrained semiconductor , but rather on the vector @xmath105 , which depends on the angle between @xmath103 and @xmath85 vectors and on the actual recombination energy @xcite : @xmath106 for recombination with the tops ( @xmath107 ) of the light - hole and heavy - hole subbands , the vectors @xmath101 and @xmath108 are given by the same expressions ( [ k0 ] ) as @xmath88 and @xmath87 , but with @xmath86 being replaced by @xmath103 .
we readily see that @xmath108 is always parallel to the growth axis , while @xmath101 is at an angle @xmath109 to this axis ( fig.[ezhik ] ) .
if the pl detection energy @xmath110 , recombination with the light - holes is only possible and , therefore , @xmath111 . when @xmath112 , the crystal valence - band splitting can be ignored and an effective direction of the pl detection , given by vector @xmath105 , coincides with the real one : @xmath113 .
so , with increase of the detection energy , the vector @xmath105 changes its direction from @xmath114 to @xmath115 , rotating clockwise through angle ( @xmath98 ) .
since both the value and direction of the @xmath80 and @xmath105 vectors are energy dependent , the hanle curves will be different for different excitation and detection energies .
however , as shown in @xcite , they can still be described by eq.([hanlebulk ] ) , provided the angle @xmath64 is replaced by the angle @xmath116 , where @xmath117 and @xmath118 are the angles between the _ z _ axis and vectors @xmath119 and @xmath96 , respectively , reckoned counterclockwise from the _ z _ axis .
the above qualitative analysis of orientation of the vectors @xmath120 and the angle between them is based on eqs . ( [ k0 ] )
valid for the @xmath121-point , @xmath122 .
the values of @xmath80 and @xmath105 for arbitrary frequencies @xmath123 and @xmath60 can be found taking into account the interband optical transitions with @xmath124 where the heavy- and light - hole states are mixed . such kind of calculations has been performed for unstrained bulk zinc - blende - lattice semiconductors @xcite , strained bulk semiconductors with the top of the heavy - hole subband lying above that of the light - hole subband @xcite , quantum - well structures in the approximation of infinitely - high barriers @xcite and for finite barriers @xcite ( see also @xcite ) , and semiconductor superlattices @xcite . in this section
we will follow the method of ref .
@xcite developed in order to calculate the optical orientation of electron spins under interband absorption of circularly polarized light normally incident on a hexagonal crystal , e.g. , cds or cdse .
in fact , in @xcite the quasicubic model of the valence band structure is used : the crystal splitting of the higher two valence subbands is obtained by applying an effective uniaxial strain to a crystal of the cubic symmetry with the @xmath24 valence band . here
we generalize the approach of ref .
@xcite to consider ( i ) the oblique incidence of the exciting light on the sample , and ( ii ) the opposite sign of the uniaxial strain resulting in the opposite sequence of the split subbands , now the light - hole subband lies above the heavy - hole subband .
the spin - polarized electrons are described by the spin - density matrix @xmath125 , where @xmath126 are the spin indices .
this matrix can be found from the balance matrix equation @xmath127 where the first and second terms describe the spin relaxation and larmor precession of the electron spins , and the right - hand side term is the generation rate of the electron spin - density matrix .
four components of the latter matrix can be calculated by using the general equation @xmath128 \sum_j m_{cs , v_n j}({\bm k } ) m^*_{cs ' , v_n j}({\bm k})\:.\ ] ] hereafter we omit common multipliers and use the following notations : @xmath129 is the electron wave vector , @xmath130 and @xmath131 are the electron energies in the conduction band @xmath132 and the valence subband @xmath133 given by @xmath134 @xmath135 @xmath136 is the uniaxial - strain axis coinciding with the crystal growth axis , the index @xmath137 enumerates degenerate states in the @xmath138 and @xmath139 valence subbands ; @xmath140 is the electron effective mass in the conduction band , @xmath141 and @xmath79 are the standard valence - band parameters entering the luttinger hamiltonian taken in the spherical approximation ( @xmath142 ) ; @xmath90 is the splitting of the @xmath24 valence band at @xmath143 called the crystal splitting . in the following we assume a value of @xmath90 to be positive in which case the indices @xmath138 and
@xmath139 correspond to the light - hole and the heavy - hole subbands , respectively , and the parameter @xmath144 coincides with the fundamental band gap @xmath145 . the opposite order of the light- and heavy - hole subbands considered in ref .
@xcite is described by negative values of @xmath90 .
the matrix elements for interband electron transitions , @xmath146 , are related to the interband matrix elements of the momentum operator @xmath147 by @xmath148 where @xmath149 is the light polarization unit vector . under oblique incidence with the incidence plane containing the axes @xmath17 and @xmath136 ( and perpendicular to the @xmath150 axis ) the initial spin @xmath80 of the photogenerated electrons
is related to @xmath151 by @xmath152 one can show that the sum over @xmath153 in eq .
( [ rhodot ] ) can be reduced to @xmath154 where @xmath155 is the 2@xmath1564 matrix of the interband matrix elements @xmath157 @xmath158 calculated between the @xmath7 conduction and @xmath24 valence bloch functions : @xmath159\:,\ ] ] @xmath160 , @xmath161 is the interband matrix element @xmath162 with @xmath163 and @xmath164 being the standard orbital bloch functions at the @xmath121 point , @xmath165 @xmath166 is the effective electron hamiltonian in the valence band @xmath24 split into two subbands of the heavy and light holes .
since the matrices @xmath167 and the electron energy spectrum are independent of the wave - vector azimuth angle @xmath168 the hamiltonian @xmath169 in @xmath170 can be averaged over @xmath168 and reduced to a diagonal matrix .
we omit the further details and present the final result for the generation matrix ( [ rhodot ] ) written as a sum of the partial contributions @xmath171 due to the optical transitions from the subbands @xmath172 . the partial contributions @xmath171 can eventually be presented in the form @xmath173 where @xmath174 is the degree of circular polarization of the light wave in vacuum , @xmath89 is the refraction angle , @xmath175 and @xmath176 are the pauli spin 2@xmath1562 matrices .
the other notations are defined by @xmath177 @xmath178 @xmath179 @xmath180 and @xmath181 are the amplitude transmission coefficients of the @xmath182- and @xmath183-polarized light , @xmath184\:,\ ] ] @xmath185\:.\ ] ] note that @xmath186 is proportional to the reduced density of states `` conduction band - valence subband @xmath187 '' . the integrals ( [ zsubn ] ) and ( [ zsubn ] ) can be readily calculated assuming @xmath188 in gaas the dimensionless parameter @xmath189 is about 0.23 . for the light propagating in the direction @xmath190
one has @xmath191\:,\ ] ] @xmath192 \hspace{0.5 cm } \mbox{for } \hspace{1 cm } \varepsilon > 1\ : .
\end{array } \right.\ ] ] here @xmath193 @xmath194 @xmath195 is the light frequency , and the function @xmath196 of a variable @xmath197 is defined by @xmath198 the photoelectron average spin components in the directions @xmath136 and @xmath17 are determined by @xmath199 where @xmath200\:,\ : r^+_x = - \frac12 \sqrt{\chi_+}\ [ 1 + \pi(\chi_+ ) ] \:,\ ] ] @xmath201 if @xmath202 , and @xmath203 \ : , \ : r^-_x = - \frac12 \sqrt{\chi_-}\ [ 1 - \pi(\chi_- ) ] \:,\ ] ] if @xmath204 .
the above equations correspond to positive @xmath90 .
however , they are also valid for negative @xmath90 if the function ( [ pi ] ) is replaced by @xmath205 the angle @xmath206 between the vector of the average spin and the axis @xmath136 equals to @xmath207 where @xmath208 .
here we take into account that , in the medium with a big index of refraction , the refraction angle @xmath89 is small even for remarkable incidence angles . under photoexcitation near the fundamental edge , @xmath209 , one has @xmath210 , @xmath211 and the optical transitions are allowed only from the upper light - hole subband . for @xmath212 , the function @xmath213 in ( [ pi ] ) tends to 1 , and according to eq .
( [ szsx ] ) the ratio @xmath214 is given by @xmath215 and @xmath206 by @xmath216 in agreement with the first equation ( [ shhslh ] ) . at the edge of the transitions from the heavy - hole subband , @xmath217 , the function @xmath218 and , therefore , @xmath219 and @xmath220 is negative .
therefore , for electrons excited from the top of the @xmath139 subband , @xmath221 , in agreement with the second equation ( [ shhslh ] ) .
it is convenient to use the reduced excitation and detection energies @xmath222 and @xmath223 .
our measurements have been made for a large excitation energy @xmath224 permitting the crystal splitting of the valence subbands to be neglected .
below we present the results of calculation for that energy .
they show that in this case @xmath225 , i.e. @xmath226 , and @xmath227 , which is in full agreement with the above qualitative analysis based on the expressions ( [ k0 ] ) .
the dependencies of the polarization @xmath228 and of the angle @xmath116 on the detection energy @xmath229 calculated for zero magnetic field and specular configuration are shown by solid lines in fig.[alfa&ro(edet)]a and fig.[alfa&ro(edet)]c .
one can see that for the detection energy @xmath230 , when recombination with the light holes with @xmath231 is only possible and the vector @xmath232 is directed at the angle @xmath233 to the @xmath136 axis , the angle @xmath234 and polarization @xmath235 . in this case ,
when increasing @xmath229 , the admixture of the heavy - hole states leads to the increase of @xmath237 and @xmath116 .
these changes of @xmath237 and @xmath116 are small when @xmath229 varies from 0 to 1 .
however , the @xmath237 and @xmath116 dependencies have abrupt humps at @xmath238 , when the transitions from the heavy - hole subband are switched on , involving the creation of electrons with opposite signs .
with further increase of @xmath229 , the quantity @xmath237 increases rapidly , then reverses its sign from negative to positive and approaches its maximal value @xmath239 at @xmath240 , while the angle @xmath116 approaches the maximal value of @xmath241 , corresponding to which are @xmath242 and @xmath243 .
those values of @xmath237 and @xmath116 remain practically invariable with @xmath229 increasing still further . the pl circular polarization ( _ a _ ) , the spin @xmath244 ( _ b _ ) , the angles ( @xmath245 ) ( _ c _ ) and @xmath64 ( _ d _ ) calculated as a function of the pl detection energy for specular ( solid ) and backscattering ( dashed ) configurations for the refraction angle @xmath246 , corresponding to the incidence angle of 48@xmath26 , and excitation energy @xmath247 in the absence of magnetic field . in _
b _ , the dependence @xmath248 is shown also for the normal incidence , @xmath34 . in the case of specular configuration the normal to the sample surface coincides with the bisectrix between the exciting beam and the direction of the pl detection . ] the angle @xmath249 ( @xmath250 ) when the detection energy is equal to @xmath251 ( shown by a vertical dashed line in fig.[alfa&ro(edet ) ] ) .
therefore at such an energy of detection the dependence of @xmath237 on @xmath229 passes through zero , although @xmath252 ( see thick solid line in fig.[alfa&ro(edet)]b ) . note that @xmath253 only when the direction of pl detection coincides with the axis of strain ( @xmath254 , @xmath255 ) ( thin solid line in fig.[alfa&ro(edet)]b ) .
since @xmath256 and @xmath257 reverse their signs as the argument changes by 180@xmath26 , the @xmath116 angle can be substituted by the angle @xmath64 , which , as shown in fig.[alfa&ro(edet)]d , is approximately equal @xmath258 at @xmath259 , @xmath109 at @xmath260 and changes drastically within the range @xmath261 , making a leap from -90@xmath26 to + 90@xmath26 at @xmath262 ( @xmath263 ) . with such definition of angle @xmath64 , expression
( [ hanlebulk ] ) for the hanle effect takes the form : @xmath264 where @xmath265 .
we consider the coupled spin - system of free and bound electrons in the model of spin - dependent recombination proposed by weisbuch and lampel @xcite , applied by paget @xcite and generalized in @xcite . in this model
the magnetic - field dependence of the pl polarization can be with high accuracy described by two lorentzians : @xmath266 { \rm sign}(\varepsilon_{\rm det}-\varepsilon_{\perp})~,\ ] ] where the first and the second term in the square brackets describe depolarization of the free and the bound electrons , respectively , @xmath267 is the angle of rotation of the mean spin of localized electrons @xmath268 in a magnetic field @xmath269 during the lifetime of their spin @xmath49 , @xmath270 is the residual polarization , observed experimentally in high magnetic field .
it is convenient to measure signs of @xmath0 and @xmath271 at @xmath272 and @xmath273 . as seen from fig.[alfa&ro(edet)]d , in the first case
@xmath274 ( as in bulk unstrained semiconductor ) , and in the second case @xmath275 .
experimental ( circles ) and calculated ( solid , dotted ) hanle curves obtained in gaas@xmath1n@xmath2 in the specular configuration set - up at high - energy excitation and high - energy recombination when @xmath276 .
@xmath35ev , @xmath277ev , @xmath246 .
inset illustrates relative orientation of the spins @xmath75 , @xmath278 and the unit vectors @xmath279 , @xmath77 . ]
we realized the first case at @xmath280 ( here , just as in all the following measurements , @xmath247 and the refraction angle @xmath246 ) .
the experimental hanle curve ( open circles in fig.[2teta ] ) has a manifestly asymmetric shape in this case .
the maximum of its narrow part ( fig.[2teta]a ) , which corresponds to depolarization of bound electrons , shifts to positive values of the magnetic field . for the employed specular geometry of experiment , to this corresponds the positive sign of @xmath51 @xcite ( note that the hanle curve , measured in the same geometry in bulk gaas , in which @xmath281 , has a maximum at @xmath282 ) .
the maximum of the broad part of the experimental curve @xmath48 determined by depolarization of free electrons is less pronounced , since it is superimposed by the depolarization curve of bound electrons , which has a greater - by - an - order amplitude in zero magnetic field .
the polarization of bound electrons , however , decreases rapidly as the magnetic field increases , and we can neglect its contribution at @xmath283 kg . the solid curve in fig.[2teta ] is calculated from eq.([hanlesplitted ] ) with the same values of depolarization curve halfwidths for bound and free electrons @xmath55 g and @xmath54 g that have been obtained from the fitting of the experimental hanle curves in fig.[0teta ] , measured under normal excitation and detection ( the same values @xmath284 and @xmath52 will be used below in approximation of experimental hanle curves measured under different experimental conditions ) .
one can see that this curve approximates the experimental dependence @xmath48 satisfactorily .
this makes it possible to extract the partial hanle curve for free electrons that is described by the first term in eq.([hanlesplitted ] ) and shown by the dotted line in fig.[2teta ] .
as the dotted line has a maximum at @xmath285 ( fig.[2teta]b ) , we can conclude that @xmath286 .
experimental ( circles ) and calculated ( solid , dotted ) hanle curves obtained in gaas@xmath1n@xmath2 in the specular configuration at high - energy excitation and the electron recombination with the light holes when @xmath287 .
@xmath35ev , @xmath288ev , @xmath246 . ] the experimental dependence @xmath48 for this case that has been realized at @xmath289 , is shown by open circles in fig.[(-1)teta ] .
the negative sign of the measured luminescence polarization results from the fulfillment of the condition @xmath290 .
solid curve in fig.[(-1)teta ] is calculated using eq.([hanlesplitted ] ) for @xmath275 .
the maxima ( in the absolute value ) on the depolarization curves for the bound ( fig.[(-1)teta]a ) and free ( dotted line in fig.[(-1)teta]b ) electrons are situated at @xmath282 , which also attests to positive signs of @xmath0 and @xmath51 . deformation - induced splitting of the light- and heavy - hole subbands allows the hanle curve asymmetry to be observed even in the backscattering configuration where the directions of detection and excitation are antiparallel , @xmath291 . to realize that possibility , the magnetic field @xmath269 , lying in the sample surface plane ,
is directed perpendicular to the exciting beam , while the sample is turned around vector @xmath269 in such a way that there is nonzero angle @xmath89 between the normal to the sample ( axis @xmath292 ) and the exciting beam ( vector @xmath279 ) , as shown in the insert in fig.[3teta ] .
the dependence of angle @xmath64 between vectors @xmath75 and @xmath278 on detection energy @xmath229 calculated for backscattering configuration and @xmath247 is shown by dashed line in fig.[alfa&ro(edet)]d .
one can see that @xmath293 if @xmath259 and @xmath294 if @xmath260 . . at @xmath259 angle @xmath295 , which makes the asymmetry of the experimental hanle curve ( open circles in fig.[3teta ] ) more pronounced than in the case of the specular configuration when @xmath296 .
the negative sign of polarization is due to the fact that @xmath297 .
solid and dotted curves in fig.[3teta ] are calculated from eq.([hanlesplitted ] ) and the second term of eq.([hanlesplitted ] ) , respectively , for @xmath298 .
it is seen that the maxima of depolarization curves both for the bound ( fig.[3teta]a ) and the free electrons ( dotted line in fig.[3teta]a ) are found in the region of positive values of the magnetic field . for a backscattering configuration
this indicates that the signs of @xmath0 and @xmath51 , which is in full agreement with the result obtained with the use of specular configuration .
experimental ( circles ) and calculated ( solid , dotted ) hanle curves obtained in gaas@xmath1n@xmath2 in the backscattering configuration at high - energy excitation and the electron recombination with light holes when @xmath299 .
@xmath35ev , @xmath288ev , @xmath246 .
] . at @xmath260 ( when deformation splitting can be neglected ) the angle @xmath300 and ,
according to eq.([hanlesplitted ] ) , the hanle curve must be symmetrical and insensitive to the sign of _ g_-factor .
indeed , the experimental hanle ( open circles in fig.[0alfa-12teta ] ) , measured at @xmath280 , is symmetric .
experimental ( circles ) and calculated ( solid , dotted ) hanle curves for gaas@xmath1n@xmath2 in the case of backscattering configuration at high - energy excitation and high - energy recombination when @xmath67 .
@xmath35ev , @xmath301ev , @xmath246 . ]
the experimental dependencies @xmath48 presented above have been measured under high - energy excitation ( @xmath302 ) and detection energy @xmath272 or @xmath259 , when angle @xmath64 remains practically invariable with varying @xmath229 . under these conditions the greatest absolute value of the angle @xmath64 and , respectively , the greatest asymmetry of hanle curve
are observed in specular configuration at @xmath272 ( @xmath303 ) and in backscattering configuration at @xmath259 ( @xmath298 )
. a greater asymmetry of hanle curve can be obtained in specular configuration under excitation and detection with only a light hole participating ( @xmath304 ) , when @xmath305 @xcite .
note that the maximum asymmetry of hanle curve can be realized both in specular and backscattering configurations under high - energy excitation and detection energy @xmath262 , when in the absence of magnetic field the vectors @xmath278 and @xmath75 are perpendicular to each other ( angle @xmath306 ) and @xmath307 , while in a magnetic field @xmath308 , where the sign + ( @xmath309 ) before the right - hand side corresponds to the specular ( backscattering ) configuration .
however , at @xmath262 the modulus of @xmath278 vector diminishes sharply and is near zero if the refraction angle does not exceed a few degrees ( @xmath310 for @xmath255 ) , amounting to @xmath311 only at angles @xmath89 close to the maximum value ( @xmath312 in gaas @xcite ) ( see fig.[alfa&ro(edet)]b ) .
experimental dependencies @xmath48 measured in gaasn consist of narrow and broad parts . the large amplitude of the narrow part bears witness to strong polarization of bound electrons , arising due to spin - dependent recombination .
we have approximated the experimental curves @xmath48 by sum of two lorentzians , one of which describes the depolarization of bound electrons , and the other , that of free electrons . at the same time , we have shown in ref.@xcite that the polarizations of free and bound electrons are summed up additively ( @xmath313 ) only in the limit of low polarization of bound electrons .
if polarization of bound electrons is strong , it becomes necessary to solve a system of nonlinear equations , coupling polarizations of free and bound electrons @xcite .
as the preliminary analysis shows , in that case a substantially better fit of the experimental hanle curve is obtained for the region of low magnetic fields , where the polarization of bound electrons dominates .
the detailed theoretical description of the compound hanle effect under an arbitrary polarization ( 100% including ) of bound electron will be presented in a separate publication . here
we note that the shape of hanle curves ( in particular their asymmetry ) , obtained from the solution of a system of nonlinear equations does not differ qualitatively from the results of approximation by two lorentzians , which makes it possible to employ the latter for determination of the signs of @xmath0 and @xmath51 .
in order to determine the sign of the electron @xmath0 factor in a semiconductor film , we have applied the method based on analysis of the asymmetrical hanle effect under oblique incidence of the exciting light onto the sample and detection of the photoluminescence at oblique emission . the relationship between the @xmath0 factor sign and direction of the asymmetrical shift of the hanle curve
is governed by the valence band structure , particularly , by its splitting due to an internal uniaxial strain in the film , and depends on the photoexcitation and photodetection energies . in the present experimental and theoretical work
we have extended the method to the case where the internal strain shifts the light - hole subband upwards relative to the heavy - hole subband .
the interband optical orientation of electron spins , circular polarization of photoluminescence , and shape of the hanle curve have been calculated in dependence on the energies of excitation and detection as well as on the angles of incidence and secondary emission .
the experimental study has been performed on uniaxially compressed gaas@xmath5n@xmath314 epitaxial films .
the observed hanle effect is a result of the magnetic - field induced depolarization of the coupled system of spin - polarized free electrons and electrons bound on deep paramagnetic centers .
the asymmetry of the resultant hanle curve has been measured at room temperature under oblique incidence of the pump radiation and detection in the specular or backscattering configurations .
positive signs of the gyromagnetic @xmath0-factors of the free and bound electrons in the gaas@xmath1n@xmath2 alloy have been deduced from comparison between experiment and theory .
we are grateful to late b.p .
zakharchenya for stimulating interest in this work and to m.m .
glazov , k.v .
kavokin and v.m .
ustinov for fruitful discussions .
partial support by the russian foundation for basic research , grants of the russian academy of sciences , and tara project of the university of tsukuba is acknowledged .
10 to be precise , @xmath315 , where @xmath316 is the angle between @xmath317 and @xmath318 .
since the angle @xmath89 between @xmath279 and @xmath318 can not exceed 16@xmath26 because of the large refractive index , approximately 3.6 , of gaas and gaasn , we can accept @xmath319 .
|
positive signs of the effective @xmath0-factors for free electrons in the conduction band and electrons localized on deep paramagnetic centers have been measured in nitrogen dilute alloy gaas@xmath1n@xmath2 at room temperature .
the @xmath0-factor signs have been determined from an asymmetry in the depolarization of edge photoluminescence in a transverse magnetic field ( hanle effect ) at the oblique incidence of the exciting radiation and oblique - angle detection of the luminescence .
the tilted spin polarization of free electrons is induced under interband absorption of circularly polarized light , and the paramagnetic centers acquire spin polarization because of spin - dependent capture of free spin - polarized electrons by these centers . the measured hanle curve is a superposition of two lines , narrow and broad , with the widths @xmath3 g and @xmath4 g , arising due to the depolarization of localized and free electrons , respectively .
the difference in the linewidths by two orders of magnitude strongly indicates much longer spin lifetime for the paramagnetic centers as compared with that for the free carriers .
the magnitude and direction of the asymmetry in the measured hanle curve have been found to depend on the partial contributions to the recombination radiation from the heavy- and light - hole subbands split by a uniaxial deformation of the gaas@xmath5n@xmath6 film grown on a gaas substrate .
we have extended the theory of optical orientation in order to calculate the excitation spectrum of the photoelectron tilted - spin polarization and the circularly - polarized luminescence spectrum taking into account that , in the strained samples under study , the light - hole subband lies above the heavy - hole one .
the results have further been used to calculate the shape of hanle curve as a function of the excitation and registration energies as well as the incidence and detection angles and to compare the theory with experiment .
|
in the * dense @xmath0 loop model * , a square lattice is tiled with the following two kinds of square tiles known as * plaquettes * , denoted symbolically by @xmath1 and @xmath2 :
|
the dense @xmath0 loop model is a statistical physics model with connections to the quantum xxz spin chain , alternating sign matrices , the six - vertex model and critical bond percolation on the square lattice .
when cylindrical boundary conditions are imposed , the model possesses a commuting family of transfer matrices .
the original proof of the commutation property is algebraic and is based on the yang - baxter equation . in this paper
we give a new proof of this fact using a direct combinatorial bijection .
|
we recently introduced tensor parametrizations for the phenomenological effective forces skyrme and gogny , by taking into account a tensor force of zero and finite range , respectively @xcite .
the tensor parameters were adjusted on top of already existing skyrme and gogny parametrizations , by employing a three - step fitting procedure inspired by refs .
@xcite and by modifying also , simultaneously , the spin orbit strength .
this work was done as an exploratory study to identify the correct signs and regions for the values of the tensor parameters , and can be viewed as a preparatory study for a global fit of all the parameters , especially in the gogny case , where much less work has been done including the tensor force . in the present work ,
we employ one of the skyrme parametrizations introduced in ref .
@xcite and show that ( even if it was not found with an adjustment of all the parameters of the force ) it properly accounts , within a mean field picture , for shell effects that were recently confirmed by experimental measurements , namely the magicity of the calcium isotopes @xmath0ca and @xmath1ca . experimental studies strongly indicate @xmath2 as a new magic number in ca isotopes due to the high energy of the first 2@xmath6 state in this nucleus @xcite .
more recently , high precision mass measurements were performed for the neutron
rich ca isotopes @xmath7ca and @xmath1ca by employing the mass spectrometer of isoltrap at cern @xcite . the found results and , in particular , the trend obtained for the two neutron separation energies @xmath8 , definitely confirmed the magicity of the nucleus @xmath0ca . for the nucleus @xmath1ca ,
the first experimental spectroscopic study on low lying states was performed very recently with proton knockout reactions at riken @xcite .
this study ( in particular , the high energy of the first @xmath9 state ) provided a robust experimental signature indicating the magic nature of the nucleus @xmath1ca .
some comparisons with theoretical calculations have been included in the two experimental studies of refs .
@xcite . in ref .
@xcite , the experimental @xmath8 energies have been compared with the results of microscopic calculations based on chiral interactions , with coupled cluster and shell model results , as well as with energy density functional ( edf ) results based on the mean field approximation . in ref .
@xcite , shell model results @xcite have been compared with the experimental data and the important role played by the neutron
proton ( n p ) tensor contribution has been investigated .
it was also mentioned in ref .
@xcite that recent calculations including three body forces @xcite provide a very good agreement with the experimental results .
it was stressed that , in the shell model calculations of ref .
@xcite , the effect of three body forces is included empirically .
this explains why the obtained results are very similar to those of refs .
@xcite , where the three - body contribution is taken into account . in this work ,
we perform hartree - fock calculations in spherical symmetry and neglect pairing correlations .
the calculations are done in coordinate space .
we do not need more sophisticated models because our objective is to isolate the genuine tensor contribution : we show that the tensor force , and its induced like particle and n
p effects , may describe the magicity of @xmath0ca and @xmath1ca within a simple mean field scheme . by analyzing ca isotopes ,
it is shown in particular that an enhancement of the magicity of @xmath0ca and @xmath1ca may be predicted within mean field calculations by including the like particle contribution generated by the tensor force .
on the other side , by analyzing @xmath2 and @xmath3 isotones , the role played by the n
p tensor contribution is investigated .
one of the skyrme parametrizations introduced in ref .
@xcite is employed , that was constructed on top of the sly5 @xcite skyrme force . in the three step fitting procedure adopted in ref .
@xcite , the first adjustment was done to tune the spin
orbit strength ( before tuning the tensor parameters ) to reproduce the neutron @xmath10 spin
orbit splitting in the nucleus @xmath11ca .
this nucleus is spin saturated and , consequently , the tensor force does not have any effect on its spectroscopic properties . after this first adjustment ,
the tensor parameters were tuned to reproduce the neutron @xmath10 spin
orbit splitting first in the nucleus @xmath12ca and then in the nucleus @xmath13ni .
details about this procedure may be found in ref .
the difference of our skyrme parameter set with respect to the tensor parametrization published by col _
@xcite ( also introduced on top of sly5 ) is that the spin
orbit strength is simultaneously modified in our case . for our skyrme set , the spin
orbit strength was reduced to 101 mev @xmath14 , with respect to the value in the original force , and the parameters responsible for the like particle and the n
p tensor effect were adjusted to the values @xmath15 and @xmath16 mev @xmath14 , respectively .
the parameters @xmath17 and @xmath18 are related to the parameters @xmath19 and @xmath20 as follows , @xmath21 where @xmath20 and @xmath19 are the strengths of the skyrme zero range tensor force in even and odd states of relative motion , respectively @xcite .
the article is organized as follows . in sec . [ like ]
we analyze the enhancement of magicity in the isotopes @xmath0ca and @xmath1ca , that is related to the like particle tensor contribution . in sec .
[ np ] , @xmath2 and @xmath3 isotones are analyzed and the n
p tensor effects are investigated . in sec .
[ masses ] the two neutron separation energies are compared with the experimental data for the nuclei @xmath22ca , @xmath0ca , and @xmath1ca . in sec .
[ concl ] conclusions are drawn .
we evaluate the single particle neutron gap for four ca isotopes , the closed
shell nuclei @xmath11ca and @xmath12ca and the two systems that we wish to analyze here , @xmath0ca and @xmath1ca .
ca isotopes are spin saturated in protons .
this means that tensor effects on neutron single particle states may be induced in practice only by the like particle neutron neutron ( n
n ) contribution along the isotopic chain .
we show in fig .
1 the single particle gaps obtained for the four isotopes with our effective skyrme force by quenching the like particle tensor strength ( dashed line ) and by switching it on ( solid line ) .
the reported gaps refer to @xmath25 , @xmath26 , @xmath2 , and @xmath3 for @xmath11ca , @xmath12ca , @xmath0ca , and @xmath1ca , respectively .
two effects may be observed .
first , a global enhancement of the single particle gaps is visible due to the tensor contribution .
this enhancement is found also for the closed
shell nucleus @xmath12ca , where the gap is increased by 1.3 mev .
the corresponding experimental value is 5.4 mev @xcite .
this means that the inclusion of the n
n tensor contribution leads to the correct shift , towards the experimental result . for the nucleus @xmath11ca , the results obtained with and without the n
n tensor effect are obviously almost the same , since this nucleus is fully spin saturated and the tensor force is not active there .
we notice that the gap variation due to the inclusion of the n
n tensor contribution is more important for @xmath0ca than for @xmath12ca and @xmath1ca , as will be explained below . in the figure , also the results obtained with the original sly5 force are reported ( blue dotted line ) .
we may observe that our new set of parameters provides a global improvement of the results : it can be seen that both single particle gaps obtained for @xmath11ca and @xmath12ca are closer to the experimental values ( red triangles ) .
it is interesting to notice that , in the case of @xmath11ca , where the tensor force does not play any role , the better agreeement with the experimental gap is entirely due to the reduced strength of the spin orbit contribution .
ca , @xmath12ca , @xmath0ca , and @xmath1ca .
the black solid line corresponds to the results obtained with the new skyrme force ; the black dashed line corresponds to the results obtained with the new skyrme force by neglecting the n
n tensor contribution .
the results obtained by using the original sly5 force are also reported ( blue dotted line ) .
the experimental values for @xmath11ca and @xmath12ca are represented by red triangles.,width=302 ] let us investigate in detail the tensor effect for each of the three isotopes where the tensor force contributes . for @xmath12ca , the neutron @xmath27 state is filled . both high-@xmath28 ( by using the terminology of ref .
@xcite ) single particle states @xmath27 and @xmath29 are pushed downwards , more strongly for the first state than for the second [ fig .
the net effect is that the gap between the two levels is increased .
ca ( a ) , @xmath0ca ( b ) , and @xmath1ca ( c ) with and without the like particle tensor contribution.,width=302 ] for @xmath0ca , the neutron @xmath29 state is filled .
the high-@xmath28 @xmath29 state is pushed downwards whereas the low-@xmath28 @xmath30 state is pushed upwards [ fig .
the single particle gap is increased . for @xmath1ca
, the neutron @xmath30 state is filled .
both low-@xmath28 single particle states @xmath31 and @xmath30 are pushed upwards , more strongly for the first state than for the second [ fig .
the gap increases also this time . in particular , the stronger enhancement of the single particle gap obtained in @xmath0ca is due to the fact that each single particle state is pushed in the opposite direction by the tensor contribution . figs .
1 and 2 clearly show that the single particle gaps are increased in @xmath0ca and @xmath1ca due to the like particle tensor contribution .
this provides an enhancement of magicity for the two isotopes , with respect to what found by neglecting the tensor term .
in particular , the gap is increased for @xmath0ca from 1.24 to 3.06 mev and for @xmath1ca from 0.45 to 1.86 mev , providing in this way nuclei that have a stronger closed shell nature . with the opposite sign of the tensor strength
, the effect would be the opposite , that is , the gap would be decreased by including the tensor contribution and would be in particular more strongly shrinked for the nucleus @xmath0ca , where the @xmath29 state would be shifted upwards and the @xmath30 state would be pushed downwards .
the presently used signs of the tensor parameters are thus the correct ones that allow us to enhance the magic character of the nuclei under study .
to analyze the effects of the n
p tensor contribution , isotonic chains containing @xmath0ca and @xmath1ca have to be analyzed .
let us start with the @xmath2 isotones @xmath0ca , @xmath1ti , @xmath13cr , and @xmath4fe . in these nuclei ,
the last occupied neutron state is the state @xmath29 , the upper level is the state @xmath30 , and the neutron gap @xmath2 thus coincides with the spin
orbit splitting of the @xmath32 neutron states .
going from @xmath4fe to @xmath0ca , the @xmath33 number is reduced from 26 to 20 .
the occupation of the proton state @xmath27 is thus decreased from 6 to 0 .
p tensor effect is expected to provide an attractive interaction between the proton state @xmath27 and the neutron state @xmath30 and a repulsive interaction between the proton state @xmath27 and the neutron state @xmath29 in a given nucleus .
such tensor contribution is thus expected to induce a reduction of the neutron @xmath34 spin
orbit splitting , that is , a reduction of the @xmath2 gap .
this effect is however expected to be weakened along the isotonic chain , as far as the occupation of the proton @xmath27 state is reduced , that is , going from @xmath4fe to @xmath0ca . with the reduction of the n
p effect , the gap is expected to increase going from @xmath4fe to @xmath0ca .
this is shown in fig .
3 , where the gap is plotted ( black solid line ) and in fig .
4 , where the involved proton and neutron single
particle energies are displayed . in fig .
4 one can see that , moving from the left to the right ( that is , towards more neutron
rich nuclei along the isotonic chain ) , the energy of the proton state moves towards lower values and the energies of the neutron states are pushed upwards , as expected in mean field calculations . on top of the global mean field evolution , we should isolate the tensor contribution . from fig .
3 , one observes that the expected effect of enhancement of the gap going from @xmath4fe to @xmath0ca looks very weak . to better understand how the tensor force acts in this particular case ,
we compare the found results with those obtained by switching off the n p tensor strength .
the corresponding values are shown in fig .
3 ( red dashed line ) . one can observe that , without the n
p tensor contribution , the values are similar but the trend is the opposite , that is , the gap is reduced when passing from @xmath4fe to @xmath0ca .
the tensor force changes this trend . to isolate all the tensor contributions ,
we have repeated the same calculations by quenching also the parameter responsible for the like particle tensor effect .
the corresponding results are shown in the figure by a green dot - dashed line .
it is clear that the like particle tensor contribution provides an enhancement of the gap but the change of trend is due only to the n
p contribution .
it is also interesting to compare the present results with the values obtained with the original sly5 force ( blue dotted line in fig .
3 ) . we can see that the trend is very similar to that obtained in our case by switching off the n
p tensor strength or both tensor strengths .
there is however a shift of the values : the results obtained with the original parametrization are located between the green dot dashed and the red dashed curves .
the effect of the reduction of the spin
orbit induces the shift from the dotted to the dot dashed curve and the inclusion of the like particle tensor strength strongly pushes the values upwards . finally , the inclusion of the n
p tensor contribution determines the change of slope and provides increasing ( instead of decreasing ) gaps going from @xmath4fe to @xmath0ca .
this is what expected following the arguments based on shell model calculations and presented in ref .
@xcite , where this analysis is done for @xmath3 isotones .
calculated for the isotones @xmath0ca , @xmath1ti , @xmath13cr , and @xmath4fe.,width=302 ] and of the neutron states @xmath29 and @xmath30 for the isotones @xmath0ca , @xmath1ti , @xmath13cr , and @xmath4fe .
, width=302 ] we repeat the same analysis for @xmath3 isotones for the nuclei @xmath1ca , @xmath13ti , @xmath4cr , and @xmath5fe .
the results are plotted in figs . 5 and 6 .
the difference with respect to the previous case is that the last occupied neutron state is now the state @xmath30 and the neutron gap @xmath3 is calculated between the states @xmath31 and @xmath30 .
now , the n
p tensor effect associated to the proton state @xmath27 induces for both neutron states an attractive interaction , stronger for the @xmath10 state than for the @xmath34 state .
this means that the gap is expected also in this case to be reduced in a given nucleus owing to the n
p tensor contribution .
again , when this contribution is weakened going from @xmath5fe to @xmath1ca , the gap is eventually expected to increase .
this effect may be seen in fig .
5 ( black solid line ) . in fig .
6 the involved neutron and proton single particle energies are represented .
again , the global mean field evolution provides a reduction of the proton energy and an enhancement of the neutron energies , when going towards more neutron
rich nuclei along the isotonic chain . on top of this general effect ,
the tensor effect has to be disentangled .
this is done by plotting in fig .
5 the results obtained by switching off the n p tensor strength ( red dashed line ) .
the gap still increases , but much less strongly than in the full calculations . by switching off
also the like particle tensor strength , the obtained values are reported by a green dot - dashed line .
the value for the nucleus @xmath5fe is not reported .
the hartree
fock calculation does not converge in this case .
this is probably due to the fact that the neutron states @xmath30 and @xmath31 become almost degenerate and cross each other at each iteration .
the values obtained with the original sly5 force are also presented ( blue dotted line ) . also this time
one observes that the slope in the results obtained with the original sly5 force is very similar to that obtained by switching off , in our case , the n
p tensor strength or both tensor strenghts .
also in this case there is a shift between the dotted curve and the dashed and dot dashed curves .
the spin orbit reduction in the new parametrization determines the shift to lower values from the dotted to the dot dashed curve .
the inclusion of the like particle tensor contribution induces a strong shift to higher energies ( dashed curve ) . finally , the inclusion of the n
p tensor contribution leads to a change of slope ( solid curve ) .
we notice that this time , even without the inclusion of the n
p tensor contribution , the mean field calculations provide an enhancement of the gap when going from @xmath5fe to @xmath1ca .
the tensor force leads however to a much stronger effect , affecting significantly the slope . as was mentioned in sec .
i , the calculations are done in spherical symmetry and the effect of possible deformations is thus neglected .
the inclusion of possible small deformations for some @xmath2 and @xmath3 isotones would probably modify quantitatively our predictions but we do not expect that the qualitative trends and the general conclusions of this work would be changed : the tensor force would always shift in the same direction the single particle energies ; the qualitative evolution of the corresponding gaps would not be expected to be stronlgy modified .
calculated for the isotones @xmath1ca , @xmath13ti , @xmath4cr , and @xmath5fe .
, width=302 ] and of the neutron states @xmath30 and @xmath31 for the istones @xmath1ca , @xmath13ti , @xmath4cr , and @xmath5fe.,width=302 ]
it was stressed in ref .
@xcite that the new sets may induce non negligeable effects on the masses of some nuclei , because the spin
orbit strength was modified without refitting also the other parameters of the force .
we have thus checked whether the values of the masses predicted for the ca isotopes under study are reasonable with the new parametrization .
we have found the following binding energies : 343.83 , 415.91 , 438.62 , and 445.21 mev for the nuclei @xmath11ca , @xmath12ca , @xmath0ca , and @xmath1c , respectively .
these values have to be compared with the binding energies obtained with the original sly5 force , that are 344.07 , 415.92 , 437.25 , and 444.94 mev for @xmath11ca , @xmath12ca , @xmath0ca , and @xmath1c , respectively .
we observe that the deviations are not very important .
the largest deviation between the new skyrme values and those obtained with sly5 is 0.3% . in ref .
@xcite , the two neutron separation energies are compared with several theoretical models .
experimentally , a clear signature of shell closure for the nucleus @xmath0ca is the important reduction of the @xmath8 value going from @xmath0ca to @xmath1ca .
this new experimental value is now available owing to the high precision mass measurement of @xmath1ca reported in ref .
@xcite : going from @xmath22ca to @xmath0ca , the value of @xmath8 remains almost the same , and it has a sudden decrease from @xmath35 to @xmath36 .
several edf results are reported in fig .
3(b ) of ref .
it is commented there that , in general , these models do not provide the correct trend for the @xmath8 energies going from @xmath37 to @xmath36 : they predict a very smooth ( almost linear ) change in the @xmath8 values .
we have thus computed the two neutron separation energies @xmath38 , where @xmath39 is the binding energy of the nucleus @xmath40 , and compared them with the experimental values for the isotopes @xmath22ca , @xmath0ca , and @xmath1ca .
this is shown in fig . 7
where also the experimental values ( including those of ref .
@xcite and the new data of ref .
@xcite ) are presented .
ca , @xmath0ca , and @xmath1ca calculated with the new skyrme set .
the experimental values are reported @xcite .
the theoretical value for the nucleus @xmath13ca is also shown . in the inset
the difference between the theoretical and the experimental values is displayed.,width=302 ] we observe that the agreement with the experimental values is quite good and that the predicted change with respect to @xmath41 is not linear .
the good agreement can also be seen in the inset of the figure , where the differences between the theoretical and the experimental values of @xmath8 are reported .
we have to mention that the correct trend for the @xmath8 values may be found also by using the original sly5 force , without including the tensor term . however , the signatures of magicity are slightly stronger with the present set of parameters . with the present modified sly5 ,
the so called two neutron shell gap , calculated as @xmath42 , is equal to 5.1 mev for @xmath0ca , to be compared with an experimental value of almost 4 mev @xcite .
the reduction of the @xmath8 value , found going from @xmath36 to @xmath43 , can not be compared with any experimental result because the mass of the nucleus @xmath13ca was not yet measured experimentally .
the experimental @xmath8 value is thus still unknown at @xmath43 .
however , our theoretical prediction , that leads to a quite significant reduction from @xmath36 to @xmath43 , is coherent with the new spectroscopic data of ref .
@xcite , that indicate a magic nature for the nucleus @xmath1ca .
one should mention however that the inclusion of possible pairing correlations in the nucleus @xmath13ca would render this system more bound and would thus provide a higher value of @xmath8 at @xmath43 .
we can conclude that the present edf results display the correct trend for the values of @xmath8 in the region from @xmath22ca to @xmath1ca ( weak change from @xmath37 to @xmath35 and significant change from @xmath35 to @xmath36 ) .
in this work , we have examined some shell effects that may be clearly related to the contributions induced by a tensor force within the mean field framework with the skyrme interaction .
the employed tensor force is of zero range .
we use one parametrization that was introduced in a previous work @xcite on top of the skyrme sly5 force @xcite . in this parameter set , the spin
orbit strength was adjusted together with the tensor parameters . with this parameter set ,
we have performed hartree fock calculations in spherical symmetry .
we are aware that many effects are disregarded by using this simple theoretical model , but our objective is to isolate the genuine effects coming from the inclusion of the tensor force in a mean field model .
the magicity of the two ca isotopes @xmath0ca and @xmath1ca has been confirmed by two recent experimental measurements @xcite .
we have shown here that the introduction of the tensor force leads to an enhancement of magicity in the two isotopes . by analyzing ca isotopes
, it is shown that the like particle tensor contribution renders these nuclei more magic than they would be predicted by neglecting it . by studying @xmath2 and @xmath3 isotones , the neutron
proton tensor effects are identified and isolated with respect to the effects coming from the like particle tensor contribution .
it is shown that the @xmath2 and @xmath3 neutron gaps are predicted to increase along isotonic chains , going from @xmath4fe to @xmath0ca , and from @xmath5fe to @xmath1ca , respectively . in the case
@xmath2 , the n
p tensor contribution is responsible for the increasing trend ( without this contribution , the gap decreases ) . in the case
@xmath3 , the gap increases even without the n
p tensor contribution , but the n
p tensor contribution determines an important change of the slope .
a check on the masses of the ca isotopes under study is performed and the two neutron separation energies are compared with the experimental data for @xmath22ca , @xmath0ca , and @xmath1ca . a good agreement is found : from @xmath37 to @xmath35 , the @xmath8 value does not change strongly , whereas it is significantly reduced from @xmath35 to @xmath36 .
this provides an evidence for the shell closure @xmath2 in ca isotopes .
the application described in this work confirms the robustness of the findings of ref .
@xcite about the signs and the values for the tensor parameters .
the employed parameter set , even if introduced with a partial adjustment of the parameters ( the other parameters of the forces are not modified , with the exception of the spin orbit strength ) , provides the correct expected shell effects .
different signs of the tensor parameters would generate un uncorrect behavior if compared with the experimental results .
the present tensor parametrization ( and its induced like particle and neutron proton effects ) allows us in particular to describe , within a simple mean field picture , the magicity of the ca isotopes @xmath0ca and @xmath1ca .
10 m. grasso and m. anguiano , phys .
c 88 , 054328 ( 2013 ) .
m. zalewski , j. dobaczewski , w. satula , and t.r .
werner , phys .
c 77 , 024316 ( 2008 ) .
m. zalewski , p. olbratowski , m. rafalski , w. satula , t.r .
werner , and r.a .
wyss , phys .
c 80 , 064307 ( 2009 ) .
a. huck , et al .
c 31 , 2226 ( 1985 ) .
a. gade , et al .
c 74 , 021302 ( r ) ( 2006 ) .
f. wienholtz , et al . ,
nature 498 , 346 ( 2013 ) .
d. steppenbeck , et al . ,
nature 502 , 207 ( 2013 ) .
m. honma , t. otsuka , t. mizusaki , riken accel .
prog . rep .
41 , 32 ( 2008 ) .
holt , t. otsuka , a. schwenk , and t. suzuki , j. phys .
g 39 , 085111 ( 2012 ) .
g. hagen , m. hjorth - jensen , g.r .
jansen , r. machleidt , and t. papenbrock , phys .
109 , 032502 ( 2012 ) .
e. chabanat , p. bonche , p. haensel , j. meyer , r. schaeffer , nucl .
phys . a 627 , 710 ( 1997 ) ; ibid .
a 635 , 231 ( 1998 ) ; ibid . a 643 , 441 ( 1998 ) .
g. col , h. sagawa , s. fracasso , and p.f .
bortignon , phys .
b 646 , 227 ( 2007 ) . t.h.r .
skyrme , phil . mag . 1 , 1043 ( 1956 ) ; t.h.r .
skyrme , nucl .
phys . 9 , 615 ( 1958 ) .
n. schwierz , i. wiedenhover , and a. volya , arxiv:0709.3525 .
t. otsuka , r. fujimoto , y. utsuno , b.a .
brown , m. honma , and t. mizusaki , phys .
87 , 082502 ( 2001 ) .
m. wang et al .
c 36 , 1603 , 2012 .
|
we investigate the magicity of the isotopes @xmath0ca and @xmath1ca , that was recently confirmed by two experimental measurements , and relate it to like particle and neutron proton tensor effects within a mean field description . by analyzing ca isotopes ,
we show that the like particle tensor contribution induces shell effects that render these nuclei more magic than they would be predicted by neglecting it .
in particular , such induced shell effects are stronger in the nucleus @xmath0ca and the single particle gaps are increased in both isotopes due to the tensor force . by studying @xmath2 and @xmath3 isotones , neutron
proton tensor effects may be isolated and their role analyzed .
it is shown that neutron
proton tensor effects lead to increasing @xmath2 and @xmath3 gaps , when going along isotonic chains , from @xmath4fe to @xmath0ca , and from @xmath5fe to @xmath1ca , respectively .
the mean field calculations are perfomed by employing one skyrme parameter set , that was introduced in a previous work by fitting the tensor parameters together with the spin orbit strength .
the signs and the values of the tensor strengths are thus checked within this specific application .
the obtained results indicate that the employed parameter set , even if generated with a partial adjustment of the parameters of the force , leads to the correct shell behavior and provides , in particular , a description of the magicity of @xmath0ca and @xmath1ca within a pure mean field picture with the effective two body skyrme interaction
. 0.5 cm
|
`` factorizations `` in the naive factorization assumption and in factorization theorem have very different meanings .
the former refers to the factorization of a process into subprocesses .
for example , a @xmath0 meson decay amplitude @xmath1 is written , in the factorization assumption , as the product @xcite , @xmath2 where the meson decay constant @xmath3 arises from the production of the meson @xmath4 from the vacuum , and the form factor @xmath5 is associated with the @xmath6 transition .
the latter refers to the factorization of perturbative and nonperturbative dynamics in a qcd process . according to factorization theorem ,
the above amplitude is expressed as @xmath7 where @xmath8 denotes the convolution over parton kinematic variables , the hard kernel @xmath9 absorbs perturbative dynamics , and the @xmath0 ( @xmath10 , @xmath4 ) meson distribution amplitude @xmath11 ( @xmath12 ) absorbs nonperturbative dynamics in the @xmath13 decay . a piece of contribution to @xmath0 meson decays is factorizable , if it respects either eq .
( [ fa ] ) in the sense of the factorization assumption , or eq .
( [ ft ] ) in the sense of factorization theorem .
below we shall use the terms factorizable '' and ' ' nonfactorizable " without specifying which sense it refers to .
both collinear and @xmath14 factorizations are the fundamental tools of perturbative qcd , where @xmath14 denotes parton transverse momenta .
we first explain these two types of theorems by considering the simplest scattering process @xmath15 as an example .
the momentum @xmath16 of the pion and the momentum @xmath17 of the out - going on - shell photon are chosen as @xmath18 the leading - order ( lo ) quark diagram , in which the anti - quark @xmath19 carries the on - shell fractional momentum @xmath20 and the internal quark carries @xmath21 , leads to the amplitude , @xmath22}{(p_2-k)^2}\nonumber\\ & = & -\frac{tr[\not \epsilon \not p_2 \gamma_\mu \not p_1\gamma_5]}{x q^2}\ ; , \label{p1a}\end{aligned}\ ] ] with the leading spin structure @xmath23 of the pion and the momentum transfer squared @xmath24 .
we have suppressed other constant factors , such as the electric charge , the color number , and the pion decay constant , which are irrelevant in the following discussion .
the trivial collinear factorization of eq .
( [ p1a ] ) reads ,
@xmath25}{x q^2}\;. \label{h0c}\end{aligned}\ ] ] the zeroth - order distribution amplitude @xmath26 is proportional to @xmath27 , implying that the parton entering the lo hard kernel @xmath28 carries the same momentum as the parton entering the distribution amplitude does .
the trivial @xmath14 factorization of eq .
( [ p1a ] ) reads @xcite , @xmath29}{x q^2+k_t^2}\;. \label{h0p}\end{aligned}\ ] ] because of the zeroth - order wave function @xmath30 , @xmath28 does not depend on the parton transverse momentum actually .
quark diagrams for @xmath31 with @xmath32 representing the virtual photon vertex.,width=302 ] the @xmath33 quark diagrams for eq .
( [ p1a ] ) from full qcd are displayed in fig .
[ fig1 ] , in which the upper line represents the @xmath34 quark .
the collinear factorization of these radiative corrections is given by @xmath35 where the first - order distribution amplitude @xmath36 is defined by the effective diagrams in fig .
[ fig2 ] @xcite .
expressions from figs .
[ fig2](c ) , [ fig2](e ) , and [ fig2](f ) are proportional to @xmath37 , where @xmath38 is the loop momentum carried by the collinear gluon .
the @xmath39-function indicates that the exchange of the collinear gluon modifies the momentum fraction of the parton entering @xmath28 from @xmath40 to @xmath41 .
the @xmath14 factorization of fig .
[ fig1 ] leads to @xcite @xmath42 where the first - order wave function @xmath43 from figs .
[ fig2](c ) , [ fig2](e ) , and [ fig2](f ) is proportional to @xmath44 . in this case
the collinear gluon exchange modifies both the longitudinal and transverse parton momenta flowing into the hard kernel .
effective diagrams for the pion wave function.,width=302 ] it is observed that @xmath28 in eq .
( [ pg11 ] ) depends on @xmath45 nontrivially in the first - order @xmath14 factorization .
being convoluted with @xmath46 , the partons entering the next - to - leading ( nlo ) hard kernel @xmath47 are still on - shell . to acquire a nontrivial @xmath14 dependence , @xmath47 must be convoluted with the higher - order wave functions @xmath48 , @xmath49 : the gluon exchanges in @xmath48 render the incoming partons of @xmath47 , ie .
, the incoming partons of the quark diagrams @xmath50 and the effective diagrams @xmath43 off - shell by @xmath51 @xcite .
we thus derive @xmath52 according to the formula , @xmath53 with the @xmath19 quark carrying the momentum @xmath54 .
this is the way to obtain a @xmath14-dependent hard kernel without breaking gauge invariance , since the gauge dependences cancel between @xmath50 and @xmath43 .
a physical quantity is wtitten as a convolution of a hard kernel with model wave functions , which are determined by methods beyond a perturbation theory , such as lattice qcd and qcd sum rules , or extracted from experimental data .
a gauge - invariant hard kernel then leads to gauge - invariant predictions from @xmath14 factorization .
factorization theorems have been applied to exclusive @xmath0 meson decays , and different approaches have been developed .
below we compare the frameworks of perturbatiove qcd ( pqcd ) @xcite , qcd - improved factorization ( qcdf ) @xcite , and soft - collinear effective theory ( scet ) @xcite . the @xmath56 transition form factor @xmath57 involved in the semileptonic decay @xmath58
is expressed , in collinear factorization , as @xmath59 with the lo hard kernel @xmath60 .
the parton momentum fractions @xmath61 and @xmath62 are carried by the spectator quarks on the @xmath0 meson and pion sides , respectively . obviously , the above integral is logarithmically divergent for the asymptotic model @xmath63 @xcite .
an end - point singularity implies that exclusive @xmath0 meson decays are dominated by soft dynamics .
that is , a heavy - to - light form factor is not calculable in collinear factorization , and @xmath57 should be treated as a soft object @xcite .
this is the basis of qcdf , and subleading corrections are added systematically @xcite .
the above treatment has been further elucidated in the framework of scet @xcite : only the 1 term in @xmath28 contains the end - point singularity , which leads to an @xmath64 , ie . , soft object @xmath65 .
the @xmath66 term does not , leading to an @xmath67 object @xmath68 with the @xmath0 meson mass @xmath69 .
therefore , at leading power in @xmath70 , the @xmath71 form factor can be split into the nonfactorizable and factorizable components , @xmath72 which have different power counting in the strong coupling constant @xmath73 : the former is of @xmath74 and the latter of @xmath33 .
the values of @xmath65 and @xmath75 have been determined from a fit to the @xmath76 data @xcite .
the formulation of the @xmath56 transition in @xmath14 factorization theorem is different .
when the parton transverse momenta are included , @xmath65 does not develop an end - point singularity , and both @xmath65 and @xmath68 are factorizable .
hence , they are of the same order in @xmath73 , and can be combined into a single term , giving @xcite , @xmath77 the end - point singularity is smeared into the large logarithm @xmath78 , and absorbed into the pion wave function @xmath79 . resumming this logarithm to all orders in the conjugate @xmath80 space @xcite , we derived the sudakov factor @xmath81 , which describes the parton distribution in @xmath80 .
since @xmath65 has been included , the large - energy symmetry @xcite is respected in pqcd .
recently , it was proposed that the end - point singularity is attributed to a double counting of soft degrees of freedom in collinear factorization @xcite .
the zero - bin subtraction removes the double counting , and leads to a modified scet formalism for @xmath65 , labelled by scet@xmath82 hereafter .
the power counting of scet@xmath55 in both @xmath70 and @xmath73 is then consistent with that of pqcd .
the regularization of the end - point singularity introduces the logarithms @xmath83 in scet@xmath55 @xcite , whose treatment has not yet been explained .
when applying the above factorization approaches to two - body nonleptonic @xmath0 meson decays , further difference appears in the treatment of annihilation amplitudes .
the @xmath84 annihilation amplitudes from the scalar penguin operators are divergent in collinear factorization , where @xmath85 is the chiral enhancement scale .
because of the end - point singularity , an annihilation amplitude has been parameterized as @xmath86 in qcdf @xcite , where @xmath87 is a hadronic scale and the free parameter @xmath88 is postulated to vary in the range @xmath89 .
it is not clear what mechanism generates the strong phase @xmath90 . with the similar zero - bin subtraction ,
an annihilation amplitude is factorizable in scet@xmath55 , but found to be almost real @xcite .
a strong phase can only be generated at loop level , ie .
, at @xmath91 . however , we notice that the residual momentum carried by the @xmath80 quark in a nonfactorizable annihilation amplitude could result in a strong phase of @xmath92 @xcite , a new mechanism not included in @xcite
. the scalar penguin annihilation amplitude is also factorizable in @xmath14 factorization with the absence of the end - point singularity .
furthermore , it was almost imaginary in pqcd @xcite , whose corresponding mechanism is similar to the bander - silverman - soni one @xcite : when the @xmath93 or @xmath94 quark in a loop goes on mass shell , a strong phase is produced . in the case of the annihilation topology for heavy - to - light decays ,
the loop is formed by the virtual particles in the lo pqcd hard kernel and the infinitely many sudakov gluons exchanged between two partons in a light meson .
the virtual particle acquires the transverse momentum @xmath14 through the sudakov gluon exchange . a sizable strong phase is then given , in terms of the principle - value prescription for the virtual particle propagator , by @xmath95 therefore , the treatment and the effect of the scalar penguin annihilation amplitude are very different in qcdf , scet@xmath82 , and pqcd . though the scalar penguin annihilation amplitude is factorizable in both pqcd and scet@xmath55 , it is almost imaginary in the former , but real in the latter .
we argue that the above different opinions can be discriminated by comparing the direct cp asymmetries in the charged @xmath0 meson decays @xmath96 and @xmath97 .
the @xmath96 decays involve a @xmath0 meson transition to a pseudoscalar meson , so the penguin emission amplitude is proportional to the constructive combination of the wilson coefficients @xmath98 , where @xmath99 is the chiral enhancement scale associated with the kaon .
the @xmath97 decays involve a @xmath0 meson transition to a vector meson , so the penguin emission amplitude is proportional to the destructive combination @xmath100 .
the annihilation effect is then less influential in the former than in the latter .
if the scalar penguin annihilation is real , both decays will exhibit small direct cp asymmetries , ie . , @xmath101 .
if the scalar penguin annihilation is imaginary , it will cause a larger direct cp asymmetry in @xmath97 , ie .
, @xmath102 . the current data @xmath103 and
@xmath104 @xcite seem to prefer an imaginary scalar penguin annihilation .
according to a naive estimate of the color - suppressed tree amplitude , the hierarchy of the branching ratios @xmath105 is expected .
however , the data @xcite @xmath106 show @xmath107 , giving rise to the @xmath76 puzzle . as observed in @xcite , the nlo corrections , despite of increasing the color - suppressed tree amplitude significantly , are not enough to enhance the @xmath108 branching ratio to the measured value . a much larger color - suppressed tree amplitude , about the same order as the color - allowed tree amplitude ,
must be obtained in order to resolve the puzzle @xcite . to make sure that the above nlo effects are reasonable
, the pqcd formalism has been applied to the @xmath109 decays @xcite , which also receive the color - suppressed tree contribution .
it was found that the nlo pqcd predictions are in agreement with the data @xmath110 @xcite .
we conclude that it is unlikely to accommodate the measured @xmath108 and @xmath111 branching ratios simultaneously in pqcd , and that the @xmath76 puzzle remains .
it has been claimed that the @xmath76 puzzle is resolved in the qcdf approach @xcite with an input from scet @xcite : the inclusion of the nlo jet function , the hard coefficient of scet@xmath112 , into the qcdf formula for the color - suppressed tree amplitude gives sufficient enhancement of the @xmath108 branching ratio , if adopting the parameter scenario
s4 `` @xcite .
it is necessary to investigate whether the proposed new mechanism deteriorates the consistency of theoretical results with other data .
the formalism in @xcite has been extended to the @xmath109 decays as a check @xcite .
it was found that the nlo jet function overshoots the observed @xmath113 branching ratio very much as adopting
that is , it is also unlikely to accommodate the @xmath76 and @xmath114 data simultaneously in qcdf .
.polarization fractions in the penguin - dominated @xmath116 decays .
[ cols="^,^,^",options="header " , ] [ tab : tab1 ] for penguin - dominated @xmath116 decays , such as those listed in table [ tab : tab1 ] @xcite , the polarization fractions deviate from the naive counting rules based on kinematics @xcite .
this is the so - called the @xmath115 puzzle .
many attempts to resolve the @xmath117 polarizations have been made @xcite , which include new physics @xcite , the annihilation contribution @xcite in the qcdf approach , the charming penguin in scet @xcite , the rescattering effect @xcite , and the @xmath118 ( the magnetic penguin ) @xcite and @xmath119 @xcite transitions .
the annihilation contribution from the scalar penguin operators improves the consistency with the data , because it is of the same order for all the three final helicity states , and could enhance the transverse polarization fractions @xcite .
however , the pqcd analysis of the scalar penguin annihilation amplitudes indicates that the @xmath115 puzzle can not be resolved completely @xcite .
a reduction of the @xmath120 form factor @xmath121 , which is associated with the longitudinal polarization , further helps accommodating the data @xcite . note that there has not yet been any measurement , which constrains @xmath121 . hence , the value of @xmath121 should have a large uncertainty .
the penguin - dominated @xmath122 decays are expected to exhibit similar polarization fractions .
this is the reason the longitudinal polarization fraction in the @xmath123 decay , which contains only the penguin contribution , is close to @xmath124 as listed in table [ tab : tab1 ] .
another mode @xmath125 , nevertheless , shows a large longitudinal polarization fraction almost unity .
this mode involves tree amplitudes , which are subdominant , and should not cause a significant deviation from @xmath126 .
though the data of @xmath127 from babar still suffer a large error @xcite , the dramatically different longitudinal polarization fractions , @xmath128 , demand a deeper understanding .
it is highly suggested to update or perform the measurement of @xmath129 .
it is also worthwhile to investigate the @xmath130 decays @xcite , whose measurement can help discriminating the various proposals for resolving the @xmath115 puzzle .
the @xmath132 decays depend on the tree amplitude @xmath133 and the qcd penguin amplitude @xmath134 .
the data of the direct cp asymmetry @xmath135 then imply a sizable relative strong phase between @xmath133 and @xmath134 , which verifies the lo pqcd prediction made years ago @xcite .
the @xmath96 decays contain the additional color - suppressed tree amplitude @xmath136 and electroweak penguin amplitude @xmath137 . since both @xmath136 and @xmath137 are subdominant , the approximate equality for the direct cp asymmetries @xmath138 is expected .
however , this naive expectation is in conflict with the data @xcite , @xmath139 leading to one of the @xmath131 puzzles .
while lo pqcd gives a negligible @xmath136 @xcite , it is possible that this supposedly tiny amplitude receives a significant subleading correction .
note that the small @xmath136 is attributed to the accidental cancellation between the wilson coefficients @xmath140 and @xmath141 at the scale of the @xmath80 quark mass @xmath142 .
in @xcite the important nlo contributions to the @xmath131 decays from the vertex corrections , the quark loops , and the magnetic penguins were calculated .
it was observed that the vertex corrections increase @xmath136 by a factor of 3 , and induce a large phase about @xmath143 relative to @xmath133 .
the large and imaginary @xmath136 then renders the total tree amplitude @xmath144 more or less parallel to the total penguin amplitude @xmath145 in the @xmath96 decays , leading to nearly vanishing @xmath146 at nlo ( it is about -8% at lo ) .
we conclude that the @xmath131 puzzle has been alleviated , but not yet gone away completely . whether new physics effects @xcite are called for will become clear when the data get precise
. more detailed discussion on this subject can be found in @xcite .
two - body nonleptonic @xmath147 meson decays are interesting , since their study can test su(3 ) or u - spin symmetry .
the framework for these decays is basically identical to that for @xmath148 meson decays .
the results of two - body nonleptonic @xmath147 meson decays from different factorization approaches can be found in @xcite for qcdf , in @xcite for scet , and in @xcite for pqcd .
roughly speaking , the branching ratios predicted by qcdf and by pqcd are similar , but the predicted direct cp asymmetries are usually opposite in sign .
the factorization approaches are systematic theoretical tools for exclusive @xmath0 meson decays , in which hadronic inputs are universal , and the hard kernels can be computed order by order .
nlo corrections have been obtained for some @xmath0 meson decay modes , and the consistency between the theoretical predictions and the experimental data is improved in general .
more need to be done in order to pin down qcd uncertainty especially for those quantities exhibiting puzzling behaviors .
higher - power corrections are another important source of qcd uncertainty , which deserves a careful investigation .
the recent development in scet@xmath55 is encouraging , whose counting rules become consistent with those in pqcd .
however , the arbitrary logarithms @xmath83 resulting from the zero - bin subtraction needs to be handled ( recall that the double logarithm @xmath149 from the smearing of the end - point singularity has been resummed in pqcd ) .
we did not review the progress on the @xmath150 puzzle appearing in the extraction of the weak phase @xmath151 from penguin - dominated @xmath0 meson decays . for the detail , refer to @xcite .
99 m. bauer , b. stech , and m. wirbel , z. phys .
c * 29 * , 637 ( 1985 ) ; * 34 * , 103 ( 1987 ) .
m. nagashima and h - n .
li , phys .
d * 67 * , 034001 ( 2003 ) .
h - n . li , phys .
d * 64 * , 014019 ( 2001 ) ; m. nagashima and h - n .
j. c * 40 * , 395 ( 2005 ) .
s. nandi and h - n .
li , arxiv:0704.3790 , to appear in phys . rev .
d. h - n . li and h.l .
yu , phys .
lett . * 74 * , 4388 ( 1995 ) ; phys .
b * 353 * , 301 ( 1995 ) ; phys .
d * 53 * , 2480 ( 1996 ) .
keum , h - n .
li , and a.i .
sanda , phys .
b * 504 * , 6 ( 2001 ) ; phys .
d * 63 * , 054008 ( 2001 ) ; y.y .
keum and h - n .
li , phys . rev . * d63 * , 074006 ( 2001 ) . c. d. l , k. ukai , and m. z. yang , phys .
d * 63 * , 074009 ( 2001 ) .
m. beneke , g. buchalla , m. neubert , and c.t .
sachrajda , phys .
83 * , 1914 ( 1999 ) ; nucl . phys . *
b591 * , 313 ( 2000 ) .
bauer , s. fleming , and m. luke , phys .
d * 63 * , 014006 ( 2001 ) .
bauer , s. fleming , d. pirjol , and i.w .
stewart , phys .
d * 63 * , 114020 ( 2001 ) .
a. szczepaniak , e.m .
henley , and s. brodsky , phys .
b * 243 * , 287 ( 1990 ) .
m. beneke and t. feldmann , nucl . phys . *
b592 * , 3 ( 2000 ) .
bauer , d. pirjol , and i.w .
stewart , phys .
d * 67 * , 071502 ( 2003 ) ; r.j .
hill , t. becher , s.j .
lee , and m. neubert , jhep * 0407 * , 081 ( 2004 ) .
bauer , d. pirjol , i.z .
rothstein , and i.w .
stewart , phys .
d * 70 * , 054015 ( 2004 ) .
t. kurimoto , h - n .
li , and a.i .
sanda , phys .
rev d * 65 * , 014007 ( 2002 ) .
j. botts and g. sterman , nucl
b325 * , 62 ( 1989 ) .
li and g. sterman , nucl . phys . *
b381 * , 129 ( 1992 ) .
a.v . manohar and i.w .
stewart , hep - ph/0605001 .
arnesen , z. ligeti , i.z .
rothstein , i.w .
stewart , hep - ph/0607001 .
j. chay , h - n .
li , and s. mishima , in preparation .
m bander , d. silverman and a. soni , phys .
lett . * 43 * , 242 ( 1979 ) .
heavy flavor averaging group , http://www.slac.stanford.edu/xorg/hfag .
li and s. mishima , phys .
d * 73 * , 114014 ( 2006 ) .
y.y . charng and h - n .
li , phys .
d * 71 * , 014036 ( 2005 ) .
pham , hep - ph/0610063 .
m. beneke and d. yang , nucl . phys .
* b736 * , 34 ( 2006 ) . m. beneke and s. jager , hep - ph/0512101 . m. beneke and s. jager , nucl . phys . *
b751 * , 160 ( 2006 ) .
m. beneke and m. neubert , nucl . phys .
* b675 * , 333 ( 2003 ) .
chen , y.y .
keum , and h - n . li , phys .
d * 66 * , 054013 ( 2002 ) .
h - n . li and s. mishima , phys . rev .
d * 71 * , 054025 ( 2005 ) .
y. grossman , int .
a * 19 * , 907 ( 2004 ) .
yang , r.m .
wang , and g.r .
lu , phys .
d * 72 * , 015009 ( 2005 ) .
p.k . das and k.c .
yang , phys .
d * 71 * , 094002 ( 2005 ) .
c.h . chen and c.q .
geng , phys .
d * 71 * , 115004 ( 2005 ) .
huang , p. ko , x.h .
wu , and y.d .
yang , phys .
d * 73 * , 034026 ( 2006 ) .
a.l . kagan , phys .
b * 601 * , 151 ( 2004 ) ; hep - ph/0407076 .
m. beneke , j. rohrer , and d. yang , hep - ph/0612290 .
p. colangelo , f. de fazio , and t.n .
pham , phys .
b * 597 * , 291 ( 2004 ) .
m. ladisa , v. laporta , g. nardulli , and p. santorelli , phys .
d * 70 * , 114025 ( 2004 ) .
cheng , c.k .
chua , and a. soni , phys .
d * 71 * , 014030 ( 2005 ) .
w.s . hou and m. nagashima , hep - ph/0408007 .
m. beneke , j. rohrer , and d. yang , phys .
lett . * 96 * , 141801 ( 2006 ) .
the babar collaboration , b. aubert , et .
al , phys .
lett . * 97 * , 201801 ( 2006 ) .
h - n . li , phys .
b * 622 * 63 , ( 2005 ) .
a. datta et al . , arxiv:0705.3915 .
li , s. mishima , and a.i .
sanda , phys .
d * 72 * , 114005 ( 2005 ) .
buras , r. fleischer , s. recksiegel , and f. schwab , eur .
j. c * 45 * , 701 ( 2006 ) ; r. fleischer , s. recksiegel , and f. schwab , hep - ph/0702275 .
s. baek and d. london , arxiv : hep - ph/0701181 . m. gronau , arxiv:0706.2156 , talk presented at flavor physics and cp violation conference , bled , 2007 .
williamson and j. zupan , phys .
d * 74 * , 014003 ( 2006 ) .
xiao , x.f .
chen , and d.q .
guo , hep - ph/0608222 ; x.f .
chen , d.q .
guo , and z.j .
xiao , hep - ph/0701146 .
a. ali et al . , hep - ph/0703162 .
j. zupan , arxiv:0707.1323 , talk presented at flavor physics and cp violation conference , bled , 2007 .
|
we compare the theoretical frameworks and the phenomenological applications of the factorization approaches to exclusive @xmath0 meson decays , which include qcd - improved factorization , perturbative qcd , and soft - collinear effective theory .
recent progress on two - body nonleptonic @xmath0 meson decays made in these approaches are reviewed .
|
the cold dark matter ( cdm ) paradigm predicts that 5%10% of each galaxy s mass is bound up in subhalos left over from the hierarchical formation process . in the local group ,
the predicted number of dark matter subhalos significantly exceeds the observed number of dwarf galaxy satellites ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the abundance of substructure depends on competition between accretion of new subhalos from the environment and destruction of old subhalos by tidal forces ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
also , the number of subhalos that `` light up '' and become visible as satellite galaxies depends on whether subhalos are able to retain their gas against photoevaporation , and on the efficiency of galaxy formation in low - mass systems ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
measuring the amount of substructure in galaxy halos , and how it varies with galaxy mass , environment , and redshift , therefore provides unique access to the astrophysics of galaxy formation on small scales .
we still know very little about the physical properties of the dark matter particle , but a number of specific models have been proposed : dark matter could be sterile neutrinos , or supersymmetric particles , or a manifestation of extra dimensions , or even a product from the decay of any of these particles ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
all of those possibilities are compatible with observations that probe the universe on large scales .
however , they make some different predictions about the amount of dark matter substructure ; for example , any type of `` warm '' dark matter can lead to a suppression of power on small scales ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
studying galaxy substructure provides the opportunity to test such models and obtain important astrophysical evidence about the fundamental nature of dark matter .
strong gravitational lensing is a simple geometric phenomenon that gives us a valuable tool for studying mass in distant galaxies , including substructure .
lensing effects are succinctly encoded in the properties of the time delay surface .
consider light emitted by a background source at angular position @xmath2 from the center of a foreground lens with ( projected ) potential @xmath3 .
is a scaled version of the two - dimensional gravitational potential .
specifically , it satisfies the poisson equation @xmath4 where @xmath5 is the surface mass density of the lens , and @xmath6 is the critical surface density for lensing . ] if the light reaches us from the angular image position @xmath7 , the excess travel time relative to a hypothetical light ray that travels directly from the source with no deflection is @xmath8 , \quad t_0 = \frac{1+z_l}{c}\ \frac{d_l d_s}{d_{ls}}\ .\ ] ] by fermat s principle , images form at stationary points of the time delay surface , i.e. , positions @xmath7 such that @xmath9 .
( this directly yields the lens equation , usually written as @xmath10 . )
the magnifications of the images are determined by second derivatives of @xmath11 ( and hence depend on second derivatives of the lens potential @xmath3 ) .
what is usually measured is the differential time delay between two images , @xmath12 .
( see the review of strong lensing by @xcite for more details . )
strong lensing provides the only way to detect substructure directly ( i.e. , by virtue of its gravity ) in galaxies outside the local group .
small mass clumps in the lens galaxy can strongly perturb lensed images .
the spatial perturbations are determined by first derivatives of the lens potential : they have angular scales of milli - arcseconds for `` millilensing '' by dark matter subhalos ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , or micro - arcseconds for `` microlensing '' by stars in the lens galaxy ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
since lensing magnifications depend on second derivatives of the lens potential , they are even more sensitive to substructure : magnification perturbations can be of order unity and are therefore very apparent , especially in 4-image lenses ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . at optical and x - ray wavelengths quasar emission regions are small enough that lens flux ratios are sensitive to both dark matter subhalos and stars ( e.g. , * ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) ; this makes it difficult to isolate millilensing and study dark matter substructure .
by contrast , at radio wavelengths the quasar source is thought to be large enough to smooth over the effects of stars and be insensitive to microlensing ( but see * ? ? ?
the flux ratios of radio lenses ( especially 4-image , or quad , lenses ) have therefore been the tool of choice for studying millilensing .
the amount of substructure needed to explain radio lens flux ratios is broadly consistent with cdm predictions @xcite .
the greatest limitation of this method is the small number of radio quads known currently ( e.g. , * ? ? ?
ccrrrc c & m1 & @xmath13 & @xmath14 & @xmath15 & @xmath16 + a1 & m2 & @xmath17 & @xmath18 & @xmath19 & @xmath20 + a2 & s1 & @xmath21 & @xmath22 & @xmath23 & @xmath24 + b & s2 & @xmath25 & @xmath26 & @xmath27 & @xmath28 [ tab : foldimg ] in this paper we show that lens _ time delays _ provide an exciting new way to probe dark matter substructure , with several distinct advantages .
as we shall see , time delays are not affected by microlensing , so we can use optical as well as radio data to probe dark matter substructure ; this is important because there will be thousands of lenses discovered in new optical surveys ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , complementing new samples of radio lenses ( e.g. , * ? ? ?
time delays are sensitive to the mass function of dark matter subhalos in a way that flux ratios are not ; so they offer a good opportunity to probe the masses of subhalos in distant galaxies . by all indications
the theory of time delay millilensing is very tractable ; having a formal theory will provide a rigorous foundation for substructure studies , and may even allow us to do some of the statistics analytically .
appropriate time delay measurements are feasible now , and truly revolutionary datasets will become available in the foreseeable future . for the purpose of concreteness , we assume a cosmology with @xmath29 , @xmath30 , and @xmath31 km s@xmath32 mpc@xmath32 , and we adopt specific values for the lens and source redshifts ( motivated by particular lens systems , as discussed below ) . modifying the cosmology or the lens and source redshifts would simply rescale the time delays through the lensing time scale @xmath33
. ccrrrc c & m1 & @xmath34 & @xmath35 & @xmath36 & @xmath16 + b & m2 & @xmath37 & @xmath38 & @xmath39 & @xmath40 + a & s1 & @xmath41 & @xmath42 & @xmath43 & @xmath44 + d & s2 & @xmath45 & @xmath46 & @xmath47 & @xmath48 [ tab : cuspimg ]
we examine the effects of substructure by comparing the properties of images produced by a galaxy with substructure to those produced by an equivalent smooth galaxy . we model the smooth galaxy using a pseudo - jaffe profile with scaled surface mass density @xmath49 where @xmath50 is the critical surface density for lensing , where @xmath51 , @xmath52 , and @xmath53 are angular diameter distances between the observer and lens , the observer and source , and the lens and source , respectively .
we have written the equation for a circular lens , but it is straightforward to obtain an elliptical model by replacing @xmath54 with an appropriate elliptical radius .
the pseudo - jaffe model is equivalent to the singular isothermal model for @xmath55 , but the density falls more quickly at large radii @xmath56 to keep the total mass finite , @xmath57 .
the scale radius @xmath58 can therefore be thought of as a sort of truncation radius , although the truncation is not sharp .
we set @xmath58 to be 300 kpc , although the particular choice has little effect on our results provided it is well outside the einstein radius ( see section [ sec : calc ] for more discussion ) .
the parameter @xmath59 is the einstein radius of the lens .
we calibrate the pseudo - jaffe model by fitting it to observed lens systems .
the fit is not quantitatively precise because we model the lens environment using a simple external tidal shear , when in truth the lens may lie in a group of galaxies that has a more complicated effect ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . for the present work ,
we seek only to obtain reasonable smooth mass models , not to replicate any observed lens in rich detail .
we create a lens with a `` fold '' image configuration similar to that of pg 1115 + 080 @xcite .
the lens galaxy redshift is @xmath60 and the source redshift is @xmath61 , so the lensing time scale is @xmath62 days ( see equation ( [ eq : tdel ] ) ) .
our reference smooth model is circular with einstein radius @xmath63 , and it has external shear @xmath64 at position angle @xmath65 ( east of north ) .
we place a source at position @xmath66 relative to the center of the galaxy , and obtain the four images listed in table [ tab : foldimg ] and shown in figure [ fig : foldimg ] below .
we also create a lens with a `` cusp '' image configuration similar to that of rx j1131@xmath11231 @xcite .
the lens redshift is @xmath67 and the source redshift is @xmath68 , yielding a lensing time scale @xmath69 days .
our reference smooth model has einstein radius @xmath70 , ellipticity @xmath71 at position angle @xmath72 ( east of north ) , and external shear @xmath64 at position angle @xmath73 .
we place a source at position @xmath74 relative to the center of the galaxy , and obtain the four images listed in table [ tab : cuspimg ] and shown in figure [ fig : cuspimg ] . while there are quantitative differences in the time delays between the fold and cusp lens examples ,
many of the scalings and conceptual conclusions remain the same .
therefore in most of the presentation we focus on the fold lens for clarity .
the exceptions are discussed in section [ sec : ordering ] .
we should comment on our labeling of the lensed images .
images are traditionally labeled by letters ( e.g. , a , b , c , d ) using criteria chosen by the discoverers ; different criteria have been used , with the result that existing labels are not very informative .
@xcite instead advocate labeling the images 1234 in arrival - time order , arguing that this labeling is robust ( at least for smooth mass distributions ; see section [ sec : ordering ] ) , and presenting rules for using the image configuration to determine the labels .
we assert that it is useful to have labels convey not only the time ordering but also the type of image as well .
we therefore use `` m '' or `` s '' to indicate ( respectively ) an image at a local minimum or saddlepoint of the time delay surface , combined with `` 1 '' or `` 2 '' to indicate ( respectively ) the leading or trailing image of each type . thus , the four images arranged in arrival - time order are m1 , m2 , s1 , and s2 . in tables [ tab : foldimg ] and [ tab : cuspimg ] we give the correspondence between our labeling scheme and the traditional letter labels for pg 1115 + 080 and rx j1131@xmath11231 . while detailed semi - analytic substructure models are now available ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , for pedagogical purposes it is attractive to adopt simple but reasonable assumptions about the mass function and spatial distribution of subhalos ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) .-body simulations , but both the original simulations and the lensing calculations are computationally expensive @xcite . ]
we assume that subhalos trace the total mass distribution , so the average surface mass density in subhalos at radius @xmath54 is @xmath75 , where @xmath76 is the substructure mass fraction .
cdm simulations predict that the subhalo mass function is approximately a power law , @xmath77 with @xmath78 ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in order to explore how subhalo masses affect time delays , we consider masses in some finite range @xmath79 .
the mean subhalo mass is @xmath80 equivalently , we may write the lower mass limit in terms of the mean mass and dynamic range @xmath81 as @xmath82 another quantity that will prove to be useful is the mean squared mass , @xmath83 when @xmath84 the mass function reduces to a dirac @xmath85-function and all subhalos have the same mass .
figure [ fig : foldimg ] shows three realizations of a subhalo population in which the substructure mass fraction is @xmath86 and the subhalos have masses between @xmath87 and @xmath88 .
for reference , this mass function has @xmath89 and @xmath90 . for a subhalo of mass @xmath91 , it is useful to define a scaled mass @xmath92 which has dimensions of area . here
@xmath93 is the einstein radius of the subhalo if it is a point mass . in most of our analysis
we do treat subhalos as point masses for simplicity .
the lens potential of a subhalo is then @xmath94 while the deflection angle is @xmath95 we also consider subhalos modeled as truncated isothermal spheres .
the truncation occurs in the three - dimensional density profile , so we have @xmath96 out to the truncation radius @xmath97 , and @xmath98 for @xmath99 .
the lensing deflection angle works out to be @xmath100 , \nonumber\\ r > r_t:\ \alpha & = & \frac{{{\hat m}}}{\pi r}\ .\end{aligned}\ ] ] it is straightforward to obtain the potential by integrating , @xmath101 , but it is not instructive to write the ( somewhat long ) expression here . to obtain a mass model with substructure , we leave a fraction @xmath102 of the dark matter mass in a smooth component , and replace a fraction @xmath76 with subhalos .
the expected number of subhalos is @xmath103 and we draw the actual number from a poisson distribution with this mean .
the subhalo positions are assigned randomly as follows .
for the pseudo - jaffe model the cumulative probability distribution for the radius is @xmath104 we pick a random value for @xmath105 uniformly between 0 and 1 , and then invert equation ( [ eq : pofr ] ) to find the random radius .
the explicit inversion is @xmath106 ( again , these equations are for a circular lens , but the extension to an elliptical mass distribution is straightforward . ) we then pick a random azimuthal angle uniformly between 0 and @xmath107 .
the pseudo - jaffe model formally extends to infinity , but we do not actually need to consider subhalos at large radii because they have little effect on the time delays . in the appendix ,
equation ( [ eq : rthresh ] ) allows us to calculate the rms error that we make on time delays if we neglect subhalos beyond some radius @xmath108 .
we use this equation to set the threshold radius to ensure that our time delay errors are less than 0.001 days .
the fact that time delays are not very sensitive to distant subhalos is the reason that our results do not depend strongly on the choice of the pseudo - jaffe `` truncation '' radius .
the subhalo masses are assigned randomly from the power law mass function .
the cumulative probability distribution for the mass is @xmath109 we pick a random value for @xmath110 uniformly between 0 and 1 , and invert equation ( [ eq : pofm ] ) to find the random mass .
( when the mass function is a @xmath85-function , all halos have the same mass and this process is unnecessary . )
the number of subhalos in a simulation varies between a few hundred and a few hundred thousand , depending on the subhalo masses and the substructure mass fraction .
we use a tree algorithm @xcite to compute the net lensing potential , deflection , and magnification quickly and efficiently .
all of this analysis is included in an updated version of the public software package _ gravlens _
@xcite .
to find the images as they are affected by substructure , we note that the shifts in the image positions are small ( e.g. , * ? ? ?
* ) , so it is efficient to start from the smooth images and perturb the positions iteratively : @xmath111 where @xmath112 and @xmath113 are the deflection vector and magnification tensor , respectively , evaluated at the current position @xmath114 .
( here @xmath115 labels the iteration step , not the image index . )
note that @xmath116 represents the source position associated with the current image position , so @xmath117 is the offset in the source plane . when the image position correctly solves the lens equation , @xmath117 vanishes and the iteration process converges .
a massive clump near one of the `` macro - images '' can in principle split it into several `` milli - images . ''
, there is formally a `` micro - image '' very close to each clump .
however , these images are highly demagnified and hence unimportant .
they are absent when the clump central density is shallower than @xmath118 . ]
we check for this possibility by comparing the parity of the recovered image against the parity of the reference image .
if we start with a minimum but recover a saddle ( or vice versa ) , we know there are extra images and we have found one . note that if we start with a minimum and recover a minimum ( or likewise for a saddle ) , that does not prove there are no extra images .
however , we use the minimum / saddle cases to estimate the fraction of simulations that yield extra images , and find it to be small .
we throw away the small number of simulations that yield extra images . once we have found the image positions , we use equation ( [ eq : tdel ] ) to compute the differential time delay between image pairs , @xmath12 .
we begin by examining our sample fold lens in the presence of substructure , as shown in figure [ fig : foldimg ] . in the bottom row of that figure
we show sample light curves for image m2 ( assuming for pedagogical purposes a simple square wave variation in the source ) ; the results for the other images are similar .
the substructure clearly changes both the flux ratio and time delay between the images . in these examples ,
the flux ratio is perturbed by 0.10.2 mag and the time delay is perturbed by some fraction of a day .
@xcite also showed ( in less detail ) that substructure can affect lens time delays , but he focused on this as a source of noise in lensing measurements of the hubble constant .
we instead propose that time delay perturbations provide a new way to detect and study substructure itself , if lens time delays can be measured well enough .
the remainder of the paper is devoted to exploring how time delay perturbations depend on the subhalo population . to that end , we start with a simple case in which all the subhalos are point masses with @xmath119 , and the substructure mass fraction is @xmath86 . for each substructure realization
we compute the time delays between the images ; we then repeat the process @xmath120 times and make histograms of the time delays , as shown in figure [ fig : hist ] . for each image pair
, the average time delay matches the prediction of the reference smooth model very well , which makes sense because the smooth model can be thought of as an average over the subhalo population .
these three time delay histograms appear to be gaussian .
( the histogram of the time delay between the close images m2 and s1 is somewhat different , as discussed in section [ sec : ordering ] . )
we attribute the gaussianity to the central limit theorem : the lens potential is a sum of many random terms , and such a sum tends toward a gaussian distribution if the mean and variance of each term are finite , which we verify in the appendix .
the standard deviation is @xmath1210.26 days for all three image pairs shown in figure [ fig : hist ] .
next we consider the effects of changing the subhalo mass ( figure [ fig : hist - m ] ) and the substructure mass fraction ( figure [ fig : hist - f ] ) . in order to place different image pairs on a common scale ,
we plot histograms of the time delay perturbation @xmath122 , defined to be the difference between the time delay with substructure and the time delay for the reference smooth model .
all of the histograms appear to be nicely gaussian , to be centered on zero , and to have widths that increase with both the subhalo mass and the substructure mass fraction .
we can quantify this last point by plotting the time delay scatter @xmath123 as a function of the subhalo mass , for different values of the substructure mass fraction , as shown in figure [ fig : mscaling ] . in the appendix we use the central limit theorem to make an initial prediction of the scaling of the time delay scatter ( see equation ( [ eq : sigf2 ] ) ) : @xmath124 to test this prediction , we use the results from simulations with @xmath86 to fit for the proportionality constant , and then plot the analytic curves alongside the simulation results in figure [ fig : mscaling ] .
the agreement is striking given the simplicity of the analytic prediction .
strictly speaking , the analytic scaling appears to overestimate the time delay scatter when the subhalo mass is large .
we can identify at least two possible explanations .
first , when @xmath76 is fixed increasing the subhalo mass decreases the number of subhalos , which may make the central limit theorem less applicable and invalidate an approximation in equation ( [ eq : sigphi ] ) .
second , as the subhalo mass increases the perturbations to the image positions also increase , which may invalidate one of our starting assumptions in the appendix .
we are working on a more sophisticated theory of time delay perturbations to address both issues .
for now , though , we consider the success of the initial analytic prediction to be very encouraging .
the scaling of the time delay scatter with subhalo mass has an important corollary : there is no measurable effect on time delays from low - mass objects , such as stars .
this means we can study millilensing at any wavelength without `` contamination '' from microlensing .
so far we have assumed that all subhalos have the same mass ; now we consider a power law mass function .
figure [ fig : mfunc ] shows the time delay scatter as a function of the dynamic range @xmath81 and power law slope @xmath125 of the mass function .
the dynamic range has a significant effect : the broader the mass function , the more scatter there is in lens time delays . the power law slope affects the time delay scatter as well , but much more modestly . for comparison we also plot the analytic scaling from equation ( [ eq : sigf1 ] ) in the appendix , @xmath126 where the ratio @xmath127 can be found from equation ( [ eq : msq ] ) .
the analytic formula does not match the simulation results perfectly , but it seems remarkably good for something so simple .
we take this as an indication that the theory of time delay millilensing will be analytically tractable . having a rigorous theory of millilensing is possible only with time delays ;
analytic results for flux ratio millilensing are elusive because flux ratios are highly nonlinear and the central limit theorem does not apply ( see the appendix ) .
so far we have treated the mass clumps as point masses , partly for simplicity and partly because we know the gravity outside any spherical mass clump is the same as that of a point mass .
however , dark matter subhalos in general have some spatial extent , and if they overlap lensed images this may be important for millilensing . in this pilot study
we mainly want to determine when the point mass approximation is reasonable and when we need to worry about subhalo size and structure .
we therefore consider a simple model of isothermal spheres that are truncated ( in three dimensions ) at some radius @xmath97 .
figure [ fig : tsis ] shows how the time delay scatter changes when we vary the truncation radius while keeping the subhalo mass fixed .
( the key scale is the dimensionless truncation radius , @xmath128 , where @xmath93 is the einstein radius of a point mass with the same mass , as defined by equation ( [ eq : mhat ] ) . )
depending on the mass , subhalos need to extend beyond 10 or more einstein radii before the internal structure becomes important .
the obvious next step is to model tidal truncation and estimate the sizes of realistic subhalos .
there are some important effects to consider : tidal forces vary with position in the parent halo ; and tidal truncation also corresponds to mass loss ( so , strictly speaking , it is not right to vary the truncation radius while keeping the subhalo mass fixed , although that is a useful pedagogical exercise ) .
these effects are incorporated into semi - analytic substructure models , so in follow - up work we will study time delay millilensing using realistic subhalo populations drawn from those models .
@xcite suggested that the arrival - time ordering of lensed images is a robust prediction of smooth lens models ( at least , those with relatively simple angular structure ) .
with the help of a few simple rules , they noted , it is usually easy to identify the two minima and two saddlepoints in a quad lens , and then to deduce the arrival - time ordering . if the situation is not immediately obvious , a simple lens model makes it clear . we have therefore chosen to use the image classification and ordering as the basis of our labeling scheme : in arrival - time order we have the leading minimum ( m1 ) , the trailing minimum ( m2 ) , then the leading saddle ( s1 ) , and finally the trailing saddle ( s2 ) . in the fold image configuration we have considered , substructure does not affect the arrival - time ordering .
the reason is that the time delay perturbations are small compared with most of the time delays themselves .
it might seem that the close image pair m2/s1 could provide an exception , because the smooth model time delay ( 0.16 day ) is shorter than some of the substructure perturbations we have seen .
however , by definition a saddlepoint in the time delay surface must be higher than a nearby local minimum , so image s1 must trail m2 and the overall ordering must be preserved .
one consequence is that the histogram for the m2/s1 time delay can not remain gaussian as the substructure perturbations grow , as shown in figure [ fig : histbc ] .
this implies that the substructure influences both images at some level , although the @xmath129 histograms are broad enough that it seems fair to say that substructure affects even nearby images quite differently .
the only way this general analysis can be violated is if a massive and concentrated clump lies close enough to images m2 and s1 to change the global topology of the time delay surface ; but such a situation would be immediately apparent from the image configuration ( see * ? ? ?
* ) .
the issue of arrival - time ordering is strikingly different for a cusp image configuration like that in rx j1131@xmath11231 .
figure [ fig : cuspimg](a ) shows the images along with selected arrival - time contours for our smooth mass model . in this example , and indeed in all smooth models of rx j1131@xmath11231 examined by @xcite , the image labeled m1 is the leading image .
however , as shown in figure [ fig : cuspimg](b ) , substructure can reverse the arrival - time order of the images m1 and m2 , such that the overall ordering is m2/m1/s1/s2 . in terms of the topology of the time delay surface
, we can still say the saddlepoint must be higher than the two minima on either side , so s1 must trail both m1 and m2 . however , there are no such restrictions on the heights ( or depths , equivalently ) of the two minima with respect to each other .
it is fairly easy for substructure to modify one or both of the minima enough to change their relative heights , and hence the arrival - time ordering of images m1 and m2 .
in fact , some 27% of realizations with modest substructure ( substructure mass fraction @xmath86 and subhalo mass @xmath130 ) predict an arrival - time reversal ( see figure [ fig : hist1131 ] ) .
we believe image reversal by substructure may already have been observed in rx j1131@xmath11231 , based on the time delays reported by @xcite .
the time ordering observed among the three bright images is m2/m1/s1 with @xmath131 days , @xmath132 days , and @xmath133 days .
these time delays present two puzzles .
first , image m2 is observed to lead image m1 , whereas smooth models predict the opposite .
second , the observed m2/s1 and m1/s1 time delays are an order of magnitude longer than predicted by smooth models .
@xcite focused their attention on the second problem , and showed that placing a single , massive ( @xmath134 ) subhalo near image s1 could explain the long m2/s1 time delay .
their models still predicted m1 to be the leading image , though .
as we have demonstrated , a _ population _ of subhalos might be able to reverse the ordering to make m2 the leading image .
it will still take a lot of modeling to draw quantitative conclusions about the substructure required to explain the image ordering and the long time delays ; we will present the modeling details elsewhere ( c. r. keeton & l. a. moustakas , in preparation ) . for
now , we mainly want to introduce the idea that the temporal ordering of images in a cusp configuration can be changed by substructure and that this effect may already have been observed .
having shown that substructure affects lens time delays , we still need to consider how well the perturbations could be detected in observed lenses . the question is whether uncertainties in the `` macromodel '' used to characterize the smooth component of the lens mass distribution could mask the effects of substructure .
one key concern is the radial profile degeneracy : especially in quad lenses , varying the radial density profile of the lens galaxy can rescale the time delays while leaving other observables unchanged ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . at first glance
, this would seem to make changes induced by substructure degenerate with changes in the radial profile .
however , the radial profile rescaling affects all images in the same way , while substructure affects each image differently .
so with more than one time delay ( as in a quad lens ) , ratios of the delays cancel the rescaling and isolate substructure effects , as shown in figure [ fig : alpha ]
. the radial profile degeneracy will therefore not be a fundamental limitation in attempts to probe substructure with time delays .
this very promising inference is currently empirical , but we expect it can be made rigorous with a full theory of time delay millilensing . since radial profile effects
do not cancel perfectly ( there are small differences in the widths of the time delay ratio histograms in figure [ fig : alpha ] ) , the theory will also be useful in understanding and correcting for these residual effects .
another source of uncertainty is the angular structure of the macromodel . while it is common to fit
observed lenses with an ellipsoidal mass distribution and external shear ( and we have used such a model to construct our mock lenses ) , real galaxies may be more complex . in particular , observed ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and simulated ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) elliptical galaxies often contain mild departures from elliptical symmetry that can be described with @xmath135 and @xmath136 multipole terms in the surface mass density .
@xcite suggested that such multipole terms might provide an alternative to clumpy substructure when fitting observed lenses .
although multipoles turn out not to provide a satisfactory explanation of observed flux ratio anomalies @xcite , it is important to consider how they might influence the analysis of time delays .
we can do this by taking our mock lenses generated with substructure and trying to fit them with multipole models . for this exercise
we adopt @xmath137 errorbars for the image positions , which is typical of observations with the _ hubble space telescope ( hst ) _ and radio interferometry , and 5% time delay uncertainties , which is similar to the precision achieved for quad lenses today ( see the compilation by * ? ? ?
* ) . in general
, we find that multipole models can formally provide a good fit to the image positions and time delays , but many of the models have unreasonably large multipole terms that lead to unrealistic galaxy shapes , as shown in figure [ fig : mpole ] . to avoid the unreasonable models we can impose some priors on the multipole terms . since the multipole moments in observed and simulated elliptical galaxies are typically at the percent level , a simple first step is to adopt gaussian priors on the multipole amplitudes @xmath138 and @xmath139 with a mean of zero and standard deviation of 0.02 .
( in the present analysis we do not impose any constraints on the orientations of the multipole terms . ) with these mild amplitude priors , multipole models fail ( i.e. , they are inconsistent with the data at more than 95% confidence ) for 27 of the 100 mock lenses we examine .
we are in the process ( j. van saders et al . , in preparation ) of deriving more sophsticated priors on multipole terms from a large sample of galaxies in the galaxy evolution from morphologies and seds survey ( gems ; * ? ? ?
* ) , and we expect that those will be even more valuable in constraining macromodels and thus aiding the identification of time delay anomalies .
it is perhaps artificial to insist that we must identify anomalous lenses on the basis of time delays alone , when flux ratios are also available to provide additional constraints .
if we fit the flux ratios along with the image positions and time delays , we find that multipole models fail ( at 95% confidence ) for 52 of 100 mock lenses even if we use generous 10% flux errorbars and do not impose any priors on the multipole terms .
interestingly , only 28 of those failures would be identified on the basis of flux ratios alone ; in 24 of the mock lenses , it is only when we fit the time delays and flux ratios simultaneously that the anomalies become clear .
we infer that time delays and flux ratios contain different and complementary information about the small - scale structure of the lens potential .
while we plan to address the issues of multipole priors and complementarity of time delays and flux ratios in detail in future work , we offer the preliminary conclusion that macromodel uncertainties can be controlled well enough that we can expect to use time delays to learn about substructure .
we have shown that substructure in lens galaxies modifies the time delays in multiply - imaged gravitational lenses .
the amplitude of time delay perturbations depends on the abundance of substructure , the mass function of subhalos , and to some extent the internal structure of subhalos as well .
this phenomenon , which we call time delay millilensing , has some very attractive properties .
first , the effects of substructure on time delays are fairly easy to understand , both conceptually and quantitatively .
second , time delay perturbations are unaffected by stellar microlensing or extinction in the lens galaxy .
third , time delay ratios are immune to the radial profile degeneracy in lens modeling . furthermore , time delay millilensing is sensitive to the mass function of subhalos in a different way than flux ratio and astrometric millilensing . for point
mass perturbers , the lensing optical depth depends on @xmath140 .
the potential of a clump scales with its mass , so time delay perturbations depend in full on @xmath141 .
( see the appendix for more details . ) for comparison , position perturbations have a scale of @xmath142 while magnification perturbations are dimensionless , which means that astrometric millilensing depends on @xmath143 while flux ratio millilensing depends on @xmath144 ( compare * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we therefore suggest that lens time delays should join image positions and flux ratios as tools for studying dark matter substructure in distant galaxies .
since the three observables measure different moments of the subhalo mass function , the combination of all three will ultimately provide a powerful way to test cdm predictions and even to explore the nature of dark matter itself , given that different dark matter candidates may produce different mass functions at subgalactic scales , as a function of lens galaxy mass and redshift . before we try to use time delays to measure substructure ,
it is vital that we understand whether systematic uncertainties in lens models could corrupt substructure constraints .
we have explicitly shown that the radial profile degeneracy is not a significant problem , as time delay ratios essentially factor out the profile dependence .
we have drawn the preliminary conclusion that uncertainties in the angular structure of the lens potential can be dealt with , using additional flux ratios data and/or independent constraints on the shapes of galaxy mass distributions .
other potential concerns may include the environment of the lens galaxy @xcite , and ( sub)structure along the line of sight @xcite .
for all of these issues , the important point is again that non - local features in the lens potential affect the images in some coordinated way , whereas substructure affects the images differently .
we also need to predict time delay perturbations for subhalo populations that are more realistic than the simple ones we have used ( for pedagogical purposes ) here .
this work is underway , using existing semi - analytic substructure models . by incorporating important physical effects such as accretion of new subhalos and tidal truncation and disruption of old subhalos ,
these models yield subhalo populations whose spatial distributions and mass functions agree well with results from @xmath145-body simulations ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
our follow - up calculations will offer greater insight into the subtleties of time delay perturbations in a broader range of realistic circumstances , and better guidance as to how well we need to measure time delays if we want to probe dark matter substructure .
we have noted that in extraordinary lenses with close triplets of images in a cusp configuration , it may be possible to detect the effects of substructure through unambiguous changes in the arrival - time ordering of the images ( compared to expectations for standard macromodels ) .
we expect , though , that the real power of time delay millilensing will lie in its application to large ensembles of lenses with measured time delays . at present , time delay uncertainties are typically at the level of 1 day ( see @xcite for a compilation )
. improvements of a factor of a few are desired : our preliminary estimates indicate that improving the uncertainties to @xmath1460.5% would yield constraints on substructure at the level of 0.30.5 dex in a given 4-image lens . @xcite have shown that coordinated global optical monitoring can yield time delays with uncertainties of less than 0.1 day .
monitoring at x - ray wavelengths , where variability can be rapid and dramatic , can yield even more precise time delays : @xcite measure a time delay of @xmath147 hours between the two close images in pg 1115 + 080 with _
chandra _ and _ xmm - newton _ observations . with some effort , such techniques
can probably be extended to a modest sample of time delays in the near future . over the longer term
, the advent of large time - domain surveys is expected to yield thousands of lensed quasars with time delays ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
even better would be a dedicated space platform capable of precise monitoring of lenses over extended periods of time ( see * ? ? ?
looking to the future , measuring precise image positions , flux ratios , and time delays in a sample of @xmath146100 quad lenses could yield substructure measurements at the 20%30% level or better .
since the measurements will span the redshift range @xmath148 , this will provide a unique opportunity to study not just the amount of dark matter substructure but its evolution as well .
we thank michael kuhlen , george meylan , arlie petters , and eduardo rozo for helpful conversations .
we thank masamune oguri and the referee neal dalal for raising the issue of modeling uncertainties associated with multipole terms .
crk thanks jonathan faiwiszewski for suggesting that the central limit theoreom might be helpful .
crk is supported by grant hst - ar-10668 from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 ; and by nsf grant ast-0747311 .
the work of lam was carried out at jet propulsion laboratory , california institute of technology , under a contract with nasa .
in this appendix we present an initial theory of the statistics of time delay perturbations .
this theory rests on three assumptions : ( 1 ) subhalos are statistically independent ( of each other ; the probability of having a subhalo may still depend on position ) ; ( 2 ) subhalos may be treated as point masses ; and ( 3 ) time delay perturbations are dominated by changes in the lens potential ( as opposed to changes in the image positions ) .
the first assumption means that all subhalo positions and masses are drawn from the same probability distribution .
this simplifies the theory , and seems reasonable for lensing since we work in projection and subhalos that are near each other on the sky are most likely well separated in space .
the second and third assumptions will be relaxed in future work , but are adequate for our present goal of understanding in a general way how the time delay scatter depends on the amount of substructure and the subhalo mass function .
the lens time delay has the form given by equation ( [ eq : tdel ] ) . by assumption # 3 , if we want to understand how substructure affects @xmath149 , it is sufficient to study the potential difference @xmath150 .
any change in the time delay is simply given by the change in the potential difference , multiplied by the time scale @xmath33 .
the potential due to a collection of @xmath145 subhalos is the sum of contributions from individual subhalos .
if the subhalos are point masses , the potential difference has the form @xmath151 if we can show that the mean and variance of @xmath152 are finite , then we can use the central limit theorem to infer that the sum @xmath153 will be approximately gaussian , and to determine the variance of @xmath153 . in order to compute the mean and variance of @xmath152
, we need to average over a subhalo s position and mass .
let @xmath154 be the probability distribution for the subhalo s position , while @xmath155 is the probability distribution for its mass .
then the average of any function @xmath156 has the form @xmath157 it is useful to note that smoothed substrucure density distribution , @xmath158 , may be written in terms of the position probability distribution : @xmath159 now in order to determine the mean and variance of @xmath152 , we need to compute the moments @xmath160 in both cases we used equation ( [ eq : ks ] ) to substitute for @xmath154 in terms of @xmath161 .
we also identified the mass integrals as yielding the averages @xmath162 and @xmath163 .
the variance is then @xmath164 at the second step , we note that @xmath165 while @xmath166 , so when there are many subhalos we may neglect the second term .
the integrands in equations ( [ eq : phi1int ] ) and ( [ eq : phi2int ] ) diverge near @xmath167 and @xmath168 , but only logarithmically , and such a divergence is still integrable . as a result , as long as @xmath158 does not diverge badly near the image positions , the integrals are well behaved , so the mean and variance of @xmath152 are finite .
this means the central limit theorem holds , and we can infer that the potential difference @xmath153 , and hence the time delay @xmath169 , will have a distribution that is approximately gaussian .
( this argument also applies to clumps that are not point masses , because then any divergence in @xmath152 will be softer than logarithmic . )
furthermore , thanks to the central limit theorem we may compute the variance in @xmath153 as the quadrature sum of the variances of all the individual @xmath152 terms or , in this case , a simple multiplication by @xmath145 : @xmath170 and of course @xmath171 . heuristically , we may summarize this result as @xmath172 where @xmath161 is indicative of the amount of substructure .
while this expression oversimplifies the dependence on the spatial distribution of subhalos , it is conceptually instructive . in our simulations , we write @xmath173 , and we keep the total mass distribution @xmath174 fixed while we change the substructure mass fraction @xmath76 . therefore it is formally correct to write @xmath175 this is our general conclusion for how the time delay scatter depends on the amount of substructure and the subhalo mass function .
note that if all subhalos have the same mass @xmath91 , this simplifies to @xmath176 it is instructive to consider whether the central limit theorem can be applied to flux ratios as well as time delays . for point
mass clumps , the magnification of an image at position @xmath7 has the form @xmath177^{-1 } , \ ] ] where @xmath178 and there are similar expressions for @xmath179 and @xmath180 .
now when we try to compute the mean and variance of each term by integrating over @xmath181 , the integrand diverges non - integrably near @xmath7 , and so the mean and variance diverge ( also see * ? ? ?
the central limit theorem therefore does not apply .
the magnification probability distribution @xmath182 may still be perfectly well defined , but it need not be gaussian .
indeed , @xmath182 can be computed analytically for a uniform distribution of point masses , and it is distinctly non - gaussian ( see 11.2 of * ? ? ?
. there may be circumstances ( say , certain types of spatially extended clumps ) for which the central limit theorem does apply to magnifications , but the theorem is not universally applicable as it is for time delays . returning to time delays ,
there is one last bit of theory that is useful .
we intuitively expect that clumps far from the images have little effect on the time delay , but we can quantify this statement .
consider for simplicity images at @xmath183 and @xmath184 , and let us examine only subhalos at radii @xmath185 .
we return to equation ( [ eq : siggen ] ) , plug in for @xmath167 and @xmath168 , and then take a taylor series expansion to lowest order in @xmath186 : @xmath187 notice that we have changed notation and written @xmath188 to highlight that this is an error term specifically , the rms error we make by neglecting subhalos beyond @xmath108 .
we have computed the error in the potential , but we can relate it to the rms error in the time delay via @xmath189 .
( we have verified equation ( [ eq : rthresh ] ) using direct simulations of subhalo populations extending to large radii . )
we have assumed for simplicity that the substructure mass distribution is circularly symmetric , @xmath190 , but the generalization to elliptical symmetry is straightforward . for any reasonable substructure distribution , @xmath191 decreases as @xmath192 , so the integral converges . not only that , but @xmath193 is a monotonically decreasing function of @xmath108 . in other words ,
the farther the subhalos are from the images , the less effect they have on the time delays .
if we can tolerate some small error in time delays , we may neglect all subhalos beyond some threshold radius @xmath108 .
once we specify the time delay error tolerance , we can solve equation ( [ eq : rthresh ] ) to compute the threshold radius @xmath108 .
chartas , g. , dai , x. , & garmire , g. p. , 2004
, in carnegie astrophys .
2 , measuring and modeling the universe , ed . w. l. freedman ( pasadena , ca : carnegie observatories ) ; http://www.ociw.edu/ociw/symposia/series /symposium2/proceedings.html
kochanek , c. s. , dai , x. , morgan , c. , morgan , n. , poindexter , s. , & chartas , g. , 2007 , in asp conf . ser .
43 , statistical challenges in modern astronomy iv , ed .
g. j. babu & e. d. feigelson ( san francisco , ca : asp ) , 43 ; arxiv : astro - ph/0609112 kochanek , c. s. , schneider , p. , & wambsganss , j. , 2004 , part 2 of gravitational lensing : strong , weak & micro , proceedings of the 33rd saas - fee advanced course , ed .
g. meylan , p. jetzer , & p. north ( berlin : springer - verlag ) ; arxiv : astro - ph/0407232
|
we show that dark matter substructure in galaxy - scale halos perturbs the time delays between images in strong gravitational lens systems .
the variance of the effect depends on the subhalo mass function , scaling as the product of the substructure mass fraction and a characteristic mass of subhalos ( namely @xmath0 ) .
time delay perturbations therefore complement gravitational lens flux ratio anomalies and astrometric perturbations by measuring a different moment of the subhalo mass function . unlike flux ratio anomalies ,
`` time delay millilensing '' is unaffected by dust extinction or stellar microlensing in the lens galaxy .
furthermore , we show that time delay _ ratios _ are immune to the radial profile degeneracy that usually plagues lens modeling .
we lay out a mathematical theory of time delay perturbations and find it to be tractable and attractive .
we predict that in `` cusp '' lenses with close triplets of images , substructure may change the arrival - time order of the images ( compared with smooth models ) .
we discuss the possibility that this effect has already been observed in rx j1131@xmath11231 .
|
disorder has a profound effect on transport in quantum systems .
one of the most striking manifestations of this statement is anderson localization @xcite , where an increasing amount of disorder can lead to a phase transition from a metallic to an insulating state , even in the seemingly simple case of noninteracting particles .
this phenomenon , first predicted over fifty years ago , has been observed directly in a number of physical systems in recent years @xcite .
another subject where there has been extensive work in recent years is the problem of quantum state transfer along one - dimensional lattices or spin chains @xcite . in this problem
the goal is to transport an unknown quantum state from one end of a spin chain to the opposite end of the chain .
previous studies have for example considered the probabilistic state transfer in a chain with uniform parameters @xcite , perfect state transfer with properly designed time - independent system parameters @xcite , perfect state transfer using externally applied time - dependent control fields @xcite and quantum state routers in branched spin chains @xcite .
one question that has not been investigated in much detail in the literature is the effect of disorder on one s ability to perform perfect state transfer @xcite .
one can say that there are some simple limiting cases and trickier , intermediate cases in relation to this question . if , for example , one has full control on single - spin parameters but there is disorder in the inter - spin coupling strengths , one is able to perform perfect state transfer by using the large amount of control resources to counter the adverse effects of the disorder ( e.g. by performing a sequence of two - spin swap operations while decoupling the rest of the chain during each such two - spin operation ) . even in this case , where intuition tells us that perfect state transfer must be possible ( and does not involve exponential scaling of resources ) , there remains the practically important question of the minimum time required for perfect state transfer . on an opposite extreme , if one uses a protocol that relies on static hamiltonian settings , e.g. as is done in ref .
@xcite , then one would intuitively infer that disorder will induce anderson localization and state transfer will not be possible for long chains .
an intermediate case is that where one has full control over the parameters at the two opposite ends of the chain , with disorder present throughout the chain . in this case
there are spatial limitations on the control resources , but there are no temporal limitations
. the question then arises whether this high degree of control in the time domain will for example allow one to properly launch a spin wave that carries the quantum state and is able to propagate through the disordered landscape of the spin chain without suffering anderson localization . in order to appreciate this question , one could note here that the problem of anderson localization is usually studied in a control - free setting where the wave packet is assumed to be a generic one by some definition and in particular not prepared for the specific disorder landscape that it is going to encounter . as a result ,
it is difficult to take intuition accumulated from past studies on anderson localization and use it to predict the effect of disorder on systems that are subjected to driving by control signals . in this paper we present results of numerical simulations on the propagation and transfer of quantum states through a disordered chain with and without control . in the case of controlled state transfer
, we use optimal control theory to determine the control signals that maximize the transfer fidelity .
we find that the control signals can help suppress the effects of small amounts of disorder and enable fast perfect state transfer up to a reasonably large disorder strength , even with a limited amount of control . for large disorder strengths
, however , anderson localization physics eventually sets in and the controlled transfer can not be improved beyond the sequential nearest - neighbor transfer protocol , and in the case of limited control resources the transfer time grows indefinitely . for intermediate values of the disorder strength
, we find large variations in the transfer fidelity for different disorder instances , indicating that different disorder patterns can have drastically different localizing effects , even for the same disorder strength .
the remainder of this paper is organized as follows : in sec .
[ sec : setup ] we describe the model system and formulate the problem to be solved . in sec .
[ sec : results ] we present the results of our numerical simulations and discuss their implications . section [ sec : conclusion ] contains concluding remarks .
we consider a spin chain described by the hamiltonian : @xmath0 where @xmath1 are the on - site energies , @xmath2 are the ( nearest - neighbor ) inter - spin coupling strengths and @xmath3 ( with @xmath4 or @xmath5 ) are the usual pauli operators for the spin at site @xmath6 .
disorder can be introduced into the problem through @xmath1 and/or @xmath2 , and anderson localization occurs in both cases , up to minor quantitative differences .
we shall assume that the control fields couple to single - spin operators , in particular assuming that some or all @xmath1 are tunable parameters .
we therefore focus on the case where disorder enters the problem through @xmath2 .
we use a uniform distribution for @xmath2 , i.e. for a disorder strength @xmath7 the values of @xmath2 are distributed uniformly between @xmath8 and @xmath9 . as our main setup for discussions
, we shall assume that the on - site energies at the beginning and end of the chain , i.e. @xmath10 and @xmath11 , are control parameters that can be adjusted without any constraints , while all other @xmath1 are fixed and uniform , i.e. @xmath12 .
the system is illustrated in fig .
[ fig : schematicdiagram ] .
spins , nearest - neighbor inter - spin coupling strengths @xmath2 and control fields applied to the first and last spins in the chain . in this system ,
the on - site energies are equal throughout the lattice , except for the edge spins , where the on - site energies @xmath10 and @xmath11 can be tuned via the control fields.,width=321 ] the objective of the control protocol is to transfer a quantum state from the first site to the last site of the chain .
the chain is therefore assumed to initially be in the state @xmath13 and the desired final state is @xmath14 where the indices label the different sites in the chain .
the ground state is used in eq .
( [ eq : psiinitial ] ) because it is typically easy to prepare .
the state @xmath15 , which can be expressed as @xmath16 , is treated as an unknown quantum superposition of the ground and excited states .
there are a number of different scenarios where the assumption of an unknown state arises .
for example , the sender could have received the state @xmath15 as a bit of quantum information from another source or obtained it as the output of a quantum information protocol , and the sender is therefore unable to know the quantum state without disturbing it .
note that the state could be entangled with other parts of the system , and a protocol that transforms eq .
( [ eq : psiinitial ] ) into eq .
( [ eq : psifinal ] ) will similarly transfer any entanglement associated with the quantum state .
the state could also be known to the sender , e.g. the state @xmath17 or @xmath18 along an axis that the sender chooses but does not reveal to anybody else until a later point in the protocol , and in order to protect this information against eavesdropping any part of the state transfer protocol outside of the sender s location must be designed to work for an arbitrary quantum state . if we start with an unknown quantum state at the first site with the rest of the chain prepared in its ground state , i.e. as in eq .
( [ eq : psiinitial ] ) , the subsequent dynamics will in general involve complex dynamics in the @xmath19-dimensional hilbert space of the @xmath20-site chain .
since this problem becomes intractable for chains of length @xmath21 , we assume , as is commonly done in this context , that the parameters are in the regime where the rotating - wave approximation is valid .
in other words , we assume that @xmath2 are much smaller than @xmath1 , such that to a good approximation we can ignore terms in the hamiltonian given in eq .
( [ eq : spinchainhamiltonian ] ) that mix states with different values of @xmath22 , and we can use the approximate hamiltonian @xmath23 , \label{eq : spinchainhamiltonianrwa}\ ] ] where @xmath24 are the spin raising and lowering operators at site @xmath6 . under this approximation , the number of excitations in the system becomes a conserved quantity . as a result , and since the global ground state @xmath25 ( with @xmath26 being the ground state of the single - site hamiltonian at site @xmath6 ) does not evolve except for acquiring a simple phase , the problem of transferring an unknown single - spin quantum state from the first to the last spin ( with all other spins starting and ending up in their ground states ) can be simplified to the problem of transferring a single excitation from the first to the last site in the chain , i.e. transforming the state @xmath27 into the state @xmath28 .
the excitation that is transferred can alternatively be thought of as a particle . in other words
, one has a natural mapping between the spin chain containing a single excitation and a system containing a single particle hopping between lattice sites . in both cases the system dynamics
is well described using the picture of wave propagation . the hilbert space required in order to study these problems is of size @xmath20 only .
the simplification of the problem from a @xmath19-dimensional to an @xmath20-dimensional hilbert space allows us to investigate relatively long chains , well above the limit of @xmath21 when dealing with @xmath19-dimensional hilbert spaces .
since the output of the protocol is the state at site @xmath20 , we in principle evaluate the performance of the protocol based on how close the state of site @xmath20 at the final time ( after tracing out the state of all the other sites ) is to the state @xmath15 , averaged over all possible single - spin states @xmath15 . as mentioned above , however , under the rotating - wave approximation the transfer fidelity
can also be evaluated through the probability @xmath29 that the initial state @xmath30 evolves to the state @xmath31 at the final time .
this probability can be expressed as @xmath32 as mentioned above , in our main simulations we shall assume that @xmath10 and @xmath11 are tunable parameters , with all the other @xmath1 fixed . in this case
@xmath10 and @xmath11 can be treated as control parameters , and we can use optimal control theory techniques to identify the control signals that maximize the transfer fidelity .
we use the grape algorithm @xcite to obtain the optimal control pulses @xcite . in each such calculation , we search for the control pulse that maximizes the probability @xmath29 .
the idea is to perform these calculations for a few different values of the pulse time .
the probability @xmath29 increases from zero at very short pulse times to values that are essentially equal to unity for pulse times that are equal to or larger than the minimum perfect - transfer time @xcite .
we shall refer to this time as @xmath33 . in all of the calculations presented here , we consider chains containing fifty sites .
we divide the total pulse time @xmath34 into 200 time steps @xcite .
we perform the grape algorithm with up to 4000 iterations , although in most cases the fidelity saturated and stopped increasing within the first few hundred iterations @xcite .
it should be noted that each pulse that we find using this technique is optimized for the particular set of parameters in a given instance of disorder and will not be optimal for other instances with the same disorder strength .
these instance - specific optimized pulses are relevant in cases where the disorder is static , e.g. disorder related to defects that appear during the fabrication of a device but do not change in time .
it is crucial to note in this context that the coupling strengths in the chain can be measured using access to only the ends of the chain @xcite , such that it is not unrealistic to assume that one has access to only the edges of the chain but is nevertheless able to have knowledge about the values of @xmath2 for the entire chain .
it is also crucial to note that the optimized pulses are independent of the quantum state that is transferred across the chain .
a natural reference point that is useful to set at this point is the time that it takes to transfer a state from one end of the chain to the opposite end using a sequence of swap operations , one step at a time .
each such swap operation takes @xmath35 , such that the transfer across the entire chain of length @xmath20 takes a total time of @xmath36 in the absence of disorder and with full control over all @xmath1 , one can perform the state transfer more quickly by tuning the parameters in such a way that one launches a spin wave that traverses the chain at the maximum possible wave speed ( keeping in mind that in such discrete systems there is a maximum propagation speed set by @xmath37 ) , leading to the result that @xmath38 in the limit of an infinitely long chain ( see e.g. ref .
as mentioned in sec . [ sec : setup ] , the probability @xmath29 of successful transfer of a single excitation across the chain is a good measure of the transfer fidelity for an unknown quantum state , and we shall use this probability in our analysis below . more specifically , we assume that at the initial time an excitation is localized at the first site in the chain , and we investigate the probability that the excitation will be found at the last site in the chain at the end of a given pulse time . in fig .
[ fig : probabilityasfunctionoftime ] we plot @xmath29 as a function of pulse time @xmath34 for a few different instances of disorder with varying disorder strength in the case where control fields are applied to only the first and last sites in the chain . as would be expected
, @xmath29 increases from zero at small values of @xmath34 to unity at a value that defines the minimum perfect - transfer time @xmath33 , and it remains equal to unity for larger values of @xmath34 @xcite . for @xmath39 , where the disorder is weak , @xmath29 jumps from almost zero at pulse time @xmath40 to almost unity at @xmath41 , which means that @xmath33 is close to @xmath42 , the limiting value in the case of vanishing disorder strength and full control over the on - site energies .
even at @xmath43 , the disorder does not seem to cause much slowing down of the transfer and the different instances of disorder do not lead to large variations in @xmath33 . only one out of five disorder instances in this case ( magenta triangles ) seems to give a transfer probability that is consistently higher than those in the other four instances , and the difference is rather small .
it should be noted , however , that @xmath33 is now close to @xmath44 , which is a factor of @xmath45 slower than that obtained in the absence of disorder .
a more dramatic change occurs when we increase the disorder strength @xmath7 to 0.5 . in three out of five disorder instances
, @xmath33 is still around @xmath44 , but in one instance there is a slowing down by a factor of two and in another instance the transfer seems to slow down by at least an order of magnitude .
these large variations can be intuitively understood by keeping in mind that some instances of the disorder will contain potential barrier landscapes that are more localizing than others . as a function of pulse time @xmath34 for a fifty - site chain and control over the two edge sites . the time is measured relative to the sequential swap transfer time @xmath44 .
the different panels correspond to different values of the disorder strength : @xmath39 ( a ) , 0.3 ( b ) and 0.5 ( c ) .
there are five sets of data points in each panel .
these correspond to five different instances of the disordered coupling strengths that we generate for each value of coupling strength.,title="fig:",width=302 ] as a function of pulse time @xmath34 for a fifty - site chain and control over the two edge sites .
the time is measured relative to the sequential swap transfer time @xmath44 .
the different panels correspond to different values of the disorder strength : @xmath39 ( a ) , 0.3 ( b ) and 0.5 ( c ) .
there are five sets of data points in each panel .
these correspond to five different instances of the disordered coupling strengths that we generate for each value of coupling strength.,title="fig:",width=302 ] as a function of pulse time @xmath34 for a fifty - site chain and control over the two edge sites .
the time is measured relative to the sequential swap transfer time @xmath44 .
the different panels correspond to different values of the disorder strength : @xmath39 ( a ) , 0.3 ( b ) and 0.5 ( c ) .
there are five sets of data points in each panel .
these correspond to five different instances of the disordered coupling strengths that we generate for each value of coupling strength.,title="fig:",width=302 ] as a function of disorder strength @xmath7 for a fifty - site chain , a pulse time @xmath46 and control over the two edge sites .
the ten different data points for each value of @xmath7 correspond to ten different instances of the randomly generated coupling strengths .
the different colors are used in order to help resolve closely spaced points but have no physical significance .
the dashed line shows the mean value of @xmath29 averaged over the ten different instances at each value of @xmath7.,width=321 ] from fig .
[ fig : probabilityasfunctionoftime ] we can see that starting from @xmath47 the transfer probability @xmath29 increases steadily until it reaches a value very close to unity . furthermore , in instances where the disorder is not causing a serious increase in @xmath33 , @xmath29 has values close to unity at @xmath48 .
we therefore perform further calculations focusing on this value of @xmath34 as a way to get a simple but rather reliable indicator for @xmath33 . in fig .
[ fig : probabilityasfunctionofdisorderstrength ] we plot @xmath29 as a function of disorder strength @xmath7 for @xmath48 . up to @xmath49
, @xmath29 remains high for the vast majority of disorder instances . between @xmath50 and @xmath51 , different instances of disorder lead to drastically different values of @xmath29 . above @xmath51 , @xmath29
quickly becomes small for the vast majority of disorder instances ( with one notable exception where @xmath52 and @xmath53 ) .
as we shall see below , the disorder strength range where the probability drops from high to low ( i.e. between 0.5 and 0.7 , roughly speaking ) is related to the range where the anderson localization length becomes much smaller than the chain length .
the above results indicate that the spatially constrained but temporally unconstrained control that is applied to the two edges of the chain is successful in suppressing the localizing effects of disorder for small values of the disorder strength , but it eventually fails for strong disorder .
a possible intuitive explanation for this phenomenon is that for weak disorder a wave packet ( or at least part of it ) is able to propagate the full length of the chain , and the control fields are then able to ensure that the full amplitude of the wave packet is collected at the last site in the chain at the end of the pulse ( with a pulse time that grows only linearly with chain length ) .
for strong disorder , the wave packet can not reach the end of the chain , and at least as far as the last site of the chain is concerned there is nothing that the control fields can do in order to catch the wave packet there . in order to identify more clearly the role of the control pulses in countering the localizing effects of disorder
, we perform a few further sets of calculations with different assumptions regarding the control resources .
first we consider the problem of state transfer without control or , in other words , the free propagation of waves .
we also analyze the spatial extension of energy eigenstates in the presence of disorder .
we then consider the case where one has control over a number of sites located equidistantly from each other ( up to the minor constraint imposed by the exact chain length ) and the case where one has full control over all the sites in the chain . as a function of time @xmath54 for an excitation that is initialized at site 1 and propagates freely in a fifty - site chain , averaged over @xmath55 disorder instances .
the different curves correspond to different values of disorder strength : from top to bottom , @xmath56 .
( b ) the long - time occupation probability @xmath57 as a function of site index @xmath58 , showing the exponential - decay behavior .
the time @xmath59 is chosen because , as can be seen from panel a , it is sufficiently long that a steady - state distribution has been reached .
the three different symbols correspond to different values of disorder strength : @xmath60 ( red squares ) , 0.4 ( green circles ) and 0.6 ( blue diamonds ) .
( c ) the inverse of the localization length @xmath61 that is extracted from the exponentially decaying long - time probability distribution as a function of disorder strength @xmath7.,title="fig:",width=245 ] as a function of time @xmath54 for an excitation that is initialized at site 1 and propagates freely in a fifty - site chain , averaged over @xmath55 disorder instances .
the different curves correspond to different values of disorder strength : from top to bottom , @xmath56 .
( b ) the long - time occupation probability @xmath57 as a function of site index @xmath58 , showing the exponential - decay behavior .
the time @xmath59 is chosen because , as can be seen from panel a , it is sufficiently long that a steady - state distribution has been reached .
the three different symbols correspond to different values of disorder strength : @xmath60 ( red squares ) , 0.4 ( green circles ) and 0.6 ( blue diamonds ) . ( c ) the inverse of the localization length @xmath61 that is extracted from the exponentially decaying long - time probability distribution as a function of disorder strength @xmath7.,title="fig:",width=264 ] as a function of time @xmath54 for an excitation that is initialized at site 1 and propagates freely in a fifty - site chain , averaged over @xmath55 disorder instances .
the different curves correspond to different values of disorder strength : from top to bottom , @xmath56 .
( b ) the long - time occupation probability @xmath57 as a function of site index @xmath58 , showing the exponential - decay behavior .
the time @xmath59 is chosen because , as can be seen from panel a , it is sufficiently long that a steady - state distribution has been reached .
the three different symbols correspond to different values of disorder strength : @xmath60 ( red squares ) , 0.4 ( green circles ) and 0.6 ( blue diamonds ) .
( c ) the inverse of the localization length @xmath61 that is extracted from the exponentially decaying long - time probability distribution as a function of disorder strength @xmath7.,title="fig:",width=264 ] in the first set of these additional calculations , we initialize the system in the state @xmath30 with an excitation ( or particle ) localized at the first site , and we simply let it evolve according to the schrdinger equation with the hamiltonian given in eq .
( [ eq : spinchainhamiltonianrwa ] ) and time - independent parameters . for
a given value of @xmath7 , we run @xmath55 different instances of the disordered hamiltonian and inspect physical observables averaged over these instances .
in particular , we inspect the average position as a function of time @xmath62 $ ] and the site occupation probabilities as functions of position and time .
the results are plotted in fig .
[ fig : freepropagation ] .
the average position @xmath63 clearly shows that the excitation propagates and possibly bounces back and forth between the ends of the chain before it converges to a constant value .
this long - time value decreases with increasing disorder strength , which suggests that it is related to the anderson localization length .
furthermore , after the transient propagation period ( whose duration also shrinks with increasing disorder strength ) , the ( averaged ) probability distribution stabilizes and follows an exponential decay function as a function of site index ( with only the first and last sites remaining slightly above the exponential fitting function ) . from this long - time probability distribution
we can extract a localization length @xmath61 .
figure [ fig : freepropagation](c ) shows a plot of @xmath64 as a function of @xmath7 .
one can see that @xmath64 increases from a value that is on the order of @xmath65 for small amounts of disorder to a value close to 1/6 for @xmath66 .
this latter value indicates that the excitation propagates at most a few sites away from its initial location , hence clearly showing anderson localization .
the reason why we obtain a finite propagation distance even for the maximum value of the disorder strength in our calculations , i.e. @xmath66 , is the fact that even for this maximum value of @xmath7 there is a reasonable probability to have a smooth landscapes up to a certain distance from the chain edge , such that the excitation propagates a few sites on average .
apart from the overall scale , the exact value of @xmath64 must be specific to the choice of the distribution function for @xmath2 values .
it should be noted here that disorder in @xmath1 would not have an upper limit , in contrast to the case with disorder in @xmath2 , and one can expect that @xmath64 will grow indefinitely with increasing disorder in that case . using the results for the localization length in fig .
[ fig : freepropagation ] , we can evaluate that the mean value of @xmath29 in fig .
[ fig : probabilityasfunctionofdisorderstrength ] has the value 0.5 when @xmath67 .
similar calculations ( whose results are not shown here in detail ) with @xmath68 and 40 give similar values for the ratio @xmath69 ( 3.1 and 3.6 , respectively ) , which gives further support to the idea that the drop of @xmath29 from near unity to near zero is correlated with the range where the localization length becomes significantly smaller than the chain length .
a point that is worth noting here is that in the strong - disorder regime @xmath70 reaches a steady - state value that is much smaller than @xmath20 , and here @xmath71 . as such , the diffusive transport regime ( where @xmath70 increases as @xmath72 ) is not realized .
if we include decoherence or time - dependent fluctuations , we would obtain such diffusive dynamics . in the weak - disorder regime , we have ballistic transport ( with @xmath73 ) during the initial transient period , as can be seen in fig .
[ fig : freepropagation](a ) . and energy level index @xmath58 .
lower panel : the ipr as a function of @xmath7 for two representative energy eigenstates , namely @xmath74 ( green dashed line ) and @xmath75 ( blue dotted line ) , as well as the average ipr for the fifty states ( red solid line ) .
one can see that all energy eigenstates progress from being delocalized in the absence of disorder ( with the ipr being comparable to @xmath65 ) to being localized with a spatial extent of only a few lattice sites when @xmath66 .
the states near the edges of the spectrum are more susceptible to disorder - induced localization than states in the middle of the spectrum.,title="fig:",width=359 ] and energy level index @xmath58 .
lower panel : the ipr as a function of @xmath7 for two representative energy eigenstates , namely @xmath74 ( green dashed line ) and @xmath75 ( blue dotted line ) , as well as the average ipr for the fifty states ( red solid line ) .
one can see that all energy eigenstates progress from being delocalized in the absence of disorder ( with the ipr being comparable to @xmath65 ) to being localized with a spatial extent of only a few lattice sites when @xmath66 .
the states near the edges of the spectrum are more susceptible to disorder - induced localization than states in the middle of the spectrum.,title="fig:",width=321 ] in addition to the localization obtained in the free propagation of initially localized states , localization effects also manifest themselves in the energy eigenstates .
these effects can be investigated through the inverse participation ratio ( ipr ) , which for a given quantum state @xmath76 is given by : @xmath77 where the index @xmath58 runs over all the sites in the chain .
roughly speaking , the ipr for a state gives the inverse of the spatial extent of the state .
note that there are @xmath20 different values of the ipr for the @xmath20 energy eigenstates of the system @xcite .
the instance - averaged ipr as a function of disorder strength and energy level index is plotted in fig .
[ fig : inverseparticipationratio ] .
the increase in the ipr mirrors that in @xmath64 , up to a factor of order unity , which shows that anderson localization manifests itself similarly in these two problems . as a function of disorder strength @xmath7 for a fifty - site chain , a pulse time @xmath48 and control over four sites ( namely sites 1 , 18 , 35 and 50 ; panel a ) or eight sites ( namely sites 1 , 8 , 15 , 22 , 29 , 36 , 43 and 50 ; panel b ) .
the ten different data points for each value of @xmath7 correspond to the same ten instances of the randomly generated coupling strengths used in fig . [ fig : probabilityasfunctionofdisorderstrength ] .
the different colors are used in order to help resolve closely spaced points but have no physical significance .
the solid red line shows the mean value of @xmath29 for the four - site - control protocol averaged over the ten different disorder instances at each value of @xmath7 .
the dotted green line shows the mean value of @xmath29 for the eight - site - control protocol .
the dashed blue line shows the mean value of @xmath29 for the case where one has control over the edge sites only , i.e. the same as in fig .
[ fig : probabilityasfunctionofdisorderstrength].,title="fig:",width=321 ] as a function of disorder strength @xmath7 for a fifty - site chain , a pulse time @xmath48 and control over four sites ( namely sites 1 , 18 , 35 and 50 ; panel a ) or eight sites ( namely sites 1 , 8 , 15 , 22 , 29 , 36 , 43 and 50 ; panel b ) .
the ten different data points for each value of @xmath7 correspond to the same ten instances of the randomly generated coupling strengths used in fig . [
fig : probabilityasfunctionofdisorderstrength ] .
the different colors are used in order to help resolve closely spaced points but have no physical significance .
the solid red line shows the mean value of @xmath29 for the four - site - control protocol averaged over the ten different disorder instances at each value of @xmath7 .
the dotted green line shows the mean value of @xmath29 for the eight - site - control protocol .
the dashed blue line shows the mean value of @xmath29 for the case where one has control over the edge sites only , i.e. the same as in fig .
[ fig : probabilityasfunctionofdisorderstrength].,title="fig:",width=321 ] next we consider the case where control fields are applied to intermediate sites located equidistantly in the chain as well as the two edge sites ( i.e. sites 1 and 50 ) .
this situation is somewhat similar to that encountered in the context of quantum repeaters , where one has multiple communication stations located at properly chosen locations between two communicating parties , such that the distance from each station to the next one is sufficiently small that quantum information can be exchanged efficiently between the two parties located at opposite ends of the setup .
the results of our calculations are plotted in fig .
[ fig : probabilityasfunctionofdisorderstrengthwithintermediatesitecontrol ] . in the first set of calculations
, we consider the situation where control fields are applied to two intermediate sites in addition to the edge sites .
the largest gains obtained with the enhanced control resources are seen between @xmath50 and @xmath78 , which is the range where the transfer probability with only edge control drops from near unity to near zero .
the gains start shrinking for stronger disorder , however , and for @xmath52 one hardly gains any increase in the transfer probability .
when we increase the number of controlled sites from four to eight , meaning that the distance between each controlled site and the next one is only seven sites , the improvement is enhanced further .
the transfer probability is now very high in all instances up to @xmath51 , and even at @xmath78 the transfer probability is higher than 0.9 in the majority of the instances .
although we now see a clear enhancement at @xmath52 compared to the cases with fewer controlled sites , with a fidelity above 95% in two of the ten instances , the transfer probability still shows a clear tendency to drop towards zero as we approach @xmath66 . as a function of disorder strength @xmath7 for the case where all the on - site energies are tunable and treated as control variables . each point in the figure
is a plot over five disorder instances used in the calculations for fig .
[ fig : probabilityasfunctionofdisorderstrength].,width=321 ] finally we perform calculations where we determine the minimum perfect - transfer time @xmath33 in the case where control fields are applied to all the on - site energies in the chain . as mentioned above , in the absence of disorder and in the limit of an infinitely long chain , @xmath38 .
our numerical calculations with @xmath71 give @xmath79 .
the 15% difference with the theoretical formula can be attributed to a combination of finite - size effects ( i.e. the finite length of the chain ) , a slightly early termination of the search algorithm ( noting here that the algorithm exhibits slow convergence when it approaches the maximum transfer probability ) and the finite distance between pulse - time values used in our calculations ( which is about 4% ) .
we performed similar calculations for finite values of disorder strength and calculated the average value of @xmath33 ( averaged over a few disorder instances ) , to which we refer as @xmath80 .
the results are shown in fig .
[ fig : averagetransfertimewithfullonsitecontrol ] .
the full control over the on - site energies allows one to strongly suppress the localizing effects of the disorder in the coupling strengths up to @xmath81 , after which @xmath80 increases rapidly .
note here that the sequential swap protocol gives a transfer time @xmath44 , providing an upper bound for @xmath80 .
in other words , even for very strong disorder , one can still perform perfect state transfer with a pulse time @xmath48 . obtaining this value
, however , indicates that one is ( almost completely ) unable to take advantage of wave propagation in order to speed up the state transfer and instead has to rely on the sequential neighbor - to - neighbor state transfer protocol .
we have considered the effect of disorder in a spin chain on the controlled transfer of an excitation or an unknown quantum state from one end of the chain to the opposite end of the chain , particularly in the case where the control fields can be applied only to the two sites at the two opposite ends of the chain .
we have found that with optimized pulses one can achieve time - efficient perfect state transfer in the vast majority of disorder instances up to disorder strengths for which the uncontrolled dynamics would allow the excitation to propagate a distance that is only a small fraction of the total length of the chain .
the application of quantum - repeater - like ideas where control fields are also applied to a number of intermediate sites enhances the transfer fidelity , allowing the protocol to give a high fidelity for rather large values of the disorder strength .
however , the usefulness of this approach decreases when the disorder strength exceeds some value that depends on the details of the protocol .
these results demonstrate that local control can counter the localizing effects of disorder even substantially away from the points where the control fields are applied
. nevertheless , at sufficiently high disorder strengths the ability to perform state transfer deteriorates and even with full on - site control one can not do much better than the neighbor - to - neighbor transfer protocol as the disorder strength approaches its maximum value .
we note here that we have only considered static disorder . in practice , dynamical disorder , or in other words temporal fluctuations of the parameters , are also present in many physical settings @xcite .
while it is possible that the optimized pulses in the presence of such fluctuations will differ from those that are optimal in the absence of fluctuations , we believe that it is unlikely that control signals can reduce the effects of an unknown time - dependent noise signal . in conclusion , our results provide theoretical understanding to the question of the competition between the constraining effects of disorder and the enabling effects of control . with the steady experimental progress in quantum coherence in spins and artificial atoms @xcite and the high interest in constructing spin chains for fundamental investigation and quantum state transfer purposes , it is likely that such systems will be constructed in the near future , which will then allow the ideas discussed here to be tested and applied to achieve high - fidelity quantum state transfer in these systems .
j. billy , v. josse , z. c. zuo , a. bernard , b. hambrecht , p. lugan , c. clement , l. sanchez - palencia , p. bouyer , and a. aspect , nature * 453 * , 891 ( 2008 ) ; g. roati , c. d. errico , l. fallani , m. fattori , c. fort , m. zaccanti , g. modugno , m. modugno , and m. inguscio , nature * 453 * , 895 ( 2008 ) ; s. longhi , laser & photon . rev . * 3 * , 243 ( 2009 ) ; i. m. georgescu , s. ashhab , and f. nori , rev . mod . phys . * 86 * , 153 ( 2014 ) .
one study investigated state transfer in the presence of fluctuations that are static during a single run of the protocol but change on a slower timescale and therefore can vary from run to run ; d. burgarth , eur .
j. special topics * 151 * , 147 ( 2007 ) .
see also l .- a .
wu , y .- x .
liu , and f. nori , phys .
a * 80 * , 042315 ( 2009 ) ; z .- m .
wang , l .- a . wu , m. modugno , w. yao , and b. shao , sci . rep .
* 3 * , 3128 ( 2013 ) ; d. de falco , d. tamascelli , j. phys . a : math .
theor . * 46 * , 225301 ( 2013 ) . note that by `` control pulses '' we mean the control signals applied over the entire duration of the transfer protocol and not short pulses that are applied at specific points in time during the protocol .
the grape algorithm searches ( using a gradient - based approach starting from a randomly generated initial guess ) for signals [ e.g. @xmath82 and @xmath83 that are piece - wise - constant functions of time , which in our case have 200 segments ( and we have verified that our results remain essentially unchanged if we increase this number to 1000 ) . except for being piece - wise constant , no further constraints are imposed on the signals . in particular ,
the algorithm does not assume or require that the signals at the different sites [ e.g. @xmath82 and @xmath83 are correlated .
the fact that in some cases the probability exhibits some non - monotonicity must be a computational artefact .
indeed , if a certain fidelity can be achieved for pulse time @xmath34 , the same fidelity can be obtained for any longer time @xmath84 by tuning the edge qubits far away from resonance with the rest of the chain for a total duration of @xmath85 at the beginning or end of the pulse , thus dividing the total pulse time @xmath84 into an active pulse time of duration @xmath34 and an idle time of duration @xmath85 .
instances of non - monotonicity were rare in our calculations , leading us to infer that numerical errors were probably at the level of a few percent in most cases .
the notable exception is the case with a very small amount of disorder .
we suspect that the large dip that appeared in all the simulations is related to the degree of matching between the natural wave propagation speed and the chain length . more specifically ,
when the pulse time is chosen to be close to the time that a maximum - speed wave needs to traverse the chain , a simple control pulse will give a high transfer probability , and the pulse search algorithm will find such pulses more easily than in the case of poor matching between the pulse time and chain length .
we also note here that our calculations exhibited slower convergence in the case @xmath39 than with higher values of @xmath7 .
r. j. bell and p. dean , discuss .
faraday soc .
* 50 * , 55 ( 1970 ) ; w. m. visscher , j. non - cryst . sol .
* 8 - 10 * , 477 ( 1972 ) ; see also j. biddle , d. j. priour , b. wang , and s. das sarma , phys .
b * 83 * , 075105 ( 2011 ) .
see e.g. t. d. ladd , f. jelezko , y. nakamura , r. laflamme , c. monroe , and j. l. obrien , nature * 464 * , 45 ( 2010 ) ; i. buluta , s. ashhab , and f. nori , rep .
. phys . * 74 * , 104401 ( 2011 ) ; r. barends , j. kelly , a. megrant , a. veitia , d. sank , e. jeffrey , t. c. white , j. mutus , a. g. fowler , b. campbell , y. chen , z. chen , b. chiaro , a. dunsworth , c. neill , p. omalley , p. roushan , a. vainsencher , j. wenner , a. n. korotkov , a. n. cleland , j. m. martinis , nature * 508 * , 500 ( 2014 ) ; p. richerme , z .- x .
gong , a. lee , c. senko , j. smith , m. foss - feig , s. michalakis , a. v. gorshkov , and c. monroe , nature * 511 * , 198 ( 2014 ) ; p. jurcevic , b. p.
lanyon , p. hauke , c. hempel , p. zoller , r. blatt , and c. f. roos , nature * 511 * , 202 ( 2014 ) ; s. viciani , m. lima , m. bellini , and f. caruso , phys .
lett . * 115 * , 083601 ( 2015 ) .
|
we investigate the effect of disorder on the transfer of quantum states across a one - dimensional lattice with varying levels of control resources . we find that the application of properly designed control signals , even when applied only to the two ends of the lattice , allows perfect state transfer up to disorder strengths that would not allow a generic quantum state to propagate the length of the lattice . at sufficiently large disorder strengths ,
however , the local control signals fail to send the quantum state from one end of the system to the other end .
our results shed light on the interplay between disorder and controlled transport in one - dimensional systems .
|
r0.5 following the successful bfkl @xcite parametrisation of the forward - jet cross - section @xmath0 at leading order ( lo ) at hera @xcite , it is possible to perform a similar study using next - to - leading ( nll ) resummed bfkl kernels .
this method can be used for forward jet production at hera in particular , provided one takes into account the proper symmetric _ two - scale _ feature of the forward - jet problem , whose scales are in this case @xmath1 for the lepton vertex and @xmath2 for the jet vertex . in this short report , we will only discuss the phenomelogical aspects and all detailed calculations can be found in ref .
@xcite for forward jets at hera and in ref .
@xcite for mueller navelet jets at the tevatron and the lhc .
we perform a saddle point approximation of the bfkl nll formalism and compare it with the h1 forward jet cross section measurements .
the bfkl nll @xcite formalism reads : @xmath3 with @xmath4 where the saddle point equation is @xmath5 .
the effective kernels @xmath6 are obtained from the nll kernel by solving the implicit equation : @xmath7 the values of @xmath8 are taken at nll @xcite using different resummation schemes to remove spurious singularities defined as ccs , s3 and s4 @xcite .
contrary to ll bfkl , it is worth noticing that the coupling constant @xmath9 is taken using the renormalisation group equations , the only free parameter in the fit being the normalisation .
one difficulty arises while fitting h1 @xmath10 data @xcite : we need to integrate the differential cross section on the bin size in @xmath11 , @xmath12 ( the momentum fraction of the proton carried by the forward jet ) , @xmath13 ( the jet transverse momentum ) , while taking into account the experimental cuts . to avoid numerical difficulties , we choose to perform the integration on the bin using the variables where the cross section does not change rapidly , namely @xmath14 , @xmath15 , and @xmath16 .
experimental cuts are treated directly at the integral level ( the cut on @xmath17 for instance ) or using a toy monte carlo .
more detail can be found about the fitting procedure in appendix a of ref .
@xcite .
the nll fits @xcite can nicely describe the h1 data @xcite for the s4 scheme ( @xmath18 per degree of freedom with statistical errors only ) whereas the s3 and ccs schemes show higher @xmath19 .
( @xmath20 and @xmath21 respectively with statistical errors only ) the fit @xmath19 are good for all schemes if one considers statistical and systematics errors added in quadrature @xcite .
the dglap nlo calculation fails to describe the h1 data at lowest @xmath22 ( see fig .
1 ) . the h1 collaboration also measured the forward jet triple differential cross section @xcite and the results are given in fig .
the bfkl ll formalism leads to a good description of the data when @xmath23 is close to 1 and deviates from the data when @xmath24 is further away from 1 .
this effect is expected since dglap radiation effects are supposed to occur when the ratio between the jet @xmath13 and the virtual photon @xmath11 are further away from 1 .
the bfkl nll calculation including the @xmath11 evolution via the renormalisation group equation leads to a good description of the h1 data on the full range .
we note that the higher order corrections are small when @xmath25 , when the bfkl effects are supposed to dominate .
by contrast , they are significant as expected when @xmath24 is different from one , ie when dglap evolution becomes relevant .
we notice that the dglap nlo calculation fails to describe the data when @xmath25 , or in the region where bfkl resummation effects are expected to appear .
r0.5 mueller navelet jets are ideal processes to study bfkl resummation effects @xcite . two jets with a large interval in rapidity and with similar tranverse momenta are considered . a typical observable to look for bfkl effects
is the measurement of the azimuthal correlations between both jets .
the dglap prediction is that this distribution should peak towards @xmath26 - ie jets are back - to - bacl- whereas multi - gluon emission via the bfkl mechanism leads to a smoother distribution .
the relevant variables to look for azimuthal correlations are the following : @xmath27 the azimuthal correlation for bfkl reads : @xmath28 where in the nll bfkl framework , @xmath29 and @xmath30 is the effective resummed kernel .
computing the different @xmath31 at nll for the resummation schemes s3 and s4 allowed us to compute the azimuthal correlations at nll . as expected , the @xmath32 dependence is less flat than for bfkl ll and is closer to the dglap behaviour @xcite . to illustrate this result
, we give in fig .
3 the azimuthal correlation in the cdf acceptance .
the cdf collaboration installed the mini - plugs calorimeters aiming for rapidity gap selections in the very forward regions and these detectors can be used to tag very forward jets .
a measurement of jet @xmath33 with these detectors would not be possible but their azimuthal segmentation allows a @xmath34 measurement . in fig .
2 , we display the jet azimuthal correlations for jets with a @xmath35 gev and @xmath366 , 8 , 10 and 11 . for @xmath3611
, we notice that the distribution is quite flat , which would be a clear test of the bfkl prediction .
similar measurements are possible at the lhc and predictions can be found in ref .
@xcite .
99 slides : ` http://indico.cern.ch/contributiondisplay.py?contribid=47&sessionid=8&confid=9499 ` l. n. lipatov , _ sov .
* 23 * ( 1976 ) 338 ; e. a. kuraev , l. n. lipatov and v. s. fadin , _ sov .
* 45 * ( 1977 ) 199 ; i. i. balitsky and l. n. lipatov , _ sov . j. nucl . phys . _
* 28 * ( 1978 ) 822 .
|
we perform a bfkl - nll analysis of forward jet production at hera which leads to a good description of data over the full kinematical domain .
we also predict the azimuthal angle dependence of mueller - navelet jet production at the tevatron and the lhc using the bfkl nll formalism .
|
it has been a long - standing problem in geometric analysis to find a good notion of a ricci flow with singularities .
the motivation comes from the fact that a ricci flow with a smooth initial condition can develop singularities without blowing up everywhere ; hence one would like to continue the flow beyond the singular time .
the presence of singularities in ricci flow is one instance of the widespread phenomenon of singularities in pde , which is often handled by developing appropriate notions of generalized solutions .
for example , in the case of the minimal surface equation , one has minimizing or stationary integral currents @xcite , while in the case of mean curvature flow one has brakke flows @xcite and level set flows @xcite .
generalized solutions provide a framework for studying issues such as existence , uniqueness and partial regularity . one highly successful way to deal with ricci flow singularities was by surgery , as developed by hamilton @xcite and perelman @xcite .
surgery served as a way to suppress singularities and avoid related analytical issues , thereby making it possible to apply ricci flow to geometric and topological problems .
perelman s construction of ricci flow with surgery drew attention to the problem of flowing through singularities . in (
* ; * ? ? ?
13.2 ) , perelman wrote , `` it is likely that by passing to the limit in this construction one would get a canonically defined ricci flow through singularities , but at the moment i do nt have a proof of that . '' in this paper we address the problem of flowing through singularities in the three - dimensional case by introducing singular ricci flows , which are ricci flow spacetimes subject to certain conditions .
we show that singular ricci flows have a number of good properties , by establishing existence and compactness results , as well as a number of structural results .
we also use singular ricci flows to partially answer perelman s question , by showing that ricci flow with surgery ( for a fixed initial condition ) subconverges to a singular ricci flow as the surgery parameter goes to zero . these results and further results in the sequel
strongly indicate that singular ricci flows provide a natural analytical framework for three - dimensional ricci flows with singularities .
although it is rather different in spirit , we mention that there is earlier work in the literature on ricci flow through singularities ; see the end of the introduction for further discussion . before formulating our first result ,
we briefly recall perelman s version of ricci flow with surgery ( which followed earlier work of hamilton ) ; see appendix [ rfsurgery ] for more information .
ricci flow with surgery evolves a riemannian @xmath0-manifold by alternating between two processes : flowing by ordinary ricci flow until the metric goes singular , and modifying the resulting limit by surgery , so as to produce a compact smooth riemannian manifold that serves as a new initial condition for ricci flow .
the construction is regulated by a global parameter @xmath1 as well as decreasing parameter functions @xmath2 , which have the following significance : * the scale at which surgery occurs is bounded above in terms of @xmath3 .
in particular , surgery at time @xmath4 is performed by cutting along necks whose scale tends to zero as @xmath5 goes to zero .
note that @xmath6 has other roles in addition to this . *
the function @xmath7 defines the canonical neighborhood scale : at time @xmath4 , near any point with scalar curvature at least @xmath8 , the flow is ( modulo parabolic rescaling ) approximated to within error @xmath9 by either a @xmath10-solution ( see appendix [ appkappa ] ) or a standard postsurgery model solution . in ricci
flow with surgery , the initial conditions are assumed to be _
normalized _ , meaning that at each point @xmath11 in the initial time slice , the eigenvalues of the curvature operator @xmath12 are bounded by one in absolute value , and the volume of the unit ball @xmath13 is at least half the volume of the euclidean unit ball . by rescaling , any compact riemannian manifold can be normalized .
perelman showed that under certain constraints on the parameters , one can implement ricci flow with surgery for any normalized initial condition .
his constraints allow one to make @xmath6 as small as one wants .
hence one can consider the behavior of ricci flow with surgery , for a fixed initial condition , as @xmath6 goes to zero . in order to formulate our convergence theorem , we will use a spacetime framework . unlike the case of general relativity , where one has a lorentzian manifold , in our setting
there is a natural foliation of spacetime by time slices , which carry riemannian metrics .
this is formalized in the following definition .
[ def_ricci_flow_spacetime ] a _ ricci flow spacetime _ is a tuple @xmath14 where : * @xmath15 is a smooth manifold - with - boundary .
* @xmath16 is the _ time function _ a submersion @xmath17 where @xmath18 is a time interval ; we will usually take @xmath19 . *
the boundary of @xmath15 , if it is nonempty , corresponds to the endpoint(s ) of the time interval : @xmath20 .
* @xmath21 is the _ time vector field _ , which satisfies @xmath22 .
* @xmath23 is a smooth inner product on the spatial subbundle @xmath24 , and @xmath23 defines a ricci flow : @xmath25 . for @xmath26 , we write @xmath27 , @xmath28}={\mathfrak{t}}^{-1}([a , b])$ ] and @xmath29)$ ] . henceforth ,
unless otherwise specified , when we refer to geometric quantities such as curvature , we will implicitly be referring to the metric on the time slices .
note that near any point @xmath30 , a ricci flow spacetime @xmath14 reduces to a ricci flow in the usual sense , because the time function @xmath16 will form part of a chart @xmath31 near @xmath11 for which the coordinate vector field @xmath32 coincides with @xmath21 ; then one has @xmath33 .
our first result partially answers the question of perelman alluded to above , by formalizing the notion of convergence and obtaining subsequential limits : [ thm_main_convergence ] let @xmath34 be a sequence of three - dimensional ricci flows with surgery ( in the sense of perelman ) where : * the initial conditions @xmath35 are compact normalized riemannian manifolds that lie in a compact family in the smooth topology , and * if @xmath36 denotes the perelman surgery parameter for @xmath37 then @xmath38 .
then after passing to a subsequence , there is a ricci flow spacetime @xmath39 , and a sequence of diffeomorphisms @xmath40 with the following properties : 1 .
@xmath41 and @xmath42 are open subsets .
2 . let @xmath43 and @xmath44 denote the scalar curvature on @xmath37 and @xmath45 , respectively .
given @xmath46 and @xmath47 , if @xmath48 is sufficiently large then @xmath49 3 .
@xmath50 is time preserving , and the sequences @xmath51 and @xmath52 converge smoothly on compact subsets of @xmath45 to @xmath53 and @xmath54 , respectively .
4 . @xmath50 is asymptotically volume preserving : let @xmath55 denote the respective volume functions @xmath56 and @xmath57 . then @xmath58 is continuous and @xmath59 , with uniform convergence on compact subsets of @xmath60 .
furthermore : 1 .
the scalar curvature function @xmath61 is bounded below and proper for all @xmath62 .
@xmath45 satisfies the hamilton - ivey pinching condition of ( [ hi ] ) .
@xmath45 is @xmath10-noncollapsed below scale @xmath63 and satisfies the @xmath7-canonical neighborhood assumption , where @xmath10 and @xmath7 are the aforementioned parameters from ricci flow with surgery .
theorem [ thm_main_convergence ] may be compared with other convergence results such as hamilton s compactness theorem @xcite and its variants ( * ? ? ?
* ; * ? ? ?
* appendix e ) , as well as analogous results for sequences of riemannian manifolds .
all of these results require uniform bounds on curvature in regions of a given size around a basepoint , which we do not have . instead
, our approach is based on the fact that in a three dimensional ricci flow with surgery , the scalar curvature controls the local geometry .
we first prove a general pointed compactness result for sequences of ( possibly incomplete ) riemannian manifolds whose local geometry is governed by a control function .
we then apply this general compactness result in the case when the riemannian manifolds are the spacetimes of ricci flows with surgery , and the control functions are constructed from the scalar curvature functions . to obtain ( 1)-(3 ) of theorem [ thm_main_convergence ] we have to rule out the possibility that part of the spacetime with controlled time and scalar curvature escapes to infinity
, i.e. is not seen in the pointed limit .
this is done by means of a new estimate on the spacetime geometry of a ricci flow with surgery ; see proposition [ curve ] below . motivated by the conclusion of theorem [ thm_main_convergence ]
, we make the following definition : [ def_singular_ricci_flow ] a ricci flow spacetime @xmath14 is a _ singular ricci flow _ if it is @xmath64-dimensional , the initial time slice @xmath65 is a compact normalized riemannian manifold and 1 . the scalar curvature function @xmath66 is bounded below and proper for all @xmath62 .
2 . @xmath15 satisfies the hamilton - ivey pinching condition of ( [ hi ] ) .
3 . for a global parameter @xmath1 and decreasing functions @xmath67 ,
the spacetime @xmath15 is @xmath10-noncollapsed below scale @xmath63 in the sense of appendix [ kappa ] and satisfies the @xmath7-canonical neighborhood assumption in the sense of appendix [ canon ] .
although conditions ( b ) and ( c ) in definition [ def_singular_ricci_flow ] are pointwise conditions imposed everywhere , we will show that @xmath15 is a singular ricci flow if ( b ) and ( c ) are only assumed to hold outside of some compact subset of @xmath68 , for all @xmath69 .
thus ( b ) and ( c ) can be viewed as asymptotic conditions at infinity for a ricci flow defined on a noncompact spacetime . with this definition , perelman s existence theorem for ricci flow with surgery and theorem [ thm_main_convergence ]
immediately imply : [ cor_existence_smooth_initial_condition ] if @xmath70 is a compact normalized riemannian @xmath0-manifold then there exists a singular ricci flow having initial condition @xmath70 , with parameter functions @xmath10 and @xmath7 as in theorem [ thm_main_convergence ] . from the pde viewpoint ,
flow with surgery is a regularization of ricci flow , while singular ricci flows may be considered to be generalized solutions to ricci flow . in this language ,
corollary [ cor_existence_smooth_initial_condition ] gives the existence of generalized solutions by means of a regularization procedure .
one can compare this with the existence proof for brakke flows in @xcite or level set flows in @xcite .
the existence assertion in corollary [ cor_existence_smooth_initial_condition ] leads to the corresponding uniqueness question : if two singular ricci flows have isometric initial conditions , are the underlying ricci flow spacetimes the same up to diffeomorphism
? an affirmative answer would confirm perelman s expectation that ricci flow with surgery should converge to a canonical flow through singularities , as it would imply that if one takes a fixed initial condition in theorem [ thm_main_convergence ] then one would have convergence without having to pass to a subsequence .
having such a uniqueness result , in conjunction with theorem [ thm_main_convergence ] , would closely parallel the results of @xcite that @xmath71-convex mean curvature flow with surgery converges to level set flow when the surgery parameters tend to zero .
the asymptotic conditions in the definition of a singular ricci flow have a number of implications which we analyze in this paper .
in addition to clarifying the structure of limits of ricci flows with surgery as in theorem [ thm_main_convergence ] , the results indicate that singular ricci flows are well behaved objects from geometric and analytical points of view . to analyze the geometry of ricci flow spacetimes , we use two different riemannian metrics .
[ spacetimemetric ]
let @xmath14 be a ricci flow spacetime .
the _ spacetime metric _ on @xmath15 is the riemannian metric @xmath72 , where @xmath73 is the extension of @xmath23 to a quadratic form on @xmath74 such that @xmath75 .
the _ quasiparabolic metric _ on @xmath15 is the riemannian metric @xmath76 . for the remainder of the introduction ,
unless otherwise specified , @xmath14 will denote a fixed singular ricci flow .
conditions ( b ) and ( c ) of definition [ def_singular_ricci_flow ] imply that the scalar curvature controls the local geometry of the singular ricci flow .
this has several implications .
* ( high - curvature regions in singular ricci flows are topologically standard ) for every @xmath4 , the superlevel set @xmath77 is contained in a disjoint union of connected components whose diffeomorphism types come from a small list of possibilities , with well - controlled local geometry .
in particular , each connected component @xmath78 of @xmath79 has finitely many ends , and passing to the metric completion @xmath80 adds at most one point for each end ( proposition [ prop_large_r_structure ] ) . * ( bounded geometry ) the spacetime metric @xmath81 has bounded geometry at the scale defined by the scalar curvature , while the quasiparabolic metric @xmath82 is complete and has bounded geometry in the usual sense the injectivity radius is bounded below and all derivatives of curvature are uniformly bounded ( lemma [ quasip ] ) .
the local control on geometry also leads to a compactness property for singular ricci flows : * ( compactness ) if one has a sequence @xmath83 of singular ricci flows with a fixed choice of functions in definition [ def_singular_ricci_flow ] , and the initial metrics @xmath84 form a precompact set in the smooth topology , then a subsequence converges in the sense of theorem [ thm_main_convergence ] ( proposition [ rfcompactness ] ) .
the proof of the compactness result is similar to the proof of theorem [ thm_main_convergence ] .
we also have global results concerning the scalar curvature and volume . *
( scalar curvature and volume control ) for any @xmath85 , the scalar curvature is integrable on @xmath86 .
the volume function @xmath87 is finite and locally @xmath88-hlder in @xmath4 for some exponent @xmath89 , and has a locally bounded upper right derivative .
the usual formula holds for volume evolution : @xmath90}}r\;{\operatorname{dvol}}_{g_{{{\mathcal m}}}}\ ] ] for all @xmath91 .
( propositions [ vol1 ] and [ holder ] ) * ( @xmath92 bound on scalar curvature ) for all @xmath93 and all @xmath4 , the scalar curvature is @xmath92 on @xmath79 .
( proposition [ prop_r_in_lp ] ) the starting point for the proof of the above results concerning volume is the fact that for ricci flow , the time derivative of the volume form is given by the negative of the scalar curvature .
to obtain global results , one is effectively forced to do integration by parts using the time vector field @xmath21 .
however , there is a substantial complication due to the potential incompleteness of the time vector field ; we overcome this by using theorem [ thm_finiteness_of_bad_worldlines ] below , together with the gradient estimate on scalar curvature .
to describe the next results , we introduce the following definitions .
[ def_worldline ] a path @xmath94 is _ time - preserving _ if @xmath95 for all @xmath96 .
the _ worldline _ of a point @xmath30 is the maximal time - preserving integral curve @xmath94 of the time vector field @xmath21 , which passes through @xmath11 .
if @xmath94 is a worldline then we may have @xmath97 . in this case , the scalar curvature blows up along @xmath98 as @xmath99 , and the worldline encounters a singularity .
an example would be a shrinking round space form , or a neckpinch . a worldline may also encounter a singularity going backward in time .
a worldline @xmath94 is _ bad _ if @xmath100 , i.e. if it is not defined at @xmath101 . among our structural results ,
perhaps the most striking is the following : [ thm_finiteness_of_bad_worldlines ] suppose that @xmath14 is a singular ricci flow and @xmath102 .
if @xmath78 is a connected component of @xmath79 then only finitely many points in @xmath78 have bad worldlines . as an illustration of the theorem ,
consider a singular ricci flow that undergoes a generic neck pinch at time @xmath103 , so that the time slice @xmath104 has two ends ( @xmath9-horns in perelman s language ) which are instantly capped off when @xmath105 . in this case
the theorem asserts that only finitely many ( in this case two ) worldlines emerge from the singularity .
see figure [ fig-1 ] , where the bad wordlines are indicated by dashed curves .
( the point in the figure where the two dashed curves meet is not in the spacetime . ) , height=384 ] a key ingredient in the proof of theorem [ thm_finiteness_of_bad_worldlines ] is a new stability property of neck regions in @xmath10-solutions .
we recall that @xmath10-solutions are the class of ancient ricci flows used to model the high curvature part of ricci flows with surgery .
we state the stability property loosely as follows , and refer the reader to section [ stab ] for more details : let @xmath15 be a noncompact @xmath10-solution other than the shrinking round cylinder . if @xmath94 is a worldline and @xmath106 is sufficiently necklike at @xmath107 , then as @xmath108 , the time slice @xmath79 looks more and more necklike at @xmath98 . here
the notion of necklike is scale invariant .
an easy but illustrative case is the bryant soliton , in which worldlines other than the tip itself move away from the tip ( in the scale invariant sense ) as one goes backward in time .
as mentioned above , theorem [ thm_finiteness_of_bad_worldlines ] is used in the proof of ( [ eqn_volume_formula ] ) and the properties of volume .
we mention some connectedness properties of singular ricci flows . *
( paths back to @xmath65 avoiding high - curvature regions ) any point @xmath109 can be joined to the initial time slice @xmath65 by a time - preserving curve @xmath110{\rightarrow}{{\mathcal m}}$ ] , along which @xmath111\}$ ] is bounded in terms of @xmath112 and @xmath113 .
( proposition [ curve2 ] ) * ( backward stability of components ) if @xmath114{\rightarrow}{{\mathcal m}}$ ] are time - preserving curves such that @xmath115 and @xmath116 lie in the same connected component of @xmath106 , then @xmath117 and @xmath118 lie in the same component of @xmath79 for all @xmath119 $ ] .
( proposition [ sameconnected ] ) finally , we alluded before to the fact that the _ a priori _ assumptions in definition [ def_singular_ricci_flow ] are really restrictions on the spacetime near infinity . *
( asymptotic conditions ) given @xmath120 and a decreasing function @xmath121 , there are decreasing functions @xmath122 and @xmath123 with the following property .
let @xmath15 be a ricci flow spacetime with normalized initial condition , on which condition ( a ) of definition [ def_singular_ricci_flow ] holds .
suppose that for each @xmath69 there is a compact subset of @xmath68 so that condition ( b ) , and the @xmath7-canonical neighborhood assumption of condition ( c ) , hold on the part of @xmath68 outside of the compact subset .
then @xmath15 satisfies definition [ def_singular_ricci_flow ] globally with parameters @xmath63 , @xmath124 and @xmath125 .
( proposition [ asympprop ] ) we now mention some other work that falls into the broad setting of ricci flow with singular structure .
a number of authors have considered ricci flow with low regularity initial conditions , studying existence and/or uniqueness , and instantaneous improvement of regularity @xcite .
ricci flow with persistent singularities has been considered in the case of orbifold ricci flow , and ricci flow with conical singularities @xcite passing to flows through singularities , feldman - ilmanen - knopf noted that in the noncompact khler setting , there are some natural examples of flows through singularities which consist of a shrinking gradient soliton that has a conical limit at time zero , which transmutes into an expanding gradient soliton @xcite .
closer in spirit to this paper , angenent - caputo - knopf constructed a rotationally invariant ricci flow through singularities starting with a metric on @xmath126 @xcite .
they showed that the rotationally invariant neckpinches from @xcite , which have a singular limit as @xmath4 approaches zero from below , may be continued as a smooth ricci flow on two copies of @xmath126 for @xmath127 .
the paper @xcite also showed that the forward evolution has a unique asymptotic profile near the singular point in spacetime .
there has been much progress on flowing through singularities in the khler setting. for a flow on a projective variety with log terminal singularities , song - tian @xcite showed that the flow can be continued through the divisorial contractions and flips of the minimal model program . we
refer to @xcite for the precise statements .
the paper @xcite has related results , but uses a viscosity solution approach instead of the regularization scheme in @xcite .
the fact that the khler - ricci flow can be reduced to a scalar equation , to which comparison principles may be applied , is an important simplifying feature of khler - ricci flow that is not available in the non - khler case .
the work in this paper is related to the question of whether there is a good notion of a weak ricci flow .
this question is subject to different interpretations , because the term `` weak '' has different meanings in different settings .
the results in this paper show that singular ricci flows give a possible answer in the @xmath0-dimensional case , and in the @xmath64-dimensional case under the assumption of nonnegative isotropic curvature .
in the general higher dimensional case , however , it remains unclear if there is a good general notion of weak ricci flow .
the remainder of the paper is broken into two parts .
part i , which is composed of sections [ sec_compactness_locally_controlled]-[sec_main_convergence ] , is primarily concerned with the proof of the convergence result , theorem [ thm_main_convergence ] .
part ii , which is composed of sections [ sec_singular_ricci_flows_basic_properties]-[asympcon ] , deals with results on singular ricci flows .
appendix [ app_background_material ] introduces and recalls notation and terminology for ricci flows and ricci flows with surgery .
we now describe the contents section by section .
section [ sec_compactness_locally_controlled ] gives a general pointed compactness result for riemannian manifolds and spacetimes , whose geometry is locally controlled as a function of some auxiliary function .
section [ surgerysec ] develops some properties of ricci flow with surgery ; it is aimed at showing that the ricci flow spacetime associated with a ricci flow with surgery has locally controlled geometry in the sense of section [ sec_compactness_locally_controlled ] .
section [ sec_main_convergence ] applies the two preceding sections to give the proof of theorem [ thm_main_convergence ] .
section [ sec_singular_ricci_flows_basic_properties ] establishes some foundational results about singular ricci flows , concerning scalar curvature and volume , as well as some results involving the structure of the high curvature region .
section [ stab ] proves that neck regions in @xmath10-solutions have a stability property going backward in time .
section [ sec_finiteness ] proves theorem [ thm_finiteness_of_bad_worldlines ] , concerning the finiteness of the number of bad worldlines , and gives several applications .
section [ sec_curvature_volume_estimates ] contains some estimates involving volume and scalar curvature .
section [ asympcon ] proves that conditions ( b ) and ( c ) of definition [ def_singular_ricci_flow ] are asymptotic conditions . in section [ pic ]
we indicate how the preceding results extend to four - dimensional singular ricci flows with nonnegative isotropic curvature .
appendix [ app_background_material ] collects a variety of background material ; the reader may wish to quickly peruse this before proceeding to the body of the paper .
appendix [ technical ] is a technical addendum about three - dimensional @xmath10-solutions .
we refer the reader to appendix [ sec_notation ] for notation and terminology .
all manifolds that arise will be taken to be orientable .
in this section we prove a compactness result , theorem [ theorem3.8first ] , for sequences of controlled riemannian manifolds . in subsection [ spacetimecompactness ]
we extend the theorem to a compactness result for spacetimes , meaning riemannian manifolds equipped with time functions and time vector fields .
we will need a sequential compactness result for sequences of riemannian manifolds which may be incomplete , but whose local geometry ( injectivity radius and all derivatives of curvature ) is bounded by a function of an auxiliary function @xmath128 .
a standard case of this is sequential compactness for sequences @xmath129 of riemannian @xmath7-balls , assuming that the distance function @xmath130 is proper , and that the geometry is bounded in terms of @xmath131 .
fix a smooth decreasing function @xmath132 $ ] and smooth increasing functions @xmath133 for all @xmath134 .
[ definition3.1first ] suppose that @xmath135 is a riemannian manifold equipped with a function @xmath128 .
given @xmath136 $ ] , a tensor field @xmath137 is _
@xmath138-controlled _ if for all @xmath139 and @xmath140 , we have @xmath141 if in addition @xmath142 is smooth then we say that the tuple @xmath143 is _ @xmath138-controlled _ if 1 .
the injectivity radius of @xmath144 at @xmath11 is at least @xmath145 for all @xmath146 , and 2 . the tensor field @xmath147 , and @xmath142 itself , are @xmath138-controlled .
note that if @xmath148 then @xmath149 is all of @xmath144 , so there are quantitative bounds on the geometry at each point of @xmath144 .
note also that the value of a control function may not reflect the actual bounds on the geometry , in the sense that the geometry may be more regular near @xmath11 than the value of @xmath150 suggests .
this creates flexibility in choosing a control function , which is useful in applications below .
suppose that @xmath151 is a complete pointed riemannian manifold .
put @xmath152 .
then @xmath147 is @xmath153-controlled if and only if for all @xmath154 and @xmath140 , we have @xmath155 on @xmath156 .
suppose that @xmath142 has constant value @xmath157 .
if @xmath158 then @xmath159 is vacuously @xmath153-controlled . if @xmath160 then @xmath159 is @xmath153-controlled if and only if for all @xmath161 , we have @xmath162 and @xmath163
there are compactness results in riemannian geometry saying that one can extract a subsequential limit from a sequence of complete pointed riemannian manifolds having uniform local geometry .
this last condition means that for each @xmath164 , one has quantitative uniform bounds on the geometry of the @xmath7-ball around the basepoint ; c.f .
* ; * ? ? ?
* theorem 2.3 ) .
in such a case , one can think of the distance from the basepoint as a control function .
we will give a compactness theorem for riemannian manifolds ( possibly incomplete ) equipped with more general control functions .
for notation , if @xmath165 is a diffeomorphism then we will write @xmath166 for both the pushforward action of @xmath167 on contravariant tensor fields on @xmath168 , and the pullback action of @xmath169 on covariant tensor fields on @xmath168 .
[ theorem3.8first ] let @xmath170 be a sequence of pointed tuples which are @xmath171-controlled , where @xmath172 and @xmath173 .
then after passing to a subsequence , there is a pointed @xmath174-controlled tuple @xmath175 and a sequence of diffeomorphisms @xmath176 such that 1 .
given @xmath177 , for all sufficiently large @xmath48 the open set @xmath178 contains the ball @xmath179 in the riemannian manifold @xmath180 and likewise @xmath181 contains the ball @xmath182 in the riemannian manifold @xmath183 .
2 . given @xmath184 and @xmath134 , for all sufficiently large @xmath48
we have @xmath185 and @xmath186 3 .
@xmath187 is connected and , in particular , every @xmath188 belongs to @xmath181 for @xmath48 large
. we will give the proof of theorem [ theorem3.8first ] in subsection [ pf3.8first ] .
we first describe some general results about controlled riemannian manifolds .
one approach to proving theorem [ theorem3.8first ] would be to imitate what one does when one has curvature and injectivity radius bounds on @xmath7-balls , replacing the control function based on distance to the basepoint by the control function @xmath142 .
while this could be done , it would be somewhat involved . instead
, we will perform a conformal change on the riemannian manifolds in order to put ourselves in a situation where the geometry is indeed controlled by the distance from the basepoint .
we then take a subsequential limit of the conformally changed metrics , and at the end perform another conformal change to get a subsequential limit of the original sequence .
let @xmath189 be @xmath153-controlled .
put @xmath190 we will only consider @xmath191 on the subset @xmath192 , where it is smooth .
the next two lemmas are about @xmath23-balls and @xmath193-balls . [ lemma3.21 ] for each finite @xmath194 $ ] , each @xmath195 and each @xmath196
, there is some @xmath197 so that the ball @xmath198 in the riemannian manifold @xmath199 is contained in the ball @xmath200 in the riemannian manifold @xmath201 .
on @xmath202 we have @xmath203 .
therefore any path in @xmath202 with @xmath23-length at most @xmath7 has @xmath193-length at most @xmath204 , so we may take @xmath205 .
let @xmath206 be a basepoint in @xmath149 .
[ ballinside ] for all @xmath207 , the ball @xmath208 in the riemannian manifold @xmath209 is contained in @xmath210 . given @xmath211 ,
let @xmath212 { \rightarrow}\psi^{-1}([0 , a))$ ] be a smooth path from @xmath206 to @xmath11 with unit @xmath193-speed and @xmath193-length @xmath213 .
then @xmath214 the lemma follows .
we now look at completeness properties of @xmath193-balls .
[ compactclosure ] for all @xmath215 , the ball @xmath208 in the riemannian manifold @xmath216 has compact closure in @xmath192 .
choose @xmath217 . from (
* ; * ? ? ?
* chapter 1 , theorem 2.4 ) , it suffices to show that any @xmath193-unit speed geodesic @xmath218 with @xmath219 , having @xmath193-length @xmath220 , can be extended to @xmath221 $ ] . from lemma [ ballinside ] , @xmath222 . hence the @xmath23-injectivity radius along @xmath223
is bounded below by @xmath224 . for large @xmath225 , the points @xmath226 form a cauchy sequence in @xmath227 .
as @xmath228 and @xmath229 are bilipschitz on @xmath230 , the sequence is also cauchy in @xmath231 . from the uniform positive lower bound on the @xmath23-injectivity radius at @xmath232 ,
there is a limit in @xmath149 .
the lemma follows .
[ lemma3.4 ] if @xmath148 then @xmath233 is complete .
this follows from lemma [ compactclosure ] and ( * ? ? ?
* ; * ? ? ?
* chapter 1 , theorem 2.4 ) . finally , we give bounds on the geometry of @xmath193-balls .
[ morebounds ] given @xmath234 , @xmath235 and @xmath236 , there exist 1 .
a smooth decreasing function @xmath237 $ ] , and 2 .
smooth increasing functions @xmath238 , @xmath140 , with the following properties .
suppose that @xmath159 is @xmath153-controlled and @xmath239
. then 1 .
@xmath240 and @xmath142 are @xmath241-controlled on the riemannian manifold @xmath242 ( in terms of the functions @xmath243 ) .
if @xmath244 then @xmath245 pointwise on @xmath208 . conclusion ( a ) ( along with the concomitant functions @xmath243 ) follows from lemma [ ballinside ] , the assumption that @xmath246 and @xmath142 are @xmath247-controlled , and the formula for the riemannian curvature of a conformally changed metric . to prove conclusion ( b ) , suppose that @xmath244 and @xmath211 .
since @xmath228 and @xmath229 are bilipschitz on the ball @xmath248 in the riemannian manifold @xmath249 , we can find @xmath250 so that @xmath251 . since we have a @xmath23-curvature bound on @xmath248 , and a lower @xmath23-injectivity radius bound at @xmath11 , we obtain a lower volume bound @xmath252 .
since @xmath23 and @xmath193 are relatively bounded on @xmath253 , this gives a lower volume bound @xmath254 .
using the curvature bound of part ( a ) and ( * ? ? ?
* ; * ? ? ?
* theorem 4.7 ) , we obtain a lower bound @xmath255 .
this proves the lemma .
put @xmath256 .
consider the tuple @xmath257 .
recall that @xmath258 . for the moment
, we replace the index @xmath48 by the index @xmath259 .
using lemma [ compactclosure ] , lemma [ morebounds ] and a standard compactness theorem ( * ? ? ?
* ; * ? ? ?
* theorem 2.3 ) , after passing to a subsequence we can find a complete pointed riemannian manifold @xmath260 , domains @xmath261 and @xmath262 , and diffeomorphisms @xmath263 so that 1 .
2 . @xmath265 .
@xmath266 has compact closure .
4 . for any compact set @xmath267
, we have @xmath268 for all sufficiently large @xmath259 .
5 . given @xmath184 and @xmath134
, we have @xmath269 for all sufficiently large @xmath259 . using lemma [ morebounds ] again , after passing to a further subsequence if necessary
, we can assume that there is a smooth @xmath270 on @xmath187 so that for all @xmath184 and @xmath134 , we have @xmath271 for all sufficiently large @xmath259 .
put @xmath272 .
we claim that if the sequence @xmath273 increases rapidly enough then the conclusions of the theorem can be made to hold with @xmath274 , @xmath275 and @xmath276 . to see this , we note that * given @xmath277 , lemma [ lemma3.21 ] implies that the ball @xmath278 in the riemannian manifold @xmath279 will be contained in @xmath181 for all sufficiently large @xmath48 . *
if the sequence @xmath273 increases rapidly enough then for large @xmath48 , the map @xmath50 is arbitrarily close to an isometry and @xmath280 is arbitrarily close to @xmath270 on @xmath181 . hence given @xmath277 , the ball @xmath281 in the riemannian manifold @xmath282 will be contained in @xmath178 for all sufficiently large @xmath48 , so conclusion ( 1 ) of the theorem holds . *
the metrics @xmath54 and @xmath283 are bilipschitz on @xmath181 .
then if the sequence @xmath273 increases rapidly enough , conclusion ( 2 ) of the theorem can be made to hold .
* conclusion ( 3 ) of the theorem follows from the definition of @xmath181 .
this proves theorem [ theorem3.8first ] .
we now apply theorem [ theorem3.8first ] to prove a compactness result for spacetimes .
[ definition3.1 ] a _ spacetime _ is a riemannian manifold @xmath284 equipped with a submersion @xmath285 and a smooth vector field @xmath21 such that @xmath286 .
given @xmath136 $ ] and @xmath287 , we say that the tuple @xmath288 is _ @xmath138-controlled _ if @xmath289 is @xmath153-controlled in the sense of definition [ definition3.1first ] and , in addition , the tensor fields @xmath16 and @xmath21 are @xmath138-controlled .
[ theorem3.8 ] let @xmath290 be a sequence of pointed tuples which are @xmath171-controlled , where @xmath172 and @xmath173 .
then after passing to a subsequence , there is a pointed @xmath174-controlled tuple @xmath291 and a sequence of diffeomorphisms @xmath292 of open sets such that 1 .
given @xmath293 and @xmath196 , for all sufficiently large @xmath48 , the open set @xmath178 contains the ball @xmath179 in the riemannian manifold @xmath294 and likewise @xmath181 contains the ball @xmath295 .
@xmath50 exactly preserves the time functions : @xmath296 3 .
@xmath50 asymptotically preserves the tensor fields @xmath297 , @xmath298 , and @xmath299 : if @xmath300 and @xmath301 is the corresponding element of @xmath302 , then for all @xmath184 and @xmath134 , we have @xmath303 for all sufficiently large @xmath48 .
@xmath45 is connected and , in particular , every @xmath304 belongs to @xmath181 for large @xmath48 .
put @xmath305 .
consider the pointed spacetime @xmath306 .
[ lemma3.18again ] for every @xmath307 , the tensor fields @xmath308 , @xmath309 , @xmath310 , and @xmath299 are all @xmath311-controlled for all large @xmath48 , in the sense of lemma [ morebounds ] .
this follows as in the proof of lemma [ morebounds ] .
we follow the proof of theorem [ theorem3.8first ] up to the construction of @xmath312 and @xmath270 . using lemma [ lemma3.18again ] , after passing to a further subsequence if necessary
, we can assume that there are smooth @xmath313 and @xmath314 on @xmath45 so that if @xmath315 then for all @xmath184 and @xmath134 , we have @xmath316 for all sufficiently large @xmath259 .
again , @xmath317 is a submersion and @xmath318 . as before ,
put @xmath319 .
let @xmath320 be the flow generated by @xmath314 ; this exists for at least a small time interval if the starting point is in a given compact subset of @xmath187 .
put @xmath321 .
then there is some @xmath322 so that @xmath320 exists on @xmath323 for @xmath324 . given @xmath325 , put @xmath326 . assuming that @xmath327 is large enough
, we can define @xmath328 by @xmath329 by construction , @xmath330 , and so conclusion ( 2 ) of the theorem holds . if @xmath327 is large then @xmath50 will be a diffeomorphism to its image . putting @xmath331 ,
if @xmath327 is large enough then @xmath181 can be made arbitrarily close to @xmath323 .
it follows that conclusions ( 1 ) , ( 3 ) and ( 4 ) of the theorem hold . in hamilton
s compactness theorem @xcite , the comparison map @xmath50 preserves both the time function and the time vector field . in theorem [ theorem3.8 ]
the comparison map @xmath50 preserves the time function , but the time vector field is only preserved asymptotically .
this is good enough for our purposes .
in this section we prove several estimates for ricci flows with surgery .
these will be used in the proof of theorem [ thm_main_convergence ] , to show that the sequence of ricci flow spacetimes has the local control required for the application of the spacetime compactness theorem , theorem [ theorem3.8 ] .
the arguments in this section require familiarity with some basic properties of ricci flow with surgery .
for the reader s convenience , we have collected these properties in appendix a.8 .
the reader may wish to review this material before proceeding .
let @xmath15 be a ricci flow with surgery in the sense of perelman ( * ? ? ?
* ; * ? ? ?
* section 68 ) .
as recalled in appendix [ rfsurgery ] , there are parameters or more precisely positive decreasing parameter functions associated with @xmath15 : * the canonical neighborhood scale function @xmath332 .
we can assume that @xmath333 .
* the noncollapsing function @xmath334 . * the parameter @xmath335 .
this has a dual role : it is the quality of the surgery neck , and it enforces a scale buffer between the canonical neighborhood scale @xmath7 , the intermediate scale @xmath336 and the surgery scale @xmath337 . * the intermediate scale @xmath338 , which defines the threshold for discarding entire connected components at the singular time . * the surgery scale @xmath339 . * the global parameter @xmath1 .
this enters in the definition of a canonical neighborhood .
for the ricci flow with surgery to exist , a necessary condition is that @xmath63 be small enough . in this section ,
canonical neighborhoods are those defined for ricci flows with surgery , as in ( * ? ? ?
* ; * ? ? ?
* definition 69.1 ) .
the next lemma gives a sufficient condition for parabolic neighborhoods to be unscathed .
[ lemma2.1 ] let @xmath15 be a ricci flow with surgery , with normalized initial condition .
given @xmath340 , there are numbers @xmath341 , @xmath342 , @xmath343 and @xmath344 , @xmath140 , with the following property . if @xmath345 $ ] and @xmath346 , put @xmath347
. then 1 .
the forward parabolic ball @xmath348 and the backward parabolic ball @xmath349 are unscathed .
@xmath350 , @xmath351 and @xmath352 on the union @xmath353 of the forward and backward parabolic balls . by (
* ; * ? ? ?
* lemma 70.1 ) , we have @xmath354 for all @xmath355 . where @xmath356 is a universal constant .
the same argument works for @xmath357 . since @xmath358 is proper on time slices ( c.f .
* ; * ? ? ?
* lemma 67.9 ) ) , it follows that @xmath359 has compact closure in its time slice . if @xmath360 then @xmath361 for all @xmath362 $ ] .
hence if @xmath363 then the forward and backward parabolic balls @xmath364 do not intersect the surgery regions . to show that the balls @xmath364 are unscathed , for an appropriate value of @xmath365 , it remains to show that @xmath349 does not intersect the time - zero slice .
we have @xmath366 .
since @xmath367 and @xmath368 , if @xmath369 then @xmath370 and the balls @xmath364 are unscathed .
the hamilton - ivey estimate of ( [ hi ] ) gives an explicit upper bound @xmath350 on @xmath371 .
using the distance distortion estimates for ricci flow ( * ? ? ?
* ; * ? ? ?
* section 27 ) , there is a universal constant @xmath372 so that whenever @xmath373 , we have @xmath374 .
then shi s local derivative estimates ( * ? ? ?
* ; * ? ? ?
* appendix d ) give estimates @xmath352 on @xmath375 .
since @xmath376 , we have @xmath377 .
the @xmath10-noncollapsing statement gives an explicit lower bound @xmath378 on a slightly smaller parabolic ball , which after reducing @xmath365 , we can take to be of the form @xmath364 .
if @xmath15 is a ricci flow solution and @xmath379 { \rightarrow}{{\mathcal m}}$ ] is a time - preserving spacetime curve then we define @xmath380 using the spacetime metric @xmath381 .
the next lemma says that given a point @xmath382 in a @xmath10-solution ( in the sense of appendix [ appkappa ] ) , it has a large backward parabolic neighborhood so that any point @xmath383 in the parabolic neighborhood can be connected to @xmath382 by a time - preserving curve whose length is controlled by @xmath384 , and along which the scalar curvature is controlled by @xmath384 .
[ boxlemma ] given @xmath385 , there exist @xmath386 and @xmath387 with the following property . if @xmath15 is a @xmath10-solution and @xmath388 then 1 .
there is some @xmath389 with @xmath390 .
the scalar curvature on @xmath391 is at most @xmath392 , and 3 .
given @xmath393 , there is a time - preserving curve @xmath394 \rightarrow p_- \left ( x_0 , t_0 , \frac34 a r(x_0 , t_0)^{- \frac12 } \right)$ ] from @xmath383 to @xmath382 with @xmath395 to prove ( 1 ) , suppose , by way of contradiction , that for each @xmath396 there is a @xmath10-solution @xmath37 and some @xmath397 so that @xmath398 on @xmath399 . by compactness of the space of normalized pointed @xmath10-solutions
( see appendix [ appkappa ] ) , after normalizing so that @xmath400 and passing to a subsequence , there is a limiting @xmath10-solution @xmath401 , defined for @xmath402 , with @xmath403 everywhere . by the weak maximum principle for complete noncompact manifolds (
* ; * ? ? ?
* theorem a.3 ) and the evolution equation for scalar curvature , there is a universal constant @xmath404 ( whose exact value is nt important ) so that if @xmath405 on a time-@xmath4 slice then there is a singularity by time @xmath406 . applying this with @xmath407
gives a contradiction .
part ( 2 ) of the lemma , for some value of @xmath78 , follows from the compactness of the space of normalized pointed @xmath10-solutions . to prove ( 3 )
, the curve which starts as a worldline from @xmath383 to @xmath408 , and then moves as a minimal geodesic from @xmath408 to @xmath382 in the time-@xmath103 slice , has @xmath81-length at most @xmath409 . by a slight perturbation to make it time - preserving , we can construct @xmath410 in @xmath411 with length at most @xmath412 . after redefining @xmath78 ,
this proves the lemma .
the next proposition extends the preceding lemma from @xmath10-solutions to points in ricci flows with surgery . recall that @xmath63 is the global parameter in the definition of ricci flow with surgery .
[ curve ] there is an @xmath413 so that if @xmath414 then the following holds
. given @xmath415 , suppose that @xmath416 , where @xmath78 is the constant from lemma [ boxlemma ] .
then for any @xmath417 , there are @xmath418 and @xmath419 with the following property .
let @xmath15 be a ricci flow with surgery having normalized initial conditions .
given @xmath420 with @xmath421 , suppose that @xmath422
. then there is a time preserving curve @xmath423{\rightarrow}{{\mathcal m}}$ ] with @xmath424 and @xmath425 so that @xmath426 for all @xmath427 $ ] .
we begin by noting that we can find @xmath413 so that if @xmath414 then for any @xmath428 with @xmath376 which is in a canonical neighborhood , lemma [ boxlemma](2 ) implies that @xmath429 on @xmath430 .
if in addition @xmath431 then for any @xmath432 , we have @xmath433 hence the parabolic neighborhood does not intersect the surgery region . to prove the proposition , we start with @xmath382 , and inductively form a sequence of points @xmath434 and the curve @xmath410 as follows , starting with @xmath435 . + * step 1 :* if @xmath436 then go to substep a. if @xmath437 then go to substep b. + * substep a :* since @xmath436 , the point @xmath438 is in a canonical neighborhood . as will be explained , the backward parabolic ball @xmath439 does not intersect the surgery region . applying lemma [ boxlemma ] and taking @xmath440 small , we can find @xmath441 with @xmath442 , and a time - preserving curve @xmath443 \rightarrow p_-(x_{i-1 } , t_{i-1 } , a r(x_{i-1 } , t_{i-1})^{- \ : \frac12})\ ] ] from @xmath444 to @xmath445 whose length is at most @xmath446 along which the scalar curvature is at most @xmath447 .
if @xmath448 then go to step 2 . if @xmath449 then the process is terminated . + * substep b :* since @xmath437 , put @xmath450 and @xmath451\}$ ] .
define @xmath452 \rightarrow { { \mathcal m}}$ ] to be the worldline @xmath453 . if @xmath448 then go to step 2 .
( note that @xmath454 . ) if @xmath455 then the process is terminated . +
* step 2 :* increase @xmath456 by one and go to step 1 .
+ to recapitulate the iterative process , if @xmath457 is large then there may initially be a sequence of substep a s .
since the curvature decreases by a factor of at least two for each of these , the number of these initial substeps is bounded above by @xmath458 .
thereafter , there is some @xmath445 so that @xmath459 .
we then go backward in time along a segment of a worldline until we either hit a point @xmath444 with @xmath460 , or we hit time zero .
if we hit @xmath444 then we go back to substep a , which produces a point @xmath461 with at most half as much scalar curvature , etc .
we now check the claim in substep a that the backward parabolic ball @xmath462 does not intersect the surgery region . in the initial sequence of substep a s ,
we always have @xmath463 .
if we return to substep a sometime after the initial sequence , we have @xmath464 either way , from the first paragraph of the proof , we conclude that @xmath465 does not intersect the surgery region .
we claim that the iterative process terminates .
if not , the decreasing sequence @xmath466 approaches some @xmath467 . for an infinite number of @xmath456 , we must have @xmath468 , which converges to @xmath469 .
consider a large @xmath456 with @xmath468 .
the result of substep a is a point @xmath461 with @xmath470 .
if @xmath456 is large then this is less than @xmath471 .
hence one goes to substep b to find @xmath472 with @xmath473 .
however , there is a double - sided bound on @xmath474 for @xmath475 $ ] , coming from the curvature bound on a backward parabolic ball in ( * ? ? ?
* ; * ? ? ?
* lemma 70.1 ) and shi s local estimates ( * ? ? ?
* ; * ? ? ?
* appendix d ) .
this bound implies that the amount of backward time required to go from a point with scalar curvature @xmath476 to a point with scalar curvature @xmath477 satisfies @xmath478 this contradicts the fact that @xmath479 .
we note that the preceding argument can be made effective .
this gives a upper bound @xmath480 on the number of points @xmath444 , of the form @xmath481 .
we now estimate the length of @xmath410 .
the contribution to the length from segments arising from substep b is at most @xmath482 .
the contribution from segments arising from substep a is bounded above by @xmath483 .
it remains to estimate the scalar curvature along @xmath410 . along a portion of @xmath410 arising from substep
a , the scalar curvature is bounded above by @xmath484 . along a portion of @xmath410 arising from substep b
, the scalar curvature is bounded above by @xmath485 .
thus we can take @xmath486 .
this proves the proposition .
finally , we give an estimate on the volume of the high - curvature region in a ricci flow with surgery .
this estimate will be used to prove the volume convergence statement in theorem [ thm_main_convergence ] .
[ volprop ] given @xmath415 , there are functions @xmath487 , with @xmath488 and @xmath489 , having the following property .
let @xmath15 be a ricci flow with surgery , with normalized initial condition .
let @xmath490 denote its initial volume .
given @xmath491 , if @xmath4 is not a surgery time then let @xmath492 be the volume of the corresponding superlevel set of @xmath358 in @xmath79 .
if @xmath4 is a surgery time , let @xmath492 be the volume of the corresponding superlevel set of @xmath358 in the postsurgery manifold @xmath493 .
then if @xmath494 $ ] , we have @xmath495 suppose first that @xmath15 is a smooth ricci flow .
given @xmath494 $ ] , let @xmath496 be the identity map . for @xmath497
, put @xmath498 let @xmath499 { \rightarrow}{{\mathcal m}}$ ] be the worldline of @xmath500
. from the ricci flow equation , @xmath501 suppose that @xmath502 satisfies @xmath503 .
then @xmath504 . let @xmath497 be the point where the worldline of @xmath11 hits @xmath65 . from ( [ gradest ] )
we have @xmath505 as long as @xmath506 .
let @xmath507 be the smallest number so that @xmath506 for all @xmath508 $ ] . since @xmath333 ,
the normalized initial conditions imply that @xmath509 . from ( [ etaeqn ] ) , if @xmath508 $ ] then @xmath510 in particular , @xmath511 from ( [ changeofr ] ) , @xmath512 hence @xmath513 on the other hand , for all @xmath514 , equation ( [ scalarsurg ] ) gives @xmath515 so @xmath516 thus @xmath517 integrating over such @xmath497 , arising from worldlines emanating from @xmath518 , we conclude that @xmath519 so the conclusion of the proposition holds in this case with @xmath520 now suppose that @xmath15 has surgeries . for simplicity of notation
, we assume that @xmath4 is not a surgery time ; otherwise we replace @xmath79 by @xmath493 .
we can first apply the preceding argument to the subset of @xmath518 consisting of points whose worldline goes back to @xmath65 .
the conclusion is that the volume of this subset is bounded above by @xmath521 , where @xmath522 is the same as in ( [ eps1 ] ) .
now consider the subset of @xmath518 consisting of points whose worldline does not go back to @xmath65 .
we can cover such points by the forward images of surgery caps ( or rather the subsets thereof which go forward to time @xmath4 ) , for surgeries that occur at times @xmath523 .
let @xmath524 be the total volume of the surgery caps for surgeries that occur at time @xmath525 .
let @xmath526 be the total volume that is removed at time @xmath525 by the surgery process . from the nature of the surgery process (
* ; * ? ? ?
* section 72 ) , there is an increasing function @xmath527 , with @xmath528 , so that @xmath529 this is essentially because the surgery procedure removes a long capped tube , whose length ( relative to @xmath530 ) is large if @xmath5 is small , and replaces it by a hemispherical cap .
on the other hand , using ( [ volumesurg ] ) , @xmath531 since surgeries up to time @xmath4 can not remove more volume than was initially present or generated by the ricci flow .
the time-@xmath4 volume coming from forward worldlines of surgery caps is at most @xmath532 .
hence the proposition is true if we take @xmath533
in this section we prove theorem [ thm_convergence_flow_with_surgery ] except for the statement about the continuity of @xmath534 , which will be proved in corollary [ cor_volume_continuous ] .
the convergence assertion in theorem [ thm_convergence_flow_with_surgery ] involves a sequence @xmath535 of ricci flows with surgery , where the functions @xmath7 and @xmath10 are fixed , but @xmath536 ; hence @xmath537 and @xmath538 also go to zero .
we will conflate these ricci flows with surgery with their associated ricci flow spacetimes ; see appendix [ rfsurgery ] .
[ thm_convergence_flow_with_surgery ] let @xmath34 be a sequence of ricci flows with surgery with normalized initial conditions such that : * the time - zero slices @xmath35 are compact manifolds that lie in a compact family in the smooth topology . *
@xmath38 .
then after passing to a subsequence , there is a singular ricci flow @xmath539 , and a sequence of diffeomorphisms @xmath540
so that 1 . @xmath41 and @xmath42 are open subsets .
2 . given @xmath541 and @xmath47 , we have @xmath542 and @xmath543 for all sufficiently large @xmath48 .
3 . @xmath50 is time preserving , and the sequences @xmath544 and @xmath52 converge smoothly on compact subsets of @xmath45 to @xmath53 and @xmath54 , respectively .
( note that by ( [ eqn_v_j_contains ] ) any compact set @xmath545 will lie in the interior of @xmath181 , for all sufficiently large @xmath48 . ) 4 . for every @xmath546
, we have @xmath547 and @xmath548 where @xmath549 , @xmath550 , and @xmath551 . 5 .
@xmath50 is asymptotically volume preserving : if @xmath552 denote the respective volume functions @xmath553 and @xmath554 , then @xmath59 uniformly on compact subsets of @xmath555 . because of the normalized initial conditions , the ricci flow solution @xmath556 is smooth on the time interval @xmath557 $ ] . as a technical device ,
we first extend @xmath556 backward in time to a family of metrics @xmath558 which is smooth for @xmath559 $ ] . to do this , as in ( * ? ? ?
* ; * ? ? ?
* pf . of theorem 1.2.6 ) , there is an explicit smooth extension @xmath538 of @xmath560 to the time interval @xmath561 $ ] , with values in smooth covariant @xmath71-tensor fields .
the extension on @xmath562 $ ] depends on the time - derivatives of @xmath560 at @xmath563 which , in turn , can be expressed in terms of @xmath564 by means of repeated differentiation of the ricci flow equation .
let @xmath565 $ ] be a fixed nonincreasing smooth function with @xmath566 } = 1 $ ] and @xmath567 .
given @xmath1 , for @xmath568 put @xmath569 using the precompactness of the space of initial conditions , we can choose @xmath63 small enough so that @xmath558 is a riemannian metric for all @xmath48 and all @xmath402
. then 1 .
@xmath558 is smooth in @xmath570 $ ] , 2 .
@xmath558 is constant in @xmath4 for @xmath571 , and 3 . for @xmath402
, @xmath558 has uniformly bounded curvature and curvature derivatives , independent of @xmath48 .
let @xmath297 be the spacetime riemannian metric on @xmath37 ( see definition [ spacetimemetric ] ) .
choose a basepoint @xmath572 .
after passing to a subsequence , we can assume that for all @xmath48 , @xmath573 where @xmath574 is the parameter of lemma [ lemma2.1 ] .
put @xmath575 , where @xmath43 is the scalar curvature function of the ricci flow metric @xmath556 . put @xmath576 . if @xmath577 then @xmath578 and @xmath579 . in particular
, @xmath580 so @xmath581 avoids the surgery region .
we claim that there are functions @xmath234
, @xmath582 so that definition [ definition3.1 ] holds for all @xmath48 .
suppose that @xmath583 . when @xmath402 there is nothing to prove , so we assume that @xmath546 . if @xmath584 $ ] then the normalized initial conditions give uniform control on @xmath558 .
if @xmath585 then @xmath586 where @xmath587 is the parameter from lemma [ lemma2.1 ] and we use the assumption that @xmath333 from the beginning of section [ surgerysec ] .
then @xmath588 so the parabolic ball @xmath589 of lemma [ lemma2.1 ] does not intersect the initial time slice . as @xmath590 from ( [ canapply ] ) , we can apply lemma [ lemma2.1 ] to show that @xmath591 is @xmath592-controlled in the sense of definition [ definition3.1first ] . note that lemma [ lemma2.1 ] gives bounds on the spatial and time derivatives of the curvature tensor of @xmath558 , which implies bounds on the derivatives of the curvature tensor of the spacetime metric @xmath593 . finally , the function @xmath309 and the vector field @xmath594 , along with their covariant derivatives , are trivially bounded in terms of @xmath297 . after passing to a subsequence
we may assume that the number @xmath480 of connected components of the initial time slice @xmath595 is independent of @xmath48 .
then the theorem follows from the special case when the initial times slices are connected , since we may apply it to the components separately .
therefore we are reduced to proving the theorem under the assumption that @xmath595 is connected . after passing to a subsequence , theorem [ theorem3.8 ]
now gives a pointed @xmath174-controlled tuple @xmath291 satisfying ( 1)-(4 ) of theorem [ theorem3.8 ] .
we truncate @xmath45 to the subset @xmath596 .
we now verify the claims of theorem [ thm_convergence_flow_with_surgery ] .
part ( 1 ) of theorem [ thm_convergence_flow_with_surgery ] follows from the statement of theorem [ theorem3.8 ] . given @xmath597 and @xmath598 , proposition [ curve ]
implies that there are @xmath599 and @xmath600 so that the set @xmath601 is contained in the metric ball @xmath179 in the riemannian manifold @xmath602 .
hence by the definition of @xmath299 , and part ( 1 ) of theorem [ theorem3.8 ] , we get ( [ eqn_u_j_contains ] ) .
pick @xmath603 . put @xmath604 . by part ( 4 ) of theorem [ theorem3.8 ] , we know that @xmath605 belongs to @xmath181 for large @xmath48
. now part ( 3 ) of theorem [ theorem3.8 ] and the definition of @xmath299 imply that @xmath606 if @xmath48 is large then it makes sense to define @xmath607 .
then for @xmath48 large , @xmath608 and so by proposition [ curve ] , there is a time - preserving curve @xmath609{\rightarrow}{{\mathcal m}}^j$ ] such that @xmath610 } r({\gamma}_j(t ) ) \right ) < c = c(r(m_\infty),t_\infty)\,.\ ] ] by part ( 1 ) of theorem [ theorem3.8 ] we know that @xmath611 for large @xmath48 . by part ( 2 ) of theorem [ theorem3.8 ] , for large @xmath48
the map @xmath50 is an almost - isometry .
hence there are @xmath612 and @xmath613 so that @xmath605 is contained in the metric ball @xmath182 in the riemannian manifold @xmath614 combining this with ( [ eqn_psi_infty ] ) and part ( 1 ) of theorem [ theorem3.8 ] yields ( [ eqn_v_j_contains ] ) .
this proves part ( 2 ) of theorem [ thm_convergence_flow_with_surgery ] .
part ( 3 ) of theorem [ thm_convergence_flow_with_surgery ] now follows from part ( 2 ) of the theorem and parts ( 2 ) and ( 3 ) of theorem [ theorem3.8 ] .
equation ( [ rest ] ) follows from ( [ scalarsurg ] ) and the smooth approximation in part ( 3 ) of theorem [ thm_convergence_flow_with_surgery ] .
let @xmath615 be the volume of the @xmath616-sublevel set for the scalar curvature function on @xmath617 .
( if @xmath4 is a surgery time , we replace @xmath79 by @xmath493 . )
let @xmath618 be the volume of the @xmath616-sublevel set for the scalar curvature function on @xmath619 .
then @xmath620 part ( 4 ) of theorem [ thm_convergence_flow_with_surgery ] now follows from combining this with ( [ volumesurg ] ) .
next , we verify the volume convergence assertion ( 5 ) .
let @xmath621 denote the sup norm on @xmath622)$ ] . by parts ( 2 ) and ( 3 ) of theorem [ thm_convergence_flow_with_surgery ] ,
there is some @xmath623 , with @xmath624 , so that @xmath625 also , if @xmath626 then @xmath627 proposition [ volprop ] implies @xmath628 combining ( [ c1 ] ) and ( [ comb2 ] ) , and taking @xmath629 , gives @xmath630 combining ( [ comb1 ] ) , ( [ comb2 ] ) and ( [ comb3 ] ) yields @xmath631 given @xmath632 , we can choose @xmath633 so that @xmath634 .
given this value of @xmath616 , we can choose @xmath635 so that if @xmath636 then @xmath637 hence if @xmath636 then @xmath638 . this shows that @xmath59 , uniformly on compact subsets of @xmath555 , and proves part ( 5 ) of theorem theorem [ thm_convergence_flow_with_surgery ] . finally , we check that @xmath45 is a singular ricci flow in the sense of definition [ def_singular_ricci_flow ] .
using part ( 3 ) of theorem [ thm_convergence_flow_with_surgery ] one can pass the hamilton - ivey pinching condition , canonical neighborhoods , and the noncollapsing condition from the @xmath37 s to @xmath45 , and so parts ( b ) and ( c ) of definition [ def_singular_ricci_flow ] hold .
we now verify part ( a ) of definition [ def_singular_ricci_flow ] .
we start with a statement about parabolic neighborhoods in @xmath45 . given @xmath639 and @xmath598 ,
suppose that @xmath640 has @xmath641 and @xmath642 . for large @xmath48 ,
put @xmath643 .
lemma [ lemma2.1 ] supplies parabolic regions centered at the @xmath644 s which pass to @xmath45 .
hence for some @xmath645 , the forward and backward parabolic regions @xmath646 and @xmath647 are unscathed . there is some @xmath648 so that when equipped with the spacetime metric @xmath649 , the union @xmath650 is @xmath225-bilipschitz homeomorphic to a euclidean parabolic region . from ( [ rest ] ) , we know that @xmath358 is bounded below on @xmath651}$ ] .
in order to show that @xmath358 is proper on @xmath651}$ ] , we need to show that any sequence in a sublevel set of @xmath358 has a convergent subsequence .
suppose that @xmath652 is a sequence with @xmath653 and @xmath654 .
after passing to a subsequence , we may assume that @xmath655 .
then the regions @xmath656 will intersect the time slice @xmath657 in regions whose volume is bounded below by @xmath658 . by the volume bound in part ( 4 ) of theorem [ thm_convergence_flow_with_surgery ] ,
only finitely many of these can be disjoint .
therefore , after passing to a subsequence , @xmath659 is contained in @xmath660 for some @xmath661 .
as @xmath660 has compact closure in @xmath662 , a subsequence of @xmath659 converges .
this verifies part ( a ) of definition [ def_singular_ricci_flow ] .
in this section we prove some initial structural properties of ricci flow spacetimes and singular ricci flows . in subsection [ subsec6.1 ]
we justify the maximum principle on a ricci flow spacetime and apply it to a get a lower scalar curvature bound . in subsection [ subsec6.2 ] we prove some results about volume evolution for ricci flow spacetimes which satisfy certain assumptions , that are satisfied in particular for singular ricci flows . the main result in subsection [ subsec6.3 ]
says that if @xmath15 is a singular ricci flow and @xmath663 { \rightarrow}{{\mathcal m}}$ ] are two time - preserving curves , such that @xmath664 and @xmath665 are in the same connected component of @xmath106 , then @xmath666 and @xmath667 are in the same connected component of @xmath668 for all @xmath669 $ ] .
we recall the notion of a ricci flow spacetime from definition [ def_ricci_flow_spacetime ] . in this section
, we will consider it to only be defined for nonnegative time , i.e. @xmath16 takes value in @xmath555 .
we also recall the metrics @xmath81 and @xmath82 from definition [ spacetimemetric ] .
let @xmath670 be the dimension of @xmath15 .
our notation is * @xmath671 , * @xmath28 } = { \mathfrak{t}}^{-1}([a , b])$ ] and * @xmath672)$ ] . in this subsection
we prove a maximum principle on ricci flow spacetimes and apply it to get a lower bound on scalar curvature .
[ maximumprin ] let @xmath15 be a ricci flow spacetime .
given @xmath673 , let @xmath674 be a smooth vector field on @xmath68 with @xmath675 . given a smooth function @xmath676 \rightarrow { \mathbb r}$ ] , suppose that @xmath677 is a proper function , bounded above , which satisfies @xmath678 suppose further that @xmath679 \rightarrow { \mathbb r}$ ] satifies @xmath680 with initial condition @xmath681 . if @xmath682 on @xmath65 then @xmath683 on @xmath68 .
as in ( * ? ? ?
* ; * ? ? ?
* pf . of theorem 3.1.1 ) , for @xmath1 , we consider the ode @xmath684 with initial condition @xmath685 .
it suffices to show that for all small @xmath63 we have @xmath686 on @xmath68 . if not then we can find some @xmath1 so that the property @xmath686 fails on @xmath68 .
as @xmath687 is proper and bounded above , there is a first time @xmath103 so that the property fails on @xmath104 , and an @xmath688 so that @xmath689 .
the rest of the argument is the same as in ( * ? ? ?
* ; * ? ? ?
* pf . of theorem 3.1.1 ) .
[ scalarlem ] let @xmath15 be a ricci flow spacetime .
suppose that for each @xmath69 , the scalar curvature @xmath358 is proper and bounded below on @xmath68 .
suppose that the initial scalar curvature bounded below by @xmath690 , for some @xmath691 .
then @xmath692 since @xmath358 is proper on @xmath68 , we can apply lemma [ maximumprin ] to the evolution equation for @xmath693 and follow the standard proof to get ( [ scalarcurv ] ) . in this subsection
we first justify a fubini - type statement for ricci flow spacetimes .
then we show that certain standard volume estimates for smooth ricci flows extend to the setting of ricci flow spacetimes under two assumptions : first that the quasiparabolic metric is complete along worldlines that do not terminate at the time - zero slice , and second that in any time slice almost all points have worldlines that extend backward to time zero .
the fubini - type statement is the following .
[ vollemma ] let @xmath15 be a ricci flow spacetime .
given @xmath694 , suppose that @xmath695 } { \rightarrow}{\mathbb r}$ ] is measurable and bounded below .
suppose that @xmath696}$ ] has finite volume with respect to @xmath697 .
then @xmath698 } } f \ : { \operatorname{dvol}}_{g_{{\mathcal m } } } = \int_{t_1}^{t_2 } \int_{m_t } f \ : { \operatorname{dvol}}_{g(t ) } \ : dt.\ ] ] as @xmath699 is bounded below on @xmath700}$ ] , for the purposes of the proof we can add a constant to @xmath699 and assume that it is positive
. given @xmath701}$ ] , there is an time - preserving embedding @xmath702 with @xmath703 ( where @xmath704 ) whose image is a neighborhood of @xmath11
. we can cover @xmath700}$ ] by a countable collection @xmath705 of such neighborhoods , with a subordinate partition of unity @xmath706 .
let @xmath707 be the corresponding map . as @xmath708 } } \phi_i \ : f \ : { \operatorname{dvol}}_{g_{{\mathcal m } } } = \int_{t_1}^{t_2 } \int_{e(x_i , t ) } \phi_i \ : f \ : { \operatorname{dvol}}_{g(t ) } \ : dt,\ ] ] we obtain @xmath709 } } f \ : { \operatorname{dvol}}_{g_{{\mathcal m } } } & = \sum_i \int_{{{\mathcal m}}_{[t_1,t_2 ] } } \phi_i \ : f \ : { \operatorname{dvol}}_{g_{{\mathcal m } } } \\ & = \sum_i \int_{t_1}^{t_2 } \int_{e_i(x_i , t ) } \phi_i \ : f \ : { \operatorname{dvol}}_{g(t ) } \ : dt \notag \\ & = \int_{t_1}^{t_2 } \sum_i \int_{e_i(x_i , t ) } \phi_i \ : f \ : { \operatorname{dvol}}_{g(t ) } \ : dt \notag \\ & = \int_{t_1}^{t_2 } \int_{{{\mathcal m}}_t } f \ : { \operatorname{dvol}}_{g(t ) } \ : dt .
\notag \end{aligned}\ ] ] this proves the lemma .
we now prove some results about the behavior of volume in ricci flow spacetimes .
[ vol1 ] let @xmath15 be a ricci flow spacetime .
suppose that 1 .
the quasiparabolic metric @xmath82 of definition [ spacetimemetric ] is complete along worldlines that do not terminate at the time - zero slice .
2 . if @xmath710 is the set of points whose maximal worldline does not extend backward to time zero , then @xmath711 has measure zero with respect to @xmath712 , for each @xmath546 .
3 . the initial time slice @xmath65 has volume @xmath713 .
the scalar curvature is proper and bounded below on time slabs @xmath68 , and the initial time slice has scalar curvature bounded below by @xmath690 , with @xmath691 .
let @xmath714 be the volume of @xmath79
. then 1 .
@xmath358 is integrable on @xmath700}$ ] .
3 . for all @xmath716 , @xmath717 } } r \ : { \operatorname{dvol}}_{g_{{{\mathcal m}}}}.\ ] ] 4 . the volume function @xmath714 is absolutely continuous .
5 . given @xmath718 ,
we have @xmath719 given @xmath720 , there is a partition @xmath700 } = { { \mathcal m}}_{[t_1 , t_2]}^\prime \cup { { \mathcal m}}_{[t_1 , t_2]}^{\prime \prime } \cup { { \mathcal m}}_{[t_1 , t_2]}^{\prime \prime \prime}$ ] , where 1 .
a point in @xmath700}^\prime$ ] has a worldline that intersects both @xmath721 and @xmath722 .
2 . a point in @xmath700}^{\prime \prime}$ ] has a worldline that intersects @xmath721 but not @xmath722 .
3 . a point in @xmath700}^{\prime \prime \prime}$ ] has a worldline that does not intersect @xmath721 . by our assumptions , @xmath700}^{\prime \prime \prime }
\subset \bigcup_{t \in [ t_1 , t_2 ] } b_t$ ] has measure zero with respect to @xmath723 ; c.f .
the proof of lemma [ vollemma ] . put @xmath724}^\prime \cap m_{t_1}$ ] and @xmath725}^{\prime \prime } \cap m_{t_1}$ ] . for @xmath726
$ ] , there is a natural embedding @xmath727 coming from flowing along worldlines .
the complement @xmath728 has measure zero .
thus @xmath729 and @xmath730 given @xmath731 , let @xmath732 \rightarrow { { \mathcal m}}_{[t_1,t_2]}$ ] be its worldline .
put @xmath733 from the ricci flow equation , @xmath734 using lemma [ scalarlem ] , @xmath735 when @xmath736 , this proves part ( 1 ) of the proposition . next , @xmath737}^{\prime } } r \ : { \operatorname{dvol}}_{g_{{{\mathcal m } } } } , \notag\end{aligned}\ ] ] where we applied lemma [ vollemma ] with @xmath738}^{\prime}}$ ] in the last step . given @xmath739 ,
let @xmath740 be the supremal extension time of its worldline .
from the completeness of @xmath82 on worldlines that do not terminate at the time - zero slice , @xmath741 thus @xmath742 , so @xmath743}^{\prime \prime } } r \ : { \operatorname{dvol}}_{g_{{{\mathcal m}}}}. \notag\end{aligned}\ ] ] part ( 3 ) of the proposition follows from combining equations ( [ add1 ] ) , ( [ add2 ] ) , ( [ add3 ] ) and ( [ add4 ] ) .
part ( 2 ) of the proposition is now an immediate consequence . by lemma [ vollemma ] and part ( 3 ) of the proposition ,
the function @xmath744 is locally-@xmath745 on @xmath60 with respect to lebesgue measure .
this implies part ( 4 ) of the proposition . to prove part ( 5 ) of the proposition , using lemma [ scalarlem ] and parts ( 1 ) and ( 3 ) of the proposition , we have
@xmath746 this proves the proposition . in this subsection
we collect a number of properties of singular ricci flows , the latter being in the sense of definition [ def_singular_ricci_flow ] .
we first show the completeness of the quasiparabolic metric .
[ quasip ] if @xmath15 is a singular ricci flow then the quasiparabolic metric @xmath82 of definition [ spacetimemetric ] is complete away from the time - zero slice .
suppose that @xmath747 is a curve that goes to infinity in @xmath15 , with @xmath748 bounded away from zero .
we want to show that its quasiparabolic length is infinite .
if @xmath748 is not bounded then the quasiparabolic length of @xmath410 is infinite from the definition of @xmath82 , so we can assume that @xmath748 takes value in some interval @xmath749 $ ] .
since @xmath358 is proper and bounded below on @xmath68 , we have @xmath750 .
after truncating the initial part of @xmath410 , we can assume that @xmath751 for all @xmath752 .
in particular , each point @xmath753 is in a canonical neighborhood .
now @xmath754 so @xmath755 the gradient estimates in ( [ gradest ] ) , of the form @xmath756 for some universal @xmath757 .
we deduce that @xmath758 since the left - hand side is infinite , the quasiparabolic length of @xmath410 must be infinite .
this proves the lemma .
the next lemma gives the existence of unscathed forward and backward parabolic neighborhoods of a certain size around a point , along with geometric bounds on those neighborhoods .
[ singforback ] let @xmath15 be a singular ricci flow
. given @xmath415 , there are numbers @xmath759 , @xmath343 and @xmath344 , @xmath140 , with the following property . if @xmath109 and @xmath760 , put @xmath761 .
then 1 . the forward parabolic ball @xmath762 and the backward parabolic ball @xmath763 are unscathed .
@xmath350 , @xmath351 and @xmath352 on the union @xmath764 of the forward and backward parabolic balls .
the proof is the same as that of lemma [ lemma2.1 ] .
the next two propositions characterize the high - scalar - curvature part of a time slice .
[ goodeps ] let @xmath15 be a singular ricci flow . for all @xmath765 ,
there is a scale function @xmath766 with @xmath767 , such that for every point @xmath109 with @xmath768 the @xmath769-canonical neighborhood assumption holds , and moreover @xmath770 is @xmath769-modelled on a @xmath10-solution .
( recall that here @xmath771 , i.e. we are suppressing the time dependence in our notation . )
the proof is similar to the proof of ( * ? ? ?
* ; * ? ? ?
* theorem 52.7 ) .
suppose that the proposition is not true .
then there is a sequence of singular ricci flows @xmath772 , along with points @xmath773 , that together provide a counterexample . after passing to a subsequence
, we extract a limiting flow , as in step 2 of the proof of ( * ? ? ?
* ; * ? ? ?
* theorem 52.7 ) . in that proof ,
the existence of a limiting flow used a point selection argument from step 1 of the proof . in the present case , because of the canonical neighborhood assumption in the definition of singular ricci flow , we do not have to perform point selection in order to extract a limiting flow .
similarly , the use of the existing canonical neighborhood assumption simplifies step 2 of the proof of ( * ? ? ?
* ; * ? ? ?
* theorem 52.7 ) .
the rest of the proof of the proposition is the same as in steps 3 and 4 of the proof of ( * ? ? ?
* ; * ? ? ?
* theorem 52.7 ) . as a further point ,
time slices are not assumed to be compact as in ( * ? ? ?
* ; * ? ? ?
* theorem 52.7 ) , and are therefore not necessarily complete . to deal with this , there are places in the proof where one applies lemma [ singforback ] to compensate for any incompleteness .
[ prop_large_r_structure ] let @xmath15 be a singular ricci flow . for any @xmath85 and @xmath774 , there exist @xmath775 and @xmath776 such that for every @xmath777 , each connected component of the time slice @xmath79 has finitely many ends , each of which is an @xmath778-horn .
morever for every @xmath779 , the superlevel set @xmath780 is contained in a finite disjoint union of properly embedded three - dimensional submanifolds - with - boundary @xmath781 such that 1 .
each @xmath782 is contained in the superlevel set @xmath783 .
2 . the boundary @xmath784 has scalar curvature in the interval @xmath785 .
3 . for each @xmath456 one of the following holds : 1 .
@xmath782 is diffeomorphic to @xmath786 or @xmath787 and consists of @xmath778-neck points .
note that here the interval @xmath788 can be open ( a double horn ) , closed ( a tube ) or half - open ( a horn ) .
@xmath782 is diffeomorphic to @xmath789 or @xmath790 and its boundary @xmath791 consists of @xmath778-neck points . 3 .
@xmath782 is diffeomorphic to @xmath792 , @xmath793 , or @xmath794 .
@xmath782 is diffeomorphic to a spherical space form other than @xmath792 or @xmath793 .
4 . in cases ( b ) and ( c ) of ( 3 ) , if @xmath795 is a subset consisting of non-@xmath778-neck points , such that for any two distinct elements @xmath796 we have @xmath797 , then the cardinality @xmath798 is at most @xmath799 in case ( b ) and at most @xmath71 in case ( c ) . 5 .
each @xmath782 with nonempty boundary has volume at least @xmath800 .
the proof is the same as in ( * ? ? ?
* ; * ? ? ?
* section 67 ) .
we now prove a statement about preservation of connected components when going backwards in time .
[ sameconnected ] let @xmath15 be a singular ricci flow .
if @xmath801{\rightarrow}{{\mathcal m}}$ ] are time - preserving curves , and @xmath115 , @xmath116 lie in the same connected component of @xmath106 , then @xmath117 , @xmath118 lie in the same connected component of @xmath668 for every @xmath119 $ ] .
the idea of the proof is to consider the values of @xmath4 for which @xmath666 can be joined to @xmath667 in @xmath79 , and the possible curves @xmath802 in @xmath79 that join them . among all such curves
@xmath802 , we look at one which minimizes the maximum value of scalar curvature along the curve . call this threshold value of scalar curvature @xmath803 .
we will argue that @xmath802 can only intersect the high - scalar - curvature part of @xmath79 in its neck - like regions .
but the scalar curvature in a neck - like region is strictly decreasing when one goes backward in time ; this will imply that @xmath803 , when large , is decreasing when one goes backward in time , from which the lemma will follow . to begin the formal proof ,
suppose that the lemma is false .
let @xmath804 $ ] be the set of times @xmath119 $ ] such that @xmath117 and @xmath118 lie in the same connected component of @xmath79 ; note that @xmath805 is open .
put @xmath806\subset s\}$ ] .
then @xmath807 since @xmath808}$ ] is a product .
also , for @xmath809 and @xmath810 $ ] close to @xmath811 , if @xmath812 is the worldline of @xmath813 at time @xmath4 then @xmath814 lies in the same connected component of @xmath79 as @xmath815 . therefore , after reducing @xmath507 if necessary , we may assume without loss of generality that @xmath816 is a worldline . for @xmath669 $ ] and @xmath47 , put @xmath817 . for @xmath818
$ ] , let @xmath819 be the set of @xmath820 such that @xmath117 and @xmath118 lie in the same connected component of @xmath821 . put @xmath822 . since @xmath823 is proper , the sets @xmath824 are compact and nested , which implies that @xmath117 and @xmath118 lie in the same component of @xmath825 , i.e. @xmath826 . by lemma [ singforback ] , there is a @xmath827 such that for any @xmath828 $ ] and any @xmath829 there is a @xmath830 , where @xmath831 is a continuous function that is nonincreasing in @xmath112 , such that the worldline @xmath832 of @xmath11 is defined and satisfies @xmath833 in the time interval @xmath834 . now for @xmath835 $ ] and @xmath836 satisfying @xmath837 , let @xmath838 denote the result of flowing @xmath839 under @xmath21 for an elapsed time @xmath840 .
then @xmath841 is well - defined and contains @xmath842 and @xmath843 in the same component , so @xmath844 this implies that @xmath845{\rightarrow}{\mathbb r}$ ] is locally lipschitz ( in particular continuous ) and that @xmath846 as @xmath847 from the right ; otherwise there would be a sequence @xmath848 along which @xmath849 is uniformly bounded above by some @xmath850 , which would allow us to construct @xmath851 whenever @xmath852 , contradicting the definition of @xmath811 .
we now concentrate on @xmath4 close to @xmath811 .
suppose that for some @xmath818 $ ] we have @xmath853}r\circ { \gamma}_0 , \max_{[t_0,t_1]}r\circ { \gamma}_1)\,.\ ] ] then by proposition [ prop_large_r_structure ] the superlevel set @xmath854 is contained in a finite union of components @xmath855 , each diffeomorphic to one of the possibilities ( a)-(d ) in the statement of the proposition .
let @xmath856 be the result of removing from @xmath79 the interior of each @xmath857 that is not of type ( a ) .
since @xmath117 and @xmath118 lie outside @xmath858 , and each @xmath782 of type ( b)-(d ) has at most one boundary component , it follows that @xmath117 and @xmath118 lie in the same connected component of @xmath856 . for @xmath859 close to @xmath4 ,
let @xmath860 be the result of flowing @xmath856 under @xmath21 .
now @xmath861 consists of @xmath9-neck points , and at such a point the scalar curvature is strictly increasing as a function of time .
hence there is a @xmath862 such that the worldline @xmath863{\rightarrow}{{\mathcal m}}$ ] of any @xmath864 satisfies @xmath865 for @xmath866 .
this implies that @xmath867 when @xmath868 is close to @xmath4 , again under the assumption ( [ bigg ] ) , which contradicts ( [ eqn_r_crit_blows_up ] ) .
we now state a result about connecting a point in a singular ricci flow to the time - zero slice by a curve whose length is quantitatively bounded , and along which the scalar curvature is quantitatively bounded .
[ curve2 ] let @xmath15 be a singular ricci flow .
given @xmath869 , there are @xmath418 and @xmath419 with the following property .
suppose that @xmath870 , with @xmath871 . then there is a time preserving curve @xmath423{\rightarrow}{{\mathcal m}}$ ] with @xmath872 and @xmath873 so that @xmath426 for all @xmath427 $ ] .
the proof is the same as that of proposition [ curve ] .
finally , we give a compactness result for the space of singular ricci flows .
[ rfcompactness ] let @xmath874 be a sequence of singular ricci flows with a fixed choice of parameters @xmath63 , @xmath7 and @xmath10 whose initial conditions @xmath875 lie in a compact family in the smooth topology .
then a subsequence of @xmath874 converges , in the sense of theorem [ thm_convergence_flow_with_surgery ] , to a singular ricci flow @xmath45 .
using proposition [ curve2 ] , the proof is the same as that of proposition [ thm_convergence_flow_with_surgery ] . in the setting of proposition [ rfcompactness ] ,
if we instead assume that the ( normalized ) initial conditions have a uniform upper volume bound then we can again take a convergent subsequence to get a limit ricci flow spacetime @xmath45 . in this case
the time - zero slice @xmath876 will generally only be @xmath877-regular , but @xmath45 will be smooth on @xmath878 .
in this section we fix @xmath385 ; the dependence of various constants on this choice of @xmath10 is implicit .
we recall the notion of a @xmath10-solution from appendix [ appkappa ] . in this section
we establish a new dynamical stability property of caps and necks in @xmath10-solutions , which we will use in section [ sec_finiteness ] to show that a bad worldline @xmath94 is confined to a cap region as @xmath4 approaches the blow - up time @xmath879 . conceptually speaking
, the stability assertion is that among pointed @xmath10-solutions , the round cylinder is an attractor under backward flow ; similarly , under forward flow , non - neck points form an attractor .
the rough idea of the argument is as follows .
suppose that @xmath880 and @xmath881 is a @xmath6-neck in a @xmath10-solution @xmath15
. then @xmath882 will also be neck - like as long as @xmath568 is not too negative .
one also knows that @xmath15 has an asymptotic soliton as @xmath108 , which is a shrinking round cylinder .
thus @xmath15 tends toward neck - like geometry as @xmath4 approaches @xmath883 , which is the desired stability property .
however , there is a catch here : the asymptotic soliton is a pointed limit of a sequence where the basepoints are not fixed , and so a priori it says nothing about the asymptotic geometry near @xmath882 as @xmath108 . to address this we exploit the behavior of the @xmath259-function .
we recall the notation @xmath884 from section [ sec_notation ] for the parabolic rescaling of a ricci flow spacetime @xmath15 .
we recall the notation @xmath885 and @xmath886 from section [ sec_notation ] for the standard ricci flow solutions .
we also recall the notion of one ricci flow spacetime being @xmath63-modelled on another one , from appendix [ appcloseness ] .
[ thm_stability_of_necks ] there is a @xmath887 , and for all @xmath888 there is a @xmath889 with the following property .
suppose that 1 .
@xmath15 is a @xmath10-solution with noncompact time slices , 2 .
@xmath890 , 3 .
@xmath891 , and 4 .
@xmath892 is a @xmath893-neck
. then either @xmath15 is isometric to the @xmath894-quotient of a shrinking round cylinder , or for all @xmath895 $ ] , 1 .
@xmath896 is a @xmath897-neck and 2 .
@xmath898 is @xmath899-close to @xmath900 , where @xmath901 .
we recall the notion of a generalized neck from appendix [ appcloseness ] .
[ cor_de_de_over_4 ] if @xmath902 as in the previous theorem , then there is a @xmath903 such that if @xmath15 is a @xmath10-solution with noncompact time slices , @xmath890 , @xmath891 , and @xmath881 is a generalized @xmath904-neck , then @xmath905 is a generalized @xmath906-neck , and @xmath907 . in this subsection
we prove some preparatory results about the convergence of @xmath259-functions with regard to a convergent sequence of three - dimensional @xmath10-solutions .
suppose that we are given that @xmath908 at some point @xmath383 in a @xmath10-solution . by compactness of the space of normalized pointed @xmath10-solutions ( see appendix [ appkappa ] ) , we obtain @xmath909 for some universal function @xmath699 . since @xmath358 is pointwise nondecreasing in time in a @xmath10-solution , we also have @xmath910 whenever @xmath911 .
[ curvbasepoint ] let @xmath912 be a sequence of pointed @xmath10-solutions , with @xmath913 , and let @xmath914 be the reduced distance function with spacetime basepoint at @xmath915 .
then after passing to a subsequence , we have convergence of tuples @xmath916 here the ricci flow convergence is the usual smooth convergence on parabolic balls , and @xmath327 converges to a locally lipschitz function @xmath917 uniformly on compact subsets of @xmath918 , after composition with the comparison map .
since @xmath919 is uniformly bounded above , the compactness of the space of normalized pointed @xmath10-solutions implies that a subsequence , which we relabel as @xmath912 , converges in the pointed smooth topology . along the curve @xmath920
, the scalar curvature is bounded above by @xmath921 . for any @xmath922 , this gives a uniform upper bound on @xmath327 on @xmath923 \subset { { \mathcal m}}^j$ ] . from ( [ ye2.55 ] ) , we get that @xmath259 is uniformly bounded on balls of the form @xmath924 . then from ( [ ye2.57 ] ) , we get that @xmath259 is uniformly bounded on parabolic balls of the form @xmath925 . using ( [ ye2.54 ] ) and passing to a subsequence , we get convergence of @xmath926 to some @xmath917 , uniformly on compact subsets of @xmath918 .
[ prop_convergence_no_curvature_bound ] let @xmath912 be a sequence of pointed @xmath10-solutions , and let @xmath914 be the reduced distance function with spacetime basepoint at @xmath915 .
suppose that @xmath927 is a sequence satisfying @xmath928 .
then after passing to a subsequence , the tuples @xmath929 converge to a tuple @xmath930 where : 1 .
the convergence @xmath931 is smooth on compact subsets of @xmath918 , and @xmath45 is either a @xmath10-solution defined on @xmath932 or the static solution on @xmath933 .
@xmath926 converges to a locally lipschitz function @xmath917 uniformly on compact subsets of @xmath918 , after composition with the comparison diffeomorphisms .
3 . the reduced volume functions @xmath934 converge uniformly on compact sets to the function @xmath935 , where @xmath936 4 . the function @xmath917 satisfies the differential inequalities ( 24.6 ) and ( 24.8 ) of @xcite .
if @xmath937 is constant on some time interval @xmath938\subset ( -\infty,0)$ ] then @xmath45 is a gradient shrinking soliton with potential @xmath917 . from ( [ ye2.57 ] ) , given @xmath922
, there is a uniform bound for @xmath327 on @xmath939 $ ] . from ( [ ye2.53 ] ) , there are bounds for @xmath43 on @xmath939 $ ] .
then we get a uniform scalar curvature bound on the balls @xmath940 , and hence on the parabolic neighborhoods @xmath941 . taking @xmath942 and @xmath943 , and applying a diagonal argument
, we can assume that @xmath944 converges to a @xmath10-solution @xmath945 .
as in the proof of lemma [ curvbasepoint ] , after passing to a subsequence we get @xmath946 for some @xmath917 defined on @xmath918 . the rest is as in @xcite .
[ lem_cyl_l_function ] let @xmath947 be a shrinking round cylinder with @xmath948 at @xmath101 , where the @xmath259-function has spacetime basepoint @xmath949 .
then : 1 . for every @xmath950 the @xmath259-function @xmath951 attains its minimum uniquely at @xmath581 .
@xmath952 , where the coordinate @xmath953 for the @xmath954-factor in @xmath885 satisfies @xmath955 , and @xmath956 3 .
@xmath957 . part ( 1 ) follows from the formula for @xmath958-length : @xmath959 with equality if and only if @xmath960 for all @xmath961 $ ] .
part ( 2 ) follows from applying lemma [ lem_classification_gradient_soliton_kappa_solutions ] in section [ subsec_gradient_shrinking_solitons ] to parabolic rescalings of @xmath15 .
part ( 3 ) is now immediate . from ( [ eqn_l_infinity_cyl ] )
, we get that @xmath917 is strictly decreasing along backward worldlines , except at its minimum value @xmath799 .
in particular , if @xmath962 then @xmath963 for all @xmath964 . by compactness , this stability property is inherited by any tuple which approximates the shrinking round cylinder .
[ lem_cyl_l_function_stability ] for all @xmath774 and @xmath965 , there is a @xmath966 with the following property .
suppose that @xmath967 is a @xmath10-solution , with @xmath259-function based at @xmath949 , so that 1 .
@xmath967 is @xmath968-close to @xmath969 on the time interval @xmath970 $ ] , for some @xmath971 , and 2 .
@xmath972 for all @xmath973 $ ] .
( here @xmath974 need not be one . )
then @xmath975 for all @xmath976 $ ] .
if the lemma fails then for some @xmath774 and @xmath965 , there is a sequence @xmath977 so that for all @xmath48 : * @xmath978 is @xmath979-close to @xmath980 on the time interval @xmath970 $ ] , for some @xmath981 . *
@xmath982 for all @xmath973 $ ] . *
@xmath983 for some @xmath984 $ ] . using ( [ eqn_l_infinity_cyl ] ) and passing to a limit ,
we obtain @xmath985 with the property that @xmath986 and @xmath987 for some @xmath976 $ ] . but this contradicts the formula @xmath988 . in this subsection
we prove a backward stability result for cylindrical regions , initially without fixing the basepoint .
we then prove a version fixing the basepoint , which will imply theorem [ thm_stability_of_necks ] .
we start with a preliminary lemma about reduced volume .
[ redvol2 ] there are @xmath989 and @xmath990 so that if @xmath15 is a @xmath10-solution defined on @xmath562 $ ] with @xmath991 then @xmath992 . using the monotonicity of the reduced volume ,
if the lemma fails then there is a sequence @xmath34 of @xmath10-solutions and points @xmath993 so that @xmath994 and @xmath995 . after passing to a subsequence ,
there is pointed convergence to a pointed @xmath10-solution @xmath996 . then @xmath997 and @xmath998 for all @xmath402 .
this is a contradiction ( * ? ? ?
* ; * ? ? ?
* pf . of proposition 39.1 ) .
[ prop_cylinder_stability_1_noncompact ] for all @xmath774 and @xmath827 , there is a @xmath999 with the following property .
suppose that @xmath15 is a noncompact @xmath10-solution defined on @xmath1000 $ ] and : * @xmath1001 is a point with @xmath1002 .
* @xmath1003 is the @xmath259-function with spacetime basepoint @xmath949 .
* @xmath1004 , and @xmath1005 is a point where the reduced distance satisfies @xmath1006 .
then one of the following holds : 1 .
the tuple @xmath1007 is @xmath778-close to a triple @xmath1008 , where @xmath1009 and @xmath917 is the asymptotic @xmath259-function of ( [ eqn_l_infinity_cyl ] ) .
note that @xmath1010 need not be at the minimum of @xmath917 in @xmath1011 .
@xmath15 is isometric to a @xmath894-quotient of a shrinking round cylinder .
if the proposition were false then for some @xmath774 there would be a sequence @xmath912 of pointed @xmath10-solutions , and a sequence @xmath1012 , such that * @xmath1013 .
* there is a point @xmath1014 with @xmath1015 , such that ( 1 ) and ( 2 ) fail for @xmath1016 . from lemma [ redvol2 ] , there are constants @xmath989 and @xmath990 so that for each @xmath48 , we have @xmath1017 . using the monotonicity of the reduced volume and the existence of the asymptotic soliton , there is a uniform positive lower bound on @xmath1018 . after passing to a subsequence
, we can find @xmath1019 so that @xmath1020 , and the reduced volume is constant to within a factor @xmath1021 on a time interval @xmath1022 $ ] where @xmath1023 . by appendix [ appkappa ] , for each @xmath48
there is a point @xmath1024 with @xmath1025 . after passing to a subsequence , by proposition [ prop_convergence_no_curvature_bound ]
, the sequence of parabolically rescaled flows @xmath1026 pointed - converges to a gradient shrinking soliton @xmath1027 that is either a @xmath10-solution or is flat @xmath933 . since the reduced volume of @xmath1027 is also bounded above by @xmath1028 , it can not be flat @xmath933 .
in addition , @xmath1029 can not be a round spherical space form , as each @xmath595 is noncompact .
also , @xmath1029 can not be a @xmath894-quotient of a shrinking round cylinder , because then @xmath617 would contain a one - sided @xmath1030 ; by the classification of noncompact @xmath10-solutions this would imply that @xmath617 is isometric to the @xmath894-quotient of a round cylinder for all @xmath4 , contradicting the failure of ( 2 )
. therefore @xmath1029 must be a shrinking round cylinder . by the same reasoning
, the asymptotic soliton of @xmath37 can not be a @xmath894-quotient of a shrinking round cylinder , so it must be a shrinking round cylinder .
it follows that the reduced volume @xmath1031 is nearly constant in the interval @xmath1032 $ ] .
but this implies that after passing to a subsequence , @xmath1033 converges to a gradient shrinking soliton @xmath1034 , which is a @xmath10-solution and whose asymptotic reduced volume is that of the shrinking round cylinder . from lemma [ lem_classification_gradient_soliton_kappa_solutions ] , this implies that @xmath1035 is a shrinking round cylinder , contradicting the failure of ( 1 ) .
we now use the stability result of proposition [ prop_cylinder_stability_1_noncompact ] , together with lemma [ lem_cyl_l_function_stability ] , to show that worldlines that start close to necks have a nearly minimal value of @xmath259 , provided one goes at least a certain controlled amount backward in time .
[ lem_basepoint_stability ] for all @xmath774 , there exist @xmath1036 and @xmath1037 such that if @xmath1038 is a pointed @xmath10-solution with @xmath1002 , and @xmath949 is a @xmath6-neck , then @xmath1039 for all @xmath1004 .
suppose the lemma were false .
then for some @xmath774 , there would be a sequence @xmath1040 of pointed @xmath10-solutions so that 1 .
@xmath1013 and 2 .
@xmath915 is a @xmath1041-neck , but 3 .
@xmath1042 for some @xmath1043 .
we can assume that @xmath1044 .
let @xmath1045 be a parameter to be determined later . by proposition [ prop_cylinder_stability_1_noncompact ] ,
there is a @xmath1046 such that for large @xmath48 and every @xmath1047 with @xmath1048 , the tuple @xmath1049 is @xmath1050-close to @xmath1051 , for some @xmath1052 .
since @xmath1053 converges to the pointed round cylinder by assumption , lemma [ lem_cyl_l_function](3 ) implies there is a @xmath1054 $ ] such that for large @xmath48 , we have @xmath1055 . put @xmath1056 \ : : \ : l_j(x_j , t)\geq 1+\hat{\epsilon}\}$ ] , so @xmath1057 .
note that there is an @xmath1058 independent of @xmath1050 such that @xmath1059 since @xmath1060 in view of the time - derivative bound on @xmath259 of ( [ ye2.54 ] ) . for large @xmath48 , in @xmath1061
we have @xmath982 for @xmath973 $ ] , but @xmath1062 . since @xmath1063 , we know from proposition [ prop_cylinder_stability_1_noncompact ] that @xmath1064 is @xmath1050-close to @xmath1065 for some @xmath1066 , again for large @xmath48
. if @xmath968 is the constant from lemma [ lem_cyl_l_function_stability ] , and @xmath1067 , then that lemma gives a contradiction .
this follows by combining proposition [ prop_cylinder_stability_1_noncompact ] with lemma [ lem_basepoint_stability ] .
the main result of this subsection , corollary [ corcloseness ] , will be used in the proof of proposition [ prop_r_in_lp ] .
we recall from appendix [ appkappa ] that if @xmath15 is a @xmath10-solution then @xmath1068 denotes the points in @xmath79 that are not centers of @xmath778-necks .
[ lemcloseness ] there is some @xmath1069 so that for any @xmath1070 , there are @xmath1071 and @xmath1072 with the following property .
let @xmath15 be a compact @xmath10-solution .
suppose that @xmath79 contains an @xmath1073-neck . then there are points @xmath1074 so that @xmath1075 is covered by disjoint balls @xmath1076 and @xmath1077 , whose intersections with @xmath1078 are nonempty .
let @xmath1079 be the parameter of ( * ? ? ?
* ; * ? ? ?
* corollary 48.1 ) .
suppose that there is some point @xmath1080 so that @xmath1081 . by the compactness of the space of pointed @xmath10-solutions
, it follows that there is an upper bound on @xmath1082 , depending only on @xmath88 .
if @xmath1073 is sufficiently small then we obtain a contradiction .
hence we are in case c of ( * ? ? ?
* ; * ? ? ?
* corollary 48.1 ) , so there are points @xmath1074 such that @xmath1083 .
if @xmath1073 is sufficiently small then a cross - section of the @xmath1073-neck separates @xmath79 into two connected components , each of which must have a cap region .
hence if @xmath1073 is sufficiently small then @xmath1076 and @xmath1077 are disjoint .
as @xmath1068 is closed , both @xmath1076 and @xmath1077 intersect @xmath1084 . [ pointedgradient ] given @xmath1085 and a compact family @xmath1086 of pointed @xmath10-solutions , with basepoints at time zero , there is some @xmath1087 such that for each @xmath1088 , there is a point @xmath1089 with @xmath1090 $ ] so that @xmath1091 is @xmath778-close to a pointed gradient shrinking soliton which is a @xmath10-solution .
suppose that the lemma fails .
then for each @xmath396 , there is some @xmath1092 so that for each @xmath1093 with @xmath1094 $ ] , there is no pointed gradient shrinker ( which is a @xmath10-solution ) that is @xmath778-close to @xmath1095 .
after passing to a subsequence , we can assume that @xmath1096 .
from the existence of an asymptotic soliton for @xmath45 , there is some @xmath1097 so that @xmath1098 is @xmath1099-close to a gradient shrinking soliton ( which is a @xmath10-solution ) .
then for large @xmath48 , there is some @xmath1100 so that @xmath1101 is @xmath778-close to the gradient shrinking soliton .
this is a contradiction .
[ corcloseness ] let @xmath1086 be a compact family of compact @xmath10-solutions that does not have any constant curvature elements .
then for each @xmath1085 , there is some @xmath1102 such that for each @xmath1088 , there is a point @xmath1103}$ ] which is the center of an @xmath778-neck . by assumption ,
there is some @xmath1104 so that no time - zero slice of an element of @xmath1086 is @xmath365-close to a constant curvature manifold . by lemma [ pointedgradient ] , for each @xmath1105 , there is some @xmath1106 such that for each @xmath1088 , there is some @xmath1103}$ ] so that @xmath1107 is @xmath1108-close to a gradient shrinking soliton ( which is a @xmath10-solution ) .
if @xmath1108 is sufficiently small , in terms of @xmath365 , then by the local stability of ricci flows of constant positive curvature , this soliton can not have constant curvature .
hence it is either a round shrinking cylinder or a @xmath894-quotient of a round shrinking cylinder .
if it is a round shrinking cylinder then as long as @xmath1109 , we are done .
if it is a @xmath894-quotient of a round shrinking cylinder then if @xmath1108 is sufficiently small , by moving the basepoint we can find a point @xmath1110 that is the center of an @xmath9-neck .
in this section we study bad worldlines ; recall from definition [ def_worldline ] that a worldline @xmath94 in a singular ricci flow @xmath15 is bad if @xmath100 . in theorem [ finitenessthm ] we prove that only finitely many bad worldlines intersect a given connected component in a given time slice .
we then give some applications .
the main result of this section is the following theorem .
[ finitenessthm ] let @xmath15 be a singular ricci flow . for @xmath85 , every connected component @xmath1111 of @xmath1112
intersects at most @xmath480 bad worldlines , where @xmath1113 . in particular , the set of bad worldlines in @xmath15 is at most countable .
we begin with a lemma .
[ lem_controlled_diameter ] for all @xmath1114 there exist @xmath1115 and @xmath1116 such that if @xmath829 and @xmath770 is @xmath778-modelled ( see appendix [ appcloseness ] ) on a @xmath10-solution of diameter @xmath1117 , then : 1 .
the connected component @xmath1118 of @xmath79 containing @xmath11 is compact and has @xmath1119 .
2 . let @xmath1120 be the ricci flow on the ( smooth manifold underlying ) @xmath1118 defined on the time interval @xmath1121 , with initial condition at time @xmath4 given by @xmath1122 , and blow - up time @xmath1123 .
let @xmath1124 be the corresponding ricci flow spacetime .
then @xmath1125 is compact for all @xmath1126 , and if @xmath1127 is the connected component of @xmath1128 containing @xmath1118 , then @xmath1127 is isomorphic to @xmath1124 .
@xmath1129 on @xmath1127 .
let @xmath1130 be the collection of pointed @xmath10-solutions @xmath1131 such that @xmath1002 , and @xmath1132 .
then every @xmath1133 has @xmath1119 , and since @xmath1130 is compact , there is a @xmath1134 such that @xmath1135 in @xmath1136 for all @xmath1133 .
part ( 1 ) now follows .
let @xmath1124 be as above .
then @xmath1137}$ ] is compact for all @xmath1138 , and by hamilton s theorem for manifolds with @xmath1139 , we know that @xmath1140 consider an isometric embedding of ricci flow spacetimes @xmath1141 that extends the isometric embedding @xmath1142 , and which is defined on a maximal time interval @xmath635 starting at time @xmath4 .
then @xmath635 can not be a closed interval @xmath1143 $ ] , since then the embedding could extended to a larger time interval using uniqueness for ricci flows .
if @xmath1144 with @xmath1145 , then since @xmath358 is bounded on @xmath1146}$ ] , by the properness of @xmath358 on @xmath1147 we may extend the embedding to an isometric embedding @xmath1146}{\rightarrow}{{\mathcal m}}_{[t,\hat t]}$ ] , contradicting the maximality of @xmath635 .
therefore there exists an isometric embedding @xmath1148 of ricci flow spacetimes , as asserted .
the image is clearly an open subset of @xmath1128 ; it is closed by ( [ eqn_r_blows_up ] ) .
this proves ( 2 ) .
( 3 ) . from the proof of ( 1 ) above , for every @xmath1133
, we have @xmath1149 on @xmath1136 . taking @xmath1150 , and @xmath778 sufficiently small
, we get that @xmath1129 in @xmath1118 , and therefore in @xmath1151 as well . by the maximum principle applied to the scalar curvature evolution equation
, we have @xmath1129 on @xmath1124 , and hence on @xmath1127 as well . in the proof
below , we take @xmath1152 .
let @xmath1153 be constants , to be determined later .
during the course of the argument below , we will state a number of inequalities involving @xmath769 and @xmath1154 ; these will be treated as a cumulative set of constraints imposed on @xmath769 and @xmath1154 , i.e. we will be assuming that each inequality is satisfied .
we recall that by proposition [ sameconnected ] , @xmath1111 determines a connected component @xmath1155 of @xmath79 for all @xmath777 .
let @xmath1156 be the collection of bad worldlines intersecting @xmath1111 .
choose @xmath1157 such that @xmath1158 , and let @xmath1159 be the set of worldlines @xmath94 which belong to @xmath1156 , where @xmath1160 .
we will show that if @xmath1161 , then @xmath1162 is bounded by a function of @xmath10 , @xmath482 and @xmath1163 ; the theorem then follows immediately .
_ step 1 . if @xmath1164 , then there exists @xmath1165 $ ] such that : _ 1 . for every worldline @xmath94 in @xmath1156
we have @xmath1166 .
2 . for every worldline @xmath94 in @xmath1159 , and every @xmath1167 $ ] with @xmath1168 ,
the pair @xmath1169 is @xmath769-modelled on a noncompact @xmath10-solution . by the compactness of the space of normalized pointed @xmath10-solutions
( see appendix [ appkappa ] ) there exist @xmath1170 and @xmath1171 such that if @xmath1169 is @xmath1172-modelled on a pointed @xmath10-solution with diameter greater than @xmath1173 , then it is @xmath769-modelled on a noncompact @xmath10-solution .
put @xmath1174 , where @xmath1175 is as in lemma [ lem_controlled_diameter ] .
let @xmath1176 be the set of @xmath1177 such that the pair @xmath770 is @xmath1178-modelled on a pointed @xmath10-solution with diameter at most @xmath1173 .
suppose first that @xmath1176 is nonempty .
lem_controlled_diameter ] implies that @xmath1179 , for all @xmath1180 ; therefore by the properness of @xmath66 , the time function @xmath16 attains a minimum value @xmath1181 on @xmath1176 .
pick @xmath1182 .
then by lemma [ lem_controlled_diameter ] , the connected component of @xmath1183}$ ] containing @xmath11 is isomorphic to the @xmath1184$]-time slab of the spacetime @xmath1124 of a ricci flow on a compact manifold with positive ricci curvature . in particular , it also coincides with @xmath1185}c_t$ ] , and therefore the curvature is bounded on the latter .
hence for every @xmath94 in @xmath1156 , we have @xmath1186 .
if @xmath1176 is empty then we put @xmath1187 ; then the conclusion of part ( a ) of step 1 still holds , so we continue . by proposition [ goodeps ]
, there exists @xmath1188 , such that for all @xmath1189 with @xmath1190 , the pair @xmath770 is @xmath1178-modelled on a pointed @xmath10-solution . by lemma [ singforback ] ,
if @xmath1191 , @xmath94 belongs to @xmath1159 and @xmath1192\cap i\cap [ 0,\hat t)$ ] , then we have @xmath1193 . hence either @xmath1194 is ( a ) @xmath769-modelled on a noncompact @xmath10-solution , or ( b ) @xmath1178-modelled on a @xmath10-solution of diameter at most @xmath1173 ; but in case ( b ) we would have @xmath1195 , which is impossible because @xmath1196 .
this proves that part ( b ) of step 1 holds when @xmath1197 .
the borderline case @xmath1198 follows by applying the previous arguments to times @xmath4 slightly less than @xmath1181 and taking the limit as @xmath1199 , using the fact that being @xmath1200-modelled on a noncompact @xmath10-solution is a closed condition .
this completes step 1 .
hereafter we assume that @xmath1164 . by part ( a ) of step 1 ,
the set @xmath1156 is the same as the set of bad worldlines intersecting @xmath1201 .
hence we may replace @xmath482 by @xmath1181 ; then by part ( b ) of step 1 , for every @xmath94 in @xmath1159 , and every @xmath1192 $ ] , the pair @xmath1169 is @xmath769-modelled on a noncompact @xmath10-solution .
provided that @xmath1202 , for all @xmath94 belonging to @xmath1159 and every @xmath1203 , the pair @xmath1169 is not a generalized @xmath1204-neck .
( here @xmath904 is the parameter from corollary [ cor_de_de_over_4 ] , and a generalized neck is in the sense of appendix [ appcloseness ] . ) _
suppose that @xmath94 belongs to @xmath1159 , and @xmath1205 is a generalized @xmath1204-neck for some @xmath1206 . by step 1
we know that @xmath1205 is @xmath769-modelled on a noncompact pointed @xmath10-solution @xmath1207 . if @xmath1208 , then @xmath1207 will be a generalized @xmath904-neck .
let @xmath1209 be as in corollary [ cor_de_de_over_4 ] .
then @xmath1210 is a generalized @xmath906-neck . if @xmath1211 , then we get that : * @xmath1212 is defined at @xmath1213 .
* @xmath1214 is a generalized @xmath1204-neck .
* @xmath1215 .
thus we may iterate this to produce a sequence @xmath1216 such that @xmath1217 for all @xmath456 .
this contradicts the fact that @xmath1160 , and completes step 2 .
hereafter we assume that @xmath1202 .
let @xmath1218 be such that if @xmath401 is a noncompact @xmath10-solution , @xmath1219 , and neither @xmath1220 nor @xmath1221 is a @xmath906-neck , then @xmath1222 let @xmath1223 be a constant , to be determined in step 4 .
provided that @xmath1224 , if @xmath1225 , and @xmath1226 then @xmath1227 .
_ let @xmath1228 be the domain of @xmath1229 , for @xmath1230 .
suppose that @xmath1231 $ ] . by steps 1 and 2
, @xmath1232 is @xmath769-modelled on a noncompact @xmath10-solution and neither @xmath118 nor @xmath1233 is a @xmath1204-neck . if @xmath1234 then using the @xmath769-closeness to a noncompact @xmath10-solution and ( [ eqn_non_necks_close ] ) , we can say that @xmath1235 implies @xmath1236 since we are assuming ( [ eqn_d_1 ] ) , a continuity argument shows that ( [ eqn_3_d0 ] ) holds for all @xmath1237 $ ] .
if @xmath1238 , then @xmath1239 , so @xmath1240 ; similar reasoning holds if @xmath1241 .
. moreover , if @xmath1242 then any geodesic from @xmath118 to @xmath1233 in @xmath1155 will lie in the set with @xmath1243 , so @xmath1244 is a decreasing function of @xmath4 .
since ( [ eqn_3_d0 ] ) implies that @xmath1245 as @xmath1246 , it follows that @xmath1227 .
this completes step 3 .
hereafter we assume that @xmath1224 .
provided that @xmath1247 , the cardinality of @xmath1159 is at most @xmath1248 .
_ take @xmath1249 , and let @xmath1250 , @xmath1251 , and @xmath1252 be as in proposition [ prop_large_r_structure ]
. with reference to step 3 , take @xmath1253 .
let @xmath1254 be such that if @xmath94 belongs to @xmath1159 , and @xmath1192\cap i$ ] , then @xmath1255 ; c.f .
the proof of step 1 . then the set @xmath1256 is contained in @xmath1257 . by step 3 , for any two distinct elements @xmath1225 we have @xmath1258 by proposition [ prop_large_r_structure ] , for all @xmath1259 we have @xmath1260
. therefore @xmath1261 .
this proves that @xmath1159 is finite , and hence a weaker version of the theorem , namely that the set of all bad worldlines is countable .
since the set of bad worldlines is countable , their union has measure zero in spacetime .
therefore we may apply proposition [ vol1 ] , to conclude that @xmath1262 if @xmath1263 , then each @xmath782 has nonempty boundary .
hence part ( 5 ) of proposition [ prop_large_r_structure ] gives a bound @xmath1264 .
this proves theorem [ finitenessthm ] .
[ cor_volume_continuous ] if @xmath15 is a singular ricci flow , then the volume function @xmath87 is absolutely continuous . by theorem [ finitenessthm ]
the set of bad worldlines is countable , and hence has measure zero . combining this with lemma [ quasip ]
, we may apply proposition [ vol1](4 ) , to conclude that @xmath714 is absolutely continuous .
theorem [ finitenessthm ] also has the following topological implications .
let @xmath15 be a singular ricci flow . 1 .
if @xmath62 and @xmath1265 is an open subset that does not contain any compact connected components of @xmath1112 , then there is a smooth time - preserving map @xmath1266{\rightarrow}{{\mathcal m}}$ ] that is a `` weak isotopy '' , in the sense that it maps @xmath1267 diffeomorphically onto an open subset of @xmath668 , for all @xmath1268 $ ] .
2 . for all @xmath62 ,
the pair @xmath1269 is @xmath1270-connected for @xmath1271 .
let @xmath1272 be the collection of connected components of @xmath1112 .
pick @xmath1273 .
let @xmath1274 be the set of bad worldlines intersecting @xmath78 . by theorem [ finitenessthm ]
the set @xmath1274 is finite , so its intersection with @xmath78 is contained in a @xmath0-disk @xmath1275 .
there is a @xmath1276 such that the worldline of every @xmath1277 is defined in the interval @xmath1278 $ ] .
hence we get a time - preserving map @xmath1279{\rightarrow}{{\mathcal m}}$ ] that is a diffeomorphism onto its image . by assumption , either @xmath78 is noncompact , or @xmath78 is compact and @xmath1280 .
therefore there is a smooth homotopy @xmath1281}$ ] ( purely in the time-@xmath482 slice ) such that @xmath1282 is the inclusion map , @xmath1283 is a diffeomorphism onto its image for all @xmath1284 $ ] , and @xmath1285 .
we define @xmath1286 on @xmath1287 $ ] by @xmath1288 , and extend this to @xmath1289 $ ] by following worldlines .
note that if @xmath1290 are distinct components of @xmath1112 then @xmath1291 is disjoint from @xmath1292 , so the resulting map @xmath1286 has the property that @xmath1293 is an injective local diffeomorphism for every @xmath1294 $ ] . ( 2 ) .
suppose that @xmath1295 , and @xmath1296 is a map of pairs , where @xmath1297 if @xmath1298 and @xmath1299 . by theorem [ finitenessthm ] ,
the hausdorff dimension of the bad worldlines is at most one .
then after making a small homotopy we may assume that @xmath1300 is smooth , and that its image is disjoint from the bad worldlines .
we can now find a homotopy through maps of pairs by using the backward flow of the time vector field , which is well - defined on @xmath1301 .
in this section we establish further curvature and volume estimates for singular ricci flows .
there are two main results . in proposition
[ prop_r_in_lp ] , we show that @xmath1302 is integrable on each time slice , for each @xmath1303 . in proposition [ holder ] we give an estimate on how the volume @xmath714 can change as a function of @xmath4 .
when combined with part ( 5 ) of proposition [ vol1 ] , it shows that @xmath714 is @xmath1304-hlder in @xmath4 , where @xmath1305 is the constant in ( [ gradest ] ) .
[ prop_r_in_lp ] let @xmath15 be a singular ricci flow . then for all @xmath93 and @xmath415 , there is a bound @xmath1306 for all @xmath494 $ ] . before entering into the details , we first give a sketch of the proof . due to the bounds on @xmath714 from proposition [ vol1 ] , it suffices to control the contribution to the left - hand side of ( [ eqn_r_in_lp ] ) from the points with large scalar curvature .
such points fall into three types , according to the geometry of the canonical neighborhoods : ( a ) neck points , ( b ) cap points , and ( c ) points @xmath1307 whose connected component in @xmath79 is compact and has diameter comparable to @xmath1308 .
if @xmath1309 is a neck point with worldline @xmath1310{\rightarrow}{{\mathcal m}}$ ] then thanks to the stability of necks going backward in time , the scale - invariant time derivative @xmath1311 will remain very close to the cylindrical value along @xmath1212 , until @xmath358 falls down to a value comparable to @xmath1312 . combining this with previous estimates on the jacobian as in section [ sec_singular_ricci_flows_basic_properties ]
, we can bound the contribution from the neck points to the left - hand side of ( [ eqn_r_in_lp ] ) in terms of the volume of the corresponding set of points in the time zero slice . to control the contribution from points of type ( b ) , we show that it is dominated by that of the neck points . to control the contribution from the points of type ( c ) we use a similar approach .
we again analyze the geometry going backward in time along worldlines , except that in this case there are three stages : one where the components are nearly round , one when they are no longer nearly round but still have diameter comparable to @xmath1313 , and one when they have a large necklike region .
we now start on the proof . with the notation of the proof of proposition [ vol1 ] ,
let @xmath1314 be the complement of the set of points in @xmath79 with a bad worldline . from theorem [ finitenessthm
] , it has full measure in @xmath79 . given @xmath1315 , let @xmath499 \rightarrow { { \mathcal m}}_{[0 , t]}$ ] be the restriction of its worldline to the interval @xmath1316 $ ] .
define @xmath1317 as in ( [ jacobian ] ) , with @xmath736 . from ( [ jacrf ] )
, we have @xmath1318 given @xmath1319 , we consider times @xmath4 in the range @xmath1320 $ ] .
let @xmath778 , @xmath1321 , @xmath616 and @xmath1322 be as in proposition [ prop_large_r_structure ] .
we take @xmath491 . from proposition [ goodeps ]
we can assume that the @xmath778-canonical neighborhood assumption holds on the superlevel set @xmath1323}^ { > \overline{r}}$ ] of the scalar curvature function .
we will further adjust the parameters @xmath778 and @xmath1322 later . for any @xmath1324 , write @xmath1325 where * @xmath1326 consists of the points in @xmath1327 that are centers of @xmath1328-necks , *
@xmath1329 consists of the points @xmath1330 so that after rescaling by @xmath1331 , the pair @xmath1332 is @xmath778-close to a pointed noncompact @xmath10-solution , and * @xmath1333 .
taking @xmath1334 , there is a compact set @xmath1335 of @xmath10-solutions so that for @xmath1336 , after rescaling by @xmath1331 the connected component of @xmath79 containing @xmath500 is @xmath778-close to an element of @xmath1335 ( c.f .
step 1 of the proof of theorem [ finitenessthm ] ) .
in particular , before rescaling , the diameter of the component is bounded above by @xmath1337 and the scalar curvature on the component satisfies @xmath1338 for an appropriate constant @xmath1339 , where @xmath1340 denote the average scalar curvature on the component . by the pointed compactness of the space of normalized @xmath10-solutions , and the diameter bound on the caplike regions in normalized pointed noncompact @xmath10-solutions
, there is a @xmath1341 so that @xmath1342 hence we can restrict our attention to @xmath1343 and @xmath1344 .
consider @xmath1345 . with @xmath1346 , @xmath1347 and @xmath899 being parameters of theorem [ thm_stability_of_necks ]
, we assume that @xmath1348 and @xmath1349 . using theorem [ thm_stability_of_necks ] and the @xmath778-canonical neighborhood assumption , there are @xmath1350 so that for @xmath1351 $ ] , the rescaled solution @xmath1352 is @xmath1353-close to @xmath900 . by reducing @xmath778
, we can make @xmath1354 arbitrarily negative .
the gradient bound ( [ gradest ] ) gives @xmath1355 as long as @xmath1356 stays in a canonical neighborhood .
if @xmath1357 then for all @xmath1358 $ ] , we have @xmath1359 and ( [ ueqn ] ) holds , so @xmath1360 for a round shrinking cylinder , the sharp value for @xmath1361 in ( [ ueqn ] ) is @xmath799 . for any @xmath1362 , if @xmath1347 is sufficiently small then we are ensured that @xmath1363 in all , @xmath1364 because of the cylindrical approximation , @xmath1365 and so there is a constant @xmath1366 such that for very negative @xmath1354 , we have @xmath1367 we now replace @xmath4 by @xmath1368 and iterate the argument .
eventually , there will be a first time @xmath1369 when we can no longer continue the iteration because the curvature has gone below @xmath616 .
suppose that there are @xmath480 such iterations .
then @xmath1370 from ( [ 2ineq ] ) , @xmath1371 so @xmath1372 put @xmath1373 . by choosing @xmath1374 sufficiently close to @xmath799 ( from above ) and @xmath1354
sufficiently negative , we can make @xmath1307 arbitrarily close to @xmath799 ( from below ) . since @xmath1375
we obtain @xmath1376 then @xmath1377 this finishes the discussion of the neck points .
let @xmath1378 be a new parameter .
given @xmath632 small , let @xmath1379 be the connected components of @xmath1380 that intersect @xmath1381 and are @xmath365-close to a constant curvature metric , and let @xmath1382 be the other connected components of @xmath79 that intersect @xmath1381 . using proposition [ sameconnected ] , a connected component @xmath1383 in @xmath1384 determines a connected component @xmath1385 in @xmath1386 for all @xmath1387 .
let @xmath1383 be a component in @xmath1382 . from ( [ oncomp ] )
, we have @xmath1388 . using the compactness of the space of approximating @xmath10-solutions
, we can apply lemma [ lemcloseness ] and corollary [ corcloseness ] . then for @xmath1108 small and @xmath1389
, there is some @xmath1390 $ ] so that @xmath1385 consists of centers of @xmath1108-necks and two caps . from ( [ ueqn ] ) , if @xmath1391 and @xmath1392 ) } \ge \overline{r}\ ] ] then @xmath1393 so @xmath1394 if @xmath1395 then ( [ needed ] ) holds and from ( [ easy ] ) , @xmath1396 . from ( [ oncomp ] ) and ( [ easy ] ) , we also have @xmath1397 since the volume element at @xmath1398 is nonincreasing as a function of @xmath1399 $ ] , we obtain @xmath1400 we now apply the argument starting with ( [ capineq ] ) to @xmath1385 . taking @xmath1401 large compared to @xmath1402 , in order to ensure many iterations in the earlier - neck argument
, we get a bound @xmath1403 this takes care of the components in @xmath1382 .
let @xmath1383 be a component of @xmath1380 in @xmath1379 .
let @xmath831 be the infimum of the @xmath687 s so that for all @xmath1404 $ ] , the metric on @xmath1385 is @xmath365-close to a constant curvature metric . for a ricci flow solution with time slices of constant positive curvature , @xmath358 is strictly increasing along forward worldlines but @xmath1405 is strictly decreasing in @xmath4 . hence if @xmath365 is sufficiently small then we are ensured that @xmath1406 if @xmath1407 has a point with scalar curvature at most @xmath1408 and @xmath365 is small then @xmath1409 if , on the other hand , @xmath1410 then we can apply the preceding argument for @xmath1411 , replacing @xmath4 by @xmath831 .
the conclusion is that @xmath1412 since @xmath1413 @xmath1414 and @xmath1415 the proposition follows from ( [ capineq ] ) , ( [ newcomb1 ] ) , ( [ newcomb2 ] ) , ( [ newcomb3 ] ) , ( [ newcomb3.5 ] ) and ( [ newcomb4 ] ) . [ holder ]
let @xmath15 be a singular ricci flow .
let @xmath1361 be the constant from ( [ gradest ] ) .
we can assume that @xmath1305 .
then whenever @xmath718 satisfies @xmath1416 and @xmath1417 , we have @xmath1418 let @xmath1419 be the set of points @xmath1420 whose worldline @xmath1421 extends forward to time @xmath1422 and let @xmath1423 be the points @xmath500 whose worldline @xmath1421 does not extend forward to time @xmath1422 .
put @xmath1424 @xmath1425 and @xmath1426 then @xmath1427 suppose that @xmath739 . let @xmath1428 be the domain of the forward extension of @xmath1421 , with @xmath1429 . for all @xmath1430 ,
we have @xmath1431 if the lemma is not true , put @xmath1432 then @xmath1433 . from the gradient estimate ( [ gradest ] ) and the fact that @xmath1434 , we know that @xmath1435 .
whenever @xmath1436 , we have @xmath1437 so there is some @xmath989 so that the gradient estimate ( [ gradest ] ) holds on the interval @xmath1438 .
this implies that ( [ crawl ] ) holds for all @xmath1439 , which contradicts the definition of @xmath1440 .
this proves the lemma .
hence @xmath1441 and @xmath1442 since @xmath1443 on the set @xmath1444 .
suppose now that @xmath1445 . for all @xmath1446 $ ]
, we have @xmath1447 if the lemma is not true , put @xmath1448 \ : : \ : r(\gamma_x(u ) ) > \frac{1}{\frac{1}{r(x ) } - \eta(u - t_1 ) } \right\}.\ ] ] then @xmath1449 and the gradient estimate ( [ gradest ] ) implies that @xmath1450 .
now @xmath1451 hence there is some @xmath989 so that @xmath1452 for @xmath1453 $ ] .
if @xmath1452 for all @xmath1454 $ ] then ( [ gradest ] ) implies that ( [ estpp ] ) holds for @xmath1455 $ ] , which contradicts the definition of @xmath1456 . on the other hand , if it is not true that @xmath1452 for all @xmath1454 $ ] , put @xmath1457 \ : : \ :
r(\gamma_x(u ) ) < r(t_2)^{-2 } \right\}.\ ] ] then @xmath1458 and @xmath1459 . equation ( [ gradest ] ) implies that @xmath1460 which contradicts ( [ reqal ] ) .
this proves the lemma .
hence if @xmath1445 then @xmath1461 so @xmath1462 thus @xmath1463 since @xmath1305 , if @xmath1464 $ ] then @xmath1465 , so @xmath1466 then @xmath1467 now suppose that @xmath1468 . for all @xmath1446 $ ]
, we have @xmath1469 if the lemma is not true , put @xmath1470 \ : : \ : r(\gamma_x(u ) ) > \frac{1}{r(t_2)^2 - \eta(u - t_1 ) } \right\}.\ ] ] then @xmath1471 . if @xmath1472 then clearly @xmath1473 .
if @xmath1474 then since @xmath1475 , there is some @xmath1476 so that @xmath1477 for @xmath1478 $ ] ; then ( [ gradest ] ) gives the validity of ( [ estppp ] ) for @xmath1478 $ ] , which implies that @xmath1473 . in either case , @xmath1479 .
now @xmath1480 hence there is some @xmath989 so that @xmath1452 for @xmath1481 $ ] .
if @xmath1452 for all @xmath1482 $ ] then ( [ gradest ] ) implies that ( [ estppp ] ) holds for @xmath1483 $ ] , which contradicts the definition of @xmath1484 . on the other hand , if it is not true that @xmath1452 for all @xmath1482 $ ] , put @xmath1485 \ : : \ : r(\gamma_x(u ) ) < r(t_2)^{-2 } \right\}.\ ] ] then @xmath1486 and @xmath1487 .
the gradient estimate ( [ gradest ] ) implies that @xmath1488 which contradicts ( [ reqal2 ] ) .
this proves the lemma .
hence if @xmath1468 then @xmath1489 so @xmath1490 thus @xmath1491 since @xmath1492 $ ] , we can apply ( [ trivial ] ) to conclude that @xmath1493 combining ( [ addup1 ] ) , ( [ addup2 ] ) , ( [ addup3 ] ) and ( [ addup4 ] ) gives ( [ volest ] ) .
in this section we show that the _ a priori _ assumptions in definition [ def_singular_ricci_flow ] are really conditions on the spacetime near infinity .
that is , given @xmath120 and a decreasing function @xmath121 , there are decreasing functions @xmath122 and @xmath123 with the following property .
let @xmath15 be a ricci flow spacetime with normalized initial condition , on which condition ( a ) of definition [ def_singular_ricci_flow ] holds .
suppose that for each @xmath69 there is a compact subset of @xmath68 so that condition ( b ) , and the @xmath7-canonical neighborhood assumption of condition ( c ) , hold on the part of @xmath68 outside of the compact subset .
then @xmath15 satisfies definition [ def_singular_ricci_flow ] globally with parameters @xmath63 , @xmath124 and @xmath125 .
if @xmath15 is a ricci flow spacetime and @xmath1494 , put @xmath1495 .
we define perelman s @xmath259-function using curves emanating backward from @xmath1496 , as in ( * ? ? ?
* ; * ? ? ?
* section 15 ) .
that is , given @xmath109 with @xmath1497 , consider a time - preserving map @xmath1498 \rightarrow { { \mathcal m}}$ ] from @xmath11 to @xmath1496 .
we reparametrize @xmath1499 $ ] by @xmath1500 .
then @xmath1501 is the spatial projection of the velocity vector of @xmath410 and @xmath1502 is computed using the metric on @xmath1503 at @xmath1504 .
let @xmath1505 be the infimal @xmath1506-length of such curves @xmath410 .
the reduced length is @xmath1507 [ box ] given @xmath1508 there are @xmath1509 and @xmath1510 with the following property . for any @xmath109 with @xmath1511 , let @xmath1512 denote the @xmath82-distance from @xmath11 to the set @xmath1513 .
then @xmath1514 with a slight variation on perelman s definition ( * ? ? ?
* ; * ? ? ?
* definition 79.1 ) , we put @xmath1515 .
we define @xmath1516 using @xmath1517 instead of @xmath1518 . applying the lower curvature bound ( [ scalarcurv ] ) ( with @xmath1519 ) , we know that @xmath1520 and so @xmath1521
hence it suffices to estimate @xmath1517 from below
. given numbers @xmath1522 , if @xmath1511 then @xmath1523 suppose first that @xmath410 leaves @xmath1496 and exits @xmath1513 at some time @xmath1524 .
if the parabolic ball were euclidean then we could say from ( [ pareu ] ) that @xmath1525 .
if @xmath1526 and @xmath1527 are small enough , depending on @xmath1496 , then we can still say that @xmath1513 is unscathed and @xmath1528 .
given such values of @xmath1526 and @xmath1527 , suppose now that @xmath410 does not exit @xmath1513 in the time interval @xmath1529 . then @xmath1530 .
now latexmath:[\[{\mathcal l}_+(\gamma ) \ge \sqrt{\delta t } \int_{\delta t}^{t_0 - { \mathfrak{t}}(m ) } \left ( r_+(\gamma(\tau ) ) + since @xmath1520 , it follows that along @xmath410 , we have @xmath1532 then @xmath1533 so @xmath1534 thus @xmath1535 the proposition follows .
[ boxx ] suppose that @xmath1536 is complete away from the time - zero slice and the time-@xmath103 slice .
given @xmath1537 , the restriction of @xmath1538 to @xmath1539 is proper and bounded below . from ( [ leqn ] ) ,
the function @xmath1538 is bounded below on @xmath1539 .
suppose that it is not proper .
then for some @xmath757 , there is a sequence @xmath1540 in @xmath1539 going to infinity with @xmath1541 for all @xmath456 .
we can choose @xmath1542 with @xmath1543 and @xmath1544 by the completeness of @xmath82 , we have @xmath1545
. then ( [ leqn ] ) gives a contradiction .. it is not hard to see that @xmath1538 is continuous on @xmath1546 . from the proof of proposition [ box ] , given @xmath1547 and @xmath1548 , the time - preserving curves @xmath1549 \rightarrow { { \mathcal m}}$ ] from @xmath11 to @xmath1496 with @xmath1550 lie in a compact subset of @xmath15 . from standard arguments (
* ; * ? ? ?
* section 17 ) , it follows that there is an @xmath1506-minimizer from @xmath1496 to @xmath11 .
since @xmath1538 is bounded below and time - slices have finite volume from corollary [ cor_volume_continuous ] , the reduced volume @xmath1551 exists .
the results of ( * ? ? ?
* ; * ? ? ?
* sections 17 - 29 ) go through in our setting .
in particular , @xmath1552 is nonincreasing in @xmath831 .
[ smalll ] suppose that @xmath1536 is complete away from the time - zero slice and the time-@xmath103 slice .
for every @xmath1553 , there is some @xmath502 with @xmath1554 .
putting @xmath1555 , we have @xmath1556 in the barrier sense ( * ? ? ?
* ; * ? ? ?
* section 24 ) . from corollary
[ boxx ] , for each @xmath1557 , the function @xmath1558 is proper and bounded above on @xmath1559 . in particular , for each @xmath1553 , the maximum of @xmath1560 exists on @xmath79 .
we want to show that the maximum is nonnegative . by way of contradiction , suppose that for some @xmath1561 and some @xmath1562 , we have @xmath1563 for all @xmath1564
. given @xmath1565 , we can apply lemma [ maximumprin ] on the interval @xmath1566 $ ] to conclude that @xmath1567 for all @xmath1568 .
however , along the worldline @xmath410 going through @xmath1496 , for small @xmath1569 we have @xmath1570 then for small @xmath831 , we have @xmath1571 .
taking @xmath1572 gives a contradiction and proves the proposition . in his first ricci flow paper
, perelman showed that there is a decreasing function @xmath1573 with the property that if @xmath15 is a smooth ricci flow solution , with normalized initial conditions , then @xmath15 is @xmath124-noncollapsed at scales less than @xmath63 ( * ? ? ?
* ; * ? ? ?
* theorem 26.2 ) .
[ newkappa ] let @xmath15 be a ricci flow spacetime with normalized initial condition .
given @xmath1574 , suppose that @xmath1575 is complete away from the time - zero slice and the time-@xmath836 slices
. then @xmath1539 is @xmath124-noncollapsed at scales less than @xmath63 .
the proof is along the lines of that of ( * ? ? ?
* ; * ? ? ?
* theorem 26.2 ) .
we can assume that @xmath1576 .
to prove @xmath124-noncollapsing near @xmath1577 , we consider @xmath1506-curves emanating backward in time from @xmath1496 to a fixed time slice @xmath1578 , say with @xmath1579 . by proposition [ smalll ] ,
there is some @xmath1580 with @xmath1554 . using the bounded geometry near @xmath11 and the monotonicity of @xmath1581 , the @xmath124-noncollapsing follows as in ( * ?
* ; * ? ? ?
* pf . of theorem 26.2 ) .
we now show that the conditions in definition [ def_singular_ricci_flow ] , to define a singular ricci flow , are actually asymptotic in nature .
[ asympprop ] given @xmath1 , @xmath1582 and a decreasing function @xmath1583 \rightarrow ( 0 , \infty)$ ] , there is some @xmath1584 with the following property .
let @xmath15 be a ricci flow spacetime such that @xmath1585 is bounded below and proper , and there is a compact set @xmath1586 so that for each @xmath1587 , 1 .
the hamilton - ivey pinching condition of ( [ hi ] ) is satisfied at @xmath11 , with time parameter @xmath113 , and 2 .
the @xmath7-canonical neighborhood assumption of appendix [ canon ] is satisfied at @xmath11 .
then the conditions of definition [ def_singular_ricci_flow ] hold on @xmath1539 , with parameters @xmath63 , @xmath124 and @xmath125 .
condition ( a ) of definition [ def_singular_ricci_flow ] holds on @xmath1539 by assumption . also by assumption , for @xmath1587 , the curvature operator at @xmath11 lies in the convex cone of ( [ hicone ] ) .
the proof of hamilton - ivey pinching in ( * ? ? ?
* ; * ? ? ?
* pf . of theorem 6.44 ) , using the vector - valued maximum principle , now goes through since any violations in @xmath1539 of ( [ hi ] ) would have to occur in @xmath225 .
this shows that condition ( b ) of definition [ def_singular_ricci_flow ] holds on @xmath1539 .
since the @xmath7-canonical neighborhood assumption holds on @xmath1588 the proof of lemma [ quasip ] shows that @xmath82 is complete on @xmath1539 away from the time - zero slice and the time-@xmath836 slice .
proposition [ newkappa ] now implies that @xmath1539 is @xmath124-noncollapsed at scales less than @xmath63 . to show that condition ( c ) of definition [ def_singular_ricci_flow ] holds on @xmath1539 , with parameters @xmath63 and @xmath124 , and some parameter @xmath1589 , we apply the method of proof of ( * ? ? ?
* ; * ? ? ?
* theorem 52.7 ) for smooth ricci flow solutions .
suppose that there is no such @xmath125 .
then there is a sequence @xmath1590 of ricci flow spacetimes satisfying the assumptions of the proposition , and a sequence @xmath1591 , so that for each @xmath1270 there is some @xmath1592 where the @xmath1593-canonical neighborhood assumption does not hold .
the first step in ( * ? ? ?
* ; * ? ? ?
* pf . of theorem 52.7 ) is to find a point of violation so that there are no nearby points of violation with much larger scalar curvature , in an earlier time interval which is long in a scale - invariant sense .
the proof of this first step uses point selection . because of our assumption that the @xmath7-canonical neighborhood assumption holds in @xmath1594 , as soon as @xmath1595 we know that any point of violation lies in @xmath1596 .
thus this point selection argument goes through . the second step in ( * ? ? ?
* ; * ? ? ?
* pf . of theorem 52.7 ) is a bounded - curvature - at - bounded - distance statement that uses hamilton - ivey pinching and @xmath124-noncollapsing . since we have already proven that the latter two properties hold , the proof of the second step goes through . the third and fourth steps in ( *
* ; * ? ? ?
* pf . of theorem 52.7 ) involve constructing an approximating @xmath124-solution .
these last two steps go through without change .
proposition [ asympprop ] shows that the @xmath125-canonical neighborhood assumption holds with parameter @xmath1597 .
we can assume that @xmath125 is a decreasing function of @xmath836 .
hence @xmath15 is a singular ricci flow with parameters @xmath63 , @xmath1598 and @xmath1599 .
in this section we extend the results of the paper to four - dimensional ricci flow with nonnegative isotropic curvature , under an assumption of no incompressible embedded spherical space forms .
the basic results about such flows are due to hamilton @xcite .
let @xmath1600 be a closed oriented @xmath64-manifold with nonnegative isotropic curvature .
suppose that under ricci flow , the isotropic curvature does not immediately become positive . from (
* ; * ? ? ?
* theorem 4.10 ) , one of the following happens : 1 .
@xmath1600 is flat .
the universal cover @xmath1601 is an isometric product of metrics on @xmath1602 and a surface @xmath1603 .
@xmath144 is biholomorphic to @xmath1604 and @xmath23 is a khler metric with positive first chern class . in these three cases ,
the ricci flow is well understood . in case
( 1 ) , it is just a static flow . in cases ( 2 ) and ( 3 )
, the ensuing ricci flow will be smooth on a time interval @xmath1605 and nonexistent thereafter .
hence we assume that @xmath1600 has positive isotropic curvature .
this condition is preserved under the ricci flow .
there is a version of the hamilton - ivey inequality for ricci flow on a @xmath64-manifold with positive isotropic curvature ( * ? ? ?
* ; * ? ? ?
* section b ) .
it has the implication that any blowup limit is a @xmath10-solution with positive isotropic curvature , and restricted isotropic curvature in the sense of ( * ? ? ?
* ; * ? ? ?
* ( 2.4 ) ) .
any such compact solution is diffeomorphic to @xmath1606 .
any such noncompact solution is diffeomorphic to @xmath1607 , or is an isometric product @xmath1608 where @xmath1609 is diffeomorphic to a spherical space form .
using @xcite , we have the following extension of lemma [ lem_classification_gradient_soliton_kappa_solutions ] .
[ 4dlemma ] let @xmath15 be a four - dimensional gradient shrinking soliton that is a @xmath10-solution , has positive isotropic curvature and restricted isotropic curvature in the terminology of ( * ? ? ?
* ; * ? ? ?
* ( 2.4 ) ) , and blows up as @xmath1610 . for @xmath568 and a point @xmath1611 , let @xmath1612 be the @xmath259-function on @xmath15 constructed using @xmath1506-geodesics going backward in time from @xmath1613 .
then there is a function @xmath1614 so that the limit @xmath1615 exists , independent of @xmath1616 , with continuous convergence on compact subsets on @xmath15 . define @xmath935 as in ( [ redvol ] ) .
then one of the following holds : 1 .
@xmath15 is isometric to the shrinking round cylinder solution on @xmath1617 , where @xmath1618 , with @xmath1619 , @xmath1620 and @xmath1621 .
2 . @xmath15 is a shrinking round @xmath1606 , with @xmath1622 , @xmath1623 and @xmath1624 .
let @xmath135 be a compact @xmath64-manifold with positive isotropic curvature and no embedded @xmath1625-injective spherical space forms .
the construction of ricci flow with surgery , starting from @xmath135 was initiated by hamilton @xcite .
there was a problem with @xcite because no canonical neighborhood result was available at the time .
the construction was revisited by chen - zhu @xcite , based on perelman s work in the three - dimensional case @xcite and a preprint version of @xcite .
the result of lemma [ 4dlemma ] , which follows from @xcite , was not available when @xcite was written . with the incorporation of lemma [ 4dlemma ] , the construction of the ricci flow with surgery is strictly analogous to perelman s work .
the results of the present paper now extend to the setting of four - dimensional singular ricci flows with positive isotropic curvature , again under the assumption that the initial time slice does not have any embedded @xmath1625-injective spherical space forms . returning to ricci flow with surgery ,
suppose that there are such embedded spherical space forms in the initial time slice .
there is still a well - defined ricci flow with surgery , but the time slices may be orbifolds with isolated singularities @xcite .
it should be possible to extend the results of the present paper to the orbifold setting , using @xcite , but we do not address the subject here .
in this section we collect some needed facts about ricci flows and ricci flows with surgery .
more information can be found in @xcite .
let @xmath1626 be a ricci flow spacetime ( definition [ def_ricci_flow_spacetime ] ) . for brevity , we will often write @xmath15 for the quadruple . in a ricci flow with surgery
, we will sometimes loosely write a point @xmath502 as a pair @xmath581 . given @xmath1627 , the rescaled ricci flow spacetime is @xmath1628 .
given @xmath1629 , we write @xmath1630 for the open metric ball of radius @xmath7 in @xmath79 .
we write @xmath1631 for the parabolic neighborhood , i.e. the set of points @xmath1632 in @xmath1633}$ ] if @xmath1634 ( or @xmath1635}$ ] if @xmath1636 ) that lie on the worldline of some point in @xmath1637 .
we say that @xmath1638 is _ unscathed _
if @xmath1637 has compact closure in @xmath79 and for every @xmath1639 , the maximal worldline @xmath410 through @xmath1632 is defined on a time interval containing @xmath1640 $ ] ( or @xmath1641 $ ] ) .
we write @xmath1642 for the forward parabolic ball @xmath1643 and @xmath1644 for the backward parabolic ball @xmath1645 .
we write @xmath885 for the standard ricci flow on @xmath1646 that terminates at time zero , with @xmath1647 .
we write @xmath886 for the standard round shrinking @xmath0-sphere that terminates at time zero .
let @xmath1648 and @xmath1649 be two ricci flow spacetimes in the sense of definition [ def_ricci_flow_spacetime ] . consider a time interval @xmath1650 $ ] .
suppose that @xmath1651 and @xmath1652 have @xmath1653 .
we say that @xmath1654 and @xmath1655 are _
@xmath63-close _ on the time interval
@xmath1650 $ ] if there are open subsets @xmath1656 with @xmath1657 , @xmath1230 , and there is a pointed diffeomorphism @xmath1658 so that * @xmath1659 has compact closure in @xmath1660 , * @xmath1661 is unscathed , * @xmath167 is time - preserving , i.e. @xmath1662 , * @xmath1663 and * @xmath1664 has norm less than @xmath63 in the @xmath1665 + 1}$]-topology on @xmath1666 .
now consider an open time interval @xmath1667 .
suppose that @xmath1668 .
after time shift and parabolic rescaling , we can assume that @xmath1669 and @xmath1670 .
in this case , we say that @xmath1654 and @xmath1655 are @xmath63-close on the time interval @xmath1671 if there are open sets @xmath1656 with @xmath1672 , @xmath1230 , and there is a pointed diffeomorphism @xmath1673 so that 1 .
@xmath1659 has compact closure in @xmath1674 , 2 .
@xmath1675 and @xmath1676 are unscathed , 3 .
@xmath167 is time - preserving , i.e. @xmath1662 , 4 .
@xmath1663 and 5 .
@xmath1664 has norm less than @xmath63 in the @xmath1665 + 1}$]-topology on @xmath1666 .
suppose @xmath1677 , @xmath1678 are as above , and @xmath1679 is a continuous function for @xmath1680 .
then we say that the triples @xmath1681 , @xmath1682 are @xmath9-close if ( 1)-(5 ) above hold , and in addition @xmath1683 we will apply this notion when the @xmath1684 s are reduced distance functions .
we define @xmath63-closeness similarly on other time intervals , whether open or half - open .
if @xmath1685 and @xmath1686 are ricci flow spacetimes then we say that @xmath1677 is _ @xmath9-modelled _ on @xmath1678 if after shifts in the time parameters so that @xmath1687 , and parabolic rescaling by @xmath1688 and @xmath1689 respectively , the resulting ricci flow spacetimes are @xmath9-close to each other on the time interval @xmath1690 $ ] .
( it is implicit in the definition that @xmath1691 and @xmath1692 ; this will be the case for us since we are interested in modelling regions of high scalar curvature . )
a point @xmath11 in a ricci flow spacetime @xmath15 is a _ generalized @xmath9-neck _ if @xmath770 is @xmath9-modelled on @xmath1693 , where @xmath401 is either a shrinking round cylinder or the @xmath894-quotient of a shrinking round cylinder .
we say that @xmath1694 is a ( strong ) _ @xmath3-neck _ if after time shifting and parabolic rescaling , it is @xmath3-close on the time interval @xmath1695 $ ] to the product ricci flow which , at its final time , is isometric to the product of @xmath954 with a round @xmath71-sphere of scalar curvature one .
the basepoint is taken at time @xmath1696 . in this case
, we also say that @xmath11 is the _ center of a @xmath3-neck_. if @xmath788 is an open interval then a metric on an embedded copy of @xmath1697 in @xmath79 , such that each point is contained in an @xmath3-neck , is called a _
@xmath3-tube _ , or a _
@xmath3-horn _ , or a _
double @xmath3-horn _ , if the scalar curvature stays bounded on both ends , stays bounded on one end and tends to infinity on the other , or tends to infinity on both ends , respectively .
( our definition differs slightly from that in ( * ? ? ?
* ; * ? ? ?
* definition 58.2 ) , where the definition is in terms of the `` @xmath3-necks '' of that paper , as opposed to the `` strong @xmath3-necks '' that we are using now . ) a metric on @xmath1698 or @xmath1699 , such that each point outside some compact set is contained in a @xmath3-neck , is called a _ @xmath3-cap _ or a _ capped @xmath3-horn _ , if the scalar curvature stays bounded or tends to infinity on the end , respectively .
let @xmath15 be an @xmath1700-dimensional ricci flow spacetime .
let @xmath1701 be a decreasing function .
we say that @xmath15 is _ @xmath10-noncollapsed at scales below @xmath63 _ if for each @xmath1702 and all @xmath109 with @xmath1703 , whenever @xmath1704 is unscathed and @xmath1705 on @xmath1704 , then we also have @xmath1706 . in the application to ricci flow with surgery , @xmath63 will be taken to be the global parameter .
we refer to ( * ? ? ?
* ; * ? ? ?
* section 15 ) for the definitions of the @xmath259-function @xmath1707 and the reduced volume @xmath1708 . for notation
, we recall that the @xmath259-function is defined in terms of @xmath1506-geodesics going backward in time from a basepoint @xmath1709 .
the parameter @xmath831 is backward time from @xmath1632 ; e.g. @xmath1710 . given @xmath1711 , a _
@xmath15 is a smooth ricci flow solution defined on a time interval of the form @xmath1712 ( or @xmath1713 $ ] ) such that * the curvature is uniformly bounded on each compact time interval , and each time slice is complete . * the curvature operator is nonnegative and the scalar curvature is everywhere positive .
* the ricci flow is @xmath10-noncollapsed at all scales .
we will sometimes talk about @xmath10-solutions without specifying @xmath10 . unless other specified , it is understood that @xmath1714
. if @xmath1694 is a pointed @xmath10-solution then we will sometimes understand it to be defined on the interval @xmath1715 $ ] .
examples of @xmath10-solutions are @xmath885 and @xmath886 .
any pointed @xmath10-solution @xmath1694 has an _
asymptotic soliton_. it is obtained by constructing the @xmath259-function using @xmath1506-geodesics emanating backward from @xmath11 . for any @xmath1716 , there is some point @xmath1717 where @xmath1718 .
put @xmath1719 .
then the parabolic rescaling @xmath1720 subconverges as @xmath1721 to a nonflat gradient shrinking soliton called the asymptotic soliton ( * ? ? ?
* ; * ? ? ?
* proposition 39.1 ) .
( in the cited reference , the convergence is shown on the ( rescaled ) time interval @xmath1722 $ ] , but using the estimates of subsection [ lestimates ] one easily gets pointed convergence on the time interval @xmath1671 . ) hereafter , we suppose that the spacetime @xmath15 of the @xmath10-solution is four - dimensional .
a basic fact is that the space of pointed @xmath10-solutions @xmath1694 , with @xmath1723 , is compact ( * ? ? ?
* ; * ? ? ?
* theorem 46.1 ) . given @xmath1724 ,
let @xmath1725 denote the points in @xmath15 that are not centers of @xmath3-necks .
we call these _ cap points_. put @xmath1726 . from (
* ; * ? ? ?
* corollary 47.2 ) , if @xmath3 is small enough then there is a @xmath1727 such that if @xmath79 is noncompact then * @xmath1728 is compact with @xmath1729 and * @xmath1730 whenever @xmath1731 , where @xmath1732 for some @xmath1733 . if @xmath15 is noncompact , and not a round shrinking cylinder , then @xmath1734
a version of the preceding paragraph that also holds for compact @xmath10-solutions can be found in ( * ? ? ?
* ; * ? ? ?
* corollary 48.1 ) .
a compact @xmath10-solution is either a quotient of the round shrinking sphere , or is diffeomorphic to @xmath792 or @xmath793 ( * ? ? ?
* ; * ? ? ?
* lemma 59.3 ) .
there is some @xmath1735 so that any @xmath10-solution is a @xmath1736-solution or a quotient of the round shrinking @xmath792 ( * ? ? ?
* ; * ? ? ?
* proposition 50.1 ) [ lem_classification_gradient_soliton_kappa_solutions ] let @xmath15 be a three - dimensional gradient shrinking soliton that is a @xmath10-solution and blows up as @xmath1610 .
for @xmath568 and a point @xmath1611 , let @xmath1612 be the @xmath259-function on @xmath15 constructed using @xmath1506-geodesics going backward in time from @xmath1613 . then there is a function @xmath1614 so that the limit @xmath1615 exists , independent of @xmath1616 , with continuous convergence on compact subsets on @xmath15 .
define @xmath935 as in ( [ redvol ] ) .
then one of the following holds : 1 .
@xmath15 is the shrinking round cylinder solution @xmath885 on @xmath1646 , with @xmath1737 , @xmath1738 and @xmath1739 .
@xmath15 is the @xmath894-quotient of the cylinder in ( 2 ) , with @xmath1740 .
the pullback of @xmath917 to the cylinder is @xmath1741 .
3 . @xmath15 is a shrinking round spherical space form @xmath1742 , where @xmath1743 , @xmath1744 , @xmath1745 and @xmath1746 .
the classification of the solitons follows from ( * ? ? ?
* ; * ? ? ?
* corollary 51.22 ) .
let @xmath1300 be a potential for the soliton , i.e. @xmath1747 and @xmath1748 there is a constant @xmath78 so that @xmath1749 . from ( * ? ?
* ; * ? ? ?
* theorem 3.7 ) and ( * ? ? ?
* ; * ? ? ?
* proposition 2.5 ) , for any sequence @xmath1750 , after passing to a subsequence there is a limit @xmath1751 with continuous convergence on compact subsets of @xmath15 . in the case of a gradient shrinking soliton , ( * ? ? ?
* ; * ? ? ?
* chapter 7.7.3 ) implies that @xmath1752 ; c.f .
* ; * ? ? ?
* example 3.3 ) .
thus the limit @xmath1753 exists and equals @xmath1754 , independent of @xmath1616 . in our case , the formulas for @xmath917 and @xmath937 now follow from straightforward calculation .
we recall some estimates on the @xmath259-function that hold for @xmath10-solutions , taken from @xcite the letter @xmath78 will denote a generic universal constant . from (
* ; * ? ? ?
* ( 2.53 ) ) , @xmath1755 from ( * ? ? ?
* ; * ? ? ?
* ( 2.54),(2.56 ) ) , @xmath1756 from ( * ? ? ?
* ; * ? ? ?
* ( 2.55 ) ) , @xmath1757 from ( * ? ? ?
* ; * ? ? ?
* ( 2.57 ) ) , @xmath1758 from ( * ? ? ?
* ( 3.7 ) ) , @xmath1759 in this section we recall the notion of a canonical neighborhood for a ricci flow with surgery , and define the notion of a canonical neighborhood in a singular ricci flow .
we mention that this rather complicated looking definition is motivated by the structure of @xmath10-solutions and the standard ( postsurgery ) solution .
let @xmath1760 be a decreasing function .
let @xmath1 be small enough so that the bulletpoints at the end of subsection [ appkappa ] hold ( with @xmath1761 ) .
let @xmath1762 and @xmath1763 be the constants in ( * ? ? ?
* ; * ? ? ?
* definition 69.1 ) . as in (
* ; * ? ? ?
* definition 69.1 ) , a ricci flow with surgery @xmath15 defined on the time interval @xmath1650 $ ] satisfies the _ @xmath7-canonical neighborhood assumption _ if every @xmath1764 with scalar curvature @xmath1765 has a canonical neighborhood in the corresponding forward / backward time slice , in the following sense .
there is an @xmath1766 and an open set @xmath1767 with @xmath1768 that falls into one of the following categories : \(a ) @xmath1769\subset{{\mathcal m}}$ ] is a strong @xmath9-neck for some @xmath1770 .
( note that after parabolic rescaling the scalar curvature at @xmath581 becomes @xmath799 , so the scale factor must be @xmath1771 , which implies that @xmath1772 . )
\(b ) @xmath168 is an @xmath9-cap which , after rescaling , is @xmath63-close to the corresponding piece of a @xmath1736-solution or a time slice of a standard solution .
\(c ) @xmath168 is a closed manifold diffeomorphic to @xmath792 or @xmath793 .
\(d ) @xmath168 is @xmath63-close to a closed manifold of constant positive sectional curvature .
moreover , the scalar curvature in @xmath168 lies between @xmath1773 and @xmath1774 . in cases ( a ) ,
( b ) , and ( c ) , the volume of @xmath168 is greater than @xmath1775 . in case ( c ) ,
the infimal sectional curvature of @xmath168 is greater than @xmath1773 .
finally , we require that @xmath1776 where @xmath1361 is the constant from ( * ? ? ?
* ; * ? ? ?
* ( 59.5 ) ) . here
the time dervative @xmath1777 should be interpreted as a one - sided derivative when the point @xmath581 is added or removed during surgery at time @xmath4 .
we use a slightly simpler definition of canonical neighborhood in the case of singular ricci flows , for definition [ def_singular_ricci_flow ] .
we do not need to consider forward / backward time slices , and in case ( b ) , we do not need to consider the case that @xmath168 is close to a time slice of a standard solution . alternatively , for a singular ricci flow , one could replace the above definition of canonical neighborhood with the requirement that every point with @xmath1778 is @xmath9-modelled on a @xmath1779-solution .
this is quantitatively equivalent to the definition above , as follows from proposition [ goodeps ] and ( * ? ? ?
* ; * ? ? ?
* lemma 59.7 ) .
we recall that there are certain parameters in the definition of ricci flow with surgery , namely a number @xmath1 and positive nonincreasing functions @xmath1780 .
the function @xmath7 is the canonical neighborhood scale ; c.f .
appendix [ canon ] .
the function @xmath10 is the noncollapsing parameter ; c.f .
appendix [ kappa ] .
the parameter @xmath1 is a global parameter in the definition of a ricci flow with surgery ( * ? ? ?
* ; * ? ? ?
* remark 58.5 ) .
the function @xmath1781 is a surgery parameter .
there is a further parameter @xmath1782 so that if a point @xmath581 lies in an @xmath63-horn and has @xmath1783 , then @xmath581 is the center of a @xmath5-neck ( * ? ? ?
* ; * ? ? ?
* lemma 71.1 ) .
one can then perform surgery on such cross - sectional @xmath71-spheres ( * ? ? ?
* ; * ? ? ?
* sections 72 and 73 ) .
perelman showed that there are positive nonincreasing step functions @xmath1784 , @xmath1785 and @xmath1786 so that if the ( positive nonincreasing ) function @xmath3 satisfies @xmath1787 then there is a well - defined ricci flow with surgery , with a discrete set of surgery times ( * ? ? ?
* ; * ? ? ?
* sections 77 - 80 ) .
in particular , we can assume that @xmath3 is strictly decreasing .
if @xmath1788 and @xmath1789 are positive functions then the @xmath1784-canonical neighborhood assumption implies the @xmath7-canonical neighborhood assumption , and @xmath1785-noncollapsing implies @xmath10-noncollapsing .
hence ricci flow with surgery also exists in terms of the parameters @xmath1790 .
consequently , we can assume that @xmath7 , @xmath10 and @xmath3 are strictly decreasing .
as in ( * ? ?
* ; * ? ? ?
* section 68 ) , the formal structure of a _ ricci flow with surgery _ is given by * a collection of ricci flows @xmath1791 , where @xmath1792 , @xmath1793 is a compact ( possibly empty ) manifold , @xmath1794 for all @xmath1795 , and the flow @xmath1796 goes singular at @xmath1797 for each @xmath1798 .
we allow @xmath1799 to be @xmath1800 . * a collection of limits @xmath1801 , in the sense of ( * ? ? ?
* ; * ? ? ?
* section 67 ) , at the respective final times @xmath1797 that are singular if @xmath1798 .
( here @xmath1802 is an open subset of @xmath1793 . ) * a collection of isometric embeddings @xmath1803 where @xmath1804 and @xmath1805 , @xmath1806 , are compact @xmath0-dimensional submanifolds with boundary .
the @xmath1807 s are the subsets which survive the transition from one flow to the next , and the @xmath1808 s give the identifications between them .
we will say that @xmath4 is a _ singular time _ if @xmath1809 for some @xmath1795 , or @xmath1810 and the metric goes singular at time @xmath1799 .
a ricci flow with surgery does not necessarily have to have any real surgeries , i.e. it could be a smooth nonsingular flow .
we now describe the ricci flow spacetime associated with a ricci flow with surgery .
we begin with the time slab @xmath1811 $ ] for @xmath1812 , which has a time function @xmath1813{\rightarrow}[t_k^-,t_k^+]$ ] given by projection onto the second factor , and a time vector field @xmath21 inherited from the coordinate vector field on the factor @xmath1814 $ ] . for every @xmath1815 , put @xmath1816 and for @xmath1795 , let @xmath1817 .
since @xmath1818 is a closed subset of the @xmath64-manifold with boundary @xmath1811 $ ] , the complement @xmath1819)\setminus ( w_k^-\cup w_k^+)$ ] is a @xmath64-manifold with boundary , where @xmath1820 .
note that the ricci flow @xmath1821 with singular limit @xmath1822 defines a smooth metric @xmath1823 on the subbundle @xmath1824 that satisfies @xmath1825 . for every @xmath1795 , we glue @xmath1826 to @xmath1827 using the identification @xmath1828 , to obtain a smooth @xmath64-manifold with boundary @xmath15 , where @xmath1829 is the image of @xmath1830 under the quotient map @xmath1831 .
the time functions , time vector fields , and metrics descend to @xmath15 , yielding a tuple @xmath14 which is a ricci flow spacetime in the sense of definition [ def_ricci_flow_spacetime ] .
recall the notion of a normalized riemannian manifold from the introduction .
our convention is that the trace of the curvature operator is the scalar curvature . from (
* ; * ? ? ?
* appendix b ) , if a smooth three - dimensional ricci flow @xmath15 has normalized initial condition then the scalar curvature satisfies @xmath1832 if follows that the volume satisfies @xmath1833 these estimates also hold for a ricci flow with surgery .
let @xmath1834 be a symmetric @xmath1835 real matrix .
let @xmath1836 denote its smallest eigenvalue .
for @xmath546 , put @xmath1837 then @xmath1838 is a family of @xmath1839-invariant convex sets which is preserved by the ode on the space of curvature operators ( * ? ? ? * ; * ? ? ? * pf .
of theorem 6.44 ) .
if a smooth three - dimensional ricci flow has normalized initial conditions then the time - zero curvature operators lie in @xmath1840 . using ( [ scalarsurg ] )
, we obtain the hamilton - ivey estimate that whenever the lowest eigenvalue @xmath1841 of the curvature operator satisfies @xmath1842 , we have @xmath1843 the surgery procedure is designed to ensure that ( [ hi ] ) also holds for ricci flows with surgery .
( perelman s definition of a normalized riemannian manifold is slightly different ; he requires that the sectional curvatures be bounded by one in absolute value ( * ? ? ?
* ; * ? ? ?
* section 5.1 ) . with his convention ,
@xmath1844 and @xmath1845 . )
the main result in this appendix is proposition [ prop_critical_transitional ] , which is an extension of proposition [ prop_cylinder_stability_1_noncompact ] to general @xmath10-solutions .
it uses similar ideas as proposition [ prop_cylinder_stability_1_noncompact ] but is a bit more complicated to state .
we include this for the sake of completeness , as it gives a general quantitative picture of the behavior of @xmath10-solitons over time .
it is not needed in the body of the paper .
suppose that @xmath15 is a @xmath10-solution defined on @xmath1000 $ ] , and @xmath1846 is the @xmath259-function with spacetime basepoint @xmath1001 .
roughly speaking , proposition [ prop_critical_transitional ] says that the reduced volume for a @xmath10-solution , as a function of time , has two types of behavior : it can remain close to the reduced volume of a gradient shrinking soliton over a long time interval , or it can transition relatively quickly between two such values . during long time intervals of the former kind , the @xmath10-solution and @xmath259-function are close , modulo parabolic scaling , to a gradient shrinking soliton and its @xmath259-function .
we now give the details .
fix @xmath1847 . by lemma [ lem_classification_gradient_soliton_kappa_solutions
] there are only finitely many @xmath0-dimensional gradient shrinking solitons that are @xmath10-solutions .
let @xmath1848 be the set of their reduced volumes , in the sense of lemma [ lem_classification_gradient_soliton_kappa_solutions ] . 1 .
if @xmath48 is odd then @xmath1856 .
if @xmath48 is even , then 1 . for every @xmath1857 with @xmath1858 ,
the rescaled triple @xmath1859 is @xmath778-close to a triple @xmath930 where @xmath45 is a gradient shrinking soliton with potential function @xmath917 , and @xmath1860 is @xmath778-close to the reduced volume of @xmath45 .
( see appendix [ appcloseness ] for the definition of @xmath778-closeness of triples ) .
2 . @xmath1861 if @xmath1854 and @xmath1862 if @xmath1863 .
3 . @xmath1860 tends to an element of @xmath1864 as @xmath108 .
then the rescaled triple @xmath1872 is @xmath769-close to a triple @xmath930 where @xmath45 is a gradient shrinking soliton with potential function @xmath917 , and @xmath1873 is @xmath769-close to the reduced volume of @xmath45 .
suppose the lemma were false .
then for some @xmath765 and @xmath827 , there would be sequences @xmath34 , @xmath1874 , @xmath1875 and @xmath1876 with @xmath1877 , @xmath1878 , @xmath1879 and @xmath1880 does not satisfy the conclusion of the lemma .
proposition [ prop_convergence_no_curvature_bound ] implies that after passing to a subsequence , we have convergence of triples to a @xmath10-solution @xmath1881 . from the assumed properties of @xmath34 ,
the reduced volume @xmath1882 of @xmath45 will be constant for @xmath1883 .
hence @xmath45 is a gradient shrinking soliton , which gives a contradiction . from the monotonicity of the reduced volume @xmath1860 ,
the existence of the asymptotic soliton and the fact that @xmath1884 , the reduced volume of flat @xmath933 ( as a gradient shrinking soliton ) , it follows that for all @xmath950 , we have @xmath1885 $ ] .
let @xmath1886 and @xmath1887 be as in lemma [ lem_nearly_constant_implies_soliton ] .
choose @xmath1888 small enough that the intervals @xmath1889\}_{1\leq j\leq k}$ ] are disjoint .
put @xmath1890 $ ] for @xmath1891 , and @xmath1892 $ ] .
we know that for each @xmath568 , there is a point @xmath1896 with @xmath1897 .
let @xmath1898 and @xmath1899 be as in lemma [ lem_nearly_constant_implies_soliton ] .
note that for every @xmath1894 , since @xmath1900 for all @xmath1901 , the contrapositive of lemma [ lem_nearly_constant_implies_soliton ] implies that for any @xmath1902 with @xmath1903 , we have @xmath1904 . iterating this @xmath1905 $ ] times gives a bound @xmath1906 the complement of @xmath1907 in @xmath1851 $ ] is a union of open intervals .
let @xmath1908 be one such interval
. then @xmath1909 takes values in an interval of length at most @xmath1910
. therefore by lemma [ lem_nearly_constant_implies_soliton ] , if @xmath96 and @xmath1911 , then for all @xmath1912 with @xmath1858 , conclusion ( 2a ) of proposition [ prop_critical_transitional ] holds .
thus we may obtain the desired partition of @xmath1851 $ ] by enlarging each nonempty interval @xmath1913 $ ] to @xmath1914 $ ] , forming the union @xmath1915 , and taking the associated partition of @xmath1851 $ ] . conclusion ( 2b ) of
the proposition follows from the construction .
conclusion ( 2c ) of the proposition follows from the finiteness of @xmath1916 , the monotonicity of the reduced volume and the existence of the asymptotic soliton . 1 .
@xmath65 is diffeomorphic to a spherical space form other than @xmath792 or @xmath1918 : @xmath1919 .
@xmath65 is diffeomorphic to the cylinder : @xmath1920 .
@xmath65 is diffeomorphic to the @xmath894-quotient of the cylinder : @xmath1921 , @xmath1922 , @xmath1920 .
@xmath65 is diffeomorphic to @xmath792 : @xmath1923 , @xmath1924 , @xmath1920 .
@xmath65 is diffeomorphic to @xmath1918 : @xmath1925 , @xmath1926 , @xmath1922 , @xmath1921 , @xmath1920 .
the relevant gradient shrinking solitons are @xmath1927 , @xmath1928 , @xmath885 and @xmath886 . from lemma [ lem_classification_gradient_soliton_kappa_solutions ] ,
their reduced volumes are @xmath1929 , @xmath1930 , @xmath1931 and @xmath1932 , respectively .
the corollary follows from the fact that the reduced volumes of the @xmath1933 s form a strictly increasing sequence , the local stability of quotients of @xmath886 , and topological restrictions .
m. gromov , j. cheeger and m. taylor , `` finite propagation speed , kernel estimates for functions of the laplace operator , and the geometry of complete riemannian manifolds '' , j. diff .
- 53 ( 1982 )
|
we introduce singular ricci flows , which are ricci flow spacetimes subject to certain asymptotic conditions .
we consider the behavior of ricci flow with surgery starting from a fixed initial compact riemannian @xmath0-manifold , as the surgery parameter varies .
we prove that the flow with surgery subconverges to a singular ricci flow as the surgery parameter tends to zero .
we establish a number of geometric and analytical properties of singular ricci flows .
|
supersymmetry ( susy ) has been considered to be the best candidate beyond the standard model ( sm ) from a viewpoint of both the hierarchy problem and the gauge coupling unification .
recent astrophysical observations showing the existence of a substantial amount of non - relativistic and non - baryonic dark matter seem to make susy even more promising .
the lightest susy particle ( lsp ) of r - parity conserving models , in most cases the lightest neutralino , can serve as a good candidate for dark matter ( dm ) . the parameter space of the constrained mssm , however , is strongly restricted by the requirement of matching the precise measurement of the dm relic density as measured by the wmap .
the mssm also suffers from a naturalness problem ( the so - called @xmath0 problem ) : why the dimensionful parameter @xmath0 of the supersymmetric higgs mass term @xmath1 has to be of ew scale .
this problem can be solved in the next - to - mssm ( nmssm ) by promoting the @xmath0 parameter to a new singlet superfield @xmath2 coupled to higgs doublets , @xmath3 @xcite .
this triple - higgs coupling term also helps to push up the mass of the lightest cp - even higgs boson , relaxing the fine - tuning necessary to comply with the lep bounds .
postulating an additional u@xmath4(1 ) gauge symmetry @xcite avoids a massless axion , or domain wall problems of the nmssm .
such a u(1)-extended mssm ( ussm ) can be considered as an effective low - energy approximation of a more complete e@xmath5ssm model @xcite , with other e@xmath5ssm fields assumed heavy .
in addition to the mssm superfields , the ussm contains a chiral superfield @xmath6 and an abelian gauge superfield @xmath7 .
thus the mssm particle spectrum is extended by a new cp - even higgs boson @xmath2 , a gauge bozon @xmath8 and two neutral inos : a singlino @xmath9 and a bino @xmath10 ; other sectors are not enlarged . as a result
the phenomenology of the neutralino sector can be significantly modified both at colliders @xcite and in cosmology - related processes @xcite . to illustrate this we consider a physically interesting scenario with higgsino and gaugino mass parameters of the order @xmath11 gev ) , and we take the interaction between the singlino and the mssm fields to be of the order of the ew scale , @xmath12 gev ) .
we assume the mssm gaugino unification relation @xmath13 and unified couplings @xmath14 , but @xmath15 will be taken as independent to investigate the impact of new states as a function of @xmath15 . for the numerical values
we take @xmath16 tev , @xmath17 tev , @xmath18 tev , @xmath19 , @xmath20 tev , neglect ( small ) @xmath21-@xmath10 mixing , and adopt the e@xmath5ssm assignment for the u@xmath4(1 ) charges @xcite .
unlike the 4x4 mssm case , the full 6x6 neutralino mass matrix can not be diagonalised analytically .
however , since the mixing between the new and mssm states is small @xmath22 compared to @xmath23 , one can perform first the diagonalisation of the 4x4 mssm and the 2x2 @xmath9-@xmath10 submatrices separately .
then the perturbative expansion of the block - diagonalisation in @xmath24 provides an excellent approximation to masses and mixings @xcite .
the mass spectrum is shown in fig.[fig : kalinowski1 ] ( left ) as a function of @xmath15 . for small @xmath25 the eigenstates ( denoted by numbers with primes )
are almost pure mssm u(1 ) and su(2 ) gauginos @xmath26 , mssm higgsinos @xmath27 , and maximally mixed u@xmath4(1 ) gaugino and singlino states , @xmath28 .
when @xmath25 is shifted to higher values , the mass eigenvalues in the new sector move apart , generating strong cross over patterns whenever a ( signed ) mass from the new block comes close to one of the ( signed ) mssm masses .
this happens at @xmath29 tev for @xmath30 and @xmath31 states , and at @xmath32 tev for @xmath33 and @xmath34 . for higher @xmath15
the @xmath34 approaches the singlino state and becomes the lsp .
evolution of ( left ) neutralino masses , ( center ) production cross sections for @xmath35 , @xmath36 and @xmath37 pairs in @xmath38 collisions , and ( right ) partial decay widths of @xmath39 ( from @xcite ) .
_ , title="fig:",width=170,height=162 ] evolution of ( left ) neutralino masses , ( center ) production cross sections for @xmath35 , @xmath36 and @xmath37 pairs in @xmath38 collisions , and ( right ) partial decay widths of @xmath39 ( from @xcite ) .
_ , title="fig:",height=170 ] evolution of ( left ) neutralino masses , ( center ) production cross sections for @xmath35 , @xmath36 and @xmath37 pairs in @xmath38 collisions , and ( right ) partial decay widths of @xmath39 ( from @xcite ) .
_ , title="fig:",width=132 ] -0.5 cm at an @xmath38 collider the production processes @xmath40 are generated by @xmath41channel @xmath42 and @xmath43 exchanges ( mass - eigenstates of @xmath44 and @xmath8 ) , and @xmath45 and @xmath46channel @xmath47 exchanges . in our scenario
@xmath48 gev , the @xmath49 mixing angle @xmath50 , and @xmath51 gev .
the @xmath15 dependence of the production cross sections for the three pairings of the two lightest neutralinos , @xmath52 and @xmath53 , is shown in fig.[fig : kalinowski1 ] ( center ) for @xmath54 gev . for small
@xmath25 the presence of @xmath43 has little influence on @xmath55 which is of similar size as in the mssm for mixed higgsino pairs .
however it significantly enhances diagonal higgsino pairs @xmath56 and @xmath57 compared with the mssm , even though the light neutralino masses are nearly identical in the two models . at and beyond the cross over with singlino , @xmath58 tev , dramatic changes set in for pairs involving the lightest neutralino . at the lhc
the neutralinos will be analyzed primarily in cascade decays of squarks or gluinos . in the ussm
the cascade chains may be extended compared with the mssm by an additional step due to the presence of two new neutralino states , for example , @xmath59 with partial decay widths significantly modified by the singlino and bino admixtures .
also the presence of additional higgs boson will influence the decay chains .
moreover , in the cross over zones the gaps between the masses of the eigenstates become very small suppressing standard decay channels and , as a result , enhancing radiative decays of neutralinos .
these decays are particularly important in the cross over at @xmath60 , where the radiative modes @xmath61 become non negligible , see fig.[fig : kalinowski1 ] ( right ) .
since the photon will be very soft , these decays will be invisible making the decay chains apparently shorter .
dependence of ( left ) the predicted relic density of dm , and ( right ) the elastic spin - independent lsp-@xmath62ge cross section . we restrict the right hand plot to 5 tev as there are no new features above this energy.__,width=226,height=188 ] -0.5 cm if the lightest neutralino ( lsp ) is expected to be the source of the relic abundance of dark matter in the universe , the predicted relic density depends on the lsp composition . in the left panel of fig.[fig : kalinowski2 ] it is shown as a function of @xmath15 @xcite . for small @xmath15
the lsp is almost an mssm higgsino and for a mass @xmath63 gev the predicted value falls below the wmap result . as @xmath15 increases , the singlino admixture increases suppressing the lsp annihilation cross section and the predicted relic density increases .
the singlino lsp predominantly annihilates via an off - shell s - channel singlet higgs , which decays to two light higgs bosons .
as @xmath15 increases , the lsp mass decreases and at @xmath64 tev it reaches @xmath65 gev making the resonant annihilation via the heavy higgs boson efficient enough to lower the relic density .
further increase of @xmath25 switches off the heavy higgs resonance and eventually the wmap value is met ( shown as a horizontal band in fig.[fig : kalinowski2 ] ( left ) @xcite ) . around @xmath66
tev the lsp becomes lighter than the light higgs .
this switches off the annihilation via an off - shell singlet higgs , @xmath67 , normally the dominant annihilation mode of a singlino lsp . as a result
the relic density rises sharply . further increasing @xmath15 decreases the lsp mass until it matches the resonant annihilation channels of the light higgs ( at around 14 tev ) and @xmath44 boson ( at around 20 tev ) . in both cases
this results in a significant dip in the relic density .
the singlino nature of the lsp is also of importance for direct dm searches .
it has a strong impact on the elastic spin - independent scattering off the nuclei , e.g. as shown in fig.[fig : kalinowski2 ] ( right ) @xcite for the @xmath62ge nucleus ( the numerical codes have been developed in @xcite ) . for small @xmath15 the two lightest neutralinos ( 3 and 4 in fig.[fig : kalinowski1 ] ) are almost pure maximally mixed mssm higgsinos .
when @xmath15 increases , the mixing with singlino lowers @xmath68 so that at @xmath69 tev the state 4 becomes the lsp . since the higgsino mixing angles are such that the elastic scattering of the state 4 is almost two orders of magnitude smaller than for the state 3
, it explains a sudden drop seen in fig.[fig : kalinowski2 ] ( right ) . at the same time
the singlino and bino admixture of the lsp increases , which explains a local maximum around 2.8 tev .
as the singlino component ( the state 5 ) of the lsp becomes dominant for higher @xmath15 values , the elastic cross section becomes smaller and smaller .
the u(1 ) extended mssm provides an elegant way of solving the @xmath0 problem . as the neutralino sector is extended , the collider phenomenology can significantly be altered and new scenarios for matching the wmap constraint can be realised .
one example , in contrast to the nmssm , is that the ussm contains regions in which predominantly singlino dark matter can fit the wmap relic density measurement without the need for coannihilation , or resonant s - channel annihilation processes , where the lsp annihilates via @xmath70 .
work supported by the polish ministry of science and higher education grant no .
1 p03b 108 30 and the ec project mtkdct-2005 - 029466 `` particle physics and cosmology : the interface '' .
jpr would like to thank alexander pukhov and andre semenov for useful advice .
s. f. king , s. moretti and r. nevzorov , phys .
d * 73 * ( 2006 ) 035009 [ arxiv : hep - ph/0510419 ] , and references therein . s. y. choi , h. e. haber , j. kalinowski and p. m. zerwas , nucl .
b * 778 * ( 2007 ) 85 [ arxiv : hep - ph/0612218 ] , and references therein .
v. barger , p. langacker , i. lewis , m. mccaskey , g. shaughnessy and b. yencho , phys .
d * 75 * ( 2007 ) 115002 [ arxiv : hep - ph/0702036 ] , and references therein .
d. jarecka , j. kalinowski , s.f.king and j. roberts , in preparation .
|
the neutralino sector of the u(1 ) extended susy is presented and some collider and cosmology - related phenomenology discussed .
|
typically , linear response theory equations are derived by adding a specific time - dependent perturbing function to the hamiltonian , which is usually harmonic in time , resulting in a set of rpa - like equations .
these equations are then solved , for a given energy , using several methods ( see for example refs .
@xcite ) to give the response of the system to a specific collective excitation mode . in this paper , a new alternate method
is presented to calculate response theory .
a specific time - dependent perturbing external piece is added to the static hamiltonian to give a time - dependent total hamiltonian , @xmath3 .
a static hartree - fock solution is then time - evolved using this @xmath3 in a time - dependent hartree - fock ( tdhf ) calculation . the time - dependent result is then fourier transformed to give the response of the system for all energies . in this scheme one recovers both the response spectrum as well as the total transition probability amplitude corresponding to a given specific collective mode .
similar analyses of the long - time evolution of tdhf equations to study collective vibrations have been utilized in the past @xcite , as well as extensions to study the damping of giant resonances @xcite .
the main advantage of this approach is that the dynamical response calculation is constructed directly on top of a static hartree - fock calculation and hence the static and dynamical calculations are calculated using the same hamiltonian description .
therefore there is a complete consistency between the static ground state of the system and the response calculations .
one can then provide a coherent description of static properties of nuclei and of dynamical properties .
this is important for example in @xmath4decay calculations of exotic nuclei , where reliable predictions are very sensitive to the deformation properties of the nucleus @xcite .
hence in this formalism consistent predictions of both the deformation and reaction rate properties are possible .
the static and dynamical hartree - fock calculations are performed using a 3-d cartesian basis - spline collocation expansion @xcite .
properties of basis - splines include the practicality associated with coordinate - space lattice grids , while also providing accurate representations of the gradient operator and a good description of the continuum . in section
ii , the time - dependent evaluation of the response theory is discussed and shown to give the total transition probability . numerical details and sample results
are presented in section iii .
the response equations can be derived from a specific time - dependent perturbation functional of the tdhf equations @xcite . to begin the proof a solution to the static schrdinger equation
is written as follows : @xmath5 a time - dependent perturbing function is added to the static hamiltonian : @xmath6 the external piece is defined as : @xmath7 f(t ) , \label{eq:3.3}\end{aligned}\ ] ] where @xmath8 is the number density operator and @xmath9 corresponds to a one - body operator to excite a particular collective mode .
the functions , @xmath9 and @xmath10 will be chosen later . at some time
@xmath11 the external piece of the hamiltonian is turned on where @xmath12 is the solution to the time - dependent schrdinger equation : @xmath13 |{\bar \psi}_s(t)\rangle .
\label{eq:3.4}\ ] ] here the subscript `` @xmath14 '' refers to the schrdinger picture .
a solution of the following form is then constructed : @xmath15 where for @xmath16 , @xmath17 . using eq .
( [ eq:3.4 ] ) , the function @xmath18 can be shown to be a solution to : @xmath19 where the superscript `` @xmath20 '' refers to the interaction picture , which reduces to the heisenberg picture when @xmath21 .
the solution to eq .
( [ eq:3.6 ] ) can be written iteratively as : @xmath22 at this point instead of using the standard route of letting @xmath40 to recover the linear response equations , we choose an alternate technique to calculate the response . in this case
we evolve the system in time and then fourier transform the result , where @xmath41 is a perturbing function .
we choose @xmath10 to be a gaussian of the following form , @xmath42 where @xmath43 is some small number @xmath44 , chosen such that we are in the linear regime .
the parameter , @xmath45 , is set to be @xmath46 , which allows for a reasonable perturbation of collective energies up to @xmath47 mev . in practice ,
our numerical calculations proceed as follows : first , we generate highly accurate static hf wave functions on the 3-d lattice .
then the external time - dependent perturbation , eq .
( [ eq:3.3 ] ) , is set up by choosing a particular form for @xmath9 , and using eq .
( [ eq:3.17 ] ) for @xmath10 .
next , we solve the tdhf equations utilizing the time - evolution operator @xmath48 , \label{eq:3.18}\ ] ] where @xmath49 $ ] denotes time - ordering .
using infinitesimal time increments , the time - evolution operator is approximated by @xmath50 , \nonumber\end{aligned}\ ] ] where the quantity @xmath51 is evaluated by repeated operations of @xmath52 upon the wave functions .
the fluctuations in the isoscalar axial quadrupole moment as a function of time are shown for @xmath0o , @xmath2s , and @xmath1ca using the skm@xmath53 interaction .
the size of the time step is @xmath54 fm / c .
a grid of @xmath55 with a cartesian box dimensioned @xmath56 is used . ] from the numerical solution of the tdhf equations , we obtain the density fluctuation as a function of time , eq .
( [ eq:3.12 ] ) .
after fourier transforming this quantity and folding with the function @xmath9 , we obtain @xmath57 in eq .
( [ eq:3.15 ] ) , from which the linear response structure function of the system is extracted .
the static and time - dependent hartree - fock calculations are performed using a collocation spline basis in a three - dimensional lattice configuration .
basis - splines allow for the use of a lattice grid representation of the nucleus , which is much easier to use than alternate basis techniques , such as multi - dimensional harmonic oscillators . also , for studies of exotic nuclei , because of weak binding , the density distributions tend to extend to large distances and hence one finds a large sensitivity to a harmonic oscillator basis due to the unphysical description of continuum states , while for a lattice grid representation one needs to simply increase the size of the box .
traditional grid representations typically use finite difference techniques to represent the gradient operator .
collocation basis - splines allow for the gradient operator to be represented by its action upon a basis function in a matrix form .
thus the collocation method gives a much more accurate representation of the gradient , while maintaining the convenience of a lattice grid and hence provides a much more accurate calculation in the end @xcite . the imaginary part of the response function corresponding to the isoscalar quadrupole moment is shown for @xmath0o using three different skyrme force parametrizations , where the skm@xmath53 , sgii and skii versions are shown in the upper , middle and lower panels , respectively .
the experimentally measured giant quadrupole resonance is around 20.7 mev . ]
we have performed the usual tests for assuring convergence with respect to the numerical box size and the number of collocation points .
typically a 7th order b - spline is used in a @xmath58 box with @xmath59 grid points .
the calculation can be performed with or without assuming time - reversal symmetry .
the collective linear response may involve particle - hole interactions which are spin - dependent and not time - reversal symmetric .
therefore , for the correct collective content to be included one should not impose time - reversal symmetry in the linear response calculations @xcite . in the results presented in this paper time - reversal symmetry
is not imposed . in a comparison
it is found that imposing time - reversal symmetry causes small shifts in the position of the collective modes on the order of @xmath60 mev .
calculations of isovector dipole ; isovector and isoscalar octupole ; and isoscalar quadrupole collective modes are performed for @xmath0o , @xmath2s , and @xmath1ca using several parameterizations of the skyrme interaction . here the parameterizations known as skii @xcite , skm@xmath53 @xcite and sgii @xcite are used for comparisons .
the exponent of the density in the density - dependent term in skii is @xmath61 , while for skm@xmath53 and sgii this exponent is @xmath62 .
this causes the skii force to produce a rather large nuclear matter incompressibility , while the skm@xmath53 and sgii forces produce more realistic compression properties .
also the skm@xmath53 and sgii forces allow for more stable static hartree - fock and tdhf computations .
the static hartree - fock calculations converge more easily and rapidly for skm@xmath53 and sgii than for ski , skii or skiii .
a time step of @xmath63 fm / c is used for the calculation .
it is found that one can perform the time evolution for up to 32768 time steps without appreciable dissipation .
the results shown here use 16384 time steps for a maximum time of @xmath64 fm / c .
fft results for @xmath0o using three different numbers of time steps .
the case when the number of time steps equals an exact power of two ( 32768 ) is clearly the better converged result , showing almost no positive values . ]
reasonable results are obtained if the parameter @xmath43 in eq .
( [ eq:3.17 ] ) is chosen to fall in the range @xmath65 . by varying the value of @xmath43 ,
the amplitude of the time - dependent density fluctuation then scales proportionally to @xmath43 , thus indicating that we are well within the linear regime of the theory .
the linear response calculations require well converged initial static hf solutions . to test for the convergence of the static hf calculation the energy fluctuation , which is the variance of @xmath34 ,
is minimized .
the energy fluctuation is defined as @xmath66 which measures how close the wave functions are to being eigenstates of @xmath34 .
this measure of convergence is very sensitive to the eigensolutions and is independent of the iteration step size .
for @xmath0o it was found that static hf solutions with energy fluctuation less than about @xmath67 provided adequate starting points for the dynamic calculation , although the smaller the energy fluctuation the better .
the dynamic calculations involve using eq .
( [ eq:3.19 ] ) to evolve the system .
since @xmath68 is an unitary operator , the orthonormality of the system is preserved , therefore it is not necessary to re - orthogonalize the solutions after every time - step .
the stability of the calculation is checked by testing the preservation of the norm of each wave function . the number of terms in the expansion of the exponent in eq .
( [ eq:3.19 ] ) is determined by requiring the norm to be preserved to a certain accuracy ( typically to @xmath69 ) .
the imaginary part of the response function corresponding to the isoscalar quadrupole moment is shown for @xmath1ca using two different skyrme force parametrizations , skm@xmath53 and sgii . ] the time - dependent perturbing part of the hamiltonian is evaluated when the exponential term in eq .
( [ eq:3.17 ] ) is greater than some small number , @xmath70 .
since it is not difficult to evaluate the action of the external part of the hamiltonian on the wave function , @xmath70 is chosen to be very small , ( @xmath71 ) . to allow the fourier transform of @xmath10 to be evaluated easily ,
it is necessary to integrate @xmath72 from @xmath73 to @xmath74 and hence we would like the entire gaussian of the perturbing function , @xmath10 , to be included into the time evolution to the desired accuracy .
the parameter @xmath75 is therefore chosen such that the complete nonzero contribution of the time - dependent perturbation is included @xmath76 for the study of the isoscalar quadrupole moment , the perturbing function @xmath9 , introduced in eq .
( [ eq:3.3 ] ) , is chosen to be the mass quadrupole moment , @xmath77 .
it turns out that other even multipole modes are also excited at the same time ( _ i.e. _ @xmath78 , @xmath79 , ... ) .
one can therefore study the effect of the coupling between the different excitation modes .
the same holds true for the odd multipoles .
this is due to the nonlinear response effects present in the tdhf time evolution @xcite .
the quadrupole collective resonances are calculated for @xmath0o using three different skyrme force parameterizations for comparisons .
two different size grids ( @xmath80 ) were used inside a ( @xmath81 fm)@xmath82 cartesian box along with periodic boundary conditions .
the imaginary part of the response function corresponding to the isoscalar quadrupole moment is shown for @xmath2s using the skm@xmath53 force . ] in fig .
[ fig1 ] the time - dependent evolution of the multipole moment defined as : @xmath83 is shown for @xmath0o , @xmath2s , and @xmath1ca using the skm@xmath53 interaction .
this figure illustrates the periodic character of the calculations with almost no damping . in this case
the smallest oscillation is about @xmath84 fm / c .
a fast fourier transform ( fft ) is used to calculate the fourier transform of @xmath85 to give @xmath86 .
the time - dependent perturbation function , @xmath87 can then be easily factored out using eq .
( [ eq:3.15 ] ) .
one can use the analytic expression for @xmath87 or use a numerical fft calculation , where the difference between the two methods ends up being negligible . in fig .
[ fig2 ] the quadrupole responses for @xmath0o are shown for the three different skyrme cases .
the imaginary part of eq .
( [ eq:4.1 ] ) is fourier transformed and divided by @xmath87 .
recalling eq .
( [ eq:3.16 ] ) , this quantity is derived to be a negative definite quantity .
the skm@xmath53 results for @xmath0o reflect this property very well in fig .
we observe a sharp peak at about @xmath88 mev and a response which is almost purely negative .
the side peaks , which are much less prominent , may not represent physical effects , but may be due to the construction of the continuum . in performing fft transformations
we find it to be important to have the number of points to be a power of two . in fig .
[ fig3 ] we demonstrate this by plotting the response function for three different choices for the total number of time steps which differ from each other by only 20 points each . as one can see the case for 32768 points results in almost no positive values , indicating good convergence .
the three cases give similar pole structures , which are near the experimental isoscalar quadrupole giant resonance .
the experimental peak is centered at an energy of @xmath89 mev with a width of about @xmath90 mev @xcite .
the resonance calculated with the sgii skyrme parametrization is closest to the experimental result , although skm@xmath53 is also fairly close .
the skii parametrization poorly represents the data , but this force is known to give a bad representation of the collective nuclear properties .
however , for the skii parametrization we were able to compare our results for @xmath0o to the continuum rpa calculations of @xcite .
we find that the difference between the two calculations is less than 2% for the l=2 , 3 , and 4 modes .
the difference may be due to the omission of the spin - orbit term in the continuum rpa calculations .
we find the closest agreement for the hexadecupole mode , where both calculations give a peak energy of 28.2 mev .
the widths of resonances are not accurately reproduced , because the continuum is included into the calculation in an approximate fashion and higher order correlations are missing in the tdhf approach . since the calculations are performed in a box , the continuum is represented in terms of discrete pseudo - continuum states , whose density is sensitive to the size of the box .
a larger box size is expected to better represent the continuum with a higher density of pseudo - continuum states .
the property represented in eq .
( [ eq:3.16 ] ) of being purely negative is not strictly reflected by the sgii nor the skii results .
one can see that this property is approximately reflected , but it is clear the convergence of the solution for these two cases is not nearly as good as for the skm@xmath53 case . the imaginary part of the response function corresponding to the isoscalar and isovector octupole , and isovector dipole are shown for @xmath0o using the skm@xmath53 force . ]
an alternate check of the calculation is the energy - weighted sum rule ( ewsr ) .
this provides a stringent test of the normalization and is derivable from the static hamiltonian .
the skm@xmath53 result gives @xmath91 of the ewsr , indicating excellent convergence .
for sgii and skii the linear response results give @xmath92 and @xmath93 of the ewsr , respectively , indicating a lack of convergence for these results .
the result for the sgii calculation is particular puzzling , since it is expected that the skm@xmath53 and sgii interactions should be similar in behavior .
the violations of the two convergence tests may be due to coupling to other collective modes , or may be due to numerical inconsistencies .
we have investigated the sgii result by varying the size of the box , while keeping the grid spacing the same .
we have found that a slightly larger box , ( -14,+14 ) fm , seems to be closer to being purely negative . in fig .
[ fig4 ] we show the quadrupole strength function for @xmath1ca using two different skyrme parameterizations . in this case
skm@xmath53 and sgii show a peak at about 17 mev .
we also note the strength of the peak .
in this case the resonance consumes all of the ewsr . in fig .
[ fig5 ] we show the same quantity for @xmath2s using the skyrme m@xmath53 force . in this case
we observe multiple structures in the giant quadrupole resonance ; the energy splitting arises from the prolate quadrupole deformation of the nuclear ground state in this system .
the resonant structure of the hexadecupole moment , @xmath78 , can also be calculated at the same time as the quadrupole moment .
this corresponds to the coupling between the quadrupole and the hexadecupole collective modes . in this case
if one were to include both @xmath94 and @xmath78 into the external time - perturbing piece of the total hamiltonian , then this result corresponds to the coupling term : @xmath95 here @xmath96 and @xmath97 . for the pure hexadecupole giant
resonance both @xmath98 and @xmath99 must be made equal to @xmath78 .
in general , mixed mode analysis produces strength functions with less pronounced peaks and can be used when computational time saving is necessary .
the imaginary part of the response function corresponding to the isoscalar and isovector octupole , and isovector dipole are shown for @xmath0o using the sgii force . ]
the octupole and dipole giant resonances are not symmetric about the @xmath100 plane and hence these nodes do not couple with the symmetric quadrupole and hexadecupole modes .
the isovector octupole moment is defined as : @xmath101 .\ ] ] in fig .
6 the isoscalar dipole , isovector octupole , and isoscalar octupole responses are shown for @xmath0o using the skm@xmath53 interaction . in fig .
7 the same quantities are again plotted using the sgii force .
the isoscalar octupole mode shows three sharp resonance structures at about @xmath102 mev , @xmath103 mev , and @xmath104 mev , for the skm@xmath53 force , whereas the sgii resonances are at the slightly higher energies .
these three peaks are also found in a spherical rpa calculation using the same effective interactions @xcite .
the spherical calculation finds an additional broad peak for skm@xmath53 and sgii centered at an energy of about @xmath105 mev .
this peak is weaker than the 3 peaks at lower energies and is not observed in the three - dimensional linear response calculation . in fig .
6 one can see that the response is approximately purely negative , but that there are some violations of this feature .
the isovector octupole response is less clear than the other collective modes .
it is also possible that resonances seen in the isoscalar octupole response may also appear in the isovector response due to couplings . in this case
the strength of the peak is expected to be most prominent in its primary channel .
this is most easily observed in fig . 7 for the sgii force .
in this case the lowest resonances at about @xmath106 mev and at @xmath107 mev are very close to the lowest peaks in the isoscalar case but with substantially reduced strength , whereas the most prominent isovector peak is at about @xmath88 mev .
the sorting of the isoscalar versus isovector mode is more complicated for the skm@xmath53 force .
the isovector dipole response resulting from the linear response calculations is not as prominent as the isoscalar responses .
this is most likely due to the absence of a strong collective dipole resonance for this nucleus . for the skm@xmath53 case ,
there are some peaks centered around @xmath108 mev , while for sgii the peaks range about @xmath109 mev .
a method for evaluating the linear response theory using tdhf is formally developed and implemented .
this method allows one to construct the dynamic calculation directly on top of the static hartree - fock calculation .
therefore , by performing a sophisticated and accurate three dimensional static hartree - fock calculation , we have a correspondingly accurate and consistent dynamic calculation .
a coherent description of static ground state properties , such as binding energies and deformations is given along with a description of the collective modes of nuclei .
a three - dimensional collocation basis - spline lattice representation is used , which allows for a much more accurate representation of the gradient operator and hence a correspondingly accurate overall calculation .
example calculations of two spherical systems ( @xmath0o , @xmath1ca ) and of a system with prolate quadrupole deformation ( @xmath2s ) are presented for the response functions corresponding to various isoscalar and isovector multipole moments .
the skm@xmath53 case for the axial isoscalar quadrupole mode gives excellent results , obeying expectations from the theory and also satisfying the energy weighted sum rule .
this work is supported by the u.s .
department of energy under grant no .
de - fg02 - 96er40963 with vanderbilt university .
some of the numerical calculations were carried out on ibm - sp supercomputers at the national energy research scientific computing center ( nersc ) .
99 g. f. bertsch and s. f. tsai , phys .
reports * 18 * , 125 ( 1975 ) .
s. krewald , v. klemt , j. speth , and a. faessler , nucl . phys . *
a281 * , 166 ( 1977 ) .
k. f. liu and g. e. brown , nucl .
* a265 * , 385 ( 1976 ) . n. van giai and h. sagawa , nucl . phys . *
a371 * , 1 ( 1981 ) .
j. blocki and h. flocard , phys .
* 85b * , 163 ( 1979 ) .
a. s. umar , m. r. strayer , r. y. cusson , p .-
reinhard , and d. a. bromley , phys .
* c 32 * , 172 ( 1985 ) .
a.s . umar and m.r .
strayer , phys .
171b * , 353 ( 1986 ) . c. simenel , ph .
chomaz , and g. de france , phys .
lett . * 86 * , 2971 ( 2001 ) . c. simenel and ph .
chomaz , phys .
c * 68 * , 024302 - 1 ( 2003 ) .
de blasio , w. cassing , m. tohyama , p.f .
bortignon , and r.a .
broglia , phys .
lett . * 68 * , 1663 ( 1992 ) .
m. tohyama and a.s .
umar , phys .
b 549 , 72 ( 2002 ) .
o. sorlin _ el . al _ , phys
* c 47 * , 2941 ( 1993 ) .
a. s. umar , j. wu , m. r. strayer and c. bottcher , j. of comp . phys . * 93 * , 426 ( 1991 ) . a. s. umar , m. r. strayer , j .- s .
wu , d. j. dean and m. c. gl , phys .
rev . * c 44 * , 2512 ( 1991 ) .
a. l. fetter and j. d. walecka , _ quantum theory of many - particle systems _ , ( st . louis : mcgraw - hill book co. , 1971 ) .
d. vautherin and d. m. brink , phys . rev .
* c 5 * , 626 ( 1972 ) .
j. bartel , p. quentin , m. brack , c. guet , and h. b. hankansson , nucl . phys . * a 386 * , 79 ( 1982 )
. k. t. knpfle , g. j. wagner , h. breuer , m. rogge , c. mayer - bricke , phys . rev
* 35 * , 779 ( 1975 ) .
reinhard and y. gambhir , ann . phys .
( leipzig ) * 1 * , 598 ( 1992 ) ; and private communication .
|
a new alternate method for evaluating linear response theory is formally developed , and results are presented .
this method involves the time - evolution of the system using tdhf and is constructed directly on top of a static hartree - fock calculation . by fourier transforming the time - dependent result the response function and the total probability amplitude are extracted .
this method allows for a coherent description of static properties of nuclei , such as binding energies and deformations , while also providing a method for calculating collective modes and reaction rates .
a full 3-d cartesian basis - spline collocation representation is used with several skyrme interactions .
sample results are presented for the giant multipole resonances of @xmath0o , @xmath1ca , and @xmath2s and compared to other calculations .
= 10000 = 10000 = 10000
|
it is still premature to conclude that the recently discovered scalar particle with a mass of around 125 gev at the lhc @xcite is the standard model ( sm ) higgs boson . among other possibilities
, it can very well be the lightest neutral higgs boson of the minimal supersymmetric standard model ( mssm ) , which is the most advocated beyond the sm scenario .
be that as it may , stringent constraints apply on the supersymmetric parameter space , which are , at least for the minimal versions , more severe than those placed from the non - observation of superparticles from direct searches at the colliders . in spite of several advantages that supersymmetry offers , like the solution of the big hierarchy problem and even the provision of a cold dark matter candidate
, we can not escape the following pertinent question : does supersymmetry even partially ease the existing tension between the sm predictions and the experimental measurements for some specific observables ?
the tension with @xmath9 of muon , namely , @xmath19 , i.e. a 3.5@xmath10 deviation of the measured value @xcite from its sm expectation @xcite , is an old intriguing one .
similarly , there is a longstanding @xmath20 discrepancy between the sm expectation and the experimental value for the @xmath21 forward - backward asymmetry @xmath22 measured on the @xmath3-peak @xcite .
recently , a 1.2@xmath10 discrepancy , but with an opposite pull to that of @xmath22 , between the sm prediction and the experimental value for @xmath23 has been reported following the latest sm calculation taking electroweak two - loop plus order @xmath24 three - loop effects @xcite .
although 1.2@xmath10 is not statistically too much significant , it is this opposite pull between @xmath5 and @xmath22 that we would like to exploit for discriminating supersymmetric models .
possible ways to ease these hiccups in a model - independent effective theory approach and the corresponding lhc signatures have recently been discussed @xcite .
but , what these tensions imply for supersymmetric models is the main concern of the present paper . for definiteness
, we will start with the parameter space of the constrained mssm , called cmssm , completely specified at the grand unification ( gut ) scale by a common scalar mass ( @xmath25 ) , a common gaugino mass @xmath26 , a universal scalar trilinear parameter ( @xmath27 ) , the ratio of the vacuum expectation values of the two higgs doublets ( @xmath28 ) , and the sign of the higgsino mass parameter ( @xmath29 ) @xcite .
we place indirect constraints on the cmssm parameter space from observables classified here under three categories : 1 .
the lightest cp - even higgs mass @xmath1 gev , reaching out to which requires the stop squarks to be very heavy ( several tev ) or having a substantial mixing between their left and right components @xcite . since all superparticle masses are interrelated in terms of the @xmath30 gut scale parameters , the higgs mass puts by far the strongest constraint on the cmssm parameter space .
2 . a set of @xmath8-physics observables , namely , @xmath31 , @xmath32 , @xmath33 , and @xmath34 .
the first two processes , being loop - induced in both sm and cmssm , put a strong constraint on the cmssm parameters from the fact that the data on them are quite precise and completely consistent with the sm expectations .
the new physics parts for the remaining ones are mediated essentially by the charged higgs boson , and here the data show some intriguing discrepancy with the sm expectations .
we include the measurement of @xmath9 of muon too in this category , which prefers the smuons and gaugino / higgsino to be light .
3 . the @xmath3-peak observables @xmath5 and @xmath11
. both are precision electroweak observables , with a pull of 1.2@xmath10 and 2.5@xmath10 respectively , when experimental data are compared with their sm predictions .
the supersymmetric contributions to these observables in the current context form the core of our analysis .
constraints from the interplay of the first two categories listed above have already been placed by several authors @xcite .
the main thrust of the present paper is to explore if it is possible to additionally satisfy the constraints listed in the third category .
the electroweak precision data was analyzed vis - - vis mssm in ref .
@xcite , but at a time when the discrepancy between the measured value of @xmath5 and the sm one - loop estimate was only about @xmath35 , and the higgs boson was not discovered .
so it is only proper to readdress the issue in the light of the present wisdom , especially when the higgs boson has finally been observed , and estimate of higher - order effects on @xmath5 has widened the discrepancy from @xmath35 to @xmath36 .
also , the opposite pull between @xmath5 and @xmath11 should be paid due attention , as these are precision observables .
we note that in cmssm the masses of the squarks and the charged higgs boson stem from a common @xmath25 .
therefore , non - observation of squarks in direct searches automatically implies that the charged higgs boson has to be heavy .
for this reason , for some of the @xmath8-physics observables where the charged higgs boson contributes as the sole new physics at tree level , its numerical impact is rather suppressed . to liberate the charged higgs boson from this unfavorable tie - up with the squarks
, we consider a slightly wider variant , called the non - universal scalar model ( nusm ) , which is characterized by two different soft scalar masses at the gut scale : @xmath37 for the matter scalars which belong to a 16-plet , and @xmath38 for the higgs scalars which belong to a decuplet , of the underlying so(10 ) @xcite .
note that even if there is a common scalar mass at the planck scale @xmath39 , the matter and higgs scalar masses at the gut scale @xmath40 can be different because of renormalization group running controlled by _ a priori _ unknown physics between @xmath39 and @xmath40 . from this perspective
, one might take @xmath38 and @xmath37 as independent parameters . in this case ,
non - observation of squarks does not necessarily imply that the charged higgs boson is heavy and @xmath41 . for nusm ,
the weak scale parameters @xmath42 and @xmath43 are not independent as at gut scale these two parameters are the same as @xmath44 . ] .
still , the masses of the charged higgs boson and the squarks in nusm are not fully uncorrelated because of the radiative corrections to the higgs mass parameters coming from the stop squarks involving large yukawa couplings .
even then , we observe that a lot of parameter space in the charged higgs mass plane which is disfavored in cmssm is resurrected in nusm . in the last part of this paper , we will consider selected benchmark points of the 19-parameter phenomenological mssm ( pmssm ) @xcite , which is a subspace of pmssm with parameters defined at the weak scale , without introducing any theoretical prejudice about their high scale behaviors .
24 such benchmark model points have been considered in ref .
@xcite , which correspond to different regions of the mssm parameter space , consistent with all other constraints like the higgs mass , dark matter relic density , and direct searches at the lhc . out of these 24 model points ,
5 have been shortlisted in @xcite .
we will analyse how these well - motivated benchmark points react to the combined effects of ( @xmath5@xmath12 ) and a few other observables .
the one - loop corrected @xmath47 coupling , including new physics , can be written as @xmath48b z_\mu \ , , \ ] ] where @xmath49 with @xmath50 and @xmath51 after incorporating the sm electroweak corrections in the @xmath52 scheme @xcite , and @xmath53 capture the higher - order effects coming from new physics . here , @xmath54 , @xmath55 , and @xmath56 .
the loop corrections induced by the superparticles can be approximated as @xmath57 the expressions for the quantities @xmath58 and @xmath59 have been adapted from appendix b of ref .
they contain contributions from top - charged higgs , stop - chargino and sbottom - neutralino loops , expressed in terms of the passarino - veltman functions @xcite . the new contribution to @xmath5
can be written as @xmath60 with @xmath61 \approx \xi \left[-\frac{60}{13 } f_l + \frac{12}{13 } f_r\right ] \
, , \label{rb - susy}\ ] ] where the right - most simplified expression in eq .
( [ rb - susy ] ) is shown only for illustration by approximating @xmath62 and @xmath63 .
_ however , this approximation is shown only for illustrative purpose_. the full expression in mssm can be found in @xcite which we have employed for our numerical analysis .. the gfitter group @xcite has recently updated the sm fit ( after the higgs discovery ) using the improved calculation of @xmath5 with the result that the discrepancy with the measured value is now 1.2@xmath10 @xcite .
the new situation is the following : @xmath64 the change in @xmath22 can similarly be written as @xmath65 \approx -\frac{5}{13}\xi \left[f_l + 5 f_r\right]\ , , \label{afb - susy}\ ] ] for which a @xmath20 discrepancy between the sm and experimental values is known to exist for a while @xcite : @xmath66 a comparative study of @xmath5 and @xmath22 is now in order .
it is not difficult to understand the tension between their supersymmetric contributions irrespective of any other constraints . in mssm ( i.e. not just in cmssm ) , both @xmath58 and @xmath59 turn out to be negative , with @xmath67 , because of the presence of the numerically dominant stop - chargino loop contribution for @xmath58 , which is absent for @xmath59 .
thus , one can have a consistent solution for @xmath5 , as can be seen from eqs .
( [ rb - susy ] ) and ( [ rb - data ] ) . on the other hand , as eq .
( [ afb - susy ] ) shows , the supersymmetric contribution to @xmath22 is always positive , while according to eq .
( [ afb - data ] ) experimental data prefers a negative contribution .
thus , not only cmssm but a large class of supersymmetric model for which @xmath68 and @xmath67 can not simultaneously satisfy @xmath5 and @xmath22 . to numerically appreciate the combined effects of @xmath5 and @xmath69 ,
we first define what is known as ` pull ' for a given observable ( @xmath70 ) , as @xmath71 where @xmath72 is the experimental central value and @xmath73 is one standard deviation experimental error .
then we introduce a quantity @xmath74 defined as @xmath75 eqs .
( [ rb - data ] ) and ( [ afb - data ] ) tell us that the pulls of @xmath5 and @xmath11 are @xmath76 and @xmath77 , respectively , so that @xmath78 .
light superparticles can improve the agreement in @xmath5 , but any such contributions can only worsen the tension in @xmath22 , as explained earlier . thus @xmath74 can not be smaller than @xmath77 .
this motivates us to look for the parameter space where @xmath74 lies between 2.5 and its sm value 2.8 ( which corresponds to completely decoupled supersymmetry ) . for illustration ,
we have chosen @xmath79 and 2.7 as two representative values , in the sense that for each case we accept the parameter space which yields @xmath74 less than that value .
we recall in this context that although @xmath5 and @xmath11 both depend on the dynamics of the @xmath47 vertex , the former is a ratio of decay widths and the latter is an asymmetry , and they are experimentally independent observables .
hence , @xmath74 is a good measure of the combined tension produced by these two observables .
the model parameters have been scanned in the following range : @xmath80 , @xmath81 ( all in tev ) , @xmath82 .
for the nusm scenario , we scan over the additional parameter @xmath38 in the range of [ 0:2 ] tev .
we have used suspect2.41 @xcite to generate the weak scale spectrum from the high scale boundary values .
the package micromegas2.4 @xcite has been utilized to extract observables like branching ratios of @xmath83 , @xmath84 and @xmath85 .
we have used the following experimental ranges for the @xmath8-physics observables @xcite , admitting @xmath86 spread around their central values , except for @xmath87 which is taken at @xmath88 confidence limit : @xmath89 -@xmath28 plane from @xmath90 gev , direct searches of gluinos and squarks at the lhc , and the lep limit on the chargino mass .
the region above each line is allowed .
the regions which satisfy @xmath5 at @xmath91 ( blue dashed lines ) , @xmath92 ( red / dark grey patch ) and @xmath93 ( green / light grey patch ) are separately shown .
( b ) right panel : allowed regions from several @xmath8-physics observables , @xmath9 of muon , all taken at @xmath86 .
the red ( dark grey ) and orange ( medium grey ) regions satisfy @xmath92 , while the yellow ( light grey ) and orange ( medium grey ) regions are allowed by @xmath9 of muon . for the @xmath8-physics
observables , the region above each line is allowed.,title="fig:",width=302 ] -@xmath28 plane from @xmath90 gev , direct searches of gluinos and squarks at the lhc , and the lep limit on the chargino mass .
the region above each line is allowed .
the regions which satisfy @xmath5 at @xmath91 ( blue dashed lines ) , @xmath92 ( red / dark grey patch ) and @xmath93 ( green / light grey patch ) are separately shown .
( b ) right panel : allowed regions from several @xmath8-physics observables , @xmath9 of muon , all taken at @xmath86 .
the red ( dark grey ) and orange ( medium grey ) regions satisfy @xmath92 , while the yellow ( light grey ) and orange ( medium grey ) regions are allowed by @xmath9 of muon . for the @xmath8-physics observables ,
the region above each line is allowed.,title="fig:",width=302 ] our results for cmssm are shown in the left and right panels of fig .
[ fig : cmssm ] in the plane of the charged higgs mass and @xmath28 . in both panels ,
we indicate the region which satisfy our criteria for @xmath74 .
_ in the left panel ( a ) _ , besides showing the effect of satisfying the higgs mass , we display the constraints coming from the non - observation of superparticles ( mainly , the gluino and the first two generations of squarks ) from direct searches at the lhc @xcite , as well as those originating from the lep limit on the chargino mass @xcite .
the region which agrees with @xmath5 within 1@xmath10 is shown by blue dashed lines .
while one might argue that a discrepancy of @xmath36 is not of much significance , we observe that softening the same to even @xmath91 requires such light sparticles , in particular the stop squarks , as to be untenable with the higgs mass constraint .
we also show regions which admit @xmath94 ( red or dark grey ) and @xmath93 ( green or light grey ) .
we reiterate that the 2.5@xmath10 pull in @xmath11 essentially controls the constraint from @xmath74 .
_ in the right panel
( b ) _ , we focus on the zone where constraints from several @xmath8-physics observables , as well as the region allowed by @xmath9 of muon , are prominent . wherever a single line is drawn for a particular observable , the space above that line is allowed from the corresponding constraint .
we make a cautionary remark that these lines separate allowed and disallowed regions where all cmssm input parameters are marginalized .
admittedly , we have not tried to simultaneously fit all the electroweak observables , however , with such large values of the sparticle masses , it is only natural to expect that the observables already consistent with the sm would not suffer any untenable tension .
the zone allowed by @xmath9 of muon is shown by yellow patch . the dominant contribution to @xmath9 of muon
comes when the bino and smuons floating inside the loops are light @xcite .
we have kept the lightest higgs boson mass at a conservative value of 123 gev , allowing for the possibility that higher order radiative corrections may contribute a further 2 to 3 gev ( setting @xmath95 gev would result in even tighter constraints ) .
even then , this rules out more region in the @xmath96 versus @xmath28 plane than the other constraints .
the reason is that one requires stop masses as heavy as several tev and a substantial mixing in the stop mass - squared matrix to yield such a higgs mass @xcite . among the flavor observables
, @xmath83 usually provides the strongest constraint @xcite , because it directly constrains the chargino mass which sits in the lower end of the cmssm spectrum .
the constraint gets more stringent at large @xmath28 because the dominant chargino - stop loop amplitude grows as @xmath97 .
the supersymmetric contribution to the branching ratio of @xmath98 grows as @xmath99 @xcite , and therefore at large @xmath28 the constraint is more stringent . as regards the process @xmath100 , the dominant new physics contribution comes from tree - level charged higgs mediation , and for very large @xmath28 provides a stronger constraint than @xmath101 .
the constraint essentially applies on the ratio @xmath102 , which is a general feature of any two - higgs doublet model @xcite . of muon , the combined @xmath8-physics data , and the requirement of satisfying @xmath92 .
the color codes are the same as in fig .
[ fig : cmssm].,title="fig:",width=302 ] of muon , the combined @xmath8-physics data , and the requirement of satisfying @xmath92 .
the color codes are the same as in fig .
[ fig : cmssm].,title="fig:",width=302 ] in fig .
[ fig : compare ] , we compare cmssm ( left panel ) with nusm ( right panel ) . in each panel
we show the impact of the combined constraints coming from the higgs mass , @xmath8-physics observables , @xmath9 of muon , and @xmath74 .
constraints from the higgs mass are similar in the two cases in spite of the fact that in nusm the charged higgs mass has its origin in @xmath38 while the squark masses are unified at a different value @xmath37 .
the reason can be traced to the weak scale sum rule that holds in both cases , namely , @xmath103 , where @xmath41 is the cp - odd higgs mass . in both cases ,
the satisfaction of @xmath104 gev requires @xmath41 to be very large ( the ` decoupling ' limit ) .
a crucial observation is that unlike in cmssm there is a small region in nusm which satisfies both @xmath94 and the higgs mass constraint .
we now comment on how the requirement of @xmath74 being less than some representative value is transmitted to an upper limit on the charged higgs mass .
this upper limit comes from the fact that the masses of the charged higgs and the stop squarks , both floating in independent triangle loops of the effective @xmath47 vertex , are intimately related through @xmath25 and @xmath105 . for moderate @xmath28 ,
one gets @xmath106 , while it is somewhat reduced for large @xmath28 due to negative contribution from bottom and tau yukawa couplings .
the squark masses of the first two generations are well approximated as @xmath107 . for the third generation
the expressions are somewhat involved due to large mixing in the squared mass matrix , with the result that one eigenvalue is lighter than any of the first two generation squark masses .
the charged higgs loop contribution to @xmath5 is numerically sub - dominant compared to the stop loop .
the limit from @xmath5 or @xmath11 , i.e. from @xmath74 , effectively applies on the stop mass , which is then translated to the charged higgs mass .
for nusm , the situation is different as the masses of the charged higgs and the stop squarks stem from different parameters , @xmath38 and @xmath37 , respectively .
still , an upper limit on the charged higgs mass arises , though it is much weaker than in cmssm .
the reason is that in nusm the charged higgs mass and the squark masses are not _ completely _ independent .
as mentioned earlier , though @xmath108 and @xmath109 originate from the same @xmath38 at high scale , their splitting at the weak scale following renormalization group running picks up the squark mass dependence ( essentially of the third generation ) .
so , beyond a certain charged higgs mass , the stop squarks become too heavy to leave any impact on @xmath74 .
this is how an upper limit is placed on the charged higgs mass in nusm , though it is weaker than in cmssm .
we now make some remarks on the impact of the observables @xmath110 and @xmath111 , defined as @xmath112 , with @xmath113 .
these ratios have recently been measured by the babar collaboration @xcite : @xmath114 which are @xmath115 and @xmath116 away from their sm estimates , respectively .
like @xmath117 , the supersymmetric contributions to these processes are completely dominated by tree - level charged higgs exchange . as has been observed in ref .
@xcite , consistency with @xmath110 and @xmath118 requires @xmath102 to be @xmath119 and @xmath120 , respectively , ruling out any _ otherwise allowed _ value of @xmath102 at 99.8% confidence level .
this conclusion would hold not only in cmssm or nusm but in any supersymmetric model having a single charged higgs .
a comment on constraints from dark matter , which we have not included in our numerical analysis , is now in order .
corners in cmssm and nuhm parameter spaces which contain points that satisfy both @xmath121 125 gev and the dark matter constraints have been explored , and one such benchmark point for cmssm parameters which satisfies lhc 2012 plus xenon100 dark matter data is : @xmath122 ( all in gev ) , and @xmath123 @xcite . however
, neither this point nor any other benchmark point can satisfy the @xmath74 criteria .
it is perhaps appropriate at this stage to perform a statistical analysis to convey the essence of our study . for this
, we consider the pmssm version of supersymmetry and exhibit how the well - motivated 5 benchmark points , shortlisted in @xcite , respond to @xmath5 , @xmath22 , @xmath110 and @xmath118 .
the 5 models , referenced with identification numbers ` 401479 ` , ` 1046838 ` , ` 2342344 ` , ` 2387564 ` , ` 2750334 ` , produce ( @xmath0 ) bino - stop coannihilation and almost invisibility of the stop , ( @xmath2 ) a pure higgsino as the lightest supersymmetric particle , ( @xmath7 ) a compressed spectrum of squarks coannihilating with bino , ( @xmath124 ) the @xmath125-funnel region with 1 tev bino and 1.4 tev squarks , and ( @xmath126 ) well - tempered neutralino , respectively .
they satisfy all the other experimental constraints , including the higgs mass ( all these points produce a large stop mixing ) , the oblique electroweak @xmath127 and @xmath128 parameters , dark matter relic density and detection cross section .
.[tab : chisq ] the pmssm fit results for 5 specified benchmark points , which predict the higgs mass and the dark matter relic density in the experimentally allowed range and conform to electroweak precision tests .
the canonical observables are @xmath32 , @xmath129 , @xmath130 , and @xmath9 of muon .
the entries in parentheses correspond to fits without the @xmath9 of muon . for details , see text after eq .
( [ pull_numbers ] ) . [ cols="^,^,^,^ " , ] now we perform a statistical analysis with the above benchmark points taking into account a limited set of observables for illustrative discussion . for the five specified benchmark points , considered to be 5 different models , we obtain the pulls for @xmath5 , @xmath22 , @xmath110 and @xmath118
roughly as @xmath131 we have in fact calculated these pulls for all the 24 benchmark points @xcite , and observed that they remain the same up to two decimal places .
note that the changes from the sm pulls are very marginal , and thus we can safely say that as far as these variables are concerned , no pmssm benchmark point performs any better than the sm .
next , we call a set of observables ` canonical ' which comprises of @xmath32 , @xmath129 , @xmath132 , and @xmath9 of muon .
we first calculate the goodness - of - fit , measured by @xmath133 per degree of freedom ( d.o.f ) , for each of the five benchmark points taking into account the observables of the canonical set .
to gain insight into how addition of new observables influences the fit , we increase their numbers in two steps . first , we add @xmath5 and @xmath22 to the canonical set , and calculate @xmath134(d.o.f ) for each benchmark point .
then , we add @xmath110 and @xmath118 on top of @xmath5 and @xmath22 , and check what happens to the @xmath134(d.o.f ) .
each such fit is done _ with _ and _ without _ the @xmath9 of muon which is a part of the canonical set .
the fit results are displayed in table [ tab : chisq ] .
we highlight the salient features that come out of this illustration . 1 . with the canonical set of observables , @xmath134(d.o.f ) les between 3.3 to 4.0 .
the fit gets even worse as we add the four new observables .
this feature holds not only for the 5 specified benchmark points , but also for the larger set of 24 such points . 2 .
the fit improves considerably if one considers only the canonical set excluding the muon @xmath9 .
however , when we include the four new observables , exclusion of @xmath9 is not much of a help in improving the goodness - of - fit .
the reason behind the above behavior of the fit is not difficult to understand .
all these benchmark scenarios correspond to heavy superparticles whose effects in virtual states are too tiny to leave significant numerical impacts on @xmath5 , @xmath22 , @xmath110 , @xmath118 and the muon @xmath9 . to sum up , supersymmetry is not any better than the sm in resolving the tensions in the aforementioned observables .
we now conclude with the following observations .
if the 125 gev scalar resonance discovered at the lhc be after all the lightest cp - even higgs boson of minimal supersymmetry , in particular of the cmssm or nusm variety , then accommodating all the _ three types _ of observables simultaneously becomes extremely difficult . as mentioned earlier ,
these three types are : ( @xmath0 ) @xmath135 gev , ( @xmath2 ) @xmath8-physics observables together with the muon @xmath9 , ( @xmath7 ) @xmath5 and @xmath11 , with their pulls combined to form @xmath74 .
effects of ( @xmath7 ) constitute the punch line of our paper .
the following tensions are to be specially noticed in the context of cmssm and nusm : ( @xmath136 ) better consistency with @xmath74 , which accounts for the combined pull of @xmath5 and @xmath11 , would require the stop squarks to be light , whereas the satisfaction of the higgs mass requires the stop squarks to be heavy .
the tension in nusm is slightly less compared to cmssm .
( @xmath137 ) the incompatibility between @xmath5 and @xmath11 is a generic feature of a large class of supersymmetric models . in the latter part of this paper
, we have considered the 19-parameter pmssm scenario and performed a @xmath133 analysis with 5 experimentally well - motivated benchmark models , which reproduce the observed higgs mass , the dark matter relic density and precision electroweak observables , all consistent with experiments .
we find that inclusions of @xmath74 , @xmath110 and @xmath118 adversely affect the fit .
we point out that models with very light sbottom squarks modify @xmath5 in the right direction @xcite , but only at the expense of growing disagreement with @xmath11 . as far as these quantities are concerned
, one must look beyond the conventional supersymmetric spectrum to find a compromise solution .
* acknowledgments : * ak was supported by csir , government of india ( project no .
03(1135)/9/emr - ii ) , and also by the drs programme of the ugc , government of india .
the research of tsr is supported by the australian research council .
99 s. chatrchyan _ et al . _
[ cms collaboration ] , phys .
b * 716 * , 30 ( 2012 ) [ arxiv:1207.7235 [ hep - ex ] ] .
g. aad _ et al . _ [ atlas collaboration ] , phys . lett .
b * 716 * , 1 ( 2012 ) [ arxiv:1207.7214 [ hep - ex ] ] ; g. aad _ et al .
_ [ atlas collaboration ] , phys .
d * 86 * , 032003 ( 2012 ) [ arxiv:1207.0319 [ hep - ex ] ] .
g. w. bennett _ et al . _ [ muon g-2 collaboration ] , phys .
d * 73 * , 072003 ( 2006 ) [ hep - ex/0602035 ] .
k. hagiwara , r. liao , a. d. martin , d. nomura , and t. teubner , j. phys .
g * 38 * , 085003 ( 2011 ) [ arxiv:1105.3149 [ hep - ph ] ] .
m. davier , a. hoecker , b. malaescu , z. zhang and , eur .
j. c * 71 * , 1515 ( 2011 ) [ erratum - ibid .
c * 72 * , 1874 ( 2012 ) ] [ arxiv:1010.4180 [ hep - ph ] ] .
j. beringer _
[ particle data group collaboration ] , phys .
d * 86 * , 010001 ( 2012 ) .
a. freitas and y. -c .
huang , jhep * 1208 * , 050 ( 2012 ) [ erratum - ibid .
* 1305 * , 074 ( 2013 ) ] [ erratum - ibid .
* 1310 * , 044 ( 2013 ) ] [ arxiv:1205.0299 [ hep - ph ] ] .
d. choudhury , a. kundu and p. saha , arxiv:1305.7199 [ hep - ph ] .
mohapatra , `` unification and supersymmetry : the frontiers of quark - lepton physics , '' springer - verlag , ny 1992 ; m. drees , r. godbole and p. roy , `` theory and phenomenology of sparticles : an account of four - dimensional n=1 supersymmetry in high energy physics , '' world scientific ( 2004 ) ; h. baer and x. tata , `` weak scale supersymmetry : from superfields to scattering events , '' cambridge , uk : univ .
( 2006 ) .
y. okada , m. yamaguchi and t. yanagida , prog .
phys . * 85 * , 1 ( 1991 ) ; j. r. ellis , g. ridolfi and f. zwirner , phys .
b * 257 * , 83 ( 1991 ) ; h. e. haber and r. hempfling , phys .
lett . * 66 * , 1815 ( 1991 ) ; j. r. ellis , g. ridolfi and f. zwirner , phys .
b * 262 * , 477 ( 1991 ) . c. strege , g. bertone , f. feroz , m. fornasa , r. ruiz de austri and r. trotta , jcap * 1304 * , 013 ( 2013 ) [ arxiv:1212.2636 [ hep - ph ] ] .
a. arbey , m. battaglia , a. djouadi , f. mahmoudi and j. quevillon , phys .
b * 708 * , 162 ( 2012 ) [ arxiv:1112.3028 [ hep - ph ] ] ; m. carena , s. gori , n. r. shah and c. e. m. wagner , jhep * 1203 * , 014 ( 2012 ) [ arxiv:1112.3336 [ hep - ph ] ] ; m. kadastik , k. kannike , a. racioppi and m. raidal , jhep * 1205 * , 061 ( 2012 ) [ arxiv:1112.3647 [ hep - ph ] ] ; j. ellis and k. a. olive , eur .
j. c * 72 * , 2005 ( 2012 ) [ arxiv:1202.3262 [ hep - ph ] ] ; s. akula , b. altunkaynak , d. feldman , p. nath and g. peim , phys .
d * 85 * , 075001 ( 2012 ) [ arxiv:1112.3645 [ hep - ph ] ] ; d. ghosh , m. guchait , s. raychaudhuri and d. sengupta , phys .
d * 86 * , 055007 ( 2012 ) [ arxiv:1205.2283 [ hep - ph ] ] ; a. fowlie , m. kazana , k. kowalska , s. munir , l. roszkowski , e. m. sessolo , s. trojanowski and y. -l .
s. tsai , phys .
d * 86 * , 075010 ( 2012 ) [ arxiv:1206.0264 [ hep - ph ] ] ; j. cao , z. heng , j. m. yang and j. zhu , jhep * 1210 * , 079 ( 2012 ) [ arxiv:1207.3698 [ hep - ph ] ] ; o. buchmueller , r. cavanaugh , m. citron , a. de roeck , m. j. dolan , j. r. ellis , h. flacher and s. heinemeyer _ et al .
_ , eur .
j. c * 72 * , 2243 ( 2012 ) [ arxiv:1207.7315 [ hep - ph ] ] ; u. haisch and f. mahmoudi , jhep * 1301 * , 061 ( 2013 ) [ arxiv:1210.7806 [ hep - ph ] ] ; a. arbey , m. battaglia , f. mahmoudi and d. martinez santos , phys .
d * 87 * , 035026 ( 2013 ) [ arxiv:1212.4887 [ hep - ph ] ] ; a. dighe , d. ghosh , k. m. patel and s. raychaudhuri , int .
j. mod .
a * 28 * , 1350134 ( 2013 ) [ arxiv:1303.0721 [ hep - ph ] ] .
s. heinemeyer , w. hollik , a. m. weber and g. weiglein , jhep * 0804 * , 039 ( 2008 ) [ arxiv:0710.2972 [ hep - ph ] ] .
p. moxhay and k. yamamoto , nucl .
b * 256 * , 130 ( 1985 ) ; b. gato , nucl .
b * 278 * , 189 ( 1986 ) ; n. polonsky and a. pomarol , phys .
d * 51 * , 6532 ( 1995 ) .
v. berezinsky , a. bottino , j. r. ellis , n. fornengo , g. mignola and s. scopel , astropart .
* 5 * , 1 ( 1996 ) [ hep - ph/9508249 ] ; m. drees , m. m. nojiri , d. p. roy and y. yamada , phys .
d * 56 * , 276 ( 1997 ) [ erratum - ibid .
d * 64 * , 039901 ( 2001 ) ] [ hep - ph/9701219 ] ; p.
nath and r. l. arnowitt , phys .
d * 56 * , 2820 ( 1997 ) [ hep - ph/9701301 ] .
m. w. cahill - rowley , j. l. hewett , s. hoeche , a. ismail and t. g. rizzo , eur .
j. c * 72 * , 2156 ( 2012 ) [ arxiv:1206.4321 [ hep - ph ] ] .
m. w. cahill - rowley , j. l. hewett , a. ismail , m. e. peskin and t. g. rizzo , arxiv:1305.2419 [ hep - ph ] .
a. djouadi , j. -l .
kneur and g. moultaka , comput .
commun .
* 176 * ( 2007 ) 426 [ hep - ph/0211331 ] .
g. belanger , f. boudjema , a. pukhov and a. semenov , comput .
commun . * 174 * 577 , 2006 [ hep - ph/0405253 ] ; g. belanger , f. boudjema , p. brun , a. pukhov , s. rosier - lees , p. salati and a. semenov , comput .
commun . * 182 * 842 , 2011 [ arxiv:1004.1092 [ hep - ph ] ] .
r. l. arnowitt , b. dutta , t. kamon and m. tanaka , phys .
b * 538 * , 121 ( 2002 ) [ hep - ph/0203069 ] ; c. beskidt , w. de boer , d. i. kazakov , f. ratnikov , e. ziebarth and v. zhukov , phys .
b * 705 * , 493 ( 2011 ) [ arxiv:1109.6775 [ hep - ex ] ] .
a. g. akeroyd and s. recksiegel , j. phys .
g * 29 * , 2311 ( 2003 ) [ hep - ph/0306037 ] .
j. p. lees _ et al .
_ [ babar collaboration ] , phys .
lett . * 109 * , 101802 ( 2012 ) [ arxiv:1205.5442 [ hep - ex ] ] .
a. arbey , m. battaglia and f. mahmoudi , phys .
d * 88 * , 095001 ( 2013 ) [ arxiv:1308.2153 [ hep - ph ] ] .
|
we study the impact of the measurements of three sets of observables on the parameter space of the constrained minimal supersymmetric standard model ( cmssm ) , its slightly general variant , the non - universal scalar model ( nusm ) , and some selected benchmark points of the 19-parameter phenomenological mssm ( pmssm ) : ( @xmath0 ) the direct measurement of the higgs boson mass @xmath1 gev at the cern large hadron collider ( lhc ) ; ( @xmath2 ) @xmath3 boson decay width in the @xmath4 channel normalized to its hadronic width ( @xmath5 ) , and the forward - backward asymmetry on the @xmath3-peak in the same channel @xmath6 ; and ( @xmath7 ) several @xmath8-physics observables , along with @xmath9 of muon .
in addition , there are constraints from non - observation of superparticles from direct searches at the lhc . in view of the recently re - estimated standard model ( sm ) value of @xmath5 with
improved higher order corrections , the measured value of @xmath5 has a 1.2@xmath10 discrepancy with its sm value , while the corresponding discrepancy in @xmath11 is 2.5@xmath10 .
mssm contributions from light superpartners improve the agreement of @xmath5 but worsen that of @xmath11 .
we project this ( @xmath5-@xmath12 ) tension _ vis - - vis _ the constraints arising from other observables in the parameter space of cmssm and nusm .
we also consider a few well - motivated pmssm benchmark points and show that pmssm does not fare any better than the sm .
sinp / tnp/2013/07 * minimal supersymmetry confronts @xmath13 , @xmath14 and @xmath15 * + gautam bhattacharyya@xmath16 , anirban kundu@xmath17 and tirtha sankar ray@xmath18 +
|
an efficient technique of measurement of attenuation of transparent optical crystals involves fabrication of monolithic resonators out of the materials and measurement of their intrinsic quality ( @xmath0 ) factors .
the attenuation of light in the resonator mode , @xmath1 , is related to the quality factor of a circular monolithic resonator as @xmath2 , where @xmath3 is the wavelength of measurement , and @xmath4 is the refractive index of the material at this wavelength .
the value of coefficient @xmath1 is determined by both absorption and scattering in the bulk material and on the surface of the resonator .
hence , @xmath1 gives a reasonable approximation of the bulk absorption if other loss mechanisms are made negligible . whispering gallery mode ( wgm )
resonators are solid state cavities that confine light in small geometrical volumes for long periods of time @xcite .
these structures are promising for many applications in nonlinear optics , rf photonics , and optical technology @xcite ranging from cavity qed @xcite to kerr optical frequency comb generation @xcite .
finesse of wgm resonators can exceed the finesse of the best fabry - perot resonators and the linewidth of the wgms can be as small as a few khz at room temperature @xcite .
various properties of the resonators are required depending on particular applications .
for example , planar on - chip integration of a microresonator calls for resonators with high enough index of refraction .
the refractive index should exceed the refractive index of fused silica substrate and preferably approach refractive index of silicon nitride or silicon .
low index materials , like magnesium and calcium fluoride , are not suitable for this purpose .
they hinders integration of kerr frequency comb generators @xcite . in this work
we demonstrate , for the first time , generation of the kerr frequency combs in sapphire resonators .
those resonators have excellent mechanical properties and represent a great potential for heterogeneous integration with photonic chips .
ultra - stable cavities are required for laser stabilization @xcite .
the ultimate stability of a monolithic microresonator is limited by fundamental thermodynamic fluctuations existing even if the resonator is kept at constant temperature @xcite .
the thermal conductivity of the resonator host material is essential to reduce those fluctuations . in addition , resonators with high thermal conductivity are less sensitive to various kinds of thermal effects resulting from slow heat spread from the mode volume to the substrate @xcite . in this work
we report on fabrication of a diamond resonator that has extremely high thermal conductivity @xcite .
tunability of wgm resonators is important in many applications .
for instance , structures containing multiple interacting resonators require coincidence of the wgm frequencies belonging to different resonators @xcite .
the fabrication of completely identical very high - q resonators is not practical , so the postproduction trimming as well as fast real time tuning of the resonant frequencies is required .
some crystalline wgm resonators possessing electro - optic effect can be tuned electrically @xcite . however , the quality ( q ) factors of those resonators are usually not very high and do not exceed one billion .
ultra - high q wgm resonators are fabricated from amorphous or centro - symmetric crystals possessing some cubic nonlinearity and zero quadratic nonlinearity @xcite .
such resonators can be tuned by either strain or heat , but can not be efficiently electrically tuned .
we demonstrate that crystalline quartz wgm resonators can bridge this gap .
quatrz resonators can simultaneously posses very high q and be tuned electrically .
we describe experimental demonstration of such resonators and discuss their properties and potential applications @xcite . in what follows
we describe our results on studies of sapphire , diamond , and quartz whispering gallery mode resonators .
we have fabricated sapphire whispering gallery mode resonators ( fig . [ fig1sapphire ] ) out of sapphire preforms and shown that pumping the resonators with continuous wave light results in generation of kerr frequency combs ( fig .
[ fig2sapphire ] ) .
the resonators have coupled q - factor exceeding @xmath5 at 1550 nm wavelength .
strong stimulated rayleigh scattering results in splitting the wgm spectra into doublets , shown at left panel of ( fig .
[ fig1sapphire ] ) .
since sapphire has high enough index of refraction ( 1.74705 ordinary and 1.73924 extraordinary at 1500 nm ) and is easily machinable , it is well suited for planar integration .
mid - ir transparency of the material makes it promising for generation of frequency combs in this spectral region . a picture of the sapphire resonator on a manufacture pin inserted into the test setup ( left panel ) , and example of the optical spectrum of the resonator ( right panel ) .
a distinct 420 khz doublet due to rayleigh scattering is clearly seen .
the wgms have bandwidths of 160 khz for the polarization perpendicular to the resonator axis and 260 khz for the polarization parallel to the resonator axis.,width=548 ] examples of optical spectra of kerr frequency combs generated in the sapphire resonator.,width=548 ]
while it is well known that diamond has an extremely broad optical transparency window , there is little consistent experimental data on its attenuation within the window .
it is relatively easy to measure transparency of fused silica because of kilometers of optical fibers fabricated from the material , however this approach is not applicable to the diamond by obvious reasons .
methods of fabrication of high quality diamond crystals @xcite has resulted in the material being used in a range of novel applications from optic to quantum information @xcite .
the material is attractive because of its exceptional physical and chemical properties , including high thermal conductivity , mechanical hardness , and a wide optical transparency window from vacuum ultraviolet through the infrared .
the best q - factor of a monolithic diamond resonator measured at 1,550 nm , up to our knowledge , is 2.5@xmath6 @xcite , which corresponds to linear attenuation coefficient @xmath7 @xmath8 .
the measurements were performed with a single crystal diamond race - track resonator built on top of a silicon dioxide / silicon substrate .
observations have included the splitting of relatively low q resonances due to stimulated rayleigh scattering , hence we can conclude that the experimentally measured q values are not limited by the bulk loss of the material .
the scattering usually results from roughness of the surface of open microresonators .
mechanical polishing allows reaching outstanding surface quality that already lead to demonstration of the highest finesse monolithic resonators ( @xmath9 , @xcite ) .
we used this technique to fabricate a diamond resonator @xcite .
the measured quality factor of the resonator was @xmath10 , which is two orders of magnitude larger compared with the best previous measurement .
we found that in our case the measured q - factor is limited by the material loss approaching @xmath11 @xmath8 and not by surface scattering .
the single - crystal diamond samples used in this experiment were synthesized with a microwave plasma - assisted chemical vapor deposition reactor operating at a frequency of 896 mhz .
the sample was produced from homo - epitaxial layer grown on a @xmath12-orientated diamond surface that had been prepared using high quality polishing techniques to minimize ra ( @xmath1320 nm ) and therefore reduce the nucleation of dislocations in the epitaxial layer .
the diamond that had approximately 20 ppb of nitrogen impurities in the solid , which results in extremely low levels of absorption across the whole transparency spectrum @xcite . this kind of a diamond was utilize to build the wgm resonator .
diamond parts for quantum computing applications have been produced in thin layers with nitrogen impurities below 1 ppb @xcite , and it can be foreseen that using the techniques exploited in the synthesis of these layers that in the future even lower absorption bulk diamond can be produced .
the fabricated resonator is shown in fig .
( [ fig1diamond ] ) .
it is shaped as a sphere with the diameter of @xmath14 mm , truncated about 130 micron below large diameter ( `` equatorial plane '' ) and suspended on a metallic rod during fabrication and measurement .
refractive index of diamond is @xmath15 at 1,550 nm .
pictures of the fabricated diamond whispering gallery mode resonator : ( a ) side view , ( b ) top view .
a silicon prism is used to couple light in and out of the resonator .
, width=548 ] measurement of the q - factor was done by means of the prism coupling method .
we used a silicon prism ( refractive index @xmath16 ) to couple light in and out of the resonator modes .
the ultimate coupling efficiency achieved in the experiment exceeds 70% .
a tunable fiber laser was utilized as a pumping light source .
the linewidth of the emitted light ( @xmath17 khz ) was small enough to perform spectroscopy of the resonator modes .
a typical measured spectrum is shown in fig .
[ fig2diamond ] .
the intrinsic q - factor was inferred from the measurement of the bandwidth of the observed resonances in the undercoupled regime .
the modes are characterized with unloaded full width at the half maximum of @xmath18 mhz .
a general criterion to distinguish surface and bulk attenuation in the open optical resonators is the equivalence of the unloaded quality factors of all the resonator modes independently on their polarization .
the modes polarized along the symmetry axis have slightly worse q - factor ( less than 10% ) , which indicates that the surface of the resonator is either slightly damaged or contaminated .
visual inspection has indicated presence of the minor residual scratches that can explain the slight difference in the q - factors because of different impact of surface scattering on q of wgms with different polarization .
nearly the same q - factor was measured at 1319 nm wavelength using diode pumped nd : yag solid state laser .
the laser light had 100 khz linewidth .
it was coupled to the resonator using the same coupling prism .
this measurement shows that the attenuation is rather broadband .
the spectrum of the resonator has distinct azimuthal and latitudinal spectral features , which indicates that the resonator has a slightly spheroidal shape .
the azimuthal free spectral range is 20 ghz .
the latitudinal spectral range depends on the polarization of the modes .
it is @xmath19 mhz for the modes polarized collinearly with the symmetry axis of the resonator , and @xmath20 mhz for the modes polarized perpendicularly to the axis .
we were able to optimize the coupling with the higher order modes by changing the launch angle for the beam of the pumping light impinging on the resonator surface .
a spheroidal resonator has two nearly equidistant mode sequences characterized with frequency intervals @xmath21 and @xmath22 @xmath23 and @xmath24 are the semi - axes of the spheroid @xcite . using result of our measurement
we conclude that non - sphericity of the resonator is approximately 1% ( @xmath25 ) .
spectrum of the resonator modes polarized perpendicular to the resonator symmetry axis .
two free spectral ranges ( fsrs)are observed .
fsr@xmath26 corresponds to azimuthal quantization of the modes , fsr@xmath27 corresponds to altitudinal quantization of the modes.,width=548 ] thermal nonlinearity is an important factor that limits the optical power range in applications of high - q resonators @xcite .
the easily observed manifestation of thermal nonlinearity is the visible extension ( or compression ) of wgm resonances observed during sweeping the probe laser : because of the thermal nonlinearity , the trace of the resonance on the screen of oscilloscope changes depending on the laser power and the speed and direction of the laser frequency scan .
this effect is produced by heating the mode volume with the light power absorbed in the material due to nonzero optical absorption .
the process can be described with two time constants , one of which is responsible for flow of heat from the mode volume to the rest of the resonator , and the other for heat exchange between the resonator and external environment .
the laser scan should be fast compared with the relaxation constants and the light power must be small to reduce the effect . in the simplest approximation
the evolution of the system can be described with a set of two equations @xcite @xmath28e = f(t ) , \\
\nonumber & & \dot \delta + \gamma \delta = \gamma \gamma \xi |e|^2,\end{aligned}\ ] ] where @xmath29 is the complex amplitude of the field in the resonator mode normalized such that @xmath30 is the total energy accumulated in the mode , @xmath31 is the optical mode half width at the half maximum , @xmath32 is the frequency of the optical pump , @xmath33 is the unperturbed resonance eigenfrequency , @xmath34 is the thermal frequency shift , @xmath35 , @xmath36 is the temperature deviation , @xmath37 stands for the external optical pump , @xmath38 characterizes thermal relaxation rate , @xmath39 is thermal diffusivity , @xmath40 is thermal conductivity , @xmath41 is material density , @xmath42 is a specific heat capacity of diamond , @xmath43 is the half - thickness of the mode in the direction of the largest field gradient , which is the radial direction for the spherical resonator ; and @xmath44 is the thermal nonlinearity coefficient , where @xmath45 is the quality factor determined by the material absorption .
the thermal nonlinearity becomes important when the power accumulated in the resonator exceeds @xmath46 .
it is easy to see from eq .
( [ xi ] ) that thermal nonlinearity of a diamond resonator is very small compared with the thermal nonlinearity of other optical materials , since the thermal conductivity in diamond is significantly larger compared with thermal conductivity of any other optical crystal .
to confirm the conclusion we performed a comparative measurement of thermal nonlinearity of two identical resonators .
one of the resonators was the diamond one , and the other was made with mgf@xmath27 .
the mgf@xmath27 resonator was fabricated so that its @xmath45 was comparable with the q of the diamond resonator . in this case @xmath47 .
it means that thermal nonlinearity becomes visible in magnesium fluoride for pump power hundred times smaller compared with the identical diamond resonator .
the thermal nonlinearity can be significantly suppressed if the resonators of high transparency material are overloaded in such a way that the loaded q becomes much smaller than the absorption - related @xmath45 ( @xmath48 ) .
if one is able to maintain extremely low levels of absorption in a resonator with transparent material with @xmath45 much larger than the one of diamond , as in the case of caf@xmath27 @xcite and mgf@xmath27 , the benefit of diamond as extreme thermal conductivity material could become irrelevant .
this , however , is a challenging engineering task , since not only the resonator material must have ultra - low loss , but absorptive contaminations must also be avoided during integration of the resonator .
the diamond resonators have an important advantage compared with resonators of other materials if the goal is to create an ultra - stable cavity .
the ultimate stability of monolithic resonators is limited by fundamental thermodynamic fluctuations existing even if the resonator is kept at constant temperature .
for instance , thermodynamic fluctuation of temperature in the mode volume results in fluctuations of the index of refraction in the mode channel that can be written as @xcite @xmath49 the mean square value of the temperature fluctuation is @xmath50 where @xmath51 is the boltzmann s constant , @xmath52 is the absolute temperature .
the spectral power density of the fluctuation of the averaged temperature of the mode of a toroidal resonator ( @xmath53 ) is @xcite @xmath54^{-1}.\end{aligned}\ ] ] the spectral density is inversely proportional to the thermal conductivity of the resonator host material .
therefore , the thermorefractive noise will be significantly smaller in a diamond resonator .
the thermal fluctuation does not depend on the loss of the resonator host material .
it shows that the existing resonator already has certain advantages over other , higher - q , crystalline resonators .
by adapting the mechanical polishing technique we have developed , we fabricated quartz wgm resonators @xcite out of cylindrical crystal quartz preforms purchased from two different vendors .
the mechanical polishing was realized with a diamond slurry .
the polishing quality is high enough to ensure negligible surface scattering of light traveling in the wgms .
extremity of the resonators is shaped into toroidal geometry .
modes in the resonators were excited using a sf11 prism coupler and a 1550 nm tunable fiber laser having 4 khz linewidth .
the modes with quality factors approaching @xmath55 ( see fig .
[ fig1quartz]a ) were detected in z - cut resonators ( the symmetry axis of the resonators coincides with the z - axis of quartz ) .
( a ) a wgm resonance belonging to a z - cut quartz wgm resonator . solid line is a lorentzian fit to the experimental data .
the optical power sent to the mode is small enough ( @xmath56w ) to avoid nonlinear processes .
the linewidth corresponds to @xmath57 .
( b ) rf return for the modulator based on the x - cut quartz resonator versus rf modulation frequency ( the ratio of the powers of the demodulated signal and the input signal ) .
the input microwave as well as the optical power at the high speed photodetector is equal to 1 mw . the rf return frequency bandwidth is @xmath58 mhz at @xmath59 db .
the loaded quality factor of the optically pumped wgm is @xmath60 ( optical bandwidth is 1 mhz ) .
the optical mode contrast is 50% .
the laser is manually tuned to the mode .
the maximum rf return corresponds to @xmath61.,width=548 ] because of the crystalline properties a z - cut quartz resonator is not tunable by an electric field applied along its symmetry axis .
we placed metal electrodes on the top and bottom surfaces of a @xmath62 m thick resonator and applied a dc voltage ( as done in @xcite ) .
no measurable voltage - dependent mode shift was observed for the voltage amplitude exceeding 200 v. to observe the electric tunability of the wgm spectrum we fabricated resonators out of x - cut quartz .
the optical quality of the x - cut crystal quartz samples in our disposal was not good enough
so the maximum q - factor of those resonators was an order of magnitude lower than that of the z - cut resonators .
we have measured the value of the pockels constant of quartz ( first measured by pockels himself @xcite ) by applying a dc voltage to the x - cut resonator s top and bottom surfaces and measuring the frequency shift of the wgms @xcite .
for instance , the te mode family shifts by @xmath63 mhz when we apply @xmath64 v to the @xmath65 @xmath66 m thick disk , so that @xmath67 , where @xmath68 hz is the carrier frequency of the pump light and @xmath69 is the index of refraction of the material at 1550 nm .
the measured value is @xmath70 pm@xmath71v .
interestingly , not only te , but also tm modes experience frequency shifts when the electric field is applied .
this basically means that the modes in our resonator do not have the distinct parallel or orthogonal direction of the electric field with respect to the symmetry axis of the resonator .
electrical tunability of quartz wgm resonators enables wgm electro - optical modulators @xcite ( feasibility of quartz electro - optical modulators have been noted previously @xcite ) .
we have fabricated such a modulator using a standard technique utilized previously with lithium - niobate wgm modulators @xcite .
we mounted a strip line rf resonator on top of the resonator sitting on a polished brass electrode , and tuned the rf resonance frequency such that it became identical to the free spectral range ( fsr ) of the optical resonator .
the strip line was placed at the resonator circumference to ensure an overlap between the rf field and the light confined in the modes .
the resultant rf strip line resonator covered approximately 190@xmath72 of the wgm resonator perimeter .
the full width at the half maximum ( fwhm ) of the rf mode was 215 mhz , the rf resonant frequency was @xmath73 ghz , hence the rf quality factor was @xmath74 . to operate the modulator we sent laser radiation to a wgm and a continuous wave rf signal to the strip line resonator .
the nonzero @xmath75 of quartz ensures the wave mixing process between light and the rf signal .
the wave mixing results in generation of optical sidebands resonant with the optical modes separated from the pumped wgm by an integer number of fsrs .
the first order sidebands grow first , followed by the higher order sidebands .
we demodulated the light escaping the resonator by means of a fast photodiode and observed the rf signal using an rf spectrum analyzer ( fig .
[ fig1quartz]b ) .
it is possible to find the parameters of the modulator using this signal .
the rf return of the modulator can be expressed as @xcite @xmath76 where @xmath77 is the numerical parameter taking into account the coupling efficiency of the light into the resonator as well as the type of the modulation ( an all - resonant modulator produces mostly phase modulated light zeroing down @xmath78 ) , @xmath79 is the rf saturation power of the modulator , @xmath80 is the resistivity of the photodiode , @xmath81 is the responsivity of the photodiode , @xmath82 is the power of the light , @xmath83 and @xmath84 are the quality factors of the optical and microwave modes , @xmath75 is the electro - optical coefficient of the resonator host material , @xmath85 and @xmath86 are the indices of refraction of the material , @xmath87 is the volume of the microwave field , @xmath88 is the overlap integral of the optical and microwave fields , @xmath89 and @xmath90 are the spatial distributions of the power of the optical and microwave fields respectively , @xmath91 , and @xmath92 if the rf frequency . to find the saturation power we use @xmath60 , @xmath93 , @xmath94 ghz , @xmath95 , @xmath70 pm@xmath71v=@xmath96 esu , @xmath97 , @xmath98 , @xmath99 @xmath100 ( the volume
is given by the 100 @xmath66 wide strip line resonator having @xmath101 length and 220 @xmath66 m thickness ) .
then we estimate @xmath102 mw .
we also know from the experiment that @xmath61 , @xmath103 mw , @xmath104 ohm , and @xmath105 a@xmath71w .
using these values and the calculated value for the saturation pump power we obtain @xmath106 . because @xmath107 is small our modulator generate mostly phase modulated light .
the electro - optical modulator based on the quartz wgm resonator will become extremely important if the quality factor of the optical resonator fabricated from the x - cut material can be significantly increased .
this will cause the saturation power of such a resonator to be low , so the modulation will be very efficient .
the modulator will then also have an extremely narrow bandwidth .
such properties are important in harmonically mode locked lasers @xcite and coupled opto - electronic oscillators @xcite .
indeed , optical etalons have been inserted into the optical loop to improve the performance of harmonically actively mode - locked lasers @xcite , and electro - optical modulators are the key parts of those lasers .
the quartz wgm modulator is able to substitute both the modulator and the etalon in the mode - locked lasers .
the low saturation power of quartz modulator will eliminate the requirement of high power rf amplifiers in oscillators , while the narrow filter function will reduce the phase noise .
a laser locked to a narrow resonance high - finesse resonator is the commonly used source of stable narrow - linewidth optical signals .
extremely narrow linewidth of the wgms in quartz resonators will allow realizing lasers with high short - term stability , while electro - optical tunability of the wgm will allow tuning or modulating the frequency of the lasers .
finally , let us discuss the nonlinear optical properties of the quartz resonators .
it is known that quartz has several raman lines @xcite and possesses a large cubic nonlinearity ( @xmath108 esu , @xmath109 times larger than fused silica @xcite ) .
therefore , we should expect to observe stimulated raman scattering ( srs ) as well as four - wave mixing in the quartz resonators as previously observed in fused silica resonators . to examine this , we amplified our laser and pumped the quartz resonators with @xmath110 mw light .
the only effect observed was the srs .
no hyper - parametric oscillations based on four - wave mixing were detected .
the reason for the suppression of the effects will be the subject of further investigations .
we have presented results concerning fabrication of high - q sapphire , diamond , and crystal quartz whispering gallery mode resonators , and demonstrated the highest reported q - factor in a monolithic resonators fabricated out of the materials .
the resonators can be useful in a variety of scientific and technical applications , including fabrication of ultra - stable optical etalons , cavity - stabilized lasers , and opto - electronic radio - frequency oscillators .
the authors acknowledge support from defense sciences office of defense advanced research projects agency under contract no .
w911qx-12-c-0067 as well as support from air force office of scientific research under contract no .
fa9550 - 12-c-0068 .
t. aoki , b. dayan , e. wilcut , w. p. bowen , a. s. parkins , t. j. kippenberg , k. j. vahala , h. j. kimble , `` observation of strong coupling between one atom and a monolithic microresonator , '' nature * 443 * , 671 - 674 ( 2006 ) .
a. a. savchenkov , a. b. matsko , v. s. ilchenko , i. solomatine , d. seidel , and l. maleki , `` tunable optical frequency comb with a crystalline whispering gallery mode resonator , '' phys .
* 101 * , 093902 ( 2008 ) .
w. liang , v. s. ilchenko , a. a. savchenkov , a. b. matsko , d. seidel , and l. maleki , `` whispering - gallery - mode - resonator - based ultranarrow linewidth external - cavity semiconductor laser , '' opt .
lett . * 35 * , 28222824 ( 2010 ) . v. n. astratov , j. p. franchak , and s. p. ashili , `` optical coupling and transport phenomena in chains of spherical dielectric microresonators with size disorder , '' appl .
. lett . * 85 * , 5508 - 5510 ( 2004 ) .
a. tallaire , a. t. collins , d. charles , j. achard , r. sussmann , a. gicquel , m. e. newton , a. m. edmonds , and r. j. cruddace , `` characterization of high - quality thick single - crystal diamond grown by cvd with a low nitrogen addition , '' diamond and related materials * 15 * , 1700 - 1707 ( 2006 ) .
r. s. balmer , j. r. brandon , s. l. clewes , h. k. dhillon , j. m. dodson , i. friel , p. n. inglis , t. d. madgwick , m. l. markham , t. p. mollart , n. perkins , g. a. scarsbrook , d. j. twitchen , a. j. whitehead , j. j. wilman , and s. m. woollard , `` chemical vapour deposition synthetic diamond : materials , technology and applications , '' j. phys .
condens . mat .
* 21 * , 364221 ( 2009 ) .
g. balasubramanian , p. neumann , d. twitchen , m. markham , r. kolesov , n. mizuochi , j. isoya , j. achard , j. beck , j. tissler , v. jacques , p. hemmer , f. jelezko and j. wrachtrup , `` ultralong spin coherence time in isotopically engineered diamond , '' nature materials * 8 * , 383 - 387 ( 2009 ) .
b. j. m. hausmann , i. b. bulu , p. b. deotare , m. mccutcheon , v. venkataraman , m.l .
markham , d. j. twitchen , and m. loncar , `` integrated high quality factor optical resonators in diamond , '' nano lett .
* 13 * , 18981902 ( 2013 ) .
s. gee , f. quinlan , s. ozharar , and p. j. delfyett , `` simultaneous optical comb frequency stabilization and super - mode noise suppression of harmonically mode - locked semiconductor ring laser using an intracavity etalon , '' ieee photon .
* 17 * , 199 - 201 ( 2005 ) .
|
different applications of crystalline whispering gallery mode resonators call for different properties of the resonator host material . we report on our recent study of resonators made out of sapphire , diamond , and quartz crystals and discuss possible applications of these resonators .
in particular , we demonstrate kerr frequency comb generation in sapphire microresonators .
|
it is widely accepted that the non - thermal radiation in blazars is produced within a relativistic jet directed towards the observer .
however , the mechanism of energization of particles and their radiation is not fully understood .
recently , simple scenarios exploiting collisions of compact objects with the jet plasma reached some attention .
in fact , many different types of compact objects ( e.g. stars , clouds or even globular clusters , fragments of supernova remnants or pulsar wind nebulae ) can be immersed within the jet , forming an obstacle for the jet plasma .
it has been proposed that particles , accelerated on the shocks appearing as a result of collisions of the jet plasma with stellar winds , can produce high energy @xmath0-rays ( see e.g. @xcite ) . in this work ,
we consider another encounter scenario in which relativistic particles , already present in the fast moving blob , interact with the dense radiation of a star within the jet .
relativistic leptons , which are isotropic in the blob , comptonize stellar radiation to gev - tev energies .
the radiation field of the star is strong enough that @xmath0-rays produced in the inverse compton process can initiate the @xmath1 pair cascade in the stellar radiation . to study such a scenario we developed a dedicated monte carlo code tracking an electromagnetic cascade in the anisotropic radiation field of a star .
we calculate the @xmath0-ray spectra emerging from such process and discuss their features . as an example
, we show that the @xmath0-ray spectrum and light curve observed from a distant flat spectrum radio quasar ( fsrq ) pks1222 + 21 during its 2010 orphan @xmath0-ray flare might be explained in terms of such a scenario .
we propose that blobs of relativistic plasma , moving in jets of active galactic nuclei ( agns ) , can encounter from time to time the luminous stars which form a quasi - spherical halo around the central super - massive black hole ( see fig .
[ fig1 ] ) .
for example , in the nearby active radio galaxy cen a , the current star formation rate ( @xmath22 m@xmath3 yr@xmath4 ) should result in the production of @xmath5 m@xmath3 of young stars during the duty cycle of the cen a active nucleus .
we can thus expect @xmath23@xmath6 stars with masses above 20 m@xmath3 @xcite .
a large number of these stars should be immersed in the jet pointing towards the observer . in the case of a simple conical jet ,
the number of stars is estimated to be of the order of @xmath283@xmath7 , where we assumed that the jet opening angle is determined by the lorentz factor of the flow @xmath8 and @xmath9 . in the case of a parabolic jet ,
the number of stars depends in a more complicated way on the parameters of the jet but is expected to be also significant since the perpendicular extend of the jet on a parsec distance scale from its base are of similar order ( see estimates in sect .
[ sec3 ] ) .
the cen a nucleus is also immersed in a quasi - spherical bulge of late type stars with the number estimated as @xmath10 @xcite .
the total number of luminous red giants can be as large as @xmath210@xmath11 , if only @xmath210@xmath12 of these bulge stars are in the red giant phase .
therefore , the interaction of a relativistic blob with the radiation field of such luminous stars within a jet of an active galaxy seems to be quite likely . in section [ sec3 ]
we explain the basic properties of the blob - star interaction model . in section [ sec4 ]
we apply such scenario to the extreme flare observed from pks1222 + 21 . in section [ sec5 ] , we summarize our conclusions .
we assume that the blob , a quasi - cylindrical region within the jet containing relativistic electrons , moves along the jet with a fixed lorentz factor @xmath13 .
the blob is characterised by its radius @xmath14 and longitudinal extend , @xmath15 ( measured in the reference frame of the star ) .
the radius of the blob has to be smaller than the local perpendicular extend of the jet @xmath16 . in the case of a simple conical jet , @xmath17 cm , where @xmath18 pc is the distance from the jet base . in the case of a parabolic jet
, @xmath16 can be modelled as , @xmath19 cm . as an example
, the parameter @xmath20 is obtained from the observation of the jet in the nearby radio galaxy m87 ( see sect . 4.4 in @xcite ) , and @xmath21 cm is the schwarzschild radius of the m87 black hole .
we assume that the longitudinal extend of relativistic electrons within the blob can be described by a gaussian distribution with a standard deviation of @xmath15 .
such a blob , when passing the stellar radiation , assumed to be point like , should produce a flare on the time scale determined by @xmath15 .
note that the emission is produced by successive layers of the blob as they enter in the fixed volume around the star , rather than from the whole blob traversing the jet .
thus , the observed duration of the emission is not going to be shortened by large values of @xmath22 . .
a specific electron in the blob moves with some impact parameter , @xmath23 , in respect to the star , at the distance @xmath24 measured along the jet .
electrons produce @xmath0-rays in the inverse compton process which can be farther absorbed in these same stellar radiation .
secondary e@xmath25 pairs can interact with the stellar photons ( @xmath26 ) initiating the ic @xmath1 pair cascade.,scaledwidth=35.0% ] relativistic electrons in the blob can suffer strong energy losses on comptonization of radiation coming from a single star provided that the star is close to the direction of a moving blob .
we assume that electrons are already injected into the blob with a power law spectrum . they are isotropic in the blob reference frame and the blob moves relativistically through the jet .
the acceleration mechanism of electrons and their maximum energies are not specified in our model .
they depend on the acceleration efficiency of electrons and of their dominant energy loss process .
we assume that electrons can reach at least sub - tev energies which in the case of the acceleration process saturated by the synchrotron energy losses would require the magnetic field within the blob below several gauss ( see e.g. eq . 3 in @xcite ) .
electrons with sub - tev energies in the relativistic blob can scatter the stellar radiation already quite deep in the klein - nishina ( kn ) regime .
therefore , @xmath0-ray photons produced in the ic process can have energies comparable to energies of the primary electrons .
these @xmath0-rays can be again absorbed in this same radiation field , initiating an ic @xmath1 pair cascade in the radiation of the luminous star .
let us consider a single electron at a specific distance , @xmath24 , from the perpendicular plane containing the star and with the impact parameter @xmath23 ( see fig .
[ fig1 ] ) .
the direction of the electron is isotropised in the blob by the random magnetic field . in order to determine whether the electron can produce a @xmath0-ray photon and whether the photon can be absorbed in the stellar radiation
, we calculate their ic scattering mean free path and the @xmath27 mean free path for specific impact distances ( see fig .
[ fig2 ] ) .
-rays in this stellar radiation ( the right panel ) as the function of the energy ( measured in the reference frame of the blob with @xmath28 for the electrons and in the reference frame of the star for the @xmath0-rays ) .
the impact distance , @xmath23 , is equal to @xmath29 ( red , solid lines ) , @xmath30 ( green dotted ) and @xmath31 ( black dashed ) . , title="fig:",scaledwidth=23.5% ] -rays in this stellar radiation ( the right panel ) as the function of the energy ( measured in the reference frame of the blob with @xmath28 for the electrons and in the reference frame of the star for the @xmath0-rays ) .
the impact distance , @xmath23 , is equal to @xmath29 ( red , solid lines ) , @xmath30 ( green dotted ) and @xmath31 ( black dashed ) . ,
title="fig:",scaledwidth=23.5% ] in the left panel we show the optical depth for the inverse compton ( ic ) process of electrons on the stellar radiation as the function of their energy measured in the blob s frame of reference . in the calculations , as an example , we assume an o type star with a temperature @xmath32 k and a radius of @xmath33 cm .
the optical depth is inversely proportional to @xmath23 .
the electrons can efficiently produce @xmath0-rays up to a distance @xmath34 cm . at the highest energies the optical depth
is diminished by the kn effect .
the right panel of fig .
[ fig2 ] shows the optical depth for the absorption of @xmath0-rays in the @xmath1 pair production process . for the assumed temperature of the star
the peak of the pair production cross section falls around 100gev .
large values of the optical depth lead to a strong absorption of such @xmath0 rays passing within the region of the inner few tens of stellar radii around the star .
some photons might however undergo a few generations of an electromagnetic cascade and thus still be able to escape this region . + in fig . [ fig3 ]
we study the dependence of the observed emission on various parameters of the relativistic electrons .
for simplicity we normalise the escaping @xmath0-ray emission to 1erg of energy injected in the blob ( measured in its frame ) .
the emission is integrated over the whole range of the observation angles .
in the top left panel , we investigate how the @xmath0-ray spectrum depends on the impact parameter of the primary electrons .
for electrons passing close to the star , the @xmath0-ray absorption in the stellar radiation makes a clear imprint on the spectrum .
interestingly , even while the optical depths for tev photons passing next to the star reach values @xmath35 , there is still a measurable flux of tev photons emitted for electrons passing that close to the star surface .
it is a combined effect of the cascading process and the fact that the impact parameters of electrons are usually smaller than those of secondary @xmath0-rays . to explain the second effect ,
let us consider a @xmath0 ray produced by an electron in a blob approaching a star ( @xmath36 , see fig .
[ fig1 ] ) at the distance @xmath24 . due to the anisotropy of the electron direction
the @xmath0 ray is expected to be produced at an angle @xmath37 with respect to the blob movement direction .
thus , the impact parameter of the @xmath0 ray , @xmath38 , might be much larger than the impact parameter of its parent electron , provided that the @xmath0 ray was emitted far from the star .
moreover , if the optical depth is of the order of @xmath39 , the @xmath0-ray might still escape the production region with similar energy , undergoing a few generations of an electromagnetic cascade if the ic process occurs in kn regime ( compare the thin and thick red curve in the top left panel of fig .
[ fig3 ] ) .
the cascading effects becomes negligible for the impact factor of an electron clearly exceeding the value of @xmath40 . in the top right panel in fig .
[ fig3 ] , we investigate the dependence of the @xmath0-ray spectra on the lorentz factor of the blob . as the electron energy
is normalised to a value calculated in the blob s frame of reference , for large values of @xmath13 the emission is magnified by @xmath41 and extends to higher energies .
as expected , the spectral shape also depends strongly on the spectral index , @xmath42 , of the power law spectrum of electrons ( see the bottom left panel of fig .
[ fig3 ] ) . for
very flat electron spectra , the visibility of the absorption dip is influenced by the maximum energies of electrons ( see the bottom right panel of fig . [ fig3 ] ) .
cross section of the blob .
the blob is moving with @xmath28 .
left panel : @xmath0-ray spectra for different radii of the blob : @xmath43 cm ( red , solid ) , @xmath44 cm ( blue , dashed ) , @xmath45 cm ( black , dotted ) .
right panel : spectra emitted within the solid angle observed at different range of observation angles @xmath46 ( red solid ) , @xmath47 ( blue dashed ) , @xmath48 ( black dotted ) , @xmath49 ( green dot - dashed ) .
the radius of the blob is equal to @xmath44 cm . , title="fig:",scaledwidth=23.5% ] cross section of the blob .
the blob is moving with @xmath28 .
left panel : @xmath0-ray spectra for different radii of the blob : @xmath43 cm ( red , solid ) , @xmath44 cm ( blue , dashed ) , @xmath45 cm ( black , dotted ) .
right panel : spectra emitted within the solid angle observed at different range of observation angles @xmath46 ( red solid ) , @xmath47 ( blue dashed ) , @xmath48 ( black dotted ) , @xmath49 ( green dot - dashed ) .
the radius of the blob is equal to @xmath44 cm . , title="fig:",scaledwidth=23.5% ] as the next step , we calculate the expected emission from an extended cylindrical blob with a given radius .
we show the dependence of the obtained spectrum on the radius of the blob ( the left panel of fig .
[ fig4 ] ) .
the spectra are normalised to a constant column energy density , namely the total energy of the electrons measured in the blob s frame is 1erg per @xmath50 cross section of the blob .
the absorption dip is clearly visible for the blobs with the radii @xmath51 .
as long as the ic scattering on the stellar radiation is the dominant energy loss process for the electrons , the emission roughly scales with the radius of the blob cylinder .
this is the effect of the inverse linear dependence of the ic optical depth on the impact parameter of the electrons ( see fig .
[ fig2 ] ) .
as the blob originates far from the star , and no other energy loss processes are taken into account , the dependence of the obtained spectrum on the longitudinal extend of the blob is trivial ( scaling with the total amount of energy available in the blob ) .
note that due to beaming effect of the blob ( anisotropy of the electron directions and their energies in the reference frame of the star ) , the @xmath0-ray spectra are strongly dependent on the observation angle .
they reach higher fluxes and larger energies for the observation angle closer to the jet axis ( see the right panel of fig .
[ fig4 ] ) .
pks1222 + 21 ( 4c+21.35 ) is a flat spectrum radio quasar ( fsrq ) at the distance corresponding to the redshift @xmath52 @xcite
. the jet moves at a small angle to the observer s line of sight with the apparent superluminal motion @xmath53 @xcite .
pks1222 + 21 has been detected by _
fermi_-lat in the gev energy range @xcite and also by agile @xcite .
first two years of the _ fermi_-lat observations of pks1222 + 21 has been studied in detail in @xcite .
two large outbursts have been observed from this source in april and june 2010 .
the @xmath0-ray spectrum is well described by the power law with the differential spectral index close to @xmath54 below 1 - 3 gev with a clear steepening above this energy ( spectral index in the range @xmath22.4 - 2.8 , @xcite ) . during the second _ fermi_-lat detected flare the source was also discovered by the magic collaboration in the energy range 70 - 400 gev @xcite .
the observed magic spectrum is well fitted by the power law with a spectral index @xmath55 .
the sub - tev emission has shown a rapid variability with the doubling time of the order of @xmath210min @xcite . a rather weak synchrotron emission of pks1222 +
21 is partially obscured by a stronger emission from the accretion disk and the dusty torus ( e.g. @xcite ) .
the isotropic power emitted in @xmath0-rays is @xmath56 erg s@xmath4 ( @xcite ) .
the mass of the black hole within pks1222 + 21 is expected to be in the range @xmath57 m@xmath3 @xcite .
the above described emission features are difficult to understand in terms of any proposed models for the @xmath0-ray production ( see e.g. discussion in @xcite ) .
here we propose that the @xmath0-ray flare can be interpreted in terms of the model discussed in section [ sec3 ] , i.e. as a result of a collision of the relativistic blob in the jet with a luminous star .
in fact the low energy spectrum , in the x - rays and the part below uv , does not change significantly between low and high @xmath0-ray states ( see e.g. fig . 9 in @xcite ) .
but the gev - tev @xmath0-ray emission changes by over an order of magnitude .
such spectral behaviour resembles emission observed during the so called orphan @xmath0-rays flares reported earlier from e.g. bl lac type active galaxy 1es 1959 + 650 ( @xcite ) . as an example
, we interpret the @xmath0-ray emission from the flare observed by the _
fermi_-lat and magic as a result of an encounter of a luminous o type star by a relativistic blob in the jet .
the time scale of the flare observed by the magic telescopes puts a strong constraint on the longitudinal shape and extend of the blob .
we assume in the calculations a gaussian distribution of the electrons with a standard deviation ( measured in the reference frame of the star ) of @xmath58 .
such a shape results in a @xmath0-ray light curve consistent with the one observed by the magic telescopes ( see the top panel of fig . [ fig5 ] ) .
-ray emission observed during the flare from the fsrq pks1222 + 21 in june 2010 by _ fermi_-lat ( empty squares ) and magic ( full circles ) , @xcite .
the absorption in the extragalactic background light is taken into account according to @xcite model .
the top panel shows the light curve above 100 gev .
the dashed vertical lines are the time range from which the spectral energy distribution ( the bottom panel ) is computed .
the parameters of the blob , star and electron distribution are given in the inset tab .
, title="fig:",scaledwidth=43.0% ] -ray emission observed during the flare from the fsrq pks1222 + 21 in june 2010 by _ fermi_-lat ( empty squares ) and magic ( full circles ) , @xcite .
the absorption in the extragalactic background light is taken into account according to @xcite model .
the top panel shows the light curve above 100 gev .
the dashed vertical lines are the time range from which the spectral energy distribution ( the bottom panel ) is computed .
the parameters of the blob , star and electron distribution are given in the inset tab .
, title="fig:",scaledwidth=43.0% ] we calculate the corresponding @xmath0-ray spectrum constructed from the photons arriving at the observer within the observation time of magic and plot it in the bottom panel of fig .
although it is possible to explain the very fast variability of the emission even with a moderate lorentz factor of the blob , strong constraints are put by the level of the observed flux . if the emission region has a radius of @xmath59 cm , and is moving with a lorentz factor of @xmath60 , we require an energy density of @xmath61 ( measured in the blob s frame of reference ) to reproduce the flux observed by magic and _ fermi_-lat ( see the bottom panel of fig .
[ fig5 ] ) .
note that such large values of the lorentz factor of the emission region in the jet have been already postulated in terms of other models in order to explain extremely short flares observed in this source ( @xcite ) or in the other sources , e.g. pks2155 - 304 ( @xcite ) . such large lorentz factors of the blob find also some observational support from the observations of the superluminal motion in pks1510 - 089 which represents similar type of blazar ( @xcite ) .
we can estimate the power of the blob in the observer s reference frame on , @xmath62 erg s@xmath4 . on the other hand ,
the eddington luminosity of the black hole in pks1222 + 21 , with the mass @xmath57 m@xmath3 , is @xmath63 erg s@xmath4 .
therefore , the blob has to contain about @xmath22@xmath64 of the eddington power .
this is quite demanding but seems not to be excluded , especially if @xmath65 .
lower values of @xmath13 require a much higher energy density in the blob ( e.g. @xmath66 for @xmath67 and @xmath59 cm ) .
the strong dependence on the @xmath13 is a combined result of the transformation of the energy density to the reference frame of the observer and the beaming of the emission in a narrower cone .
note that a larger radius of the blob will lower the energy density constrain , e.g. for @xmath68 cm we obtain @xmath69 for @xmath60 and @xmath70 for @xmath67 , at the assumption that there is no competing energy loss process of the electrons at such a large distance from the star .
we propose that blobs of relativistic plasma , moving in jets of agns , can encounter from time to time luminous stars which form a quasi - spherical halo around the central super - massive black hole .
the transit time of a star through a conical jet can be estimated as @xmath71 yr , where the velocity of the star is @xmath72 cm s@xmath4 , @xmath73 m@xmath3 is the black hole mass in units of the solar mass , and @xmath74 is the gravitational constant . at parsec scale distance from the jet base
the jet transit time is similar for a conical and a parabolic , m87-like , jet .
therefore , we do not expect significant effects of the jet shape on our final results .
the star passing the inner jet at parsec distance from the smbh can be responsible for a sequence of outbursts as different blobs reach its position .
such high activity period can last from months to years .
in fact , a single blob may be also responsible for the observed sequence of flares ( e.g. as observed in mrk 421 , @xcite or pks2155 - 304 , @xcite ) , if it meets on its path a few stars within a jet .
luminous stars are also characterised by stellar winds which pressure can balance the pressure of the blob plasma . as a result a shock structure appears around the star .
the shock radius around the star in a conical jet can be determined from ( see e.g. @xcite ) , @xmath75 cm , where @xmath76 and @xmath77 km s@xmath4 are the mass loss rate and the velocity of the stellar wind , and @xmath78 erg s@xmath4 is the power of the blob in the observer s frame . since the surface area of this shock , @xmath79 , is typically much smaller than the surface area of the considered blob , the presence of the shock should not essentially influence presented above calculations of the @xmath0-ray emission from the blob .
our model predicts the appearance of flares which emission is limited mostly to the @xmath0-ray energy range .
therefore , it can provide the mechanism for the appearance of the so - called orphan @xmath0-ray flares observed occasionally from relatively nearby agns of the bl lac type ( e.g. 1es 1959 + 650 , @xcite or mrk 421 , @xcite ) but also from the fsrqs ( e.g. pks1222 + 21 , @xcite ) . in order to show applicability of this scenario we model the case of the emission from the very strong @xmath0-ray flare observed from the distant fsrq pks1222 + 21 in 2010 .
our model is able to reproduce the observed light curve and spectral energy distribution measured by the _
fermi_-lat and magic , although the energy requirements are quite demanding .
note that in the case of blazars with orphan flares which are about an order of magnitude closer to the observer ( i.e. 1es 1959 + 650 and mrk 421 ) , this energy requirements should be about two orders of magnitude lower or the lorentz factors of the blob might be closer to values typically reported from radio observations .
we thank the referee for useful comments . this work is supported by the grant through the polish narodowe centrum nauki no .
2014/15/b / st9/04043 .
99 ackermann , m. et al .
2014 , apj , 786 , 157 aharonian , f. , et al .
2007 , apjl , 664 , l71 aleksi , j. et al .
2011 , apj , 730 , l8 araudo , a.t . , bosch - ramon , v. , romero , g.e .
2013 , mnras , 436 , 3626 asada , k. , nakamura , m. 2012 apj 745 , l28 barkov , m.v . ,
aharonian , f.a . , bosch - ramon , v. 2010 , apj , 724 , 1517 bednarek , w. , protheore , r.j .
1997 , mnras , 287 , l9 bednarek , w. , banasiski , p. 2015 apj 807 , 168 bosch - ramon , v. , perucho , m. , barkov , m. v. 2012 , a&a , 539,69 bosch - ramon , v. 2015 a&a 575 , a109 buckley , j.h .
, et al . 1996 apj , 472 , l9 ciprini , s. 2009 , atel .
# 2349 de la cita , v.m . ,
bosch - ramon , v. , paredes - fortuny , x. , khangulyan , d. , perucho , m. 2016 , a&a , submitted ( arxiv:1604.02070 ) domnguez , a. , et al .
2011 , mnras , 410 , 2556 farina , e.p .
, decarli , r. , falomo , r. , treves , a. , raiteri c.m .
2012 , mnras , 424 , 393 fraija , n. , cabrera , j. i. , bentez , e. , & hiriart , d. 2015 , in proceedings of `` swift : 10 years of discovery '' congress ( rome , 2 - 4 december 2014 ) , pos(swift 10)152 homan , d.c .
, ojha , r. , wardle , j.f.c . ,
roberts , d.h . , aller , m.f . ,
aller , h.d . , hughes , p.a .
2001 , apj , 549 , 840 hooimeyer , j.r.a . ,
schilizzi , r.t . ,
miley , g.k . , barthel , p.d .
1992 , a&a , 261 , 5 iafrate , g. , longo , f. , dammando , f. 2010 , atel .
# 2687 jorstad , s.g . ,
marscher , a.p . , mattox , j.r . ,
wehrle , a.e . ,
bloom , s.d . ,
yurchenko , a.v .
2001 , apjs , 134 , 181 jorstad , s.g .
2005 , aj , 130 , 1418 krawczynski , h. , et al .
2004 , apj , 601 , 151 longo , f. , giroletti , m. , iafrate , g. , 2009 , atel .
2021 , 1 osterbrock , d.e . , pogge , r.w .
1987 , apj , 323 , 108 sitarek , j. , bednarek , w. 2010 , mnras , 409 , 662 tanaka , y.t .
2011 , apj , 733 , 19 wykes , s. , hardcastle , m.j . ,
karakas , a.i . , vink , j.s .
2014 , mnras , 442 , 2867
|
we propose that @xmath0-rays in blazars can be produced during encounters of relativistic blobs of plasma with radiation field produced by luminous stars within ( or close to ) the jet .
the blob is expected to contain relativistic electrons which comptonize stellar radiation to the gev - tev energies .
produced @xmath0-rays can initiate the inverse compton @xmath1 pair cascade in the stellar radiation .
we propose that such a scenario can be responsible for the appearance of the so - called orphan @xmath0-ray flares .
we show that the relativistic blob / luminous star collision model can explain the appearance of the extreme orphan @xmath0-ray flare observed in the gev and sub - tev energy range from the flat spectrum radio quasar pks1222 + 21 .
[ firstpage ] galaxies : active galaxies : individual ( pks1222 + 21 ) stars : massive radiation mechanisms : non - thermal gamma - rays : general
|
one of the primary goals of heavy - ion collision experiments at ultra - relativistic energies is to study the thermodynamic aspects of strongly interacting matter at high temperature and density . while at low density and high temperature , present experimental as well as lattice data seem to indicate a smooth cross over @xcite from hadronic to a quark gluon matter , at high density and low temperature the system is expected to have a first order transition @xcite .
so the first order phase transition at high densities and low temperature should end at a critical end point ( cep ) - a second order phase transition point - as one moves towards high temperature and low density region , in the phase diagram of strongly interacting matter @xcite .
several experimental programmes have been launched to study the phase transition of strongly interacting matter . at present , relativistic heavy - ion collider ( rhic ) at brookhaven national laboratory and the large hadron collider ( lhc ) at cern are providing us plethora of data . in near future the facility for
anti - proton and ion research ( fair ) at gsi will also provide us a large amount of information .
the low temperature and high @xmath1 region will be studied in beam energy scan ( bes ) programme of rhic and the compressed baryonic matter ( cbm ) programme at fair .
those experiments would also explore the first order line of the phase diagram along with the location of the cep .
the unveiling of the nature of phase transition requires a proper understanding of quantum chromo dynamics ( qcd ) , the theory of strong interactions .
unfortunately , the non - perturbative nature of the phenomena inhibits the use of first principle qcd for the study of strongly interacting matter at extreme conditions . in this regard lattice qcd ( lqcd ) provides the most direct approach to study qcd at high temperature @xcite .
however lqcd has its own restrictions due to the discretization of space - time .
furthermore , at finite chemical potential , lqcd faces the well known sign problem .
in contrast , effective models @xcite provide a simpler alternative for the study of the strongly interacting matter in the non - perturbative domain .
some of these models have been quite successful in describing physics of strongly interacting matter .
for example , polyakov loop extended nambu - jona - lasinio model ( pnjl ) has been used to study various aspects of physics of strongly interacting matter at high temperatures and found to reproduce the zero density lattice data quite successfully @xcite . on the other hand ,
hadron resonance gas ( hrg ) model @xcite has been very successful in describing the hadron yields in central heavy ion collisions from ags up to rhic energies @xcite . a reliable way to understand
the physics of the phase transition of strongly interacting matter is to study the correlations and fluctuations of conserved charges .
susceptibilities are related to fluctuations via the fluctuation - dissipation theorem .
a measure of the intrinsic statistical fluctuations in a system close to thermal equilibrium is provided by the corresponding susceptibilities . at finite temperature and chemical potential fluctuations of conserved charges
are sensitive indicators of the transition from hadronic matter to quark - gluon plasma ( qgp ) .
moreover , the existence of the cep can be signalled by the divergent fluctuations .
for the small net baryon number , which can be met at different experiments , the transition from hadronic to qgp phase is continuous and the fluctuations are not expected to lead to any singular behaviour .
computations on the lattice have been performed for many of these susceptibilities at zero chemical potentials @xcite .
it has been shown that at vanishing chemical potential the susceptibilities rise rapidly around the continuous crossover transition region .
since the prediction of hrg model for the system at freeze - out is in very good agreement with the experiments , it would be interesting to study the susceptibilities as well as higher moments using this model and its interacting version _
i.e. _ excluded volume hadron resonance gas ( evhrg ) model @xcite .
in fact as the higher moments are expected to be more sensitive to the phase transition , any deviation of experimental observation from the model results may be taken as an indication of new phenomena .
hrg model is based on the dashen , ma and bernstein theorem @xcite which shows that a dilute system of strongly interacting matter can be described by a gas of free resonances .
the attractive part of the hadron interactions is taken care of by these resonances . on the other hand , both the long range attraction as well as
the short range repulsion are important for the description of strongly interacting matter @xcite . moreover ,
near critical temperature hrg calculations tends towards hagedorn divergence which may be due to the absence of repulsive interaction @xcite .
this repulsive part is incorporated through the excluded volume effects in the hrg @xcite and is commonly known as evhrg model .
evhrg equation of states have also been used for the hydrodynamical models of nucleus - nucleus collision @xcite . in the present work we will be discussing our results of susceptibilities and correlations using evhrg model .
recently fluctuations of conserved charges , using hrg model , have been studied in ref .
@xcite where actual experimental acceptances in terms of pseudo - rapidity and transverse momentum are considered . in ref .
@xcite , higher moments of net - proton multiplicity have been studied using evhrg model .
our aim in the present work is twofold .
firstly we would like to have a physical understanding of the evhrg model results vis - a - vis hrg , lattice and experimental data .
secondly , we would like to study the high density sector to be explored in beam energy scan at rhic and cbm experiment at fair . here
we present the temperature ( @xmath0 ) and baryonic chemical potential ( @xmath1 ) dependence of susceptibilities of different conserved quantities such as net - baryon number , net - strangeness and net - charge up to order four using evhrg model .
baryon - strangeness and charge - strangeness correlation functions have been evaluated at different @xmath0 and @xmath1 .
different values of baryon and meson radii have been used to study their effect on susceptibilities and correlations .
we have discussed experimental observables in the framework of hrg as well as evhrg models .
the product of moments of distribution of conserved quantities are related to the ratio of different order of susceptibilities @xcite such as @xmath5 , @xmath6 and @xmath7 where @xmath8 is the mean , @xmath9 is the standard deviation , @xmath10 is the skewness , @xmath11 is the kurtosis of the distribution of conserved quantities and @xmath12 are the @xmath13 order susceptibilities . in general , higher order susceptibilities are more sensitive to the large correlation length and hence the critical point @xcite .
this implies that any memory of the large correlation length retained in the thermal system at freeze - out would be reflected in the behaviour of higher moments .
we have studied the energy dependence of product of moments for net - proton , net - kaon and net - charge and compare our result for net - proton and net - charge with experimental data of fluctuation along with the transverse momentum and pseudo - rapidity acceptance .
the paper is organized as follows .
hrg model and its extension to evhrg are introduced in sec .
[ sec : hrg ] . in sec .
[ sec : fluctuation ] we have first discussed the fluctuations of different conserved charges and correlations among them at zero @xmath1 ( finite @xmath0 ) and compared them with the lattice data . then we have discussed the finite density scenario along with a comparison with the experimental data .
finally we have summarized our results in sec .
[ sec : conclusion ] .
in this section we are going to discuss hrg and evhrg model very briefly .
the detailed discussions can be found in refs .
the grand canonical partition function of a hadron resonance gas @xcite can be written as @xmath14 where sum is over all the hadrons .
@xmath15 refers to ideal _
i.e. _ , non - interacting hrg . for particle @xmath16 , @xmath17,\ ] ] where @xmath18 is the volume of the system , @xmath19 is the degeneracy factor , @xmath0 is the temperature , @xmath20 is the single particle energy , @xmath21 is the mass and @xmath22 is the chemical potential .
@xmath23 are respectively the baryon number , strangeness and charge of the particle , @xmath24 being corresponding chemical potentials .
the @xmath25 and @xmath26 sign corresponds to fermions and bosons respectively .
partition function depends in general on five parameters
. however , only three are independent , since @xmath27 and @xmath28 can be found from conservation of different quantum numbers like baryon number , charge and strangeness @xcite . the partition function is the basic quantity from which one can calculate various thermodynamic quantities of the thermal system created in heavy ion collisions .
the partial pressure @xmath29 , the particle density @xmath30 , the energy density @xmath31 , and the entropy density @xmath32 can be calculated using the standard definitions , @xmath33 , \end{split}\end{aligned}\ ] ] @xmath34\pm1},\ ] ] @xmath35\pm1}e_i , \end{split}\end{aligned}\ ] ] @xmath36 .
\end{split}\end{aligned}\ ] ] since eqs .
[ eq : p ] - [ eq : s ] determine the thermodynamic properties of the system , those are called equations of state ( eos ) of the system .
let us now discuss the interacting hrg or evhrg model .
in contrast to the point like hadrons in hrg , in evhrg model , hadronic phase is modelled by a gas of interacting hadrons , where the geometrical size of the hadrons are explicitly incorporated as the excluded volume correction @xcite to approximate the short - range repulsive hadron - hadron interaction .
excluded volume corrections were first introduced in @xcite but it was thermodynamically inconsistent . a thermodynamically consistent excluded volume correction was first proposed in @xcite . in evhrg model pressure can be written as @xmath37 where for @xmath16th particle the chemical potential is @xmath38 where @xmath39 is the volume excluded for the @xmath16 th hadron with hard core radius @xmath40 . in an iterative procedure one can get the total pressure .
pressure @xmath41 is suppressed compared to the ideal gas pressure @xmath42 because of the smaller value of effective chemical potential .
the other thermodynamic quantities like @xmath30 , @xmath43 , @xmath44 can be calculated from eqs .
[ eq : p_new ] - [ eq : mu ] as @xmath45 @xmath46 @xmath47 this correction scheme is thermodynamically consistent _
i.e. _ eos after corrections obey the relation @xmath48 in this work we have incorporated all the hadrons listed in the particle data book @xcite up to mass of @xmath49 gev . in fig .
[ fig : eos_mub ] we have shown @xmath50 and @xmath51 as a function of @xmath1 at fixed @xmath0 . for completeness one should also plot these quantities as a function of @xmath0 at fixed @xmath1 .
however such relations are already given in ref .
@xcite and we do not repeat those here .
our results show that there is almost no effect of interaction till @xmath52 gev in eos . beyond @xmath52 gev
we see quite a substantial change in these quantities .
the change is more pronounced at higher temperatures .
one can see from this figure that at large @xmath1 the energy , entropy and number density in evhrg model are suppressed by a factor of @xmath53 , compared to hrg , if we take the radii of all the hadrons to be @xmath54 fm .
this is expected as the finite radius acts as repulsive interaction between hadrons .
if we increase the size of the baryons further to @xmath55 fm the suppression is even more .
however the thermodynamic quantities are less sensitive to the mesonic radii as can be seen from fig .
[ fig : eos_mub ] .
the plot for @xmath56 fm is almost same as that for @xmath57 fm , @xmath58 fm and is suppressed compared to @xmath59 fm .
this is an expected result as the system is dominated by baryons at high @xmath1 .
the difference of the evhrg model , as compared to hrg , is governed by the radius parameter .
the electromagnetic charge radii of hadrons have been measured by different groups @xcite .
for example , the radii for @xmath60 , @xmath61 , @xmath62 and @xmath63 are around 0.8 fm , 0.9 fm , 0.7 fm and 0.6 fm respectively .
one can also define a strong interaction radii @xcite which comes out to be around the same values . in accordance with these results ,
a value of 0.8 fm for baryons and 0.62 fm for mesons were proposed earlier @xcite . on the other hand , braun - munzinger _
et al . _
@xcite argued that a more realistic approach is to incorporate repulsive behaviour of the nn potential using hard - core radius ( @xmath64 0.3 fm ) as obtained from nucleon - nucleon scattering @xcite .
the corresponding meson radius should not exceed that of baryons .
a similar hard - core radius of 0.2 - 0.3 fm has also been proposed in ref . @xcite to explain the proton - proton scattering data .
it has also been shown earlier that to justify a hydrodynamic approach to heavy ion collisions within the hadron phase the hard - core hadron radius should be @xmath65 fm in evhrg model @xcite . in the present study we have taken an approach similar to @xcite and have used different baryon ( @xmath66 ) and meson ( @xmath67 ) radii between 0.2 - 0.3 fm .
fluctuations and correlations of conserved charges such as net electric charge , baryon number and strangeness have been considered as probes of hadronization and thermalization of the system created in nuclear collisions @xcite .
fluctuation are considered to be important signatures for the existence of cep on the phase diagram as it would lead to large fluctuations in the thermodynamic quantities .
moreover , fluctuations are expected to show distinctly different behaviour in a hadron resonance gas and a qgp . derivatives of the grand canonical partition function ( @xmath68 ) with respect to chemical potential define susceptibilities which experimentally become accessible through event - by - event analysis of fluctuations in observable quantities such as baryon number , electric charge , strangeness and others .
the @xmath13 order susceptibility is defined as @xmath69 where @xmath70 is the chemical potential for conserved charge @xmath71 . for our purpose @xmath72 ( baryon ) ,
@xmath10 ( strangeness ) and @xmath73 ( electric charge ) . in fig .
[ fig : chi^2_temp ] we have shown temperature dependence of pressure and second order susceptibilities for various conserved charges at zero chemical potentials ( @xmath74 ) . from fig .
[ press_temp ] , where we plot pressure as a function of temperature , one can see that the continuum limit lattice data from @xcite agrees within error - bar with hrg and evhrg model upto @xmath75 gev .
the second order susceptibilities are found to increase rapidly with increase of temperature . near @xmath76 gev magnitude of @xmath77
is almost double compared to @xmath78 and the magnitude of @xmath79 is almost zero at this temperature . as at low temperature fluctuations of a particular charge are dominated by lightest hadrons carrying that charge .
the dominant contribution to @xmath79 at low temperatures comes from protons ( lightest baryon ) , while @xmath78 receives leading contribution from kaons ( lightest strange hadron ) and @xmath77 from pions ( lightest charged hadron ) .
since pion is lighter compared to proton and kaon , magnitude of @xmath77 is more than that of @xmath79 and @xmath78 . in evhrg
we have considered different values of @xmath66 and @xmath67 as shown in fig .
[ fig : chi^2_temp ] .
it can be seen that there is almost no effect of interaction till @xmath80 gev in fluctuations .
the reason for this is that the effective degree of freedom does not increase much up to this temperature and therefore correction due to excluded volume is small . above @xmath80
gev we find quite a substantial change in second order susceptibilities .
it can be seen from this figure that at large @xmath0 , second order fluctuations are reduced by almost @xmath81 compared to the non - interacting hadrons , if we take radii of all the hadrons to be @xmath54 fm .
if we increase radii of hadrons further to @xmath55 fm , the suppression is even more .
we have compared our result with lqcd data @xcite .
it can be seen that up to @xmath82 gev , @xmath79 is in good agreement with lqcd if we consider radii of all hadrons to be @xmath54 fm whereas @xmath77 is in good agreement with lqcd for @xmath58 fm and @xmath57 fm .
the meson radius plays an important role for @xmath78 and @xmath77 but not for @xmath79 which can be seen from the figure .
this is an expected result since in @xmath78 and @xmath77 both the baryons and mesons contribute whereas in @xmath79 only baryons contribute .
one should , however , note that the dependence of @xmath79 on @xmath67 is not completely negligible .
we will address the issue later . in non - interacting hrg model , under boltzmann
approximation , @xmath83 and @xmath84 as only baryons with baryon number one contribute to various susceptibilities @xcite .
in contrast , in case of higher order susceptibilities , electric charge and strangeness is expected to show larger values as hadrons with multiple charge / strangeness get larger weight .
a similar behaviour is expected for evrhg model as well . in fig .
[ fig : chi^4_temp ] we have shown variation of fourth order susceptibilities with temperature at @xmath85 .
the nature of all the fourth order susceptibilities is similar to second order susceptibilities . as expected , magnitudes of fourth order susceptibilities are larger compared to that of second order susceptibilities for strangeness and electric charge .
however this is not true for baryon number fluctuations .
although contributions in @xmath86 are only from baryons , these quantities also depend on size of mesons as can be seen from fig .
[ chi_b2_temp ] and fig .
[ chi_b4_temp ] . this dependence can be understood from eqs .
( [ eq : p_new ] ) and ( [ eq : mu ] ) which shows the dependence of chemical potential on hadronic radii through pressure . hence ,
even for baryon number susceptibilities there may be small difference in magnitudes of susceptibilities at different @xmath67 for evhrg .
we compare our result with lqcd data ( @xmath87 ) @xcite .
the lqcd data for @xmath88 and @xmath89 are in good agreement with hrg model up to @xmath90 gev .
whereas for @xmath91 , lqcd data is lower compared to both hrg and evhrg .
it has been argued in @xcite that for @xmath92 , @xmath93 and @xmath43 , though lattice qcd results agree at lower temperatures , they are expected to becomes larger than those in hrg at higher temperatures due to deconfinement . on the other hand ,
the difference between lattice and hrg for the higher susceptibilities may have effects due to the difference in hadronic masses .
it has also been reported that the thermodynamic quantities calculated in lattice qcd may agree well with the hrg model if the masses of the hadrons in the model are tuned appropriately to take into account the discretization errors in the hadron spectrum present in the lattice calculations @xcite . in the present study evhrg
is found to show a better agreement with the lattice data in the continuum limit .
correlations among different conserved charges may act as probes of the structure of qcd at finite temperature . in qgp , as baryon number as well as electric charge are carried by different flavours of quarks , a strong correlation is expected between b - q , q - s as well as b - s . on the other hand ,
in the hadronic sector presence of baryons and mesons would generate an entirely different type of correlations between these quantities .
hence , these correlations are expected to show changes across the phase transition which are characteristics of the changes in the relevant degrees of freedom .
correlation functions , @xmath94 , are defined by @xmath95 where @xmath71 and @xmath96 correspond to conserved charges _
b , s , q _ and @xmath97 s are chemical potentials of corresponding conserved charges . in this section
we show temperature dependence of various correlation functions at @xmath98 .
we also compare our result with lqcd data @xcite .
[ fig : chi_bsq11_temp ] shows the variation of correlations between conserved charges with temperature around @xmath99 .
since at very low temperature main contribution comes from pions ( baryon number as well as strange quantum number @xmath100 ) , all the correlations remain zero .
the next particle to be excited is kaon and as a result @xmath101 becomes non zero around @xmath102 gev .
the leading contribution to @xmath101 in the hadronic sector is due to charge kaons which has same sign for charge and strange quantum number .
so it remains positive and increases with temperature .
other correlations pick up non - zero values above @xmath76 gev when baryons start populating the system .
@xmath103 is proportional to the product of baryon number and strange quantum number .
so here most of the contribution is due to lightest baryon @xmath104 which has baryon number @xmath105 and strange quantum number @xmath106 .
other contributing particles , such as , @xmath107 , @xmath108 etc . also have relative negative sign between baryon number and strange quantum number . as a result @xmath103 remains negative .
this negative value increases with @xmath0 due to the increase in the population of the strange baryons .
again , hrg results show sharp increase compared to those from evhrg model . due to the contribution from mesons ( mainly kaons ) in @xmath101 ,
meson radius plays a dominating role and an increase in meson radius produces larger suppression in evhrg model . on the other hand
, baryon radius plays a major role for both @xmath103 as well as @xmath109 .
the dominant contribution in @xmath109 at low temperature is due to proton and antiproton and for both of those baryon number as well as charge carry the same sign .
so the value of @xmath109 remains positive and shows a sharp increase with temperature in hrg , evhrg being suppressed due to the effect of hard core radius . in the quark sector
strange quarks ( antiquarks ) carry @xmath110 ( -@xmath110 ) baryon number and the relative sign between @xmath111 and @xmath10 is always negative . as a result lattice result for @xmath103 show a similar trend as that of hrg . on the other hand
, @xmath101 is always positive as charge as well as strangeness for strange quark is negative . though the lattice values at lower temperature ( @xmath112 gev ) is close to hrg and evhrg , @xmath103 as well @xmath101 for lattice increase faster than those for hrg as strange quark mass is much smaller compared to strange baryons . at high temperature quark
masses decrease ( which will eventually become zero at stefan - boltzmann limit ) and lattice results would start saturating whereas hrg results would keep on increasing . as a result lattice values
are suppressed compared to hrg values both for @xmath101 and @xmath103 . in the case of @xmath109 , evhrg ( @xmath67=0.2 fm and @xmath66=0.3 fm )
results agree well with the lattice data for @xmath113 gev .
but at higher temperatures , the weighted sum of charges of massless quarks vanishes so the lattice results would start saturating with the further increase in temperature whereas hrg results would keep on increasing similar like @xmath101 and @xmath103 .
the lqcd results of charge correlations has been found to deviate from the ideal gas limit even at twice the transition temperature and the large contributions of flavour fluctuations are expected to be the cause of these deviations @xcite . in order to analyse the experimental results , suitable ratios
have been proposed to eliminate these contributions @xcite .
three ratios , namely , baryon - strangeness correlation coefficient @xmath114 @xcite , electric charge - strangeness correlation coefficient @xmath115 and baryon - electric charge correlation coefficient @xmath116 can be written as @xcite @xmath117 figure [ fig : c_bs_qs_temp ] shows variation of @xmath114 , @xmath115 and @xmath116 with temperature at @xmath85 .
we compare our result with lqcd data @xcite . at low temperature @xmath114
is much lower than one because denominator contains mostly kaon whereas contribution to numerator comes from baryons , mostly @xmath118 , which is less populated than kaon .
it increases with temperature but remains below 1 due to larger contribution from mesons .
if all the hadronic radii are same , the interaction effect gets cancelled . on the other hand , for larger baryonic radii ,
baryon population gets suppressed and @xmath114 value for evhrg becomes lower than hrg . in case of simple qgp , or more generally , for a system where the quark flavours are uncorrelated , this value will be unity .
since lattice @xmath103 is higher , lattice @xmath114 values remain above the hrg ( and evhrg ) and saturates at high temperature . at very low temperature ,
contribution to numerator of @xmath115 come from charged kaons only while the contributions to denominator come from all the kaons .
so @xmath115 comes out to be around @xmath119 .
with increase of temperature lambda starts populating the system and @xmath115 becomes close to @xmath120 .
there is no effect of interaction in most of the temperature range except at very high temperatures . here
, a higher baryonic radii ( compared to meson radii ) induces a small increase in @xmath115 .
interaction effects get cancelled if radii are same for all hadrons . @xmath114 and @xmath115 are related through the equation @xmath121
so in lattice studies @xcite at low temperature , where @xmath114 is very small @xmath115 is near 1.5 , same compared to hrg ( and evhrg ) and at high temperature tends to one . at low temperatures ,
contribution to numerator of @xmath116 come from protons only while the contributions to denominator come from both protons and neutrons . as
a result @xmath116 is close to @xmath122 .
it decreases slowly with increase in @xmath0 as at high @xmath0 heavier neutral baryons like @xmath118 start contributing to the denominator whereas only charged baryons contribute to numerator . for both hrg and evhrg
@xmath116 is very close to each other due to the cancellation of interaction effects . in lattice , @xmath116 approaches zero at high @xmath0 as the system is in qgp phase .
however at lower temperature it approaches the hrg / evhrg values . at very high densities and low temperatures ,
strongly interacting matter is expected to undergo a first order phase transition which ends at cep as one moves towards higher @xmath0 and lower @xmath1 along the phase boundary . at finite @xmath1 due to the expected stronger divergence at the phase boundary and at cep
, study of higher order moments is extremely important at finite @xmath1 .
in fact , not only the magnitude but the signs of third moments have also been proposed as a signature for cep @xcite . in this section
we have discussed the results of fluctuations and correlations at large @xmath1 . in fig .
[ fig : chi_b_mub ] - [ fig : chi_q_mub ] .
we show variation of @xmath123 , @xmath124 , @xmath125 , @xmath126 with @xmath1 for a fixed temperature and at @xmath127 where @xmath128 . here
we have considered @xmath76 gev and @xmath112 gev for the purpose of illustration . for each temperatures
we show variation of the quantities with @xmath1 in hrg and evhrg model with different baryon ( @xmath66 ) and meson ( @xmath67 ) radii . the effect of interaction is found to be more pronounced at high temperatures .
it can be seen from these figures that at high @xmath0 and large @xmath1 the magnitude of fluctuations are suppressed by a factor of @xmath53 or more , compared to hrg , if we take the radii of all the hadrons to be @xmath54 fm . at high @xmath1 fluctuations
are very sensitive to @xmath66 but not to @xmath67 as in this region system is dominated by baryons .
similar behaviour was obtained in fig .
[ fig : eos_mub ] as well .
@xmath129 and @xmath130 are proportional to the odd power of strange quantum number of the particles .
dominant contribution to @xmath131 comes from @xmath118 , the lightest strange baryon which has strange quantum number @xmath106 ( at @xmath132 strange mesons do nt contribute to @xmath129 and @xmath130 because particle and anti - particle terms are equal in magnitude but opposite in sign ) . as a result ,
@xmath129 and @xmath130 remain negative ( fig .
[ chi_s1_mub ] and fig .
[ chi_s3_mub ] ) .
as mentioned earlier , there are contributions from multiple charged hadrons in electrical charge and strangeness sector only , baryon number being always one .
hence , at high @xmath1 region magnitudes of @xmath125 and @xmath126 are found to be almost double compared to that of @xmath123 and @xmath124 respectively for @xmath133 ( fig .
[ fig : chi_s_mub ] and fig .
[ fig : chi_q_mub ] ) , though the change in @xmath86 remains small as shown in fig .
[ fig : chi_b_mub ] . at
very small @xmath1 and given @xmath0 ( 0.1 or 0.15 gev ) system is dominated by mesons and @xmath86 is small . with increase in @xmath1 system
will be populated by proton , neutron as well as hyperons and @xmath86 will increase .
this increase will be sharper for hrg due to the absence of repulsive interaction . at finite @xmath1 ( fixed @xmath0 ) due to larger contribution from effective chemical potentials ,
the effect of interaction , compared to zero @xmath1 case , is more prominent on correlations as shown in fig .
[ fig : chi_bsq11_mub ] .
the plots are given for different @xmath67 and @xmath66 .
it can be seen that at low @xmath1 , @xmath103 and @xmath109 are zero whereas @xmath101 is non - zero as pions and kaons are main contributors in this range of parameters . at large @xmath1 , baryons start populating the system and both b - s and b - q correlations increase more sharply compared to @xmath134 as kaons remain the main contributor in @xmath134 and charge hyperon population is much smaller .
one can also see that there is no effect of interaction till @xmath135 gev in @xmath103 and @xmath109 .
however at high @xmath0 , @xmath101 is affected by interaction even at low @xmath1 . in fig .
[ fig : chi_c_bs_qs_mub ] we show variation of @xmath114 , @xmath115 and @xmath116 with @xmath1 at a fixed @xmath0 and @xmath136 . at low @xmath1 , kaon population is dominating in the denominator of @xmath114 and it is less than unity . on the other hand increase in @xmath1 facilitates the strange baryon population and @xmath114 increases with increasing @xmath1 .
moreover , the effect of large @xmath1 becomes more pronounced for lower @xmath0 .
so that @xmath114 at @xmath0=0.1 gev is larger than that at @xmath137 gev for large @xmath1 ( @xmath138 gev ) .
similarly at low @xmath1 , @xmath115 is around @xmath119 as it gets contribution from mainly charged kaons in the numerator and both charged and neutral kaons in denominator . with increase in @xmath1
, @xmath118 starts populating the system so that @xmath115 keeps decreasing . since this effect is more pronounced for larger @xmath0
, @xmath115 is lower for larger @xmath0 at large @xmath1 ( @xmath139 gev ) . at low @xmath1 , for small @xmath0 , @xmath116 gets contribution from protons and neutrons in the denominator and only protons in the numerator .
so it is close to @xmath122 as shown earlier in fig .
[ c_bq_temp ] . at larger @xmath0 ( @xmath140 or @xmath112 gev )
, @xmath118 starts contributing in the denominator and @xmath116 becomes less than @xmath122 .
similarly as @xmath1 increases , population of protons , neutrons and lambda increases . as a result @xmath116 becomes less than @xmath122 . at high @xmath1
, @xmath116 saturates as rate of increase becomes same for different species . for hrg and evhrg
@xmath116 is close to each other due to cancellation of interaction effect .
experimentally measured moments such as mean ( @xmath8 ) , standard deviation ( @xmath9 ) , skewness ( @xmath10 ) and kurtosis ( @xmath11 ) of conserved charges are used to characterize the shape of charge distribution .
the standard deviation corresponds to the width of the distribution , the skewness is a measure of the asymmetry of the distribution and kurtosis describes peakness of the distribution .
products of moments are related to susceptibilities ( @xmath141 ) by the following relations , @xmath142 the advantage of using the above mentioned products of moments is that they are expected to be independent of the volume of the system . on the other hand , our discussion on the variation of @xmath141 with @xmath0 ( or @xmath1 ) shows that the cancellation of the volume effect , depending on the degrees of freedom involved , will become less as one increases the order of susceptibility in the numerator ( or denominator ) . in nucleus - nucleus collision experiments ,
beam energy ( @xmath143 ) is varied to scan the phase plane . at a particular @xmath143 ,
if cep is reached , the signature of cep could survive during the time evolution of the system .
hence along the freeze - out line one would expect a non - monotonic behaviour of the quantities given in eq .
[ moment_product ] @xcite . to compare the products of moments with experiment one has to know freeze - out @xmath0 and @xmath97 at a particular @xmath143 .
freeze - out curve @xmath144 can be parametrized by @xcite @xmath145 where @xmath146 gev , @xmath147 gev@xmath148 , @xmath149 gev@xmath150 .
the energy dependence of the @xmath151 can be parametrized as @xcite @xmath152 where @xmath153 and @xmath154 are listed in table [ tablelabel_muq ] .
.parametrization of chemical potentials @xmath70 along the freeze - out curve . [ cols="^,^,^ " , ] the @xmath0 - @xmath1 parametrisation with @xmath143 is obtained by fitting the different particle ratios obtained experimentally at different @xmath143 .
so for evhrg model , in general , the parameters are expected to be different for different hadronic radii . in our study , considering all the particle ratios , we found the parameters to be same as @xcite for @xmath155 . on the other hand , for @xmath156 case , for the radii used here , the parameter are found to be within @xmath157 gev of the above fit .
moreover , phenomenological freeze - out condition of fixed energy per nucleon about @xmath120 gev , calculated in our model , matches quite well with the above mentioned freeze - out parametrization .
hence we have used the above mentioned freeze - out parametrization for comparison with the experimental results .
it may also be noted here that as @xmath143 increases from 7.7 gev to 200 gev , @xmath0 increases from around @xmath158 gev to @xmath159 gev and corresponding @xmath1 varies between @xmath160 gev to @xmath161 gev .
corresponding @xmath28 and @xmath27 are expected to vary between @xmath162 gev to @xmath163 gev and @xmath164 gev to @xmath165 gev @xcite . in terms of transverse momentum ( @xmath166 ) and rapidity ( @xmath167 ) and azimuthal angle ( @xmath168 ) @xmath169 and @xmath170
can be written as @xmath171 and @xmath172 , where @xmath173 .
one can write a similar expression in terms of @xmath166 and pseudo - rapidity @xmath174 as well .
these prescriptions have been used to set the momentum and rapidity acceptance range to compare the present results with the experimental data . in fig .
[ fig : np_roots ] we have shown energy dependence of @xmath175 , @xmath3 and @xmath4 for net - proton . we compare our result with experimental data of net - proton fluctuations for @xmath176 central au - au collisions measured at star @xcite .
experimental data is measured at mid rapidity @xmath177 and within the transverse momentum range @xmath178 gev .
same acceptances range has been used for different hadronic radii in the present manuscript . at low energy , @xmath175 is almost unity and its value increases with increase of @xmath143 as can be seen from the fig .
[ chi_2_1_np_roots ] . both hrg and evhrg model
give almost same result irrespective of the value of radii . at low energy @xmath3 for hrg
is almost unity and its value decreases with increase of @xmath143 as can be seen from the fig .
[ chi_3_2_np_roots ] . at low @xmath143 , @xmath3 for evhrg is less than that of hrg model and suppression increases with increase of radii .
however , at high @xmath143 , both hrg and evhrg model give almost same result .
experimental data of @xmath3 can be described well with evhrg model for radii of hadrons between @xmath179 fm . at low energy @xmath4 ( fig .
[ chi_4_2_np_roots ] ) in hrg model is slightly less than unity and its value reaches to unity as we move to high energy .
there is prominent suppression of @xmath4 in evhrg model at low @xmath143 .
the @xmath4 of net - proton matches within error - bar with evhrg model at @xmath180 gev and @xmath181 gev but at intermediate energies evhrg model over estimates the experimental data .
deviation of experimental data of @xmath3 and @xmath4 for net - proton at intermediate energies ( @xmath182 gev and @xmath183 gev ) may be an indication of existence of quark degrees of freedom or different physics process which is not included in hrg / evhrg model . in ref .
@xcite , star data for net - proton fluctuations is already compared with evhrg model in full space .
energy dependence of fluctuations for net - proton using lqcd can be found in ref .
though at high energies , where @xmath1 is small , lqcd agrees well with the experimental data , at low energies _
i.e. _ for higher @xmath1 it fails . in fig .
[ fig : nk_roots ] we have shown energy dependence of @xmath175 , @xmath3 and @xmath4 for net - kaon .
we have chosen transverse momentum and pseudo - rapidity within the range @xmath184 gev and @xmath185 in our calculation . at low energy @xmath175 for net - kaon
is slightly more than unity and its value increases rapidly with increase of @xmath143 and result is almost same for both hrg and evhrg model .
radii of hadrons practically have no effect on the results as shown in fig .
[ chi_2_1_nk_roots ] .
@xmath3 for net - kaon decreases with increase of @xmath143 as can be seen from fig .
[ chi_3_2_nk_roots ] .
value of @xmath3 in evhrg model is suppressed compared to hrg model and suppression increases with increase of radii of hadrons . in fig .
[ chi_4_2_nk_roots ] energy dependence of @xmath4 for net - kaon is shown . in this case ,
volume effect is more prominent , as discussed earlier .
@xmath4 for net - kaon increases with increase of @xmath143 and then saturates .
value of @xmath4 is suppressed in evhrg model compared to hrg model and suppression increases with increase of radii of hadrons . at all energies
@xmath4 for net - kaon lie between @xmath186 to @xmath187 in both hrg and evhrg . in fig .
[ fig : nc_roots_2 ] we have shown energy dependence of @xmath175 , @xmath3 and @xmath4 for net - charge .
we have chosen transverse momentum and pseudo - rapidity within the range @xmath184 gev and @xmath185 in our model calculation .
we have compared our results with experimental data of net - charge fluctuations for @xmath176 central au - au collisions measured at star @xcite .
figure [ chi_2_1_nc_roots ] shows @xmath175 , for net - charge , increases rapidly with increase of @xmath143 and compared to experimental values our results are suppressed at higher energies . @xmath3 for net - charge decreases with increase of @xmath143 as can be seen from fig .
[ chi_3_2_nc_roots ] .
experimental data of @xmath3 for net - charge matches within error - bar with hrg / evhrg model at @xmath188 gev and @xmath189 gev but at other intermediate energies hrg / evhrg model over estimate the experimental data .
figure [ chi_4_2_nc_roots ] shows @xmath4 for net - charge slowly decreases with increase of @xmath143 and then saturates . at all
@xmath143 , @xmath4 for net - charge lie between @xmath190 to @xmath191 in both hrg and evhrg model which is in agreement with experiment within error - bar at @xmath192 gev .
calculations for ratios of higher order fluctuations of electric charge using lqcd can be found in ref .
we have shown fluctuations ( @xmath193 ; @xmath194 ) of various conserved charges like net baryon number , net strangeness and net charge at finite temperatures and chemical potentials using interacting and non - interacting hadron resonance gas model and compared them with the lattice as well as experimental data .
we can draw following inferences from the present study : * in general hrg results are larger compared to evhrg , difference being higher for higher temperatures and densities .
* at high temperatures and @xmath195 , compared to hrg , second order fluctuations from evhrg fits better to lqcd continuum data for radius of hadrons between @xmath196 fm . *
the lqcd data ( @xmath87 ) for @xmath197 and @xmath198 are in good agreement with hrg / evhrg model up to @xmath199 gev .
whereas , both hrg and evhrg model overestimates lqcd data for @xmath200 . *
correlations at @xmath201 and @xmath98 shows much stronger dependence on the degrees of freedom involved .
lattice continuum data for both @xmath103 and @xmath101 are closer to hrg results ( higher than evhrg ) at lower @xmath0 but rises less sharply and becomes less than hrg for higher @xmath0 . on the other hand , @xmath109 calculated in lattice is close to evhrg results at lower @xmath0 .
initially it increases with @xmath0 but then decreases beyond @xmath202gev as discussed earlier . *
the correlation ratios from both hrg and evhrg disagree with lqcd continuum data though the trends are qualitatively similar .
this difference could be attributed to the fact that lqcd incorporates transition from hadronic to quark degrees of freedom , whereas only hadronic degrees of freedom are present in hrg or evhrg model . at high @xmath1 ,
the magnitudes of susceptibilities as well as correlations are higher correspond to @xmath85 case . * the susceptibilities as well as correlations
are found to increase with @xmath1 at fixed temperatures for both hrg and evhrg , the effect of interaction being larger for higher @xmath0 . *
since low energy collisions usually correspond to larger @xmath1 , the effect of repulsive interaction , as present in evhrg , is distinguishable for lower energies .
moreover this difference is more pronounced for higher order susceptibilities . * with increase of @xmath143 ,
chemical potential decreases sharply whereas temperature increases slowly .
therefore , @xmath203 and @xmath204 decrease with increase of @xmath143 whereas @xmath205 and @xmath206 decrease slowly and then saturate . as a result @xmath175 increases with increase of @xmath143 , @xmath3 decreases with increase of @xmath143 whereas @xmath4 remains almost constant in all energies . *
although the variations of @xmath175 and @xmath3 with @xmath143 seem to describe the experimental data well , higher moment @xmath4 shows large deviations .
it may be an indication of quark degrees of freedom .
it would be interesting to look at the other higher moments as they are supposed to have stronger dependence on correlation length and hence will be more sensitive to the critical fluctuations . in general , evhrg seems to have a better agreement with the lqcd continuum data at low @xmath0 and @xmath195 .
but the comparison with the experimental data does not provide us with a clear preference between hrg and evhrg .
it may be noted that in figs.[chi_s2_temp ] and [ chi_bs11_temp ] lqcd results are larger than all the hrg results in the crossover temperature range of @xmath207 gev .
the lqcd may become closer to the hrg by the inclusion of non - pdg additional strange beryons which , though not observed experimentally , are predicted by the quark - model and also observed in the lqcd spectrum @xcite .
here we would like to mention that the , grand canonical ( gc ) description , as used for the present study is usually considered as the relevant description for the systems in heavy ion collision .
but if the conserved quanta , under consideration , is small then one need to take care of the charge conservations and the present description will not be appropriate .
more specifically , it has been found that the results for scaled variances vary for different ensembles though the mean multiplicities from different ensembles tend to agree with each other in the infinite volume limit @xcite .
moreover , micro canonical ensemble seems to give a better agreement with the data @xcite .
these findings definitely suggest the importance of the conservation laws . we plan to study these aspects in our future work .
the work is funded by csir , ugc ( drs ) and dst of the government of india .
the authors thank star collaboration for providing the experimental data .
we would like to thank anirban lahiri , sarbani majumder , xiaofeng luo , nihar ranjan sahoo and swagato mukherjee for useful discussion .
p. n. meisinger and m. c. ogilvie , phys . lett .
* b 379 * , 163 ( 1996 ) ; nucl . phys .
* b * ( proc . suppl . ) * 47 * , 519 ( 1996 ) ; k. fukushima , phys . lett .
* b 591 * , 277 ( 2004 ) ;
e. megias , e. r. arriola and l. l. salcedo , phys . rev . * d 74 * , 065005 ( 2006 ) ; * 74 * , 114014 ( 2006 ) ; j. high energy phys . * 01 * 073 ( 2006 ) ; c. ratti , m. a. thaler and w. weise , phys . rev .
* d 73 * , 014019 ( 2006 ) ; s. k. ghosh , t. k. mukherjee , m. g. mustafa and r. ray , http://prd.aps.org/abstract/prd/v73/i11/e114007[phys . rev . * d 73 * , 114007 ( 2006 ) ] .
s. mukherjee , m. g. mustafa and r. ray , http://prd.aps.org/abstract/prd/v75/i9/e094015[phys .
d 75*,094015 ( 2007 ) ] ; s. k. ghosh , t. k. mukherjee , m. g. mustafa and r. ray , http://prd.aps.org/abstract/prd/v77/i9/e094024[phys . rev . *
d 77 * , 094024 ( 2008 ) ] ; a. bhattacharyya , p. deb , a. lahiri and r. ray , http://prd.aps.org/abstract/prd/v82/i11/e114028[phys . rev . *
d 82 * , 114028 ( 2010 ) ] ; http://prd.aps.org/abstract/prd/v83/i1/e014011[*d 83 * , 014011 ( 2011 ) ] .
s. r. amendolia _ et al .
_ , nucl phys .
* b 277 * , 168 ( 1986 ) ; phys .
* b 178 * , 435 ( 1986 ) ; g. simon , c. schmitt , f. borkowski and v. walther , nucl . phys . * a 333 * , 381 ( 1980 ) ; t. udem
_ et al . _ ,
lett . * 79 * , 2646 ( 1997 ) ; p. mergell , u. g. meissner and d. drechsel , nucl . phys . * a 596 * , 367 ( 1996 ) .
|
we present temperature ( @xmath0 ) and baryonic chemical potential ( @xmath1 ) dependence of higher order fluctuations and correlation between conserved charges in excluded volume hadron resonance gas ( evhrg ) model .
products of moments , such as ratio of variance to mean ( @xmath2 , product of skewness and standard deviation ( @xmath3 ) , product of kurtosis and variance ( @xmath4 ) , for net - proton , net - kaon and net - charge have been evaluated on the phenomenologically determined freeze - out curve .
further , products of moments for net - proton and net - charge have been compared with the experimental data measured by star experiment .
the dependence of the model result on the hadronic radius parameter has also been discussed .
|
knowledge of pressure - induced transformations for materials under extreme conditions , which requires accurate understandings of the thermophysical properties into new and complex region , has gained much interest in a large number of scientific and technological domains @xcite .
carbon dioxide ( co@xmath0 ) is one kind of the most important molecules in planetary science and chemical ( or explosive ) process .
for instance , the magnetic fields and evolution of planets are closely related to the equation of states ( eos ) and electrical properties of planetary interiors @xcite , which are composed of species containing c , h , n , o , etc . ;
chemical reactions of carbonaceous molecules ( such as co@xmath0 , co ) , nitrogen ( n@xmath0 ) , oxygen ( o@xmath0 ) and their mixture at high densities and temperatures are focused due to their presence in reacted chemical explosives @xcite . from these aspects ,
clearly , an accurate description of thermodynamical properties of co@xmath0 at high pressures and temperatures is fundamental and proves itself to be indispensabe for developing realistic models of planetary interiors and reacted explosives .
solid phase diagram of carbon dioxide up to 80 gpa and molecular - nonmolecular transition have already been probed through diamond - anvil cells and confocal micro - raman spectroscopy @xcite .
solid specimens were reported to be stable until the formation of _
a_-carbonia ( an amorphous extended phase ) , which can only be revealed around 50 gpa at elevated temperature or above 65 gpa at ambient temperature @xcite . here
, we concentrate on liquid specimens , which were first single shocked to 29 gpa and double - shocked to 56 gpa by schott @xcite , then single shocked to 71 gpa by nellis _
_ @xcite using a two - stage light - gas gun .
theoretically , intermolecular potential method was applied to study the eos of co@xmath0 under dynamic compressions , where the onset of decomposition was reported to be around 30 gpa ( 4500 k , 17 @xmath1/mol ) along the hugoniot @xcite . despite that the predicted eos were in accord with experimental data , the electronic structure , which has been proved to be important in determining the dynamical , electrical , and optical properties of molecular fluids under extreme conditions @xcite , is still lacking due to the intrinsic approximations of this method . on the other hand , quantum molecular dynamic ( qmd ) simulations , where quantum - mechanical treatments are executed by combining classical molecular dynamics for the ions and density functional theory ( dft ) for electrons ,
have already been proved to be successful in describing the thermodynamical properties of mono atomic molecules ( he ) @xcite , diatomic molecules ( h@xmath0 , n@xmath0 , o@xmath0 ) @xcite , and hydrocarbons ( c@xmath2h@xmath2 ) @xcite under extreme conditions . a fully quantum - mechanical description of co@xmath0 under shock compressions
is highly recommended to be presented and understood because of the following aspects : ( i ) co@xmath0 undergos dramatic transformations under shock compressions , such as dissociation and recombination of molecules , during which carbonaceous molecules combined with oxygen ( o@xmath0 ) might be produced , and electron spin polarization should be seriously considered at this stage due to the fact that spin - triplet state governs the ground state of oxygen @xcite ; ( ii ) electrical conductivity of molecular fluid , where electronic structure is of dominance , can be greatly influenced by thermo - activated electronic states under extreme conditions ; ( iii ) temperature , which is an important parameter in experimental determination of eos , is usually difficult to be measured because of the uncertainty in determining the optical intensity for ultraviolet part of the spectrum , and qmd simulations are powerful tools to provide predictions . in the present work ,
dft - based qmd simulations have been used to investigate thermophysical properties of co@xmath0 under extreme conditions .
the eos and pair correlation functions ( pcf ) are determined through qmd simulations at equilibrium .
dynamic conductivity @xmath3 is calculated by kubo - greenwood formula , from which the dc conductivity ( @xmath4 ) is determined .
then , the dielectric function @xmath5 and reflectivity are extracted .
the rest of this paper is organized as follows .
the simulation details are briefly described in sec .
[ sec - method ] ; the pcf , which is used to study the dissociation of co@xmath0 , and the hugoniot curve are given in sec . [ sec - eos ] ; in sec .
[ sec - dynamic ] , nonmetal - metal transition and optical properties are discussed . finally , we close our paper with a summary of our main results in sec .
[ sec - conclusion ] .
the vienna ab - initio simulation package ( vasp ) @xcite , which was developed at the technical university of vienna , has been employed to perform simulations for carbon dioxide .
the elements of our calculations consist of a series of volume - fixed supercells including @xmath6 atoms , which are repeated periodically throughout the space . by involving born - oppenheimer approximation ,
electrons are quantum mechanically treated through plane - wave , finite - temperature ( ft ) dft @xcite , where the electronic states are populated according to fermi - dirac distributions at temperature @xmath7 . the exchange - correlation functional is determined by generalized gradient approximation ( gga ) with the parametrization of perdew - wang 91 @xcite .
the ion - electron interactions are represented by a projector augmented wave ( paw ) pseudopotential @xcite .
isokinetic ensemble ( nvt ) is adopted in present simulations , where the ionic temperature @xmath8 is controlled by noe thermostat , and the system is kept in local thermodynamical equilibrium by setting the electron ( @xmath7 ) and ion ( @xmath8 ) temperatures to be equal .
electron spin polarization has also been taken into account , due to the possible existence of oxygen molecules in shocked co@xmath0 .
the plane - wave cutoff energy is selected to be 600 ev so that the pressure is converged within 5% accuracy .
@xmath9 point and 4@xmath104@xmath104 monkhorst - pack scheme * k * points are used to sample the brillouin zone in molecular dynamics simulation and electronic structure calculation , respectively , because eos ( conductivity ) can only be modified within 5% ( 15% ) for the selection of higher number of * k * points . 81 atoms ( 27 co@xmath0 molecules ) are included in the cubic supercell .
the densities selected in our simulations range from 13 to 37.55 @xmath1/mol and temperatures between 218 and 10000 k , which highlight the regime of the hugoniot .
all the dynamic simulations are lasted for 3 @xmath11 6 ps , and the time steps for the integrations of atomic motion are 0.5 @xmath11 1 fs according to different densities ( temperatures ) .
then , the subsequent 1 ps simulations are used to calculate eos as running averages .
-@xmath12 ( left panel ) and @xmath13-@xmath14 ( right panel ) hugoniot curves of shocked carbon dioxide .
inset is the ( @xmath15 , @xmath16 ) diagram . here
, filled red circles denote the present qmd results , while the dashed and dotted lines denote the intermolecular potential results @xcite with and without accounting for molecular dissociation , respectively .
the experimental data reported in ref . @xcite ( open circles ) and ref .
@xcite ( open squares ) are also shown for comparison . ]
accurate understandings of the electrical and optical properties of co@xmath0 depend on a precise description of dynamical properties , such as eos .
the eos have been examined theoretically through the rankine - hugoniot equations , which follow from conservation of mass , momentum , and energy across the front of shock waves .
the equations describe the locus of points in ( @xmath17 , @xmath14 , @xmath12)-space satisfying the relation as follows : @xmath18 @xmath19 @xmath20 where @xmath17 , @xmath14 , @xmath12 denote internal energy , pressure , volume , and subscripts 0 and 1 present the initial and shocked state , respectively . in eqs .
( [ equ_hugoniot2 ] ) and ( [ equ_hugoniot3 ] ) , @xmath15 is the velocity of the shock wave and @xmath16 corresponds to the mass velocity of the material behind the shock front . in the present work ,
the initial density of co@xmath0 is @xmath21=1.17 g/@xmath1 ( @xmath22=37.55 @xmath1/mol ) and the liquid specimen is controlled at a temperature of 218 k , where the internal energy is @[email protected] ev / molecule .
the initial pressure @xmath25 can be treated approximately as zero compared to the high pressure of shocked states along the hugoniot .
the hugoniot points are obtained as follows : ( i ) smooth functions are used to fit the internal energy and pressure in terms of temperature at sampled density ; ( ii ) then , hugoniot points are derived from eq .
( [ equ_hugoniot1 ] ) .
the results are summarized in table [ h_data ] .
.hugoniot pressure and temperature points derived from qmd simulations at a series of molar volumes .
[ cols="^,^,^",options="header " , ] [ h_data ] comparisons between our simulated hugoniot curve and results from experimental measurement and intermolecular potential method are shown in fig .
[ fig_hugoniot ] ( left panel ) , where our results show good agreement with experiment .
the qmd simulated pressure and temperature along the hugoniot curve for carbon dioxide shows a systematic behavior , except for the region with a small break in slope around 40 to 50 gpa , where structural transitions characterize the eos of molecular fluid .
although the @xmath14-@xmath12 curve by intermolecular potential method is consistent with experiments , the predicted temperature remains too high compared with our calculations [ see fig .
[ fig_hugoniot ] ( right panel ) ] . along the hugoniot .
the atomic structure , where carbon and oxygen atoms are denoted by gray and red balls respectively , and the relative iso - surface of charge density ( blue regimes ) are also provided in the insets .
( a ) @xmath12 = 19 @xmath1/mol , t = 1374 k ; ( b ) @xmath12 = 16 @xmath1/mol , t = 3873 k ; ( c ) @xmath12 = 15 @xmath1/mol , t = 4175 k ; ( d ) @xmath12 = 13 @xmath1/mol , t = 7955 k. ] dynamical properties and structural transitions of co@xmath0 under extreme conditions have been examined through pcf , which represents the possibility of finding a particle at a distance @xmath26 from a reference atom .
figure [ fig_rdf ] shows pcf and charge density distribution ( at equilibrium ) of four different molar volume points , and the structural transitions of co@xmath0 are clearly reflected by pcf . at low pressures [ fig . [ fig_rdf](a ) ] , ideal molecular fluid is suggested by pcf distribution of c - o , which exhibits a main peak around 1.16 . with the increase of pressure along the hugoniot ,
this main peak in c - o pcf reduces in amplitude and get broadened , which is attributed to the thermo - induced vibrations of c - o bonds around the equilibrium length .
dissociation of co@xmath0 , which leads to soften behavior of the hugoniot [ see fig .
[ fig_hugoniot ] ( left panel ) ] , has been found at about 40 to 50 gpa .
it is indicated that carbon atoms tend to form connections between each other [ fig . [ fig_rdf](c ) ] , then , with further increase of pressure , hybrid c - c bonds ( peak around 1.45 , which lies between typical _
_ sp__@xmath27 and _ _ sp__@xmath28 hybrid c - c bond length ) , diatomic oxygen ( peak around 1.23 ) and mono - atomic oxygen are formed in the shocked system [ fig .
[ fig_rdf](d ) ] .
furthermore , we have examined the charge density difference between spin - up and spin - down electrons , which is as small as negligible , and the results suggest that electron spin polarization of the shock produced mono - atomic and diatomic oxygen are suppressed at this stage .
great controversies have been raised since the detection of nonmetal - metal transition of such diatomic molecules as hydrogen and oxygen @xcite in isentropic compressions .
the links between nonmetal - metal transition and dissociation of molecules under dynamic compression are of particular significance . here
, we examine the dynamic conductivity of co@xmath0 according to kubo - greenwood formula : @xmath29 \cr & \times|\langle\psi_{j,\textbf{k}}|\nabla_{\alpha}|\psi_{i,\textbf{k}}\rangle|^{2}\delta(\epsilon_{j,\textbf{k}}-\epsilon_{i,\textbf{k}}-\hbar\omega),\end{aligned}\ ] ] where @xmath30 is the volume of the supercell .
the @xmath31 and @xmath32 summations range over @xmath6 discrete bands included in the calculation .
the @xmath33 sum is over the three spatial directions .
@xmath34 describes the occupation of the @xmath31th band , with the corresponding energy @xmath35 and the wavefunction @xmath36 at @xmath37 .
@xmath38 is the @xmath37-point weighting factor . along the hugoniot ,
dynamic conductivity have been calculated as the average of thirty atomic snapshots at equilibrium .
the dc conductivity ( @xmath4 ) , which follows the static limit ( @xmath39 ) of @xmath40 , is then extracted and plotted in fig .
[ fig_conductivity ] . at pressures below 20 gpa
, @xmath4 can be neglected , and the results suggest an insulating molecular fluid .
thermal activation of conductivity is raised from 20 gpa , then metallization ( @xmath41 1000 @xmath42@xmath43 ) of co@xmath0 , which is accompanied with dissociation of molecules , is reached at 40 to 50 gpa .
emissivity of spectrum , which is closely related to optical reflectivity , is important in determining temperature in experiment .
optical reflectivity can be derived from dynamic conductivity .
the imaginary part @xmath44 can be obtained from the kramers - kronig relation : @xmath45 where @xmath14 is the principal value of the integral . then dielectric functions can be derived from the two parts of the conductivity : @xmath46 @xmath47 the real part @xmath48 and the imaginary part @xmath49 of the refraction index have the following relations with dielectric function : @xmath50},\end{aligned}\ ] ] @xmath51}.\end{aligned}\ ] ] the index of refraction is useful for evaluating optical properties such as the reflectivity @xmath52 and absorption coefficient @xmath53 : @xmath54^{2}+k(\omega)^{2}}{[1+n(\omega)]^{2}+k(\omega)^{2}}.\end{aligned}\ ] ] we have examined the optical reflectance at three wavelengths ( 404 , 808 , and 1064 nm ) spanning the visible spectrum , and the results are shown along the hugoniot in fig . [ fig_reflectivity ] , which clearly shows the pressure - induced change in reflectivity ( from 0.05 to 0.35 @xmath11 0.55 ) . for the pressure between 20 and 40 gpa ,
the increase in reflectance is retarded , then sharp increase exists around 40 to 50 gpa due to the nonmetal - metal transition . the results could be inspected by future experiments .
in summary , we have studied the thermodynamical properties of carbon dioxide under extreme conditions .
the eos obtained from qmd simulations are used to determine the hugoniot curve , which shows good agreement with available experimental data .
our calculations indicate that decomposition of carbon dioxide begins at 40 to 50 gpa along the hugoniot , where carbon atoms intend to form connections between each other , then diatomic and monoatomic oxygen are produced .
the nonmetal - metal transition , as well as the change in optical reflectance , are also found in the present work .
|
quantum molecular dynamic simulations are introduced to study the dynamical , electrical , and optical properties of carbon dioxide under dynamic compressions .
the principal hugoniot derived from the calculated equation of states is demonstrated to be well accordant with experimental results .
molecular dissociation and recombination are investigated through pair correlation functions , and decomposition of carbon dioxide is found to be between 40 and 50 gpa along the hugoniot , where nonmetal - metal transition is observed .
in addition , the optical properties of shock compressed carbon dioxide are also theoretically predicted along the hugoniot .
|
the universe undergoes its last major transition at @xmath8 as the first stars and early galaxies finish the process of reionizing the previously neutral intergalactic medium ( igm ) .
observations have only recently cracked the @xmath21 gyr universe with the discovery of galaxies tracing early star formation @xcite , quasars providing ionizing photons @xcite , and the complete absorption by the igm of all lyman-@xmath22 photons , i.e. the @xcite trough , in @xmath23 quasar spectra @xcite .
a new type of high-@xmath17 observation is the recent detection of metal - line absorbers in the most distant quasar spectra , probing the nucleosynthetic products of early star formation .
* hereafter bsrs ) surveyed 9 @xmath24 quasars and found 4 @xmath3 systems in the most distant object in their sample , j1148 + 5251 .
@xcite , @xcite , & @xcite discovered multiple @xmath2 absorbers at similar redshifts ( @xmath9 ) . despite the metals presumably being formed within galaxies ,
these absorbers are most readily explained as arising in the diffuse igm .
but such absorbers are rare and perhaps highly inhomogeneous : for instance , @xcite finds no @xmath2 along four lines of sight ( loss ) observed at higher signal - to - noise ( @xmath25 ) between @xmath26 .
the simple fact that metals exist in the early igm begs the question , how did they get there ?
population iii ( i.e. metal - free ) and early stars have been proposed to enrich a significant volume of the early universe , as early haloes harboring primordial galaxies have small physical scales and shallow potential wells ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the relatively invariant amount of @xmath2 absorption between @xmath27 @xcite and extended to @xmath9 @xcite has been interpreted to mean that the majority of igm metals were injected at @xmath23 ( e.g. * ? ? ?
but the most recent data set from @xcite and @xcite shows a decline of a factor of @xmath28 at @xmath29 compared to @xmath30 , suggesting that igm metallicity is increasing at early epochs , assuming typical ionization conditions do not evolve much .
furthermore , @xmath31 @xmath2 absorbers are spatially correlated with lyman break galaxies ( lbgs ) , suggesting ongoing enrichment @xcite , although these measurements are also consistent with metals being injected by dwarf galaxies at @xmath32 coupled with subsequent growth of clustering @xcite .
* hereafter od06 ) explored @xmath2 absorbers in cosmological hydrodynamic simulations , employing various heuristic prescriptions for galactic outflows to enrich the @xmath33 igm .
they determined that prescriptions based on momentum - driven winds were best able to reproduce @xmath2 observations .
od06 also showed that an increasing ionization correction for @xmath2 , from both cosmic evolution and energy injection from outflows , could mask a comparable increase of @xmath34 in the true igm metallicity from @xmath35 . hence widespread early enrichment from some exotic stellar population
is not required to explain the relatively invariant @xmath2 absorption over this redshift range .
another important question is , how do early galaxies relate to the igm metal - line absorbers at @xmath23 ?
* hereafter dfo06 ) explored reionization - epoch galaxies in the simulations of od06 , finding that strong outflows are required to match the observed luminosity function ( lf ) of @xmath9 @xmath36-band dropouts @xcite .
@xcite further found that the star formation histories from outflow simulations can reproduce broad - band spectral energy distributions ( seds ) of galaxies such as the lensed @xmath37 galaxy abell 2218 kesr @xcite .
the success of these models lies in the high mass loading of early galactic superwinds , which pushes the majority of high-@xmath17 metals nucleosynthesized in massive stars into the igm ( * ? ? ?
* oppenheimer & dav 2008 , hereafter od08 ) .
typically , many times more mass is driven out of galaxies than forms into stars .
hence if our models are correct , observations of igm metal - line absorbers may be the best way to account for the majority of metals in the @xmath38 universe .
finally , what can igm metal - line absorbers reveal about the topology of reionization ?
fifth year data from the _ wilkinson microwave anisotropy probe _ ( wmap , * ? ? ?
* ) indicate the universe was reionized at @xmath39 if it proceeded instantaneously .
numerical simulations suggest that reionization is a protracted process whereby individual sources create @xmath40 bubbles at @xmath41 that propagate outward until overlap by @xmath42 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
observations of @xmath43 absorption are divided over when reionization ends , with @xcite showing a discontinuity in the igm ionization rate at @xmath8 suggesting individual ionization bubbles have finally finished overlapping , while @xcite finds that the @xmath44 optical depth distribution shows no evidence of a sudden end to reionization at this epoch .
metal lines provide an additional handle on how reionization progresses , allowing the bulk of cosmic volume to be probed in absorption despite a fully saturated gunn - peterson trough . in this paper
we consider the above questions by studying @xmath45 metal - line absorbers in cosmological hydrodynamic simulations that incorporate our favored momentum - driven wind model . in this wind model ,
the outflow speed and mass loss rate are tied to galaxy mass using a heuristic prescription as expected for momentum - driven winds @xcite .
this wind model has enjoyed a number of successes when compared to a wide range of observations at @xmath46 ; most relevant is that it fairly uniquely matches observations of @xmath47 @xmath2 absorbers ( od06 ) .
the scaling of wind speed with galaxy velocity in this model follows the relation observed in local starburst outflows ( e.g. * ? ? ?
* ; * ? ? ?
we note that such scalings may arise from other physical scenarios besides momentum ( or radiation ) driven outflows ( e.g. * ? ? ?
* ) ; here we merely employ the scalings , without reference to the underlying physical driving mechanism .
we employ state - of - the - art gadget-2 cosmological hydrodynamic simulations with a quarter billion particles run to @xmath14 ; we describe details in 2 .
the salient predicted galaxy and igm properties are examined in 3 , and compared to our findings in dfo06 . in
4 we consider various cases for the ionization background , including the extreme cases of fully reionized and fully neutral , as well as a background with a variable intensity intended to represent ionized bubbles around individual star forming sources . 5 presents a plethora of simulated observations which are compared to available @xmath48 metal - line observations , as well as predictions for comparison with future data .
we then study the physical and environmental parameters of simulated absorbers in 6 , with a focus on how absorbers relate to galaxies and when they were injected into the igm .
7 provides a summary . throughout
we use @xcite abundances when calibrating metallicities relative to solar .
we employ our modified version of the n - body + smoothed particle hydrodynamics code gadget-2 @xcite to run three high - resolution @xmath49 particle simulation to explore the high redshift universe .
our simulations self - consistently enrich the igm via galactic outflows .
a complete description of our code , including our wind model , can be found in 2 of od06 with further modifications described in 2 of od08 ; however we also provide some of the fundamental details involving star formation , enrichment , and feedback here .
star formation ( sf ) is modeled using a subgrid recipe where each gas particle above a critical density is treated as a set of cold clouds embedded in a warm ionized medium , similar to the interstellar medium ( ism ) of our own galaxy @xcite .
gas particles eligible for sf undergo instantaneous self - enrichment from type ii supernovae ( sne ) .
the instantaneous recycling approximation of type ii sne energy to the warm ism phase self - regulates sf resulting in convergence in star formation rates ( sfrs ) when looking at higher resolutions .
the sfr is scaled to fit the disk - surface density - sfr observed in the local universe by @xcite , the accretion rate determines the sf as we argue in [ sec : lumfunc ] . ] .
star formation below 10 @xmath50 , assumed not to result in type ii sne , is decoupled from their high mass counterparts using a monte carlo algorithm that spawns star particles .
the metallicity of a star particle remains fixed once formed ; however , since type ii sne enrichment is continuous while stars are formed stochastically , every star particle invariably has a non - zero metallicity .
the chemical yields at high redshift are dominated by type ii sne , for which we use the metallicity - dependent yields from @xcite as described in od08 .
these yields vary less than 15% for the 4 species we track ( carbon , oxygen , silicon , & iron ) as long as @xmath51 , which applies to effectively all sf in our cosmological simulation .
we do not consider exotic enrichment yields from zero metallicity , pop iii , or very massive stars ( vmss , e.g. * ? ? ?
* ; * ? ? ?
* ) , and assume a @xcite initial mass function ( imf ) throughout . although it may be that such exotic stars have an effect on the igm particularly in terms of metal - line ratios ( e.g. * ? ? ?
* ; * ? ? ?
* ) , the yields used here provide a standard baseline for comparison .
the baseline yields are [ o / c]=0.27 and [ si / o]=-0.16 if we consider the @xcite @xmath52 yields assuming @xcite abundances ; the carbon mass yield ejected by sne is @xmath53 the total stellar mass in a chabrier imf
. the integrated gas - phase abundance ratios of the @xmath54 d16n256vzw simulation box are [ o / c]=0.23 and [ si / o]=-0.13 reflecting the @xcite yields inputted in our simulations . to allow the reader to scale our results to a different set of theoretical type ii sne yields we provide table [ table : yields ] listing the integrated yields assuming a @xcite imf slope with @xmath55 , where @xmath56 is the stellar imf mass fraction in the range indicated in the second column for a chabrier imf for stars @xmath57 , which we assume are the stars that undergo type ii sne . ] .
we have interpolated the published yields to @xmath52 ( i.e. @xmath58 ) using their various metallicity models , because this is near the median stellar metallicity at @xmath54 ( @xmath59 ) .
there exists little spread for the oxygen and silicon yields between @xcite , @xcite , and @xcite despite different sne mass progenitor ranges .
38% less carbon is produced from the @xcite yield models , which is significant when we consider a measurement such as @xmath15 ( [ sec : omegac4 ] ) .
lccccccc reference & mass range & @xmath60 & [ c / h ] & [ o / h ] & [ si / h ] & [ o / c ] & [ si / o ] + + * chieffi & limongi ( 2004)*@xmath61 & 13 - 35 @xmath50 & @xmath62 & 0.13 & 0.40 & 0.24 & 0.27 & -0.16 + woosley & weaver ( 1995 ) & 12 - 40 @xmath50 & @xmath62 & -0.08 & 0.39 & 0.25 & 0.47 & -0.14 + portinari et al .
( 1998 ) & 10 - 120 @xmath50 & @xmath62 & -0.05 & 0.43 & 0.17 & 0.48 & -0.26 + portinari et al .
( 1998 ) & 30 - 120 @xmath50 & @xmath62 & -0.21 & 0.35 & -1.54 & 0.56 & -1.89 + heger & woosley ( 2008 ) & 10 - 100 @xmath50 & @xmath63 & 0.02 & 0.20 & -0.29 & 0.18 & -0.49 + d16n256vzw @xmath54 sph & & & & & & 0.23 & -0.13 + + @xmath64 assuming a chabrier imf slope and @xmath55 ( i.e. 18% of stellar imf is covered by the mass range in column 2 and only this mass range contributes to type ii sne yields . ) .
+ @xmath61 we apply this set of metallicity - dependent yields in our simulations .
[ table : yields ] we note that there may be reasons to favor a more top - heavy or bottom - light imf at high redshifts ( e.g. * ? ? ?
* i.e. @xmath56 increases ) . however ,
since our simulations are well - matched to observed @xmath8 rest - uv lfs ( as we will show in 3 ) , and since metals mostly come from the same massive stars responsible for uv emission , our total metal production should be reasonably well - constrained .
along these lines we show the yields using the @xmath65 from @xcite in table [ table : yields ] ; silicon appears to be significantly under - produced in sne ejecta . for comparison
we include yields from @xcite from zero metallicity stars .
these yields show less difference relative to @xcite than the top - heavy @xcite ( except for oxygen ) , perhaps indicating just how hard it is to clearly identify the nucleosynthetic signatures of pop iii stars .
the implementation of winds follows @xcite , using a monte carlo ejection mechanism governed by two parameters , the outflow velocity @xmath66 and the mass loading factor @xmath67 , where the mass loading factor is defined as the outflow mass rate ( @xmath68 ) in units of the galaxy s sfr , i.e. @xmath69 sfr .
the wind model we employ here follows scalings of @xmath66 and @xmath67 expected for momentum - driven winds ; this fairly uniquely matches a broad range of observations including igm enrichment between @xmath70 traced by @xmath2 ( od06 ) and at @xmath71 traced by @xmath72 @xcite , the lfs of @xmath54 galaxies ( dfo06 ) , the galaxy mass - metallicity relations @xcite , and the enrichment and entropy levels in intragroup gas @xcite .
the scalings depend on galaxy velocity dispersion ( @xmath73 ) : @xmath74 where @xmath75 is the luminosity factor , which is the luminosity of the galaxy in units of the critical ( or sometimes called eddington ) luminosity of the galaxy , and @xmath76 provides a normalization for the mass loading factor . here
we randomly select @xmath77 $ ] for each sph particle we kick , following observations of local starbursts by @xcite , and we take @xmath78 .
@xmath73 is estimated from the galaxy s mass following @xcite , where the mass is obtained from an on - the - fly friends - of - friends galaxy finder .
the simulations adopt the cosmological parameters based on the 5-year wmap results @xcite , namely @xmath79 , @xmath80 , @xmath81 , @xmath82 km s@xmath83 mpc@xmath83 , @xmath84 , and @xmath85 .
the value of @xmath86 slightly exceeds the favored wmap data alone , but agrees better with the combined results including type ia sne and baryonic acoustic oscillation data .
a greater @xmath86 results in appreciably more sf at high-@xmath17 , which is important since sf drives the enriching outflows .
our runs with this cosmology are named the _ d - series _ ; the primary run is named d16n512vzw using our standard naming convention , indicating the box spans 16 comoving @xmath87s with 512@xmath88 each of gas and dark matter particles .
the _ vzw _
suffix indicates our implementation of the momentum - driven wind model as described in od08 ( with one inconsequential exception : we no longer impose an upper limit due to sn energy limitations on @xmath66 ) .
the gas particle mass is @xmath89 , which translates to a minimum galaxy baryonic mass of @xmath90 assuming a galaxy is resolved with 32 particles ( which we will demonstrate later ) .
the softening length is set to 0.6 comoving @xmath91 .
two more simulations , d8n512vzw and a d32n512vzw , are also run to explore numerical resolution convergence .
the respective softening lengths are 0.3 and 1.2 @xmath91 , and the minimum galaxy baryonic mass resolutions are @xmath92 and @xmath93
. we will show that the primary enrichers of the igm are galaxies with masses @xmath94 , which are statistically sampled best in the @xmath95 box .
the d16n512vzw simulation was run to @xmath96 , however we only consider outputs down to @xmath14 here .
this simulation was run on the intel 64-bit abe cluster at the national center for supercomputing applications on 128 processors , requiring 32,500 processor hours .
the d8n512vzw and d32n512vzw simulations were run only to @xmath97 and @xmath98 respectively , each consuming approximately half the above processor hours .
we run the spline kernel interpolative denmax ( skid ) group finder on the @xmath998 , 7 , 6 , and 5 outputs to identify galaxies .
in this section we explore key physical and observational predictions of our new simulations and relate them to findings from our previous lower - resolution simulations .
this is intended to demonstrate continuity with our previous works , as well as to provide a more global context for our exploration of high-@xmath17 metal - line absorbers .
we begin by presenting the @xmath54 galaxy stellar mass ( @xmath100 ) function in our two simulation boxes , and relating this to earlier papers where we fit observables best with a momentum - driven wind model .
we show that the 16 @xmath87 box is the best one for studying high-@xmath17 igm metals , despite the fact that it is not large enough to sample the space density of the most luminous @xmath8 galaxies .
finally , we show global evolutionary trends in star formation , wind properties , and metal enrichment .
in dfo06 we explored the properties of reionization - epoch galaxies ( @xmath101 ) , finding that the _ mzw _ wind model most capable of reproducing observations ( specifically * ? ? ? * ) compared to the constant wind ( _ cw _ ) and no wind ( _ nw _ ) cases .
the _ mzw _
model is similar to the _ vzw _ model , the only difference being that the wind speed at a given @xmath73 is constant in _ mzw _ while it has a random spread in _
vzw_. the key point is that @xmath67 varies with @xmath73 the same way in both models , and as we showed in dfo06 , @xmath67 governs high-@xmath17 sf by removing gas that would otherwise form into stars .
the constant wind model is that of @xcite , with @xmath102 km / s and @xmath103 .
we plot the stellar mass function of @xmath54 galaxies ( @xmath104 ) in figure [ fig : massfunc ] for the d8n512vzw ( orange ) , d16n512vzw ( black ) , and d32n512vzw ( green ) simulations .
we compare it to the w8n256vzw simulation ( dot - dashed magenta ) , which is part of the w - series simulations explored in dfo06 .
the w8n256vzw simulation has nearly the same resolution as the d16n512vzw simulation , but uses the first - year wmap results @xcite .
this cosmology ( @xmath105 , @xmath106 , @xmath107 , @xmath108 km s@xmath83 mpc@xmath83 , @xmath109 ) produces almost double the stars over @xmath110 owing to its higher @xmath86 . in the d16n512vzw simulation
, we find 1.83% of baryons in galaxies and 0.13% in stars ; the respective numbers in the w8n256vzw simulation are 3.10% and 0.26%
. stellar mass function of galaxies in the d8n512vzw ( orange line ) , d16n512vzw ( black line ) , and d32n512vzw ( green line ) boxes in units of @xmath111 ( comoving ) .
corresponding dotted vertical lines show the galaxy mass resolution limit . for comparison
we show the mass function of our older w8n256vzw model ( dot - dashed magenta line , from dfo06 ) , which is higher due to the different cosmology .
the solid red line is a least - squares fit to our three simulations ( equation [ eqn : m*func ] ) .
we define the characteristic @xmath100 as where the fit matches the slope of a @xmath112 power law , which is at @xmath113 .
the bottom panel shows a histogram of the stellar mass .
72% of the mass is in galaxies with @xmath114 . ]
we use @xmath104 to argue that we are resolving the galaxies responsible for metal enrichment .
stellar mass can be used as a direct proxy for the integrated metal production if two conditions are met : ( i ) all metals ( those not trapped in stellar remnants ) are produced in short - lived , massive stars ( i.e. type ii sne ) , which is a fair assumption given that longer lived stars ( agb stars ) have not had time to enrich much by @xmath115 ; and ( ii ) stellar mass is a direct proxy for the integrated number of type ii sne .
the stellar mass contained in gadget-2 star particles consists only of long - lived , non - sn stars , so the mass in star particles is a good record of the integrated type ii sne contribution .
we find that a two - dimensional polynomial fit to the logarithmic stellar mass function provides a good , simple fit : @xmath116 = -8.70 + 2.95 { { \rm log}}m _ * - 0.255 ( { { \rm log}}m_*)^2.\ ] ] shown as the red line in figure [ fig : massfunc ] , this fit obtains a slope of @xmath117 at @xmath113 , making this the characteristic @xmath100 at @xmath8 where the most mass per logarithmic @xmath100 bin is held .
we can not fit the mass function over 4 dex effectively with a schechter function , because the high mass end does not display an exponential cutoff and the low - mass end shows a gradual turnover owing to filtering by the cosmic uv background ( e.g. * ? ? ?
* ; * ? ? ?
integrating equation [ eqn : m*func ] , we find that 72% of mass is in galaxies with @xmath114 . to illustrate the stellar mass distribution , the bottom panel of figure [ fig : massfunc ] shows the fractional mass in each bin .
the median @xmath100 in terms of integrated mass is @xmath118 for the d16n512vzw simulation , and @xmath113 when including all three simulations .
the d32n512vzw box is better able to account for the cosmic variance of more massive galaxies , but its mass resolution is comparable to the median stellar mass , so it can not account for early enrichment from small galaxies .
the d8n512vzw run , conversely , does not produce many galaxies above @xmath119 , so it does not sample the high - mass end sufficiently .
the d16n512vzw run is well - suited to resolve the relevant galaxy mass range , so we employ this run from here on in .
a high-@xmath17 galaxy s sfr is more easily measured than its stellar mass , since the former can be estimated from optical data while the latter requires near - infrared data .
our simulations predict a tight relation between sfr and @xmath100 at @xmath54 , where @xmath120 this nearly linear relation is similarly tight in our simulations at all redshifts ( dfo06 , see their figure 4 ; * ? ? ?
* ; * ? ? ?
* ) , with a low scatter , and is mostly insensitive to outflow model @xcite .
this relation arises from the dominance of smooth , filamentary cold mode gas accretion , where the sfr is regulated by the dynamical timescale for gas infall in slowly - growing halo potentials ( e.g. * ? ? ?
* ) ; therefore the sf law we scale to @xcite has little effect on the sfr , chemical production , and feedback .
mergers are a sub - dominant process for fueling galaxies , particularly at early times and in smaller galaxies @xcite , which is why the scatter is small . in figure
[ fig : lumfunc ] we show our d16n512vzw simulated rest - frame 1350 uv luminosity function at @xmath54 , which is effectively a sfr function ( including extinction ) .
we compare to data from @xcite and @xcite .
we compute simulated galaxy luminosities by passing their sf histories through the @xcite population synthesis model , and applying a tophat filter around 1350 .
dust extinction is critical in this comparison ; we apply two extinction laws to our simulated galaxies : ( i ) the solid line represents a constant dust extinction , where @xmath121 , which leads to 50% flux attenuation at 1600 using the @xcite law ( we will use this value for our local ionizing field ) ; and ( ii ) the dashed line uses the metallicity - dependent extinction law calibrated from local galaxies introduced in @xcite and employed in dfo06 , which predicts less attenuation for the lower mass galaxies . in both cases the agreement with data is reasonable .
this shows that the star formation rates , and hence metal production rates , in our simulated galaxies are plausible .
rest - uv luminosity function ( 1350 magnitudes ab ) from our d16n512vzw run .
solid line assumes uniform dust extinction of @xmath121 , dashed line uses the @xcite metallicity - dependent extinction .
these are compared to observations from @xcite ( blue triangles ) and @xcite ( magenta squares ) .
the agreement is good in the overlap region .
the majority of sf remains below current detection limits , which implies that the sources of high-@xmath17 igm metals have yet to be detected .
the range of @xmath122 for parent galaxies of @xmath54 @xmath2 absorbers are demarcated by the dotted green lines .
median logarithmic stellar masses at a given magnitude are listed along the bottom . ]
numbers across the bottom of this figure show the median @xmath123 $ ] corresponding to a given rest - uv magnitude . the characteristic @xmath100 from equation [ eqn : m*func ] corresponds to @xmath124 , which is roughly one magnitude below current detection limits .
the green dotted lines indicate the magnitude range of the parent galaxies of @xmath54 @xmath2 absorbers , which we identify in [ sec : galabs ] .
hence the originating galaxies of currently observed high-@xmath17 @xmath2 systems are mostly below today s detection thresholds .
figure [ fig : sfr_wind ] ( top panel ) shows evolution of the cosmic sfr density ( sfrd ) per comoving volume , indicating a rise of a factor of three over the range @xmath6 .
the values are in the range of data , though the observations span a large range , in part owing to differences in the adopted limiting luminosity in each case . for example , @xcite finds @xmath125 more sfrd when they reduce the limiting luminosity by a factor @xmath126 for the z - band dropouts ( @xmath127 ) .
most of the @xmath54 stellar mass and sf in our simulations remain below the detection limits of current surveys . at @xmath54
, we find 31% of the sfrd arises from galaxies with @xmath128 , 39% from @xmath129 , and the remaining 30% from masses below this . at @xmath130 , 61% of the sfrd
arises in galaxies with @xmath131 .
we note that @xcite examined the evolution of the cosmic sfrd and found that our momentum - driven wind model provided the best match to the observed evolution from @xmath132 , of the models considered there . with mass loading factors @xmath133 .
the energy in outflows is @xmath134 of the available supernova energy at these redshifts , assuming a chabrier imf .
one sigma dispersions are indicated by dashed lines .
the baryonic mass of galaxies responsible for winds ( bottom panel ) increase 6@xmath135 between @xmath136 ( bottom panel ) .
, title="fig : " ] -0.13 in the adopted dependence of momentum - driven wind scalings from galaxy mass results in the median wind speed growing and median mass loading factor declining towards lower redshifts ( od08 ) .
this has direct ramifications for reionization - era galaxies , since sfrs are attenuated by a factor of @xmath137 relative to the no - wind case .
this is because @xmath138 of infalling gas that would otherwise be converted into stars , instead is ejected in outflows .
this factor assumes that outflows do not return to galaxies , which is a fair assumption at high redshift because outflow material generally does not recycle back into galaxies within a hubble time at these epochs ( od08 ) .
the lower panels of figure [ fig : sfr_wind ] show that high-@xmath17 wind properties have a relatively weak redshift dependence from @xmath139 : @xmath66 increases by 20% ( second panel ) and @xmath67 declines by 17% ( third panel ) , despite an increase of 6@xmath135 in the average baryonic mass of a galaxy driving the winds ( bottom panel , i.e. weighted by @xmath68 ) .
the reason is that @xmath73 is proportional to @xmath140 , and the declining @xmath141 turns out to roughly cancel the increase in @xmath142 during this epoch , resulting in only modest evolution of wind properties .
fiducial values during reionization in our model are therefore @xmath143 and @xmath144 .
the energy budget of momentum driven winds remains roughly 20% of the total sn energy assuming a chabrier imf ( fourth panel ) .
high-@xmath17 winds require substantial energy input , but the wind speeds are relatively mild compared to low-@xmath17 outflows from lirgs and ulirgs ( e.g. * ? ? ?
* ; * ? ? ?
as before , we find that substantial mass ejection is a key feature of our wind model : the amount of material ejected at these epochs is many times that which forms into stars .
one reason why enriching the igm is easier at high-@xmath17 is that the physical scale of the universe is small .
od08 showed that in our wind model , winds propagate a similar physical distance ( @xmath145 kpc ) for all redshifts and galaxy masses ; the comoving volume a galaxy can enrich therefore goes as @xmath146 .
figure [ fig : rho_z ] illustrates the rapid growth of metallicity at igm overdensities from @xmath6 .
the gradually rising @xmath147 trend in place at all four redshifts is one of the defining signatures of the momentum - driven wind enrichment .
in contrast , a wind model with constant @xmath66 ( e.g. * ? ? ?
* ) will create a flatter metallicity - density relationship with either a sharp drop - off at lower overdensity ( cf .
figure 10 of od06 ) or a rise in metallicity if superwinds are even stronger ( cf .
figure 8 of @xcite ) signifying efficient enrichment of voids . under our model
the igm metallicity grows between @xmath6 by a factor of @xmath148 at an overdensity ( @xmath149 ) of 80 and @xmath150 at @xmath151 . from the d16n512vzw simulation .
@xmath60 is the total mass fraction of metals ; @xmath152 .
a clear gradient is established early on , and as time passes the diffuse igm is enriched more quickly than high - density regions .
_ bottom : _ the cumulative volume filling factor of metals versus metallicity , indicating an inhomogeneous distribution of metals with only 1% of the universe enriched to @xmath153 by @xmath54 , despite some metals spreading to near the cosmic mean density . ] despite some metals reaching cosmic mean densities , most of the volume remains unenriched , especially the voids .
the cumulative volume filling factors of metals in the bottom panel of figure [ fig : rho_z ] show this , as even regions with @xmath154 fill only about 1% of the simulation volume at @xmath14 . filling factors increase only modestly to lower @xmath17 ( od08 ) , which is somewhat remarkable given that these models can match most available igm metal - line data .
hence the presence of diffuse igm metal absorbers does not necessarily indicate widespread metal distribution .
we will return to this point , and the inhomogeneous enrichment it implies , later .
regions of galactic overdensities ( @xmath155 ) are generally enriched rapidly to @xmath156 or higher .
the metallicity does not grow much at such overdensities , which is reflected in the lack of predicted evolution in the mass - metallicity relation from @xmath157 in dfo08 ( see their figure 5 ) .
the fraction of cosmic volume enriched to @xmath158 is 0.01 - 0.02% , which is actually a factor of 1.5 - 3@xmath135 _ higher _ than that at @xmath159 @xcite .
it is astounding to consider that the volume fraction of the universe enriched to around solar metallicity may change very little between @xmath160 ; the key difference is the corresponding regions at low redshift contain orders of magnitude more baryons .
this is a result of our momentum - driven wind model based on locally observed outflows .
@xcite showed these winds efficiently enrich igm overdensities at high-@xmath17 , some to @xmath161 , while by @xmath162 most of these metals recycle back into galactic environments and later winds often can not escape out of galactic haloes ( od08 ) .
many of the trends noted above can be seen in figure [ fig : snaps60 ] , which shows the locations of galaxies in the upper left and the metallicity - density distribution in the upper right in a 25 @xmath163-thick slice extending across the d16n512vzw simulation box .
gray indicates zero metallicity in these frames , which corresponds to the vast majority of the igm , and shows the nascent cosmic web .
most of the volume remains unenriched at these redshifts , while regions within early galaxies become enriched up to a tenth solar or more .
clustering between early galaxies and enriched portions of the igm is strong .
this illustrates that the galaxy - absorber connection at high-@xmath17 is extremely tight , as we investigate further in
[ sec : galabs ] .
to make predictions for observable metal - line absorbers , we must obtain ionization corrections for relevant metal ions .
this requires that we have a model for the photo - ionization rate at each point in the simulation volume .
ideally , we would like to self - consistently track the local ionization spectrum in these simulations . but this must await a full time - dependent , non - equilibrium radiative transfer implementation , which we are in the process of developing @xcite but do not yet have .
hence for this work , we will consider three ad hoc cases for the ionization background ( dubbed _ no field , hm2001 , _ and _ bubble _ ) that span a plausible range .
note that our simulations are originally run with a uniform @xcite ionizing background , and we are only imposing these backgrounds in post - processing .
this obviously does not properly capture the dynamical effects of different pressure forces in our three cases .
however , since dark matter is dynamically dominant in the relatively low - density regions giving rise to igm absorbers , post - processing the background to obtain ionization corrections is a good approximation . *
no field : * our first case assumes a fully neutral igm , where no ionizing background exists at wavelengths below the lyman limit , while lower energy photons stream freely . this extreme case may be applicable if metal injection precedes ionization fronts , or more realistically if the igm recombines in regions enriched and ionized at earlier times .
oxygen remains in its ground state due to the @xmath3 ionization potential ( 13.618 ev ) being barely higher than the @xmath43 ionization potential , while @xmath164 ( 11.260 ev ) and @xmath165 ( 8.151 ev ) are completely ionized to the second level . +
* hm2001 : * our second case assumes reionization is complete and a spatially - uniform background applies everywhere .
we use the @xcite background assuming a contribution from quasars and 10% of photons above the lyman limit from star forming galaxies generated by their cosmic ultraviolet background ( cuba ) program ; at @xmath23 , star - forming galaxies dominate the ionizing flux .
we divide the total flux by a factor of 1.6 as we did in od06 in order to match the summed flux decrement of the @xmath44 forest down to @xmath166 .
an assumption of a spatially - uniform ionization background is probably far from correct at @xmath23 , but this is intended to represent another extreme case like _ no field_. ionization fractions for metal lines are calculated using cloudy - generated @xcite ionization tables as a function of density and temperature , as described in od06 . + * bubble : * the last case assumes that the ionizing flux of the nearest galaxy dominates , which may be the most realistic scenario during reionization .
this is applicable if metals remain near enough to their parent galaxies to be dominated by the local ionizing intensity , which we will show is often true since metals typically reside within 30 kpc of the originating galaxy .
of course , detailed radiative transfer effects of escaping ionization and subsequent recombinations will play a critical role , so this scenario should be considered a crude proxy for a patchy reionization model . to implement the _ bubble _ field , we consider the ionizing field from only the nearest galaxy , weighted by the inverse square law .
we assume an average sed per stellar mass by taking the average @xcite stellar synthesis spectrum of _ all _ the stars in the simulation box for a given redshift , accounting for their ages and metallicities .
we then calculate the impinging flux using the stellar mass and distance of the nearest galaxy , which is identified for each sph particle .
note that the same ionizing flux per unit stellar mass is assumed to emanate from each galaxy at a given epoch ; we do not account for variations in nearest galaxy seds , which is another approximation required to make the problem computationally tractable .
we attenuate the emission from galaxies by dust extinction .
the assumed extinction has a fixed value of @xmath121 , as shown in figure [ fig : lumfunc ] to broadly match the observed rest - uv lf . to calculate how the emergent sed is attenuated
, we use the theoretical dust absorption curve from the x - ray to the far - ir of @xcite , using their grain size distribution of silicate and carbonaceous grains found to fit the small magellanic cloud ( smc ) bar .
we use this curve because @xmath167 lbgs exhibit the smaller grain size distributions of smc - like dust @xcite .
note that we can not directly apply the empirical @xcite extinction law because it extends only down to 1200 , while we are especially interested in the attenuation at energies where @xmath2 and @xmath5 are ionized ( i.e. @xmath168 ) .
for each sph particle , the impinging flux is then the dust - attenuated emission divided by the distance from the nearest galaxy squared .
we then use cloudy to generate the resulting metal ionization fractions , which is then used to obtain the optical depth for any ion along the los as described in od06 .
we plot the _
hm2001 _ field at @xmath130 and 6 in figure [ fig : ionbkgd ] , along with the _ bubble _ field , where we assume a @xmath169 galaxy at 10 kpc ( typical values ) for concreteness .
the difference between the solid and dotted orange lines for the _ bubble _ case at @xmath54 illustrates the effects of dust attenuation .
dust absorption in all of the @xcite models sharply peaks at 600 - 900 , causing a dip .
there is a clear lyman break , which in our cases arises mostly from stellar atmospheres , and secondarily from dust .
we note that this break is not nearly as strong as that observed by @xcite in lyman break galaxies at @xmath167 , perhaps indicating that there is a fairly small escape fraction of ionizing photons ; for instance , @xcite assume a global 10% escape fraction .
however , because these data are from lower-@xmath17 and higher-@xmath100 galaxies , it may not be applicable here . for simplicity , we do not assume an overall escape fraction .
( magenta ) and @xmath54 ( orange ) are compared to the _ bubble _ field arising from a @xmath54 @xmath170 galaxy at 10 kpc .
the thin solid orange line represents the unattenuated @xcite - computed sed at @xmath130 to compare with the dust - attenuated ( thick dotted orange line ) sed . for reference , ionization potentials for carbon , oxygen , and silicon
are indicated , and the relative cross - sections from @xmath43 and @xmath171 as a function of wavelength are shown along the bottom . ] comparing the _ bubble _ and _ hm2001 _ fields at @xmath54 , the strength of the lyman break is much less in the _ bubble _ case .
this is mostly because our typical galaxy is younger and more metal - poor than that assumed by @xcite ( @xmath172 , 0.5 gyrs ) , which leads to a harder ionizing source .
two secondary reasons for greater hardness are : ( i ) the _ bubble _ lyman - break attenuation due to dust is less than the _
attenuation applied ( @xmath173 vs. an overall 10% escape fraction ) ; and ( ii ) the dust attenuation declines at the ionization potentials of @xmath2 and @xmath5 ( shown ) .
this leads to greater ionization fractions at these higher states in the _ bubble _ field , which has observable implications as we will show throughout [ sec : observe ] . before delving into the observables ,
we present here a primer for understanding metal - line observations resulting from our two assumed ionization backgrounds , the _ hm2001 _ and the _ bubble _ fields .
the _ no field _ case is straightforward ( i.e. ionization fractions of unity for @xmath1 , @xmath3 , and @xmath4 ) , but the often nuanced behavior of the other two fields has some perhaps unexpected consequences .
we consider metals in @xmath174 phase space exploring both low and high - ionization species at two redshifts that span the redshift range we consider .
figure [ fig : ioncomp ] plots colored contours corresponding to the _ hm2001 _
( left panels ) and the _ bubble _
( right panels ) backgrounds in @xmath174 phase space for @xmath1 ( solid blue ) and @xmath2 ( dot - dashed red ) at @xmath130 and @xmath14 ( upper and lower panels , respectively ) .
each contour step represents an increase of 20% in the ionization fraction for the corresponding ion .
overlaid are thick gray logarithmic metal density contours ( darker contours indicate higher metal densities ) at the respective redshifts .
overlapping color and gray contours represent regions of phase space where absorption in that ion will arise for a given ionization field .
the _ hm2001 _ contours are the same at all spatial locations within the simulation box at a given redshift , so the resulting behavior is simpler to understand .
the corresponding contours for both @xmath1 and @xmath2 move to higher overdensity by almost a factor of 10 from @xmath6 , owing to the ionization parameter @xmath175 increasing .
@xmath175 is defined in this paper as the number density of photons capable of ionizing to a particular state ( e.g. ionizing @xmath176 to @xmath2 ) divided by the number density of a particular atomic species ( carbon in this case ) .
@xmath175 increases @xmath177 because the physical densities decline by a factor of 3.4 from hubble expansion , and the ionization intensity increases by @xmath178 between @xmath6 .
@xmath1 evolves to trace a smaller fraction of phase space at higher densities ; however the metal density is greater at these higher densities and at later times , so the total amount of @xmath1 is expected to be fairly stable from @xmath6 .
the contours extending to higher densities at @xmath179 k in figure [ fig : ioncomp ] correspond to @xmath2 in collisional ionized equilibrium ( cie ) .
this overlaps few metals , because @xmath2 is an efficient coolant and sph particles evolve rapidly through this region , much like we find for @xmath72 at @xmath71 @xcite .
( upper ) and @xmath2 ( lower ) ionization fractions as a function of overdensity at @xmath180 k for the _ hm2001 _
( dashed lines ) and _ bubble _ ( dotted lines ) at @xmath130 ( magenta ) and @xmath14 ( green ) .
hm2001 _ field evolves more as quasars provide increasing intensity above 47.9 ev at later times .
the _ bubble _
field shows less evolution , and generally @xmath2 traces lower overdensities . _
lower panel : _ the mean carbon spectral ratio @xmath181 for our two ionization fields as a function of redshift .
the star forming galaxy - dominated _ bubble _ field is significantly harder than the _
hm2001 _ field with quasars , because _
_ is dominated by galaxies at high-@xmath17 and the attenuation above 47.9 ev is greater than the _ bubble _ attenuation due to dust .
this results in overall lower values for @xmath182 in the _
hm2001 _ case . ] as most igm metals remain near @xmath183 k , we plot the @xmath1/c and @xmath2/c ionization fractions ( @xmath184 , @xmath182 ) at this temperature in figure [ fig : rho_ionf ] ( top panels ) , showing the _ hm2001 _ curves moving upward in overdensity by the factor of 10 from @xmath6 .
@xmath2 should dramatically increase as the @xmath130 field is too weak to ionize @xmath2 , while by @xmath14 the gray and @xmath2 contours move to overlap as more metals enrich lower overdensities while @xmath175 increases . a second effect illustrated in figure [ fig : rho_ionf ]
is that the maximum @xmath182 grows as a result of the harder background with more quasar contribution at @xmath14 relative to @xmath130 ; more photons exist at the ionization potential of @xmath176 ( 47.9 ev or 259 ) to create @xmath2 relative to the @xmath164 potential ( 11.3 ev or 1101 ) to make @xmath1 , resulting in a larger density range where @xmath2 resides .
we define and plot in figure [ fig : rho_ionf ] ( bottom panel ) the ratio @xmath185 , which we call the `` carbon spectral ratio '' .
this has significant implications for @xmath1 and @xmath2 observations .
the carbon spectral ratio is @xmath186 greater for the _ bubble _ versus the _ hm2001 _ field at @xmath130 .
we now focus on the _ bubble _ field behavior .
the example fields signified by the right panels in figure [ fig : ioncomp ] demonstrate the expected ionization in @xmath174 phase space at 10 physical kpc from a @xmath187 galaxy .
these contours shift right as @xmath100 increases and left as @xmath188 increases due to the inverse square law ; e.g. the contours are equivalent to the ionization field from a @xmath189 galaxy at 100 kpc .
we choose this field strength because this is the representative of where our simulated metals typically reside . like the _ hm2001
_ field , our implementation of the _ bubble _ field depends on only two variables : @xmath175 and @xmath190 .
the difference is that @xmath175 depends on a combination of density , stellar mass , and galaxy distance .
if we consider the _ bubble _ case shown in figure [ fig : ioncomp ] at @xmath130 , there exists moderate overlap with the metal contours between @xmath191 mean overdensity , which are reasonable densities at 10 physical kpc from our simulated @xmath187 galaxies . but note that the calculated incidence of intersecting at least this close to one of 332 @xmath192 galaxies in the d16n256vzw box is @xmath193 , so a high-@xmath17 quasar spectrum would be lucky to intersect one such region .
the contours move to higher overdensity for larger galaxies ( or at closer distances ) , but these are even less likely .
meanwhile , by @xmath14 , the space density of @xmath194 galaxies is @xmath195 higher and more metals have enriched igm overdensities while the ionization contours have not evolved too much .
this is why we obtain an increase in @xmath2 absorption from @xmath6 , as we show in [ sec : omega_hiion ] . the greater carbon spectral ratio in the _ bubble _ field leads to greater @xmath182 at photo - ionized temperatures , as shown in figure [ fig : rho_ionf ] . at @xmath180
k , @xmath182 maximizes at 0.67 at @xmath130 and extends over 3.2 decades of density where @xmath196 , compared to @xmath197 and 1.6 decades with @xmath196 for the _ hm2001 _ field .
the _ bubble _
hardness declines as stars age and grow more metal rich by @xmath14 , ( maximum @xmath198 ) , and rises for _ hm2001 _ as the quasar density increases ( @xmath199 , cf . the carbon spectral ratio evolution in the bottom of figure [ fig : rho_ionf ] ) .
the greater _ bubble _ hardness allows significantly more @xmath2 , especially at the highest redshifts we explore .
@xcite also finds @xmath182 increases at @xmath200 at @xmath201 when the dust attenuated flux from a @xmath202 galaxy at 100 kpc is included . in the end , this results in less evolution for @xmath2 absorption in the _ bubble _ model relative to _ hm2001 _ , as we show in [ sec : omega_hiion ] .
our _ bubble _ evolution is greater than the @xcite example , because the metals are 5 - 50 kpc from ionizing galaxies .
a key point applicable to all metal - line igm observations is that while it is well known that taking ratios of different photo - ionized species can determine spectral hardness ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , a single species ionization correction alone can vary dramatically due to the spectral shape .
it is risky therefore to assign a fiducial ionization correction , especially considering we know so little about the ionization background for high - ionization species ; as a result , an observable such as @xmath15 could easily change by a factor of a few , as we stress in [ sec : omegac4 ] . to visualize the spatial distribution of metal absorbers
, we illustrate the absorption column densities in two - dimensional maps of the 25 @xmath163 slices in the bottom panels of figures [ fig : snaps60 ] assuming a uniform _
_ background .
@xmath1 is confined to overdense regions close to galaxies , while @xmath2 is a more capable tracer of the igm volume enriched , albeit mostly at column densities that are below current observational limits . the scarcity of pink in the bottom right frame indicating column densities similar to those observed by @xcite and @xcite indicates just how rare detectable @xmath2 is .
we demonstrate evolution using a zoom - in to the second most massive galaxy in our box in figure [ fig : snapsevol ] .
this visualization shows the evolution of the surrounding @xmath203 region using the same scales as figure [ fig : snaps60 ] . a clear bipolar outflow emanates perpendicularly to a filament feeding the growth of a @xmath204 galaxy ( at @xmath54 ) .
strong @xmath1 absorption traces the overdense regions around the galaxy at all epochs , while @xmath2 evolves to trace the extended igm enrichment ; observed @xmath2 evolves faster than @xmath1 between @xmath6 .
the smaller galaxies in the bottom right of this frame , @xmath205 , are the most prolific igm enrichers at high-@xmath17 due to their large numbers and high mass loading factors .
observations of the high-@xmath17 igm are leaping forward with the advent of near - ir spectroscopy , along with the discovery of quasars at @xmath23 , with an eye to the tantalizing possibility of using gamma - ray bursts ( grbs ; * ? ? ? * ) to briefly illuminate igm absorbers at very high redshifts .
we present here predictions for high-@xmath17 metals line transitions occurring redward of @xmath44 , where the quasar continuum is free of @xmath43 absorption .
we consider five species already observed at @xmath206 with rest wavelengths ( @xmath207 ) redward of @xmath44 : @xmath1 ( 1334.5 ) , @xmath2 ( 1548.2 , 1550.8 ) , @xmath3 ( 1302.2 ) , @xmath4 ( 1260.4 ) , and @xmath5 ( 1393.8 , 1402,8 ) at our four chosen redshifts : @xmath130 , 7 , 6 , and 5 .
three ionization fields introduced in [ sec : ioncase ] are considered : _ no field _ , _
hm2001 _ , and _
bubble_. we use our quasar absorption line spectral generator specexbin ( see 2.5 of od06 for details ) to produce continuous angled loss through our simulation volume , each covering @xmath208 for each of the four redshifts .
we make an `` ideal''-quality sample of 70 loss , assuming a high @xmath25 ratio of 50 for each @xmath209 resolution element , and a moderate instrumental resolution of 15 @xmath163 .
this is comparable to what is achievable on current optical spectrographs on @xmath210 meter class telescopes , which can trace @xmath1 out to @xmath211 ( e.g. bsrs ) . at higher redshifts ,
near - ir data will be required . the current generation of near - ir spectrometers have resolutions of @xmath212 and a @xmath213 per resolution element for @xmath9 quasars ( r. simcoe , private communication ) .
the next generation should achieve @xmath214 with much higher s / n using 20 - 30 meter class telescopes , e.g. the near - infrared multi - object spectrograph ( irmos ) on the thirty meter telescope ( tmt ) @xcite .
our simulated spectra can be used both to compare to current optical spectra , as well as make predictions for future instruments .
we first consider column density distributions for three of the five ionization species .
we fit voigt profiles using the automated fitter autovp @xcite , generally with little uncertainty since metal lines are typically unsaturated and unblended
. then we sum column densities into a cosmic mass density as a function of redshift for each species .
we present predictions for equivalent widths next , for comparisons with large samples of @xmath23 metal absorbers that may obtained in the near future .
we then consider some simple absorber ratios , emphasizing what can be inferred about the ionizing conditions .
lastly , we show some sample absorption line profiles and explain the velocity profiles of systems . along the way
, we compare to published @xmath215 metal absorber data , and offer some interpretations based on our models .
we consider the observed completeness limits assuming @xmath25=10 and 20 per resolution element at @xmath216 , and we confirm that the results we show in our high quality spectra hold at lower resolution and @xmath25 .
the column density distribution ( cdd ) , @xmath217 , is the most basic metal absorber counting statistic . here
@xmath218 is the number of lines with column density between @xmath219 and @xmath220 , and @xmath221 is absorption pathlength .
following general practice among observers , we sum together all individual voigt profile _
components _ within 100 @xmath163 of each other into _
systems_. this minimizes differences due to vagaries of voigt profile fitting , and allows more straightforward comparisons among data sets with varying resolution and signal - to - noise .
for @xmath221 we use @xmath222 @xmath221 for our adopted @xmath223 cosmology is almost double that in an einstein - desitter universe ( the @xmath224 term is negligible at these redshifts ) .
we scale other observations to our assumed cosmology where necessary .
figure [ fig : cdd ] shows cdds for oxygen and carbon using different choices for the ionizing field at @xmath130 ( left ) and @xmath54 ( right ) .
the first ion considered is @xmath3 in the top panels , which has a unique association with @xmath43 because these two species remain in charge - exchange equilibrium @xcite due to their neutral ionization potentials being only 0.15% apart in energy . @xcite and
@xcite hypothesize the existence of an `` @xmath3 forest '' tracing the neutral metals that have enriched dense igm regions via early enrichment .
higher ionization states easily recombine at the high densities of the early universe .
dozens , even hundreds of lines with @xmath225 should be detectable if the igm is neutral and enriched with a large filling factor of metals , but bsrs find very few @xmath3 lines .
no @xmath3 systems above @xmath226 are found either , despite such predictions by @xcite .
our _ no field _ case ( upper panels of figure [ fig : cdd ] ) shows that @xmath3 lines should be common at @xmath130 , despite our metal filling factors being under 1% .
we predict line frequencies of 1.9 and 0.6 @xmath227 for @xmath228 and @xmath229 respectively at @xmath130 .
the corresponding frequencies rise to 5.4 and 1.9 @xmath227 by @xmath54 .
@xmath3 provides the clearest contrast of any species between a neutral and ionized igm ; the _ bubble _ field yields frequencies of 0.4 and 0.3 @xmath227 above @xmath230 at @xmath130 and 6 respectively .
this clear contrast may already have been observed by bsrs , as all of their @xmath3 absorbers detected at @xmath23 are along one of their three loss probing these redshifts .
this could suggest a patchy reionization scenario ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) where individual loss probe large - scale structures at different ionization states .
we fit a power law of the form @xmath231 to the @xmath3 distribution by summing systems into two bins ( @xmath232 and @xmath233 ) and finding the best - fit line .
we list @xmath234 $ ] and @xmath22 for each cdd in figure [ fig : cdd ] if there are at least 10 systems in each bin ; otherwise we list system frequencies per @xmath235 of the lower and upper bins ( positive numbers ) .
we set @xmath236 for the normalization .
the _ no field _
@xmath54 cdd has @xmath237 over these column densities , which translates to 85% as many systems in the upper @xmath3 bin compared to the lower bin . a power law where @xmath238 indicates more metals at higher column densities , meaning this power law must turn over so that the summed @xmath3 is finite
. power law slopes can yield information about the nature of the metallicity distribution and ionizing source , and provide an easy way to compare datasets to simulations . not only does the _ bubble _ field create fewer @xmath3 lines at every column density , it has a much steeper power law , which we can not calculate accurately as there are too few absorbers in the higher bin .
oxygen and hydrogen are easily ionized out of their neutral states by any ionizing background above 1 rydberg .
@xmath1 cdds shown in the middle panels of figure [ fig : cdd ] recreate similar behavior to @xmath3 for the _ no field _ case , but has a much higher frequency for the ionization field cases since the @xmath164 ionization potential lies below 1 ryd , unlike @xmath3 .
the @xmath54 frequency of the lines with @xmath239 is 4.5 , 0.8 , and 0.4 @xmath227 for the _ no field _ , _ hm2001 _ , and _ bubble _
fields , respectively .
furthermore , the slopes of the cdds between @xmath240 may help constrain the ionization field , being @xmath237 , @xmath241 and @xmath242 , respectively .
the _ bubble _
field produces fewer strong @xmath1 lines than _ hm2001 _ , because the metals at @xmath54 still reside close enough to the local ionizing source that carbon is ionized to higher states by the greater intensity above the lyman limit .
@xmath2 has a higher frequency of lines with @xmath243 ( 0.9 vs. 0.2 @xmath227 ) and a shallower slope ( @xmath244 vs. @xmath245 over @xmath246 ) for the _ bubble _ field compared to the _
hm2001 _ field in the bottom panels of figure [ fig : cdd ] .
there is no @xmath2 in the _ no field _ case , as shocks from structure formation rarely heat the gas to temperatures where @xmath2 is collisionally ionized at these early epochs .
we plot a black square from the three observed @xmath2 absorbers of @xcite where @xmath247 and include the additional pathlength from the sdss j0002 + 2550 los observed only by @xcite ( total @xmath248 ) .
we find an observed frequency of @xmath249 for these absorbers averaging @xmath250 . to compare directly to our models , we ( i ) multiply the frequency of our @xmath2 absorbers by the @xcite sensitivity function ( their figure 6 ) , and ( ii ) estimate the @xmath251 @xmath2 frequency by taking a log - weighted average of our @xmath54 and @xmath252 model frequencies , because strong @xmath2 absorbers rapidly increase in number from @xmath253 ( especially for _ hm2001 _ ) .
our @xmath247 frequencies are 1.2 and 0.4 @xmath227 for the _ bubble _ and _ hm2001 _ models respectively .
the uniform field is consistent with the observed frequency , while the local ionizing bubble actually over - predicts strong @xmath2 absorbers .
we actually prefer the the _ bubble _ model , because we will argue igm metals are not completely ionized and that a significant fraction of them are in lower ionization species , either recombined or shielded from uv flux .
we also plot the maximum allowed frequency of @xmath254 systems from the null @xcite detection ( upside - down triangle ) ; this is the column density where they calculate they have 50% completeness .
we appear to over - predict the amount of these weaker absorbers in either model , although by how much will have to wait for future observations .
the long dashed cyan line is the case where a @xmath153 metallicity floor is added to the simulation ; we discuss this more in
[ sec : ew ] .
the fact that @xcite finds no weak @xmath2 lines in the largest pathlength yet sampled at @xmath24 indicates a large volume of the igm is not enriched much above @xmath154 .
for comparison with lower @xmath17 , we plot @xcite power law fit to @xmath2 absorbers at @xmath255 .
there exists less @xmath2 at all column densities in our simulations , which agrees with our reproduction of a declining @xmath15 at @xmath48 as we discuss next .
we now consider the summed mass densities as a way to illustrate the evolutionary differences between ionization fields .
the evolution of @xmath2 for the two fields differs since the weak _ hm2001 _ field at @xmath130 is incapable of ionizing overdensities containing carbon up to @xmath2 , creating very few strong absorbers , while the _ bubble _ field has a harder spectrum early on ( cf .
red contours in upper panels of figure [ fig : ioncomp ] ) .
an ion s global mass density may be obtained by integrating the total column density of their systems and dividing by pathlength , as follows : @xmath256 where @xmath257 is the hubble constant , @xmath258 is the speed of light , @xmath259 is the critical density of the universe , and @xmath260 is the atomic weight of the given species . for comparison to the data , we count systems over two decades of column density chosen to correspond to the species detectability .
we count lines for @xmath2 over @xmath261 , and scale other species column density range using the ratio of the oscillator strength of a line ( @xmath262 , the strongest of a doublet ) times its rest wavelength ( @xmath263 ) for the considered species versus that of @xmath2 .
for example , the range considered for @xmath3 works out to be @xmath264 , because the resulting @xmath265 ratio is @xmath266 ; @xmath3 has a @xmath267 weaker oscillator strength and a @xmath268 smaller wavelength .
@xmath3 is harder to observe due to its weaker oscillator strength , so the column densities considered are higher .
conversely , silicon species have high oscillator strengths , meaning that column densities below @xmath269 are achievable for @xmath4 ( @xmath270 ) and @xmath5 ( @xmath271 ) .
the range for @xmath1 is @xmath272 .
nature creates a convenient conspiracy whereby greater oscillator strengths make up for lower type ii sne yield abundances , leading to similarly strong lines for our five considered ionic species .
error bars are calculated using the @xcite method ( their equation 4 ) , and do not include systematics such as cosmic variance .
we also apply this error estimation method to the bsrs dataset , which results in much larger error bars than they published as a result of them including only the uncertainty in their summed column densities .
the resulting @xmath273 s for the five species and different fields are plotted in figure [ fig : omega ] .
we also plot the total metal density of the corresponding element , summed by counting the gas phase mass of that species in the entire simulation box at the given redshift ( @xmath274 , black dash - dot line ) . with no ionization correction in the _ no field _
case , the ions @xmath1 , @xmath3 , and @xmath4 directly trace the increase seen in metal density of @xmath275 from @xmath6 .
note , however , that the ion density should be identical to the metal density if absorption lines of all column density are counted , yet the ion density is @xmath276 less than the metal density .
where are the unaccounted metals ?
the answer is that since the cdds have slope @xmath238 , they are in high - column systems that are extraordinarily rare . hence using a volume - weighted method as quasar loss to obtain a handle on a mass - weighted quantity such as ion density is challenging , more so for low ionization species , which preferentially trace high overdensities .
the @xmath277 determined by bsrs favors the _ no field _ case over either ionization field , considering their sample from @xmath278 quasars .
this suggests that at least some of the @xmath8 metal - enriched igm is neutral .
that bsrs found four @xmath3 systems ( all with @xmath279 ) toward sdss j1148 + 5251 and not more than one in any of the other 8 high-@xmath17 loss ; this is unlikely to be a statistical fluctuation ( 0.18% chance according to bsrs ) .
the _ no field _ case produces a frequency of 9.2 @xmath227 of such @xmath3 lines , which is in agreement with the j1148 los and at least @xmath280 higher than either of our simulated ionization fields .
hence our simulations agree with bsrs s interpretation that the j1148 los possibly probes a long stretch of igm with incomplete reionization , while the other loss of bsrs are consistent with being completely reionized .
how one obtains a relatively neutral los along the 200 comoving mpc spanned by the j1148 los while most loss are reionized is not easy to understand , but we can not address this directly with our current simulations owing to lack of volume .
if reionization proceeds across such metal - enriched regions extending below @xmath54 , a dramatic decline in @xmath277 is expected by @xmath14 .
the _ no field _ case is barely favored by the @xmath281 data of bsrs , but @xmath282 appears closer to predictions of an ionized igm . the fact that observed @xmath1 , @xmath4 , and @xmath3 arise together with similar profile shapes suggests that each of these ionic species may be dominant , consistent with the _ no field _ assumption .
if so , the lack of consistency among the three low ions may be a result of metal yields differing those of @xcite we employ , which is one reason we provide alternative yields in table [ table : yields ] .
@xmath282 and @xmath281 show less difference between the _ no field _ and two ionizing cases than @xmath277 in figure [ fig : omega ] , because the first two species have higher ionization potentials that are not as easily ionized by the addition of a field .
overall , the low ionization species paint a scenario in which there are still significant patches of neutral igm gas at @xmath8 along at least some loss .
hence in principle one could use such ions to trace the patchiness of reionization .
we plan to investigate this possibility using simulations that more accurately track the ionization field using radiative transfer @xcite .
@xmath15 has been measured in several studies at @xmath200 , with differing results .
@xcite finds @xmath15 is consistent with remaining constant from @xmath283 , which was extended by the measurements of @xcite out to @xmath284 .
conversely , the null detections by @xcite of @xmath2 in the highest @xmath25 data at @xmath26 supports a significantly declining @xmath15 at @xmath200 .
the measurements of @xcite over the largest pathlength ( @xmath285 ) centered at @xmath286 also finds a similar @xmath15 value . exploring a range of wind models ,
od06 explained the near constancy in @xmath15 at @xmath287 by a decreasing global ionization fraction of @xmath2 counterbalancing the increasing enrichment of the igm .
therefore , the stability of @xmath15 does not indicate that the igm must be enriched at @xmath23 .
instead , the ionization fraction decreases with time for several reasons : ( i ) decreasing physical densities due to hubble expansion , and ( ii ) the increasing _ hm2001 _ ionization field strength from @xmath16 result in @xmath2 tracing higher overdensities and less of the carbon in the diffuse igm ; while ( iii ) increasing energy input into the igm primarily from galactic outflows at later times push metals to hotter temperatures where @xmath2 does a poor job of tracing carbon ( see @xcite for a simple breakdown of this evolution ) .
od06 predicted @xmath15 rises by a factor of two from @xmath288 using the _ hm2001 _ field , although the new cosmology we now use with a reduced @xmath86 significantly curtails high-@xmath17 sf resulting in a greater rise that agrees better with @xcite and @xcite .
figure [ fig : omega ] shows that @xmath15 continues this growth trend between @xmath6 , for both ionization fields we consider . for the _ hm2001 _
field , the @xmath2 ionization correction falls by a factor of 10 ( from 220 to 21 ) between @xmath6 .
this is a stunning drop and occurs because of the ionization behavior explored in
[ sec : ionbehave ] : the uniform background grows stronger while physical densities decline resulting in a greater ionization parameter allowing metals to be ionized to @xmath2 by @xmath14 .
in essence , some of the same trends that cause the @xmath2 ionization correction to increase from @xmath16 ( od06 ) , cause it to decrease from @xmath289 .
the _ bubble _ model predicts a slower evolution in @xmath15 at @xmath200 , because this field does not evolve as much as the _
hm2001 _ uniform field .
metals around galaxies are ionized by a nearly unevolving field between @xmath290 , so an increase of @xmath15 is more likely to reflect an actual increase in igm metallicity near galaxies .
from @xmath6 , @xmath15 increases by @xmath291 while @xmath292 increases by @xmath293 .
as future @xmath15 observations achieve higher redshift above 6 , we predict to see a decline corresponding to the evolution in the global metal density if the _ bubble _ field is correct .
meanwhile , the physical and ionization conditions make @xmath294 the ideal time for @xmath2 to trace metals in the igm for either field , as the ionization correction reaches a minimum at these epochs .
@xmath295 mirrors the behavior for @xmath15 for the two fields we consider , with smaller differences .
the lower ionization potential of @xmath296 causes @xmath5 to trace higher overdensities , which can track some of the metals already injected into the igm at @xmath130 with the _ hm2001 _ field .
the evolution of @xmath295 is nearly identical to that of @xmath15 for the _ bubble _
field for the reasons mentioned above .
the trends in our simulated high - ionization @xmath273 s appear consistent with those that are observed .
our predicted @xmath15 values with either ionization field over - estimates the most recent observations at @xmath29 ; however we argue that not all metals are ionized by such a field , and that the observed value falling between the extremes of the _ bubble _ field and the _ no field _ case ( i.e. no @xmath15 ) is encouraging .
while the above agreement with @xmath15 data for our ionizing field models is nice , it is worth pointing out that there are many possible systematics in the modeling that can impact predictions .
here we point out some of the ways our predictions of @xmath15 may be uncertain .
* ionization background : * besides the possibility that not all metals are ionized , there is the uncertainty about the nature of the ionization field where it exists .
the intensity of the ionization background at the @xmath176 ionization potential is unobservable and uncertain , yet is vital for the correct determination of the @xmath2 ionization correction , as we stress in [ sec : ionbehave ] .
the carbon spectral ratio of the _ bubble _ field makes @xmath15 higher than the _
hm2001 _ field ; in fact , the _ bubble _ field produces a nearly optimally high value of @xmath15 at @xmath54 . on the other hand , @xcite notes that reprocessing of radiation between 3 and 4 ryd by resonant line absorption of the @xmath171 lyman series could reduce by @xmath297@xmath298 the photons capable of ionizing @xmath176 to @xmath2 .
@xcite suggests this could be an explanation for the strong evolution they find in @xmath15 ( @xmath299 increase between @xmath26 and @xmath300 ) if @xmath171 reionization proceeds over this redshift range , increasing @xmath176 ionizing radiation , and resulting in a lower ionization correction for @xmath2 at @xmath301 .
* wind model : * while we have investigated a plausible wind model for enriching the igm , based on momentum - driven wind scalings , this may not be the only possible outflow scenario that can reproduce observations .
varying wind speeds and mass loading factors in od06 led us to some useful intuition when compared to @xmath2 absorption data : too high wind speeds overheat the igm while too low wind speeds do not sufficiently enrich filaments ; also , too high mass loading suppresses sf so much that too few metals are produced while too low mass loading results in insufficient metals in the igm .
but while constraints at @xmath302 are fairly tight , at higher @xmath17 the possibility of more exotic processes such as pop iii stars or the lack of dust may impart fundamental changes to outflows . without a more detailed physical understanding of the physics of outflow driving ,
it is difficult to ascertain the systematic uncertainties in our wind model . * mini - halos and population iii stars : * the very first stars and galaxies are not resolved in our simulation , as they likely form within mini - halos having virial temperatures below @xmath303k ( e.g. * ? ? ?
if such systems dominate the early enrichment or ionization budget , our modeling will be incomplete .
however , @xcite suggests that they do not dominate the ionization budget , and @xcite estimated that they increase the global gas metallicity by only @xmath304 , which is @xmath305 less than the global gas metallicity at @xmath54 in our simulations .
hence unless they have much higher yields than expected , it seems unlikely that metal production in mini - halos should be important at @xmath45 .
* unresolved galaxies : * for galaxies in halos with virial temperatures above @xmath303k , our simulation nearly resolves the jeans mass at @xmath306 . using the formula for jeans length from 6 of od06 , we find a jeans mass of @xmath307 at @xmath308 assuming @xmath309 and @xmath180k , which is nearly the dynamical mass limit of a resolved galaxy in our simulation .
the flattening of the stellar mass function at @xmath310 reflects the effect of the photoionization of regions containing forming galaxies ( i.e. the filtering mass ) .
for this reason , our resolution - converged fit to the stellar mass function ( equation [ eqn : m*func ] ) suggests that we are not missing a large population of lower mass galaxies in the d16n512vzw box . * yields : * another uncertain factor that could alter @xmath15 is the sne yields .
we use the yields of @xcite , which do not change much at low-@xmath60 for carbon . in our simulations
the sfr efficiency is so high in early galaxies that the average stellar metallicity in star particles reaches @xmath311 by @xmath312 , which minimizes the importance of low-@xmath60 yields .
it has been suggested , however , that patches of low - metallicity sf persist to late epochs ( e.g. * ? ? ?
* ; * ? ? ?
* ) , which if true means our simulations are over - mixing metals .
integrating the carbon yields over the various stellar imfs listed in table [ table : yields ] does indicate as much as a factor of two difference with the yields we use at the higher end .
note also that the yields calculated by @xcite for rotating stars at solar metallicity can more than double the total carbon yield versus non - rotating models .
* initial mass function : * even if metal - free stars are unimportant , it is possible that there is an overall shift towards a more top - heavy imf at high redshifts .
@xcite finds the evolution in the @xmath100-sfr calibration could be explained by an increasingly top - heavy imf from @xmath313 , which suggests very few low - mass stars if extrapolated to @xmath8 .
@xcite presents constraints indicating a top - heavy imf at early epochs that explain the frequency of galactic carbon - enhanced metal poor stars .
imf variation concerns are mitigated in large part by noting that our simulations match the observed uv lf , which indicate that they are producing roughly the correct number of massive stars .
hence to first order the amount of carbon produced will be correct . but
depending on the nature and form of an imf variation , it remains possible that the overall carbon yield per unit high - mass sf could still vary somewhat .
as an example , the @xcite yields using a contribution only from 30 - 120 @xmath314 stars produces half as much carbon as the @xcite yields ( table [ table : yields ] ) .
* cosmological parameters : * given the uncertainties on cosmological parameters in the 5-year wmap results @xcite , the measurement of @xmath86 ( @xmath315 ) imparts the greatest uncertainty in regards to high-@xmath17 sf . as mentioned in
[ sec : galmass ] , the primary difference causing the factor of two increase in stellar mass density at @xmath54 in the w - series used in dfo06 over the d - series used here is @xmath86 ( 0.90 compared to 0.83 ) .
similarly , we find @xmath316 fewer @xmath54 stars in a @xmath317 @xmath318 test run where we reduce @xmath86 from 0.83 to 0.75 holding all other parameters constant .
given the error bars on @xmath86 , a factor of two difference in sf and chemical nucleosynthesis at @xmath319 is possible , which will lead to similar uncertainty in @xmath15 if the metals are injected into the igm the same way .
this is the main reason why the d - series results in a larger increase in @xmath15 from @xmath320 compared to the published w - series predictions in od06 . to summarize , a number of possibilities exist to alter @xmath15 by a factor of a couple .
an important constraint is the number of weaker lines ( @xmath321 ) , which our models may be over - predicting given the comparison to the largest sample from @xcite that can detect such absorbers at @xmath29 ; this could be a sign that superwinds have a smaller filling factor in the igm than our simulations predict , or that more diffuse metals are less ionized ( see [ sec : masslimit ] ) .
lower carbon yields , possibly from a top - heavy imf , could potentially lower @xmath15 , while a different ionization background is more likely also to decrease @xmath15 compared to the _ bubble _ field . the determination of @xmath86 also can potentially alter @xmath15 by a factor of two .
the good agreement seen in @xmath15 evolution should be taken as an encouraging but not definitive demonstration of the validity of our models .
we present cumulative rest equivalent width ( ew ) distributions at @xmath130 and 6 for all five species in figure [ fig : ew ] .
this is intended to facilitate future observational comparisons , as well as present a more realistic assessment of the discriminatory power of high-@xmath17 metal absorber statistics from lower resolution near - ir spectrographs .
many of the trends already discussed earlier in this section are reproduced here .
the average weaker ionizing flux of the _ hm2001 _ field at @xmath54 results in more strong @xmath1 absorbers and fewer strong @xmath2 absorbers compared to the _ bubble _ field ; the opposite is true for weak absorbers .
the same trends are apparent in @xmath4 and @xmath5 , and also at @xmath130 for @xmath2 , although the _ bubble _ field produces more high ionization absorbers at all equivalent widths here .
@xmath3 best distinguishes neutral and ionized gas .
we note that ews , particularly at the high end , may be influenced by the phenomenon of turbulent broadening . in @xcite
we found turbulent broadening necessary to reproduce the observed low-@xmath17 @xmath72 @xmath322-parameter distribution .
the purported turbulence occurs far below the mass scale resolved by our simulation , and could be associated with energy injection by outflows @xcite . since high-@xmath17 absorbers are universally young and associated with recent outflows , turbulence may result in a higher ew ( for saturated lines ) , which would affect the high end ew distribution .
@xcite observed velocity differences in @xmath323 @xmath2 profiles from lensed quasar los pairs at sub - kpc scales , which indicate more turbulence than we expect in most low-@xmath17 @xmath72 absorbers .
fortunately , turbulence should not much affect the weak end , since turbulent broadening will leave ews mostly unaltered .
the other observables we have considered such as @xmath15 all deal with column densities , which should remain unaffected by turbulent broadening as long as lines are not saturated .
@xmath322-parameters are the observable most likely affected by turbulent broadening , and may explain some of the broad @xmath2 profiles seen by @xcite and @xcite , although @xcite suggests that these may be multi - component systems with smaller line widths if observed at higher resolution .
we also consider the case of a metallicity floor of @xmath153 added everywhere in the simulation box with the uniform _
ionization field at @xmath54 for @xmath2 ( long dashed cyan line in the upper panel of figure [ fig : ew ] ) .
note that for a chabrier imf and @xcite yields , obtaining this metal floor would involve turning 0.05% of baryons into stars , equivalent to all the stars formed by @xmath324 , and distributing this uniformly among all baryons , which seems implausible .
hence this scenario requires a more exotic origin , possibly resulting from very early fast winds powered by high - yield vmss .
such a scenario is similar to that assumed in semi - analytical enrichment models of @xcite & @xcite .
@xcite calculated a complete filling factor of metals from pre - galactic outflows in their most extreme cases , while @xcite found as much as 40% of the @xmath54 volume is enriched to @xmath325 primarily by sn - driven outflows from early galaxies .
the observational signature of such a metal floor is dramatic at the weak end of the @xmath2 equivalent width distribution . today
s instruments should be able to integrate down to ew=30 m to statistically distinguish between our standard simulation and a metal floor : 16.9 @xmath227 versus 5.7 @xmath227 .
we also show this scenario in the cdd plot in the bottom left of figure [ fig : cdd ] ( long dashed cyan line ) .
the highest @xmath25 observations yet at @xmath24 @xcite find no weak components , indicating that a significant fraction of the igm is not enriched far above @xmath326 .
data is currently inconclusive as to the extent of volume enrichment at these low levels , but sitting on the brightest quasar at @xmath23 for several nights on an 8 - 10 meter telescope should produce the necessary @xmath25 of @xmath327 to adequately test this scenario .
@xmath2 is by far the best species for testing such a scenario , because it is the most effective tracer of metals around mean overdensity at @xmath54 ; @xmath5 is ionized to @xmath328 at the same overdensities and the metallicity floor does affect the weak absorber frequency . finally , we note a contrast in the @xmath2 distributions at high-@xmath17 to the @xmath329 ew distribution for @xmath72 . the faint end turnover of the latter shows very little sensitivity to the volume filling factor of metals @xcite .
part of the difference is that @xmath72 is not tracing the lowest overdensities where metals reside at low-@xmath17 ; the @xmath2 contours overlap the metals at the lowest overdensities as shown in figure [ fig : ioncomp ] .
this is precisely the reason why @xmath2 holds so much potential for mapping the true extent of high-@xmath17 metals .
another observationally - accessible statistic is aligned absorber fractions .
we consider absorbers of different ions to be aligned if both _
components _ are stronger than 50 m and are within 15 @xmath163 of each other ( @xmath330 ) .
we examine the _ no field _ first in figure [ fig : align_neutral ] .
the alignment fractions are listed in the corners , where the percentage corresponds to the likelihood of finding an absorber of the species on the opposite axis aligned with the absorber species of the near axis . for the neutral case ,
the alignment percentages are above 95% in every case .
this lends support to the idea that the low ionization species observed by bsrs , which appear aligned in every case for @xmath3 observed above @xmath331 , are tracing metals where there is no ionizing flux above the lyman limit .
these systems occur far too frequently to be arising within the ism of intervening galaxies according to bsrs
. this could be evidence that patchy reionization is still occurring at these redshifts , or alternatively that dense cloudlets entrained in winds are self - shielded from the ionizing background .
the latter scenario would predict that there be high ionization species in the same systems , possibly offset , indicating multi - phase outflows as observed locally ( e.g. * ? ? ?
* ; * ? ? ?
* ) and seen in galactic - scale simulations ( e.g. * ? ? ?
* ) . therefore @xmath2 and @xmath5 could be very useful additional constraints .
the high alignment fractions are helped by the fact that these three species have similar ews for typical type ii sne yields .
aligned absorbers fall along the linear relations in each subplot of figure [ fig : align_neutral ] , where the ratios in column densities equal the ratios in type ii sne @xcite yields .
although we use a component separation criterion of @xmath330 , the alignment statistics are nearly identical when @xmath332 and @xmath333 .
this is generally the case in all alignment statistics considered here , which indicates that the species are tracing the same underlying gas .
this contrasts with , for instance , low-@xmath17 @xmath334 alignment fractions which are markedly larger at higher @xmath335 @xcite , because @xmath72 traces physically distinct gas from the @xmath44-forest @xcite . the alignment between low and high ionization species in figure [ fig : align_ionize ]
indicate many fewer aligned absorbers .
@xmath1 is aligned with @xmath2 45 - 62% of the time , while @xmath2 is found aligned 11 - 23% of the time with @xmath1 .
finding @xmath1 where @xmath2 exists should occur much less frequently than the other way around according to the visualizations in figures [ fig : snaps60 ] and [ fig : snapsevol ] ; the small filling factor of @xmath1 is more often coincident with @xmath2 , while @xmath2 can also trace the more extended metals .
this is an ideal ratio to measure the shape of the ionization background , since abundance ratios are irrelevant and the carbon spectral ratio that is measured covers a large ionization range from 11.3 to 47.9 ev .
there exist stronger @xmath2 components aligned with @xmath1 for the _ bubble _ field relative to the _ hm2001 _ , due to the greater hardness of the former field . in the one component
@xcite could measure both at @xmath336 in sdss j1030 + 0524 , they found no aligned @xmath1 . a sample size of at least a dozen strong @xmath2 components exploring aligned @xmath1 should begin to build enough statistics to constrain the shape and intensity of the ionizing background .
@xmath4 and @xmath5 alignment provides a similar opportunity as the carbon species to measure the shape of the background , and the alignment fraction is even higher than for carbon due to the lower ionization potential of @xmath296 ( 33.5 ev ) compared to @xmath176 .
bsrs actually does find aligned @xmath5 in three of their components lying close together in the @xmath337 system of sdss j0231 - 0728 .
this system indicates possible non - uniform ionization , because the very aligned profiles of @xmath1 , @xmath3 , and @xmath4 in components about 100 @xmath163 away from the aligned @xmath338 is a strong indicator of the _ no field _ case , while @xmath5 requires an ionizing field .
bsrs mentions that aligned @xmath2 is also present according to @xcite , but the low resolution of that observation makes it difficult to determine the component alignment ; we expect it to be aligned to the components exhibiting the @xmath5 .
the presence of low and high ions nearby suggests a galactic ism as might be seen in a damped ly@xmath22 system , as forwarded by bsrs .
@xmath1 is usually found with @xmath3 , but the reverse is not true if there is an ionization field above the lyman limit .
the bluest component of the @xmath337 bsrs system mentioned above is consistent with a non - detection of @xmath3 with the observed @xmath1 and @xmath4 , supporting some sort of ionization field .
finding the fraction of @xmath1 absorbers with aligned @xmath3 is one of the most sensitive indicators of the three different ionizing cases we explore .
finally , we do not show the incidence of @xmath3 aligned with @xmath2 , because this case rarely occurs under an ionizing field due to the large difference in ionization potentials .
this is apparent in sdss j1148 + 5251 as @xcite finds no @xmath2 despite the 4 @xmath3 absorbers ( bsrs ) .
overall , the alignment fractions offer another way to constrain the nature of the ionization field .
current data are in line with expectations from our models , though this is not highly constraining .
larger samples will elucidate the nature of individual absorbers , and possibly allow one to measure the carbon spectral ratio which could give clues as to the nature of ionizing sources .
finally in this section , we consider the appearance of systems in absorption line spectra . as an illustration
, we choose a series of loss through the region indicated in figure [ fig : snaps_z60_gal2 ] , which is the same region shown evolving in figure [ fig : snapsevol ] centered on the second most massive galaxy in the simulation box .
we choose two loss that go exactly through the center of this @xmath339 galaxy and show the system profiles for our five ions and our three fields in figure [ fig : sysspec_b0 ] .
the @xmath340-direction spectrum is more along the direction of the local filament , while the @xmath341-direction is nearly along the axis of the bipolar outflow .
both show extremely strong systems , but the @xmath341 spectrum is remarkably wide ( 800 @xmath163 ) owing to peculiar velocities imprinted by the enriching outflows . comparing the physical parameters of the gas in the bottom
three panels without peculiar velocities ( cyan ) and with peculiar velocities ( red ) shows that the velocity structure dominates the system profiles of both loss .
this galaxy has a maximum @xmath66 just over @xmath342 , therefore @xmath343 on each side of the @xmath341 spectrum is sensible as winds are launched with an intentional spread of velocities , plus the los is slightly offset from the outflowing axis .
region shown in figure [ fig : snapsevol ] at @xmath54 with intersecting loss , of which we show the absorption line profiles in figures [ fig : sysspec_b0 ] and [ fig : sysspec_b30 ] .
this snapshot spans 950 physical kpc , and the hubble parameter is 651 @xmath163 mpc@xmath83 . ] none of the observed systems in the data ( or in our 70 loss ) have profiles as strong and as wide as the systems in figure [ fig : sysspec_b0 ] , because not surprisingly lines of sight directly though the centers of galaxies are rare .
if we assume similar profiles arise within 3 kpc of the 10 galaxies in our simulation box over @xmath344 , then the frequency of intersecting such a system occurs once in @xmath345 ! a more relevant consideration for the spectra in figure [ fig : sysspec_b0 ] is to take the blue - shifted half and model the outflows in the spectra of lbgs at lower redshift ( @xmath346 ) , where such galactic superwinds have been observed to be common and observations are more obtainable @xcite ; this is beyond the scope of this paper .
we note that the faster moving materials absorb more in the lower ionization species with the weaker _
hm2001 _ background .
more typical examples of quasar absorption line systems are shown in figure [ fig : sysspec_b30 ] where two loss have impact parameters , @xmath322 , of 30 kpc to the same galaxy .
velocity profiles of absorbing ions are still dominated by peculiar velocities , and the los closest to the outflow axis ( @xmath341 direction ) produces a wider system .
this system is perhaps most qualitatively similar to the two wide systems observed in low ionization species by bsrs where @xmath1 , @xmath4 , and in one case @xmath3 show a total velocity spread of @xmath347 .
this supports the idea of neutral / low - ionization gas entrained in outflows , self - shielded from the ionization background ; this probably makes more sense for the @xmath337 system than patchy reionization extending to this redshift .
a similar velocity spread is observed by @xcite in his @xmath2 @xmath348 system .
although an alignment within 30 kpcs occurs once in @xmath349 ( if we count the ten most massive galaxies in the d16n256vzw box ) , we will argue in [ sec : galabs ] that such systems are qualitatively similar to systems observed around lower mass galaxies at lower impact parameters .
the frequency of intersecting such qualitatively similar systems as those shown in figure [ fig : sysspec_b30 ] is @xmath350 if we consider all galaxies @xmath351 , and @xmath352 .
two such systems should have been intersected by now given the combined pathlength of @xcite & @xcite , which compares well to three strong @xmath2 systems being observed .
three differences when scaling to lower stellar mass are ( i ) @xmath335 of systems are smaller as the outflow velocities are smaller , ( ii ) metallicities will be lower due to the mass - metallicity relation , and ( iii ) the mass loading factors should be higher at lower mass , which counterbalances ( ii ) . in figure [
fig : syswidth ] we show how median @xmath335 varies as a function of column density in our 70 los @xmath54 sample . here
@xmath335 is defined as the difference between the central velocities of the two furthest components in a system ( @xmath353 when there is a one component system ) .
one @xmath73 boundaries ( dashed lines ) show the range of velocity spreads in each case .
this can be compared to a sample of the same spectra without peculiar velocities , shown as long dashed orange lines . while this is not strictly a fair comparison as systems without peculiar velocities will not necessarily fall into the same column density bin , this nevertheless clearly illustrates that the stronger observed systems are dominated by peculiar velocities . at high column densities , greater than 50% of a system s width owes to the peculiar velocities from outflows . with future samples ,
such a plot can be used to infer conclusions about the imprints of outflows on high-@xmath17 metal - line profiles .
we do not make a direct comparison now , because of the paucity of observed systems and differences in both @xmath25 and resolution , which may affect such a comparison . ) versus column density for @xmath2 _ hm2001 _ and _ bubble _ fields ( top and middle ) , and the @xmath3 _ no field _ case ( bottom ) .
solid lines indicate the median and dashed lines show 1 @xmath73 dispersions .
@xmath335 is defined as the velocity difference between the central velocities of the two furthest components in a system .
the orange long - dashed lines show @xmath335 if we make our loss without peculiar velocities ( i.e. the spatial distribution contributes to the system width ) .
the majority of the system width is due to peculiar velocities , where outflowing metals broaden the profile . ]
the main points here are that ( i ) we produce systems qualitatively similar to those observed in the data at a frequency that seems reasonable ; and ( ii ) the velocity widths of the strongest observed systems are likely to be dominated by peculiar velocities associated with outflowing gas .
the latter suggests that plots of @xmath335 vs. column density may provide constraints on outflow kinematics from high-@xmath17 galaxies .
as our simulations form galaxies and enrich the igm together , we are able explore the relationship between galaxies and metal absorbers , along with their physical and environmental properties . specexbin tracks for each absorber the physical properties of density , temperature , and metallicity ; the environmental properties of the distance and stellar mass of the nearest galaxy ; plus the last time an sph particle was launched in a wind ( i.e. its age " ) .
we also track an sph particle s sfr if it is in a galaxy , or @xmath66 at which it was launched if in an outflow .
unless otherwise mentioned , we use the _ bubble _ model , because ( i ) the ionizing flux from the local galaxy should dominate at the distance of the metals given the reasonable assumptions for the transmission above the lyman limit , and ( ii ) this field produces a shallower column density distribution , which seems to be supported by the data .
note , however , that the _ no field _ approximation may best explain some absorbers , most notably those in the j1148 + 5251 los .
this may help remedy the overestimated _ bubble _
@xmath15 if more metals are at lower ionization states as we suggest in [ sec : masslimit ] .
we also address the physical environment of the _ hm2001 _ field , which may be most appropriate at @xmath354 , in [ sec : physenvuniform ] we concentrate on @xmath2 absorbers at @xmath54 , examining first their physical state and then their evolutionary status , and bring them together by exploring the high-@xmath17 galaxy - absorber connection .
we consider an absorber s connection to the stellar mass of its originating galaxy , which in our models is a good proxy for the more observationally - accessible sfr .
this leads us to a suggestion of possibilities of why we may produce too many weak absorbers at high-@xmath17 .
we then switch to lower ionization species represented by @xmath1 , and lastly extend our analysis back to @xmath130 .
figure [ fig : c4_z6_4panel ] shows four absorber properties plotted against each other for @xmath54 @xmath2 absorbers in the _ bubble _ field . in the upper left panel are the same type of logarithmic metallicity contours considered in figure [ fig : ioncomp ] .
@xmath2 has the smallest global ionization correction at @xmath7 , because the ionization fractions of this species are ideally aligned with the range of densities at which metals reside at this redshift ( @xmath355 ) .
histograms in the bottom left panel show that stronger absorbers trace higher overdensities .
temperatures indicate predominantly photo - ionized @xmath2 .
outflows typically heat gas to @xmath356 k around galaxies , but the gas quickly cools to warm igm temperatures due to efficient cooling in this range .
this appears to contradict our previous work in od06 , where we claimed a significant fraction of @xmath8 @xmath2 ( 43% ) is collisionally ionized , but the difference arises because in od06 we quoted a mass - weighted fraction ( from particles ) , while here we are computing a volume - weighted quantity ( from spectra ) . for @xmath357 , our present models agree well with od06 , comparing figure 9 in od06 with the upper left of figure [ fig : c4_z6_4panel ] .
regions with @xmath358 are mostly forming stars at @xmath54 , and the frequency of metal systems where @xmath359 includes sph particles above the sf threshold is @xmath360 . of the 54 components with @xmath361 ,
only 12 have any traces of sf .
ironically , only one of these 12 lines has its line center density at @xmath358 ; even when loss probe galactic environments , the @xmath2 absorbing gas is primarily intergalactic and @xmath362 ! this may be an unphysical amount of @xmath2 , but this one absorber comprises 88% of the total @xmath15 along all loss .
this example illustrates the reason for the discrepancy between the volume - derived @xmath15 of @xmath261 absorbers versus the mass - derived @xmath15 , as discussed in [ sec : omega ] . ] .
hence our models clearly predict that the vast majority of typical @xmath2 absorbers are primarily intergalactic at @xmath54 , and the rest mostly trace halo outskirts .
the average carbon metallicities at @xmath363 and @xmath364 are @xmath365 and @xmath366 @xmath367 respectively .
this is about @xmath368 the igm metallicities at the corresponding overdensities , using the simulation - averaged metallicity - density relation in figure [ fig : rho_z ] .
hence @xmath2 absorbers are tracing a clumpy , inhomogeneous distribution of metals with a very low volume filling factor , @xmath369 .
this is generic in our models , and is even seen at the present day for @xmath72 absorbers
the metallicity inhomogeneity is even more extreme at @xmath130 : @xmath370 for @xmath371 , while the simulation - averaged metallicity is @xmath372 ( a factor of @xmath373 ) .
this shows that metal - line absorbers trace an increasingly clumpy distribution of metals at early times going back from @xmath374 , a trend that extends to low-@xmath17 igm metals as traced by @xmath72 @xcite . in summary , @xmath2 absorbers seen at @xmath8
are mostly expected to be true intergalactic absorbers , not gas within galaxies or galactic halos .
galaxy outflows reach a larger comoving volume at early times ( od08 ) , but in actuality they are enriching a small fraction of the volume in a highly patchy fashion .
we now consider how an absorber s age relates to its environment , as a way of characterizing in which regions igm metals originate .
an age - density anti - correlation exists in @xmath54 @xmath2 , with a significant spread , as shown in the bottom left panel of figure [ fig : c4_z6_4panel ] .
this is the same trend we see in low-@xmath17 @xmath72 @xcite where stronger absorbers trace higher densities that have been enriched more recently . in the case of @xmath2 ,
absorbers over @xmath375 are more likely to trace metals ejected on average 100 myr ago , while weaker absorbers trace metals 100 - 300 myr ago from earlier galaxies when they were at lower mass and metallicity .
the age - density relation can be understood by considering the age - galaxy distance ( @xmath376 , shown in physical kpc ) relation in the bottom right panel of figure [ fig : c4_z6_4panel ] .
older absorbers have had time to travel greater distances reaching lower overdensities .
taking @xmath376 divided by age , we can plot the average velocity these winds need to travel to get as far as they do , which we call the effective wind speed , @xmath377 .
the gray solid line shows @xmath378 and the dashed lines on the left and right sides are 30 and 300 @xmath163 respectively . comparing this to the average wind speed at launch @xmath379 ( cf .
figure [ fig : sfr_wind ] ) , we see that outflowing gas typically slows @xmath380% by the time it reaches the igm . as argued in od08
, this slowing is primarily hydrodynamical as outflows run into a dense igm , though gravity plays a non - trivial role .
the good age-@xmath376 correlation indicates that these absorbers trace metals leaving galaxies for the first time .
this also is expected from the evolutionary picture of feedback from @xmath381 presented in od08 , where we found that metals more often than not recycle ( i.e. return to a galaxy at later time and are shot out in another wind / form into stars ) .
metals recycle on a timescale of 1 - 3 gyr reaching a median farthest distance of @xmath145 proper kpc from their parent galaxies .
@xmath2 is tracing metals on their first journey into the igm , because the ages and distances in figure [ fig : c4_z6_4panel ] are nearly all less than the recycling times and turn - around distances .
these absorbers primarily reside in the igm and not in halos , because the halo radius ( from * ? ? ?
* ) is @xmath382 physical kpc for all @xmath54 galaxies in our simulation ; nearly all absorbers are at distances above this value in figure [ fig : c4_z6_4panel ] .
hence @xmath2 absorbers are tracing igm metals on their first journey outward from galaxies at @xmath54 . as another way of considering absorber origin , we plot the ratio of the effective velocity to the launch velocity ( @xmath383 ) versus the distance from a galaxy ( left panel ) and the launch velocity ( right panel ) in figure [ fig : c4_z6_vw_vratio ] .
this ratio should not exceed one , given that wind particles are slowed as they travel outward .
a few absorbers at greater than one are mostly due to the rare instances when the neighboring galaxy is not the originating galaxy for a wind .
the clearest trend is that @xmath384 shows a general decline with @xmath376 . for absorbers with @xmath385 , the strength of most components observed by @xcite and @xcite ,
the median ratio is nearly unity at 3 - 5 kpc , 0.8 at 10 kpc , and 0.5 by 50 kpc .
note that our outflow prescription ( following * ? ? ?
* ) turns off hydrodynamic forces for a time that does not exceed that required to cover 4 physical kpc ( at @xmath54 ) at @xmath66 .
this shows that , as soon as hydro forces are turned on , outflows begin slowing down as they travel further from their parent galaxy .
several other physically interesting trends are also evident in the right panel of figure [ fig : c4_z6_vw_vratio ] .
first , stronger absorbers arise from more massive galaxies with greater @xmath66 ; more massive galaxies enrich the igm with higher metallicity outflows due to the mass - metallicity relationship at high redshift ( @xmath386 ; dfo06 ) .
second , winds from massive galaxies slow down quicker as indicated by lower @xmath383 at higher @xmath66 ; this arises because massive galaxies tend to be surrounded by denser gas ( od08 ) . while we have explored numerous relations between absorber physical and environmental parameters at @xmath54
, there is as of yet no over - riding correlation to explain @xmath2 absorber strengths , as each parameter exhibits considerable scatter versus @xmath387 .
here we argue that the _ fundamental relation _ governing metal absorbers is the stellar mass , @xmath100 , of the neighboring galaxy versus the distance to that galaxy , @xmath376 , and most other correlations are merely a consequence .
the neighboring galaxy is defined as the galaxy with the greatest impinging flux on an absorber ( i.e. the nearest galaxy weighted by the inverse square law ) .
figure [ fig : c4_z6_mgal_dgal ] shows the @xmath388 relation . as is evident , the relation between @xmath376 and @xmath387 at a given @xmath100 shows a strong correlation . vs. @xmath389 elucidates the fundamental and close connection between galaxies and absorbers at high-@xmath17 .
it reveals a strong correlation between @xmath387 and @xmath188 for a given mass galaxy .
the ionizing conditions are identical in the _ bubble _ field along the gray dotted lines , explaining much of this correlation .
evolutionary tracks ( solid gray lines for winds emanating from a @xmath390 , @xmath391 , & @xmath392 galaxy ) tracing an absorber back to its parent galaxy assuming a constant @xmath378 and secular galaxy growth . currently observable @xmath54 @xmath2 absorbers are launched from galaxies between @xmath393 , but are observed around galaxies with @xmath110 .
the color key for absorber strength in figure [ fig : c4_z6_4panel ] applies here . ]
we plot gray dotted lines corresponding to the relation @xmath394 representing the constant flux levels resulting from the _ bubble _ field due to the inverse square law .
if all physical conditions are equal , absorbers along these lines should be identical .
indeed , this seems to hold reasonably well .
hence the key deterministic relation for absorber strength is distance from galaxy of a given stellar mass , more so than any of the physical parameters explored in the previous sections .
the paucity of absorbers below the bottom - most gray dotted line is only a selection effect poorly sampling these regions in a volume - averaged measurement ; very strong absorbers exist here if loss are fortunate enough to sample them ( e.g. the extreme systems displayed in figure [ fig : sysspec_b0 ] ) .
absorbers are closely related to galaxies at high-@xmath17 moreso than any other time because ( i ) metals are on their first journey into the igm , ( ii ) they are still in close proximity to their parent galaxies , and ( iii ) galaxy evolution is relatively simple to understand at high-@xmath17 as galaxies are mostly evolving in isolation .
we use these assumptions to construct the solid gray `` evolutionary tracks '' that metals take into the igm , assuming constant @xmath395 and an exponentially increasing @xmath100 with time , which approximately falls out of the integration of equation [ eqn : sfr_m * ] .
although both of these assumptions are not exactly true , this treatment reveals some important behaviors .
first , `` isochrones '' travel parallel to the x - axis of figure [ fig : c4_z6_mgal_dgal ] by construction if @xmath377 is assumed to be independent of @xmath389 , meaning the absorbers of the same age are found at a similar distance .
this of course an oversimplification , because @xmath377 slows down as time passes and winds with higher @xmath66 slows down faster .
however , this explains why there are very few weak absorbers ( @xmath396 ) around @xmath128 galaxies ; metals simply have not had time to travel far enough to the low - density regions where weak @xmath2 absorbers would arise .
second , tracing the evolutionary tracks back to the bottom reveals the mass of the launch galaxy for each metal absorber .
dfo06 plotted the mass - metallicity relationship of @xmath54 and @xmath130 galaxies , finding that gas metallicity increases by @xmath397 dex per decade of @xmath100 .
although the over - riding trend for absorber strength depends on the ionizing flux in the _ bubble _ field , a more subtle trend ( hardly visible ) is that absorbers grow stronger rightward along the same dotted gray line primarily as a result of the mass - metallicity relationship .
however , this effect is about half as strong as expected ( @xmath398 dex ) , and should be counter - balanced by increased mass loading factors of smaller galaxies . also , consider that @xmath377 is not really constant , which limits us from over - interpreting this plot .
the only candidate observed high-@xmath17 galaxy-@xmath2 absorber connection is along the j1030 + 0524 los , where @xcite notes the observed systems at @xmath399 may be related to an excess of @xmath36-band dropouts observed by @xcite along this quasar los .
in our models , most absorbers between @xmath261 are associated with galaxies where @xmath110 , which have corresponding 1350 ab magnitudes ranging between @xmath400 and @xmath401 ( see the luminosity range bounded by green dotted lines in figure [ fig : lumfunc ] ) , and sfrs from @xmath402 . the closest of the j1030 absorbers lies about 50 arcseconds away from a galaxy , i.e. an impact parameter of 300 kpc , which is too far for our winds to reach ( e.g. @xmath403 for 300 myr ) ; plus the magnitudes of these galaxies appear to be brighter than any of the galaxies in the d16n256vzw simulation box .
this may be an extended overdense region of space where smaller , currently unobservable galaxies are primarily responsible for the enrichment .
the uniform _
_ background case may be more appropriate below @xmath54 , and actually better explains the observed frequency of strong @xmath2 absorbers at @xmath250 , although it is more discrepant in over - estimating weaker absorbers .
we show the galaxy - absorber connection plot in figure [ fig : c4_z6_mgal_dgal_uniform ] for this field and note the following differences .
first , the distance and mass of the neighboring galaxy matters less for the characteristics of these absorbers ; the dotted lines do not have any meaning , but are shown for relative contrasts with figure [ fig : c4_z6_mgal_dgal ] . as this field is weaker near galaxies , there are fewer strong absorbers because carbon is in lower ionization states .
for the same reason , there are more weak @xmath2 absorbers at greater distances from galaxies , because these metals are otherwise ionized to @xmath404 by the _ bubble _ field . for te _ bubble
_ model . here
the gray dotted lines do not have any significance , but are shown for reference .
fewer strong absorbers exist near galaxies and more weak absorbers exist far from galaxies due to the weaker ionizing field in such environments . one important difference with the _ bubble _ field is that strong absorbers are only associated with @xmath405 galaxies , and these absorbers trace enriched low - density regions further from galaxies .
the color key for absorber strength in figure [ fig : c4_z6_4panel ] applies here . ]
a potentially accessible distinguishing characteristic of the two fields involves components with @xmath406 , which are only found around @xmath405 galaxies in the uniform case .
these absorbers arise in the situations like those illustrated in figure [ fig : snaps_z60_gal2 ] , where a bipolar outflow copiously enriches the low density igm outside of filaments where the _
@xmath2 ionization correction is lowest .
the strongest absorbers actually arise from overdensities less than 10 , which is the opposite of the behavior in the _ bubble _ model .
both our backgrounds produce a significant amount of weak absorbers ( @xmath407 ) , contributing non - trivially to @xmath15 ( 30% in the _ bubble _ case and 70% in the _
hm2001 _ case ) .
the two observational @xmath54 @xmath2 cdd constraints shown in figure [ fig : cdd ] ( bottom right ) suggest a possibly shallower distribution than either of our models produce , with the @xmath2 mass density coming almost exclusively from strong systems .
the @xcite data point is a conservative upper limit , and already provides a powerful constraint that we interpret as indicating a significant volume of the igm is not enriched far above @xmath408 .
however , if we consider that they observe no @xmath2 absorbers between @xmath371 and assume their completeness is 75% over this column density range ( this seems reasonably conservative using their figure 3 ) , the difference with our models is even more stark .
we predict 3.3 and 3.9 such absorbers should be detected at @xmath409 for the _ bubble _ and _ hm2001 _ fields respectively over the @xcite adjusted pathlength of @xmath410 .
a possible explanation is that the small galaxies in our simulation do not produce @xmath2 absorbers for some reason .
the evolutionary tracks in figure [ fig : c4_z6_mgal_dgal ] can be used as a stellar mass cut for which galaxies produce @xmath2 systems .
for example , if the originating @xmath100 must be @xmath411 , then only absorbers right of the middle evolutionary track contribute . systems below @xmath412 are more than halved , while leaving the stronger end unaffected if applying this cut .
the power law slope now approaches @xmath413 ( i.e. equal numbers of absorbers in low and high density cdd bins ) .
why would low - mass haloes not produce @xmath2 absorbers ?
a possible explanation is that the highly mass - loaded outflows from small galaxies self - shield metals leading to them being observed in lower ionization states .
thus the over - excess of @xmath2 absorbers and @xmath15 in either ionization background may be counter - balanced by the bsrs low - ionization systems , which are underestimated when applying these backgrounds to all metals .
a second possibility for us overestimating weak @xmath2 absorbers is that our outflows are traveling too far into the igm .
for example , there exist few very weak , @xmath414 , absorbers around galaxies @xmath415 in figure [ fig : c4_z6_mgal_dgal ] because winds have not had time to enrich regions at @xmath416 kpc .
if wind speeds are even lower , metals will populate less of the @xmath417 phase space at large @xmath188 , where they otherwise would produce weaker @xmath2 absorbers .
our discussion in this section on the galaxy - absorber connection may be currently inaccessible to observations and somewhat speculative , as only a handful of absorbers at @xmath8 are known thus far and identifying the nearest galaxy may have to await a next - generation facility .
our purpose is to emphasize the @xmath388 relation as the key one for understanding the origin and subsequent evolution of high-@xmath17 metal - line absorbers .
as these early galaxies push a majority of their metals into the igm , the metals are still close enough that the high-@xmath17 galaxy - absorber connection tightly establishes the absorber properties .
this is a fundamental prediction of our model , and if proved correct , offers a novel way to probe feedback processes in the reionization epoch .
we now examine the physical conditions of @xmath1 absorbers , as a representative of lower ionization species . in [ sec : ionbehave ] we showed @xmath1 traces higher overdensities closer to galaxies ; therefore we might expect this species to trace metal - enriched outflows at an earlier stage .
this turns out to be only partially true . in figure
[ fig : c2_z6 ] we plot @xmath1 absorbers in @xmath174 phase space demonstrating that @xmath1 absorbers do not trace the same extent of this phase space as @xmath2 .
our visualization in figure [ fig : snaps60 ] shows @xmath1 tracing less volume than @xmath2 .
another difference is there is a stronger correlation with density for @xmath1 , much more so than for @xmath2 .
the reason for this is that the @xmath1 ionization fraction falls off rapidly toward lower overdensities as shown in figure [ fig : ioncomp ] while the @xmath2 ionization fraction peaks at overdensities corresponding to intergalactic metals .
@xmath1 absorption strength depends much more on the physical state ( density ) than @xmath2 , which depends more on its environmental state ( proximity to and mass & metallicity of its parent galaxy ) .
the ionization field has little effect here , because density is the primary determinant for @xmath1 .
@xmath1 tracing higher overdensities translates into this species tracing regions closer to the launch galaxy , as the range in @xmath188 in figure [ fig : c2_z6 ] is less than that for @xmath2 .
@xmath282 arises from absorbers at the same ages traced by the @xmath2 absorbers ( compare y - axis histograms in the center panel of figure [ fig : c2_z6 ] and bottom left panel of [ fig : c4_z6_4panel ] ) .
@xmath1 is tracing older metals that remain closer to their parent galaxies by virtue of either lower @xmath66 or more hydrodynamic slowing .
the primary reason is the lower @xmath66 : the average @xmath66 of a @xmath1 absorber is 150 @xmath163 versus 250 @xmath163 , while the @xmath383 ratio is not different than that for @xmath2 ; hence @xmath1 is arising from slower winds around less massive galaxies . despite @xmath1 tracing a different set of winds , 62% of @xmath1 absorbers over 50 m are aligned with @xmath2 in the _ bubble _ field .
rare strong @xmath1 absorbers trace some of the youngest outflows ( @xmath418 myrs ) , which are sometimes still within the galactic halos of @xmath54 galaxies . in the right panel of figure [
fig : c2_z6 ] , @xmath1 strength is determined more plainly by @xmath188 independent of @xmath100 , which is a result of this ion s greater dependence on density compared to @xmath2 .
the galaxy - absorber connection is _ not _ as important for @xmath1 absorber strength .
if @xmath419 is mostly independent of @xmath100 , the selection effects of loss probing a random volume finds the average @xmath1 absorber around a smaller galaxy .
the median @xmath389 of the @xmath420 histogram in the bottom of this panel is @xmath421 , corresponding to a sfr of 0.03 @xmath422 .
@xmath1 is probing a different set of metals than @xmath2 , which itself is tracing faster winds arising from more massive galaxies ( @xmath423 ) enriching a larger volume .
we are already delving into the hypothetical here by considering the physical and environmental parameters of absorbers and galaxies not even discovered yet , so we take it one step further by considering what @xmath2 absorbers trace at @xmath130 .
if our suggestion that @xmath2 absorbers are tracing metals on their first journey into the igm , metals should be observed even closer to galaxies at higher redshift .
this indeed is the case in figure [ fig : c4_z8 ] for @xmath130 absorbers ; the absorbers occupy slightly higher overdensity ( left panel ) than at @xmath54 due to winds not reaching mean overdensities yet , while absorbers are closer to their parent galaxies as a consequence of having less time to travel ( middle panel ) .
these absorbers have @xmath377 more similar to @xmath66 signifying these outflows are just leaving their galaxy .
in fact , the ratio @xmath383 for the absorbers between with @xmath385 for @xmath424 kpc is 0.73 at @xmath130 versus 0.62 at @xmath54 .
such absorbers at @xmath130 trace outflowing gas more directly , whereas at @xmath54 some absorbers at this distance trace outflows that have already stalled or are returning to galaxies .
the right panel displays our galaxy - absorber connection plot with dotted lines corresponding to constant _ bubble _ ionization flux , while gray lines are evolutionary tracks assuming @xmath425 and the sfr is doubled from equation [ eqn : sfr_m * ] as in dfo6 at @xmath130 versus @xmath54 .
the more rapid assembly of galaxies at this epoch is noticeable by greater curvature in these evolutionary tracks ; more galactic evolution likely has occurred in an absorber s associated galaxy despite the younger age of the universe .
the mass - metallicity holds at this redshift ( dfo06 ) and appears to play a role in the increasing @xmath387 for more massive galaxies .
we explore high redshift metal - line absorbers using a state - of - the - art , @xmath426 particle gadget-2 cosmological simulation enriching the igm with star formation - driven galactic outflows .
our galactic outflows follow the predicted relations of momentum - driven winds , where outflow velocity scales with galaxy velocity dispersion and the mass loading factor scales inversely , leading to higher mass loading from small galaxies . our purpose is to reproduce igm metal - line observations at @xmath7 while providing predictions for future observations out to @xmath130 , as well as to investigate the physical nature of high-@xmath17 metal absorbers and its connection to galaxies .
we explore five species ( @xmath1 , @xmath2 , @xmath3 , @xmath4 , & @xmath5 ) using three different ionizing backgrounds : fully neutral , fully reionized , and a patchy bubble model based on the flux of the nearest galaxy .
our key results and predictions are as follows : + ( i ) metal lines already discovered at @xmath48 are primarily tracing the diffuse igm , not gas in galaxies .
some @xmath2 lines already discovered could be tracing metals at overdensities not much higher than the mean density of the universe .
\(ii ) we can reproduce the observed trends of @xmath15 at @xmath427 with steady enrichment of the igm filling only 1% of the igm with metals by @xmath14 .
the recent samples of @xcite and @xcite that find lower @xmath15s suggests that the filling factor could be even less .
we also propose that patchy ionization conditions where some metals are ionized by their local galaxy and others are not ionized may best explain the observed trends in both @xmath2 and lower ionization absorbers .
the determination of @xmath15 is also highly dependent on the shape of the ionizing field ( especially above 47.9 ev ) , sne yields , the @xmath86 parameter , and the number of @xmath321 absorbers .
\(iii ) the global gas - phase metallicity increases by a factor of @xmath428 from @xmath6 .
the evolution of @xmath15 generally reflects this change , but the exact amount of evolution depends on the form of the ionization background .
the uniform @xcite evolving background results in an increase in @xmath15 of @xmath429 , while the _ bubble _ field shows only a @xmath291 increase reflecting the actual increase in gas - phase metals .
the greater ionization intensity around 47.9 ev in the latter background , which is a result of the dust attenuation law we apply along with the proximity of metals to galaxies , results in @xmath195 greater @xmath15 at @xmath130 .
\(iv ) @xmath2 is the ideal tracer for metals at low overdensities at @xmath54 , and its detection holds the best option at high-@xmath17 to determine the volume filling factor of metals .
@xmath2 has the lowest global ionization correction at @xmath7 ( i.e. @xmath430c@xmath431 is at its maximum ) . adding a metallicity floor of @xmath153 will have dramatic effects on the numbers of weak @xmath2 absorbers , just below thresholds of currently published results . integrating for several nights on a single bright @xmath23 quasar could determine if a significant portion of the igm volume is enriched to this level .
the results of @xcite already indicate that a significant volume of the igm is unlikely to be enriched much above @xmath153 , and possibly there exists too many weak @xmath2 absorbers in our model without this floor .
\(v ) the velocity profiles of systems are dominated by the peculiar velocities of outflowing gas .
we reproduce the large @xmath335 s ( @xmath432 ) observed in strong @xmath3 and @xmath2 systems .
the hubble expansion of spatially distributed metals alone is far too small to create these observed system profiles .
\(vi ) the galaxies responsible for the enrichment of the igm have stellar masses in the range of @xmath433 and corresponding rest - frame uv magnitudes of @xmath400 and @xmath401 ( 1350 ab ) when considering @xmath2 , and a factor of ten less when considering @xmath1 .
most absorbers should still be near their originating galaxies ( 5 - 50 physical kpc for @xmath54 @xmath2 and 2 - 20 kpc for @xmath1 absorbers ) .
we stress the need to identify such galaxies , which will require the likes of jwst and 30-meter telescope facilities , in order to study the high-@xmath17 galaxy - absorber connection .
identifying the environments of strong @xmath2 components ( @xmath434 ) can constrain the ionizing source : within @xmath435 kpc of @xmath436 galaxies for the _ bubble _ field , and at @xmath437 kpc around @xmath438 galaxies for the _ hm2001 _ field .
\(vii ) the abundance of @xmath3 absorbers observed by @xcite in the sdss j1148 + 5251 sight line is consistent with metals residing in a neutral igm ( as also argued by those authors ) .
we can not produce such a frequency of @xmath3 within our models with ionization extending above the lyman limit .
the excellent alignment with @xmath1 and @xmath4 in this dataset also strongly supports this .
the fact that most sight lines show much less @xmath3 suggests the possibility that reionization is still inhomogeneous over large scales at @xmath38 .
it is unlikely to be an effect of inhomogeneous enrichment , since smaller galaxies with relatively low clustering lengths are primarily responsible for enriching the igm at these epochs .
another possibility for broad @xmath3 systems observed down to @xmath439 are metals entrained in outflows , which remain shielded from the ionization background as they travel into the igm .
\(viii ) aligned absorber ratios are a powerful tool to determine the physical parameters of the metal - enriched igm and the nature of the ionization field .
aligned absorbers with both components above 50 m are rare at @xmath54 , unless the igm is neutral .
this makes the @xcite finding of frequent low - ionization aligned absorbers all the more supportive of either a neutral ( probably recombined ) portion of the igm , or metals entrained in self - shielded outflows as the travel into the igm .
+ our analysis in od08 indicated that most metals injected into the igm recycle back into galaxies on a typical timescale of @xmath440 gyrs . hence at @xmath48 , metals should primarily be on the outward leg of their first journey into the igm .
this indeed turns out to be the case when we look in detail at the evolutionary state of absorbers .
this has consequences for the galaxy - absorber connection , in that this relationship is closest at high-@xmath17 because metals are still in relatively close proximity of their parent galaxies .
absorbers trace gas on an outward trajectory still holding signatures of their galactic outflow origin , while possibly being primarily irradiated by their parent galaxy s radiation field .
the properties of @xmath2 absorbers depend most on their evolutionary relation to their parent galaxy ( proximity , mass , & metallicity ) , while lower ionization species depend more on the physical state of the gas ( density , primarily ) .
this is a result of the different @xmath174 phase space traced by these ions .
another tenet of od08 is that metals travel a relatively constant physical distance ( @xmath441 kpc ) from their parent galaxies , meaning early outflows have the greatest chance of enriching a large volume of the universe .
however this journey takes significant time as wind outflows travel at an average of @xmath442 taking @xmath443 gyr to get to this distance , or more than the age of any stars at @xmath54 .
therefore , the volume of the universe enriched to significant levels ( i.e. 0.01 - 0.1 @xmath367 ) grows by an order of magnitude over the relatively brief time ( 575 myr ) covered between @xmath6 .
this era can be called the `` age of volume enrichment .
'' however metals do not exhibit the expansionist zeal of the mongol empire when it comes to enriching the igm ; the vast majority of the igm ( as much as 90% ) still remains unenriched in our simulations run to @xmath162 @xcite .
the outlook for the study of very high-@xmath17 metal lines in the igm looks promising as long as sources ( quasars and grbs ) are discovered to provide light bulbs for the near - ir spectrometers that will be on - line in the near future .
relatively low resolution and @xmath25 observations can provide key constraints as to the filling factors of metals , the velocities and mass loading factors of the winds that put the metals there , and of course the ionization state of the igm .
in contrast to metal lines between @xmath16 , we expect significant evolution in metal - line properties between @xmath6 , a prediction that may have already been observed by @xcite and @xcite . further down
the line is the study of the intriguing high-@xmath17 galaxy - absorber connection , as this relation is closer than at lower redshift .
the jwst and 20 - 30 meter class telescopes are required to study this realm as galaxies at @xmath444 with sfrs below 0.5 @xmath422 , which are primarily responsible for most igm metal lines at @xmath445 . to sum it all up ,
absorption line spectroscopy applied to the early universe should yield significant insights into early galaxy formation , and whether such galaxies enrich the igm in a manner similar to lower redshift systems , or in some exotic fashions not considered here .
we suspect the former .
very useful and encouraging conversations with george becker , max pettini , emma ryan - weber , and rob simcoe were fundamental to this paper . we also thank them for reading over this paper and providing important improvements .
we much appreciate the referee s comments , which added significantly to the paper .
we additionally thank karl gordon , david fanning , and berkeley zych .
the simulations were run on the intel 64 linux cluster abe supercluster at the national center for supercomputing applications .
the authors thank r. kennicutt for gracious hospitality at the institute of astronomy in cambridge , england where much of the work on this paper was done .
support for this work was provided by nasa through grant number hst - ar-10946 from the space telescope science institute , which is operated by aura , inc . under nasa contract nas5 - 26555 .
support for this work , part of the spitzer space telescope theoretical research program , was also provided by nasa through a contract issued by the jet propulsion laboratory , california institute of technology under a contract with nasa .
support was also provided by the national science foundation through grant number dms-0619881 .
abel , t. , bryan , g. l. , & norman , m. l. , 2002 , science , 295 , 93 adelberger , k. l. , steidel , c. c. , shapley , a. e. , & pettini , m. 2003 , apj , 584 , 45 adelberger k. l. , shapley a. e. , steidel c. c. , pettini m. , erb d. k. , & reddy n. a. 2005 , apj , 629 , 636 aguirre a. , schaye j. , kim t. , theuns t. , rauch m. , & sargent w. l. w. , 2004 , apj , 602 aguirre , a. , dow - hygelund , c. , schaye , j. , & theuns , t. 2008 , submitted to apj asplund m. , grevesse n. , & sauval a. j. , 2005 , in barnes t. g. iii , bash f. n. , eds , asp conf .
vol . 336 .
cosmic abundances as records of stellar evolution and nucleosynthesis .
soc . pac .
, san francisco , p. 25
barth , a. j. , martini , p. , nelson , c. h. , & ho , l. c. 2003 , apj , 594 , l95 becker .
g. d. , sargent , w. l. w. , rauch , m. , & simcoe , r. 2006 , apj , 640 , 69 ( bsrs ) becker g. d. , rauch m. , & sargent w. l. w. 2007 , apj , 662 , 72 becker g. d. , rauch m. , & sargent w. l. w. 2009 , astro - ph / arxiv:0812.2856 , submitted to apj bouwens r. , illingworth g. , blakeslee j. , & franx m. , 2006 , apj , 653 , 53 bouwens , r. j. , illingworth , g. d. , franx , m. , & ford , h. 2007 , apj , 670 , 928 bouwens , r. j. , illingworth , g. d. , franx , m. , & ford , h. 2008 , apj , 686 , 230 bromm v. & larson r. b. 2004 , ara&a , 42 , 79 bruzual , g. & charlot , s. 2003 , mnras , 344 , 1000 bunker , a. j. , stanway , e. r. , ellis , r. s. , mcmahon , r. g. 2004 , mnras , 355 , 374 calzetti , d. , armus , l. , bohlin , r. c. , kinney , a. l. , koornneef , j. , & storchi - bergmann , t. 2000 , apj , 533 , 682 cen , r. , nagamine , k. , & ostriker , j. p. 2005 , apj , 635 , 86 cen , r. & riquelme , m. a. 2008 , apj , 674 , 644 chabrier g. , 2003 , pasp , 115 , 763 chieffi , a. & limongi , m. 2004 , apj , 608 , 405 dalla vecchia , c. & schaye , j. 2008 , mnras , 387 , 1431 dav , r. , hernquist , l. , weinberg , d. h. , & katz , n. 1997 , apj , 477 , 21 dav , r. , finlator , k. , & oppenheimer , b. d. 2006 , mnras , 370 , 273 ( dfo06 ) dav , r. & oppenheimer , b. d. 2006 , mnras , 374 , 427 dav , r. 2008 , mnras , 385 , 147 dav , r. , oppenheimer , b. d. , & sivanandam , s. 2008 , mnras , submitted dickinson , m. , et al .
2004 , apj , 600 , l99 eikenberry , s. , et al .
2006 , spie , 6269 , 188 fan , x. , et al .
2001 , aj , 122 , 2833 fan , x. , narayanan , v. k. , strauss , m. a. , white , r. l. , becker , r. h. , pentericci , l. , & rix , h .- w .
2002 , aj , 123 , 1247 fan , x. , et al .
2001 , aj , 125 , 1649 fan , x. , et al .
2006 , aj , 132 , 117 ferland , g. j. , korista , k. t. , verner , d. a. , ferguson , j. w. , kingdon , j. b. , & verner , e. m. 1998 , pasp , 110 , 761 finlator , k. , dav , r. , papovich , c. , & hernquist , l. 2006 , apj , 639 , 672 finlator , k. , dav , r. , & oppenheimer , b. d. 2007 , mnras , 376 , 1861 finlator , k. & dav , r. 2008 , mnras , 385 , 2181 finlator , k. ozel , f. , & dav , r. 2008 , astro - ph / arxiv:0808:3578 , submitted to mnras finlator , k , ozel , f. , dav , r. , & oppenheimer , b. d. 2009 , to be submitted to mnras fontana , a. , poli , f. , menci , n. , nonino , m. , giallongo , e. , cristiani , s. , & dodorico , s. 2003 , apj , 587 , 544 fujita , a. , martin , c. l. , mac low , m .-
m . , new , k. c. b , & weaver , r. 2008 , astro - ph / arxiv:0803.2892 , submitted to apj furlanetto , s. r. & loeb , a. 2003 , apj , 588 , 18 furlanetto , s. & oh , p. 2005 ,
mnras , 363 , 1031 gnedin , n. y. 2000 , apj , 535 , 530 greif , t. h. , johnson , j. l. , bromm , v. , & klessen , r. s. 2007 , apj , 670 , 1 gunn , j. e. & peterson , b. a. 1965 , apj , 142 , 1633 haardt , f. & madau , p. 2001 , in clusters of galaxies and the high redshift universe observed in x - rays , recent results of xmm - newton and chandra " , xxxvith rencontres de moriond , eds .
neumann & j.t.t .
heger , a. & woosley , s. e. 2002 , apj , 567 , 532 heger , a. & woosley , s. e. 2008 , submitted to apj hinshaw , g. 2008 astro - ph / arxiv0803.0732 hirschi , r. 2007 , a&a , 461 , 571 hopkins , a. m. 2004 , apj , 615 , 209 iliev , i. t. , mellema , g . , pen , u .- l . ,
merz , h. , shapiro , p. r. , & alvarez , m. a. 2006 , mnras , 369 , 1625 iliev , i. t. , mellema , g . , shapiro , p. r. , & pen , u .- l .
2007 , mnras , 376 , 534 kawai n. et al . , 2006 , nat , 440 , 184 kennicutt , r. c. 1998 , apj , 498 , 541 kere , d. , katz , n. , weinberg , d. h. , & dav , r. 2005 , mnras , 363 , 2 kere , d. , katz , n. , fardal , m. , dav , r. , & weinberg , d. h. 2008 , astro - ph / arxiv0809.1430 , submitted to mnras kim , s. et al . , 2008 , astro - ph / arxiv:0805.1412 , submitted to apj kneib , j .-
p . , ellis , r. s. , santos , m. r. , & richard , j. 2004 , apj , 568 , l75 lidz , a. , mcquinn , m. , zaldarriaga , m. , hernquist , l. & dutta , s. 2007 , apj 670 , 39 madau , p. , ferrara , a. , & rees , m. j. 2001 , apj , 555 , 92 madau , p. & haardt , f. 2009 , astro - ph / arxiv:0812.0824 , submitted to apj martin , c. l. 2005a , apj , 621 , 227 martin , c. l. 2005b , in asp conf .
331 , extra - planar gas , ed .
r. brown ( san francisco : asp ) , 305 mo , h. j. , mao , s. , & white , s. d. m. 1998 , mnras , 295 , 319 murray , n. , quatert , e. , & thompson , t. a. 2005 , apj , 618 , 569 navarro , j. f. , frenk , c. s. , & white , s. d. m. 1997 , apj , 490 , 493 oh , s.
mnras , 336 , 1021 oppenheimer , b. d. & dav , r. a. 2006 , mnras , 373 , 1265 ( od06 ) oppenheimer , b. d. , dav , r. , & finlator , k. 2007 eas , 24 , 157 oppenheimer , b. d. & dav , r. a. 2008 , mnras , 387 , 587 ( od08 ) oppenheimer , b. d. & dav , r. a. 2009 , submitted to mnras osterbrock , d. e. 1989 , astrophysics of gaseous nebulae and active galactic nuclei .
university science books , mill valley , p. 42
pettini m. , shapley a. e. , steidel c. c. , cuby j .- g . , dickinson m. , moorwood a. f. m. , adelberger k. l. , & giavalisco m. 2001 , apj , 554 , 981 pettini , m. , madau , p. , bolte , m. , prochaska , j. x. , ellison , s. l. , & fan , x. 2003 , apj , 594 , 695 porciani , c. & madau , p. 2005 , apj , 625 , l43 portinari , l. , chiosi , c. , & bressan , a. 1998 , a&a , 334 505 rauch , m. , sargent , w. l. w. , & barlow , t. a. 2001 , apj , 554 , 823 rupke , d. s. , veilleux , s. , & sanders , d. b. 2005 , apjs , 160 , 115 ryan - weber , e. v. , pettini , m. , & madau , p. 2006
, mnras , 371 , l78 ryan - weber , e. v. , pettini , m. , madau , p. , & zych , b. 2009 , astro - ph / arxiv:0902.1991 , accepted to mnras scannapieco , e. , ferrara , a. , & madau , p. 2002 , apj , 574 , 590 schaye , j. , aguirre , a. , kim , t .- s .
, theuns , t. , rauch , m. , & sargent , w.l.w .
2003 , apj , 596 , 768 shapley , a. e. , steidel , c. c. , pettini , m. , & adelberger , k. l. 2003 , apj , 588 , 65 shapley a. e. , steidel c. c. , pettini m. , adelberger k. l. , & erb d. k. , 2006 , apj , 651 , 688 simcoe , r. a. , sargent , w. l. w. , & rauch , m. 2004 , apj , 578 , 737 simcoe , r. a. 2006 , apj , 653 , 977 songaila , a. 2001 , apj , 561 , l153 songaila , a. 2005 , aj , 130 , 1996 springel , v. & hernquist , l. 2003 , mnras , 339 , 289 springel , v. & hernquist , l. 2003 , mnras , 339 , 312 springel , v. 2005 , mnras , 364 , 1105 storrie - lombardi , l. j. , mcmahon , r. g. , & irwin , m. j. 1996 , mnras , 283 , l79 strickland , d. k. , heckman , t. m. , weaver , k. a. , hoopes , c. g. , & dahlem , m. 2002 , apj , 568 , 689 tegmark , m. et al . , 2004 , phys . rev .
d , 69 , 103501 thom , c. & chen , h .- w .
2008a , submitted to apj thompson , r. i. , eisenstein , d. , fan , x. , dickinson , m. , illingworth , g. , kennicutt , r. c. 2006 , apj , 647 , 787 thoul a. a. & weinberg d. h. 1996 , apj , 465 , 608 tornatore , l. , ferrara , a. , schneider , r. 2007 , mnras , 382 , 945 tripp , t. m. , semback , k. r. , bowen , d. v. , savage , b. d. , jenkins , e. b. , lehner , n. , & richter , p. 2008 , apj , accepted to apjs tumlinson , j. 2006 , apj , 641 , 1 vijh , u. p. , witt a. n. , & gordon k. d. 2003 , apj , 587 , 533 weingartner , j. c. , & draine , b. t. 2001 , apj , 548 , 296 wise , j. h. & abel , t. 2008 , apj , 685 , 40 woosley , s. e. & weaver , t. a. 1995 , apjs , 101 , 181 wright , e. 2006 , pasp , 118 , 1711 yan , h. , & windhorst , r. a. 2004 , apj , 612 , l93
|
intergalactic medium ( igm ) metal absorption lines observed in @xmath0 spectra offer the opportunity to probe early feedback processes , the nature of enriching sources , and the topology of reionization .
we run high - resolution cosmological simulations including galactic outflows to study the observability and physical properties of 5 ions ( @xmath1 , @xmath2 , @xmath3 , @xmath4 , @xmath5 ) in absorption between @xmath6 .
we apply three cases for ionization conditions : fully neutral , fully reionized , and a patchy model based on the flux from the nearest galaxy .
we find that our simulations can broadly fit available @xmath7 igm metal - line data , although all observations can not be accommodated with a single ionization condition .
variations in @xmath3 absorbers among sight lines seen by becker et al .
( 2006 ) suggest significant neutral igm patches down to @xmath8 .
strong @xmath2 absorbers at @xmath9 may be the result of ionization by the galaxy responsible for that enrichment , although the identification of the neighboring galaxy will have to wait to confirm this .
our outflows have typical speeds of @xmath10 km / s and mass loading factors of @xmath11 . such high mass loading is critical for enriching the igm to the observed levels while sufficiently curtailing early star formation to match the observed rest - frame uv luminosity function .
the volume filling factor of metals increases during this epoch , but only reaches @xmath12 for @xmath13 by @xmath14 .
detectable absorbers generally trace inhomogeneously - distributed metals residing outside of galactic halos .
@xmath2 is an ideal tracer of igm metals at @xmath7 , with dropping global ionization fractions to either higher or lower redshifts .
this results in a strongly increasing global @xmath2 mass density ( @xmath15 ) from @xmath6 , in contrast to its relative constancy from @xmath16 .
our simulations do not support widespread early igm enrichment from e.g. population iii stars , as this would overpredict the numbers of weak @xmath2 absorbers in the latest data .
high-@xmath17 absorbers arise from metals mostly on their first outward journey , at distances 5 - 50 physical kpc , and often exhibit broad profiles ( @xmath18 ) as a result of outflowing peculiar velocities in the strongest systems .
the galaxies responsible for early igm enrichment have typical stellar masses of @xmath19 , and star formation rates @xmath20yr .
future facilities will be able to study the high-@xmath17 galaxy - absorber connection in detail , revealing a wealth of information about feedback processes in the reionization epoch .
[ firstpage ] intergalactic medium , galaxies : formation , galaxies : high - redshift , early universe , cosmology : theory , methods : numerical
|
it is postulated that in the collision of two nuclei at high - energy , color strings are formed between projectile and target partons .
these color strings decay into additional strings via @xmath3 production , and ultimately hadronize to produce the observed hadrons.@xcite in the collision process , it is possible for partons from different nucleons to begin overlapping in the plane transverse to the direction of travel of the colliding nuclei .
this overlap probability increases with increasing energy and atomic number of the colliding nuclei.@xcite it is this overlap that leads to cluster formation in the transverse plane , similar to the disk model of 2-d percolation theory.@xcite at a critical cluster density , a cluster will form that spans the entire system .
this is referred to as the maximal cluster and marks the onset of the percolation threshold .
the quantity @xmath1 , the percolation density parameter , can be used to describe overall cluster density .
it can be expressed as @xmath4 with n the number of strings , s the total nuclear overlap area , and @xmath5 the transverse area of the discs / strings . at some critical value of @xmath6 ,
the percolation threshold is reached .
@xmath7 is referred to as the critical percolation density parameter . in two dimensions , for a uniform string density , and in the continuum limit , @xmath7 = 1.175.@xcite in terms of phenomena that are considered `` localized '' within a percolating system , such as j/@xmath8 suppression , the critical value of @xmath7 is predicted to be 1.72.@xcite + to calculate the percolation parameter , @xmath1 , a parameterization of pp events at 200 gev is used to compute the @xmath9 distribution @xmath10 where a , @xmath11 , and n are parameters fit to the data .
this parameterization can be used for nucleus - nucleus collisions if one takes into account the percolation of strings by@xcite @xmath12 in pp collisions at 200 gev , the quantity @xmath13 , due to the low probability of string overlap in pp collisions .
once the @xmath9 distribution for nucleus - nucleus collisions is determined , the multiplicity damping factor can be defined in the thermodynamic ( continuum ) limit as @xcite @xmath14 which accounts for the overlapping of discs , with @xmath15 corresponding to the fractional area covered by discs .
the quantity @xmath16 is equal to @xmath17 .
the data utilized in this analysis was acquired by the star experiment at rhic .
only data from @xmath18 1 unit of pseudorapidity with greater than 10 fit points in the main star detector , the time projection chamber , was used .
the events studied were minimum bias events , with standard star centrality cuts as defined by the trigger detectors and overall event multiplicities , which increase with increasing collision centrality .
the longitudinal ( z ) vertex was constrained to within @xmath18 30 centimeters of the primary event vertex .
the distance of closest approach ( dca ) of primary tracks to the event vertex was less than 3.0 cm .
all soft , charged particles with transverse momentum less than 1.2 gev were considered .
the collision systems studied includes : au+au at 200 , 62.4 , and 19.6 gev and pp , d+au , and cu+cu at 200 gev .
the percolation density parameter , @xmath1 , has been determined for several colliding systems and energies.@xcite these results have been compared to a predicted value of the critical percolation density , @xmath7 .
if the critical value of @xmath1 is exceeded , it is expected that the percolation threshold has been reached .
this would then indicate the formation of a maximal cluster that spans the system under study .
all results are shown with statistical errors only , which are smaller than the data points .
+ figure 1 is a plot of the quantity f(@xmath1 ) versus the percolation density parameter , ( @xmath1 ) , for central collisions .
the critical percolation threshold is indicated as that for overlapping strings with a non - uniform density ( `` local '' value).@xcite there is a general increase in @xmath1 with increasing system size and energy .
the predicted value of @xmath7 lies well above that of 19.6 gev au+au and 200 gev d+au collisions .
the result of 62.4 gev au+au is located near to @xmath7 . only 200 gev au+au lies well above this predicted percolation threshold .
it is expected that the overlap probability for string clusters increases as the system size and energy increase .
therefore , a larger percolation density should be obtained at 200 gev au+au compared to lower energies and other , lighter systems .
one can also consider the percolation density as a function of centrality in au+au collisions .
the centrality expressed in terms of the number of participating nucleons ( @xmath19 ) as found from monte carlo glauber calculations.@xcite larger values of @xmath19 correspond to more central collisions .
figure 2 shows @xmath1 as a function of the number of participant nucleons in au+au collisions at 200 and 62.4 gev .
it is shown here that for all centralities , except for the most peripheral bins , all 200 gev au+au collisions lie above the critical percolation threshold .
the value of @xmath1 increases with increasing collision centrality , an expected indication of additional string overlap in more central collisions . for 62.4 gev au+au
, all centralities lie at , or below , the critical percolation density .
this is true even for the most central collisions at this energy .
another interesting feature is that the value of @xmath1 appears to achieve a maximum value of approximately 1.7 in central 62.4 gev au+au collisions , whereas in central 200 gev au+au collisions @xmath1 continues to increase .
this could potentially indicate a saturation of the cluster overlap area , and hence the percolation parameter in central 62.4 gev au+au collisions . finally , fig .
3 presents the fractional area ( @xmath15 ) covered by clusters as a function of @xmath1 .
the data is shown for all centralities of 62.4 and 200 gev au+au collisions , as well as 200 gev cu+cu .
the 200 gev cu+cu result spans a region between that of mid - central to central 62.4 gev au+au and mid - peripheral 200 gev au+au . unlike the au+au results , the cu+cu data is not corrected for efficiency and detector acceptance
it has been predicted by satz@xcite that the study of cu+cu collisions at rhic energies provides one of the best possibilities to study the onset of deconfinement due to the fact that for heavier nuclei ( such as pb+pb ) at top rhic energy , all centralities are above the percolation threshold .
as shown in fig . 3 , 200 gev cu+cu and 62.4 gev au+au appear to probe the region around the critical percolation threshold .
in summary , we have presented the preliminary results for the percolation density parameter , @xmath1 , at rhic for several collision systems as a function of collision centrality .
percolation theory describes a phase transition as occurring at a critical percolation density , @xmath7 .
these results show that for light systems ( d+au ) or heavy systems at low energies ( au+au at 19.6 gev ) , the calculated value for @xmath1 lies well below the predicted critical value . for au+au collisions at 200 gev
, @xmath1 lies well above the predicted percolation threshold
. however , 62.4 gev au+au data appears to lie at or below this threshold . as a function of centrality in au+au collisions at 62.4 and 200 gev , almost all centralities except the most peripheral for 200 gev are above the percolation threshold , while for the lower 62.4 gev events , all centrality bins lie just at , or below , @xmath7 .
results for cu+cu collisions at 200 gev have been compared to 62.4 and 200 gev au+au .
the cu+cu results span the regime between central 62.4 and peripheral 200 gev au+au .
the onset of this collective behavior in traditional percolation theory is indicative of a phase transition .
this could indicate the creation of an intermediate phase between a qgp and the hadronic state .
|
multiparticle production at high energies is described in terms of color strings stretched between the projectile and target . as the string density increases ,
overlap in the transverse plane leads to cluster formation . at some critical density
a macroscopic cluster appears , spanning the entire system .
this marks the percolation phase transition .
data from @xmath0 = 200 gev p+p , d+au and au+au collisions at rhic has been analyzed using the star detector to obtain the percolation density parameter , @xmath1 .
the particle p@xmath2 spectrum from 200 gev p+p data is parameterized using a power law .
values of the fit parameters are used in the d+au and au+au analysis . for 200 gev au+au collisions , the value of @xmath1 is found to lie above the critical percolation threshold , while for other collision systems and energies , it is below the critical value .
this supports the idea of string percolation , which at high enough string density is a possible mechanism to explore the hadronic phase transition to a quark - gluon plasma .
|
the field of heavy ion collisions is a very active one , straddling high energy and nuclear physics . at the upper energy limit of this flourishing area of research
, the goal is to eventually produce and study a new state of matter in the laboratory : the quark - gluon plasma ( qgp ) .
that strongly interacting matter in conditions of extreme energy densities undergoes a phase transition is in fact a prediction of qcd @xcite . to confirm whether the phase transition indeed occurs in relativistic heavy ion collisions and
that the qgp is formed , one needs a clear signal as a signature for the qgp .
many approaches have been suggested to elucidate the existence of this elusive state of matter , but unfortunately no single measurement can yet be singled - out as a `` smoking gun '' candidate .
instead , it appears that many complementary experimental data will require simultaneous analysis @xcite .
one class of observables that appears especially attractive is that of electromagnetic signals .
this owes to the fact that such probes essentially suffer little or no final state interactions and thus constitute reliable carriers able to report on the local conditions at their emission site .
indeed , the calculated emission rates for photons and lepton pairs have been shown to strongly depend on the local density and temperature .
those facts were established some time ago and several experiments specializing in the measurement of electromagnetic radiation are either running right now or being planned . in this paper
, we shall be concerned about lepton pairs and calculations of their production . with respect
to photon measurements , they constitute an important related observable .
the extra degree of freedom associated with the invariant mass makes them valuable tools in the study of annihilation reactions , for example .
the emission of lepton pairs from the qgp and also from hot hadronic has been studied by many authors @xcite .
we will not discuss here the lepton pairs with high invariant masses ( @xmath1 ) , they constitute a fascinating story of their own and the literature on the subject is considerable .
rather , we shall address the issue of softer pairs and that of the reliability of calculations of their emission .
recently , a lot of attention has focused on measurements from the ceres / na45 collaboration @xcite .
several theoretical calculations have attempted to reproduce this experimental data @xcite .
because of the exciting potential of such measurements to reveal new physics , extra care has to be taken in their theoretical modeling . in the very soft sector ,
relevant for the ceres experiment , different theoretical calculations for the fundamental processes have in fact been confronted wih each other with a striking degree of agreement @xcite .
it has also been shown that this data supports the fact that the very soft sector of the lepton pair spectrum is relatively independent of the dynamical approach used to model the evolution of the four - volume of the colliding system .
however , the disagreement between models and data has led to different interpretations @xcite .
another interesting aspect of the ceres analysis is that the region of invariant mass @xmath2 also shows an `` anomalous '' excess of lepton pairs , when compared with estimates based on hadronic electromagnetic decays .
note that this excess is seen only in nucleus - nucleus events , the proton - nucleus results are well reproduced assuming hadronic sources @xcite .
adopting the same line of reasoning as that used in most of the calculations done to interpret the low mass sector of the ceres experiment , it is natural to ask which hadronic reactions could contribute significantly to the mass region in question .
some work on this aspect was performed in the past where it was found that in a hot environment of mesons , the reaction @xmath3 should be an important source of dileptons @xcite ( from here on we shall discuss only dielectrons ) . such studies concerned with the contribution of the @xmath4 to the lepton pair spectrum are based on effective chiral lagrangians containing @xmath5 , @xmath6 and @xmath4 .
our purpose in this paper is the following : we review several different effective lagrangians for the @xmath7 vertex , and we calculate lepton pair production from the @xmath3 reaction , using the vector meson dominance model ( vmd ) .
we shall follow a philosophy similar to that in ref .
@xcite , but we go further by including a survey of other known and used interactions .
we shall also stress the importance of conforming to empirical measurements , and we highlight the off - shell properties .
our work is organized as follows : after introducing the different models and mentioning their origins , we compare with each other the lepton results obtained and comment on the restrictions imposed on the effective interactions by the hadronic phenomenology .
we also will show results from an experimentally - extracted spectral function analysis of lepton production processes .
we then conclude .
in our survey of the @xmath7 interaction , we in turn will consider the following lagrangians : 1 .
an effective chiral lagrangian based on the @xmath8 linear @xmath9 model @xcite .
2 . a @xmath10 chiral model for pseudoscalar , vector , and axial - vector mesons @xcite .
an effective chiral lagrangian where the vector mesons are introduced as massive yang - mills fields of the chiral symmetry @xcite .
an effective lagrangian previously used in connection with photon emission rates @xcite .
an effective lagrangian used to address the issue of form factors in the bonn potential @xcite .
as one can realize , many interactions claim to address the @xmath11 vertex .
this variety owes to the fact that there is currently no unique way of implementing the chiral symmetry in an approach uniting pseudoscalars , vectors and pseudovectors .
therefore , many versions of a given coupling can be obtained , with symmetry requirements sometimes yielding to specific practical concerns .
note that all of the above lagrangians ( except one ) have been used in the literature to calculate the lepton - producing reaction we are considering here .
in such investigations , a popular practice consists of adjusting any free parameter in the above interactions to reproduce the experimental decay width @xmath12 . in cases where more than one free parameter is available
, one can also fit @xmath13 , when electromagnetic phenomena are of some relevance .
however , another well - defined empirical quantity exists , regarding the decay @xmath14 . the ratio of @xmath15-wave to @xmath16-wave ( eigenstates of the relative orbital angular momentum in the exit channel ) amplitude has been measured and is known to be @xmath17 @xcite .
theoretically , the calculation of this ratio is accomplished by expanding the decay amplitude in the helicity basis , using the standard helicity representation for the polarization vectors . on the other hand
, one can expand the amplitude in spherical harmonics .
working in the rest frame of the decaying particle , in this case the @xmath4 , one can then match the two representations in terms of their helicity content .
it is probably fair to say that this aspect of hadronic phenomenology has not received the attention it should have in the heavy ion community by users of effective hadronic lagrangians .
we will now proceed as advertised earlier and compare in turn the interactions described above in terms of their contribution to the process @xmath18 .
however , before we turn to this specific application , it is instructive to recall why the process in question is not determined uniquely by choosing an interaction and fitting on - shell properties . for our illustrative purpose ,
consider the process @xmath19 .
the most general vertex for this strong decay can be written as @xmath20 where @xmath21 is a form factor , with @xmath22 and @xmath23 being the the @xmath4 and @xmath5 four - momenta . for on - shell @xmath4 decay ,
three of the above terms are identically zero , owing to the transversality condition @xmath24 .
the on - shell form factors are usually chosen such that ( some of ) the experimental constraints discussed above are satisfied . however in the reaction under scrutiny here , @xmath25 , the @xmath5 meson is not on its mass shell and the extrapolations of form factors to this region are not unique . to evaluate the importance of such effects is the purpose of this work .
note however that off shell , for our reaction , there are still only two terms that are relevant in eq .
( [ vertex ] ) .
this is because the lepton electromagnetic current is conserved .
more specifically , the other term of eq .
( [ vertex ] ) that could in principle contribute to our reaction is @xmath26 .
however , the corresponding matrix element for electron - positron pair production , once the spin sums are carried out , will look like ( the lepton masses have been set to zero , for simplicity ) @xmath27 where @xmath28 is the four - momentum of the electron ( positron ) .
note that now @xmath29 .
the above expression , eq .
( [ prob ] ) , is zero .
the effective lagrangian for the @xmath11 vertex constructed by ko and rudaz @xcite is based on the @xmath8 linear @xmath9 model , with the @xmath5 vector meson and the @xmath4 axial - vector meson included as phenomenological gauge fields associated with a local @xmath30 chiral symmetry .
this local chiral symmetry is then broken to global chiral symmetry .
the @xmath31 vertex is generated by the following interaction lagrangian@xcite , @xmath32.\end{aligned}\ ] ] the @xmath33 decay width can be obtained as @xmath34 \
, , \ ] ] where @xmath35 is the momentum of the @xmath6 meson evaluated in the rest frame of the @xmath4 , and @xmath36 + { \kappa_{6 } g^{2 } f_{\pi } \over m_{\rho}^2}~p \cdot q , \nonumber \\
g_{a_{1 } \rho \pi } & = & - { \kappa_{6 } g^{2 } f_{\pi } \over m_{\rho}^2}\ .\end{aligned}\ ] ] the ratio of @xmath15-wave to @xmath16-wave amplitudes for this final state is @xmath37}{f_{a_{1}\rho\pi } ( e_{\rho}+2m_{\rho})+g_{a_1\rho\pi } m_{a_1}|\vec{q}|^2 } \ .\ ] ] this chiral formulation admits a @xmath38 direct contact term and as such , deviates from the `` traditional '' vector meson dominance approach .
including this , the radiative decay width one obtains for @xmath39 is @xmath40 here is an opportune place for a short digression on the vector meson dominance model ( vmd ) .
a recent discussion of its different representations can be found in ref .
we refer the reader to this reference , and those therein , for details . in short ,
the vector meson - photon vertex can either be a constant or the electromagnetic field strength tensor can couple to the rho field strength tensor to yield a @xmath41-dependent vertex , where @xmath42 is here the photon invariant mass @xcite . in the latter case ,
an additional direct photon contact term is then necessary as the vector meson - photon mixing vanishes for real photons .
the approaches described in this work invoke both versions of vmd .
we have implemented the appropriate one for each case and also verified that gauge invariance was verified in the electromagnetic sector .
returning to the lagrangian under scrutiny , following refs .
@xcite we choose the parameters @xmath43 , and @xmath44 guided by phenomenological considerations .
two choices of those numbers are displayed in table [ table1 ] .
parameter set i was used previously to address dilepton production in the reaction we are concerned with here @xcite .
however , this choice produces a wrong sign for the @xmath45 ratio and its @xmath46 can be somewhat improved .
this is achieved by parameter set ii , see table [ table1 ] .
.two parameter sets for the model of ref .
@xcite , and the associated phenomenology . [ cols="^,^,^,^ " , ] the other simple effective lagrangian which considers the @xmath31 interaction was used to formulate a meson - exchange model for @xmath47 scattering by janssen , holinde and speth@xcite .
it is @xmath48 .\ ] ] for @xmath49 mev , the coupling constant @xmath50 gev@xmath51 follows from the usual fit of the @xmath4 decay width .
the @xmath45 ratio here is @xmath52 .
a comparative summary of the different on - shell properties and predictions is shown in table [ table2 ] .
in this section we shall compute rates of lepton pair emission with the effective interactions introduced earlier . in this study
we rely on relativistic kinetic theory to provide an idealized dynamical framework . because we are mainly interested in comparisons between different models ,
this line of reasoning is entirely adequate .
we also set any chemical potential to zero , for simplicity .
the temperature chosen for our baseline study is t = 150 mev .
in general , the dilepton production rate from the annihilation of two particles @xmath53 and @xmath54 can be written as@xcite @xmath55 where @xmath56 is an overall degeneracy factor , @xmath57 is the distribution function of the incoming particles at temperature @xmath58 , @xmath59 is the relative velocity of the two particles , @xmath60 and @xmath9 is the dilepton production cross section for the reaction @xmath61 .
using boltzmann distributions , one performs the phase space integrations and obtains @xmath62 where @xmath63 is a modified bessel function , and @xmath64 is the kinematic triangle function @xmath65 the cross section @xmath66 depends only on the square of the invariant mass of the dilepton @xmath67 and is easily related to the square of a spin - averaged scattering amplitude .
this in turn can be written as @xmath68 where , @xmath69 is the fine structure constant , and @xmath70 is a leptonic tensor given by @xmath71 and @xmath72 is a hadronic tensor for the process . note that @xmath73 .
the hadronic tensor is calculated from the meson vertex function of the appropriate feynman diagram with the relevant lagrangian .
reaction , at @xmath58=150 mev .
the curves labels are as follows : ( a ) ref .
@xcite , ( b ) ref .
@xcite , ( c ) ref .
@xcite , ( d1 ) and ( d2 ) ref .
@xcite parameter sets i and ii ( see table [ table1 ] ) , ( e ) ref .
@xcite.,width=377 ] we plot in fig .
[ result150 ] the rates for dielectron production from a gas of hot mesons , at a temperature of t = 150 mev .
we concentrate on @xmath74 reactions .
all of the different interactions enumerated previously have been considered and their contribution appears here .
a striking feature is that the rates calculated with those lagrangians span two - and - a - half orders of magnitude .
recall that all of them ( except one ) have been used in the literature to perform dilepton calculations very similar to the one done here .
the features illustrated in fig .
[ result150 ] are essentially temperature - independent . we have verified this by also performing calculations at a temperature of t = 100 mev . we conclude from this that off - shell effects are indeed quite important . considering the curves labeled ( d1 ) and ( d2 ) , one can verify by looking again at table [ table2 ] that a modest change in the @xmath46 calculated on shell can result in an important variation in off - shell behaviour .
this tells us that although the on - shell @xmath46 can be used as a goodness - of - fit parameter , it is far from being enough to specify unambiguously which interaction to use in situations like the one at hand .
one has to proceed with caution .
we now turn to an approach which should have the off - shell effects buit - in , in order to help us deciding upon an interaction to use .
above , we have found large differences in lepton producing rates using different hadronic lagrangians to model the hard vertices .
this result is worrisome , as many of those interactions have been used in the past for dilepton calculations , and there is no doubt that they will be used again in the future .
we then turn to an alternative approach , with the hope that this will assist us in the selection of an appropriate theory . in the context of heavy ion reactions ,
it has been argued recently that lepton pair production cross sections could in fact be determined from the spectral functions extracted from the inverse process : @xmath75 @xcite .
we refer the reader to the appropriate reference for the complete details .
it will suffice here to state that the rate for the emission of dilepton pairs of invariant mass @xmath76 , at temperature @xmath58 in a given reaction is given by @xmath77 where @xmath69 is the usual fine structure constant , @xmath78 is a modified bessel function , and @xmath79 is a zero - temperature spectral function .
this quantity is extracted from @xmath80 annihilation data through @xmath81 we then follow the same approach as that of a current algebra calculation @xcite to extract the @xmath74 contribution from @xmath82 data .
one can then obtain @xmath83 , which can then be used in eq .
( [ rho ] ) to extract the spectral function . note that this procedure is straightforward and relatively free from ambiguity , numerical or otherwise .
it is tantamount to a direct evaluation of the dilepton cross sections by the same theoretical methods .
we plot on fig .
[ totcomp150 ] the rates previously shown on fig .
[ result150 ] , accompanied by the rate evaluated with the method outlined above .
note that this latter calculation does not run over the entire invariant mass range covered in the plot , as we have chosen not to stretch the validity of the theoretical approach which was deemed optimal in the invariant mass range covered by the double line @xcite .
this coverage is sufficient for us to make our point . considering that the spectral function is constrained by data
, we use this analysis as an extra tool to discriminate between the various interactions considered in this work .
we observe that the spectral function results are overshot by the interaction from ref .
they are underpredicted by the rates related to the lagrangians in refs . @xcite and @xcite , and by rates obtained with parameter set ii of the interaction from ref .
the interactions from refs .
@xcite and @xcite ( parameter set i ) yield results in good agreement with the spectral function determination . even though both lagrangians seem to exhibit a satisfactory off - shell behaviour , going back to table [ table2 ] one observes that the one from ref .
@xcite also produces an excellent @xmath46 for on - shell hadronic properties .
it appears that , with the parameters used in this study and considering both on - shell and off - shell behaviours , this interaction achieves the compromise we were seeking .
it is also a satisfying one from a theoretical point of view @xcite .
note in closing that the rates above were determined in the narrow - width approximation for the @xmath4 .
while at first sight this limit seems unreasonable , it has been shown to affect little the magnitude of the thermal rates @xcite .
its main effect is to soften the threshold of this specific reaction .
we have calculated dilepton emission from @xmath74 reactions at finite temperature using several different lagrangian found in the literature .
we have found widely different results .
one might argue that some of those interactions were derived for entirely different purposes , therefore comparing them on the basis of dilepton emission seems vaguely unappropriate .
one should however remember that all of the lagrangians ( except that of ref .
@xcite ) _ have _ been used previously in such exercises . up to now , a critical comparison of their results was lacking . with the help of an experimentally - constrained spectral function , combined with a quantitative analysis of on - shell properties
, we were able to select an adequate interaction .
it is clear that a companion study to this one will consider the rates for photon emission .
this work is in progress and will be reported on later .
there are however indications that for photons , the differences arising from the use of different lagrangians will be less severe @xcite .
this probably owes to the fact that the mass shell condition for real photons has a restraining effect
. it should be clear that our goal was not an exhaustive survey of the parameter space relevant to each of the models we have discussed .
we viewed these interactions as representative of what is currently on the market .
one could in principle devise a new completely phenomenological interaction .
our study highlights the important issue of what needs to be done in order to give credibility to results obtained in an environment as potentially complex as that of high energy heavy ion collisions .
our research is supported in part by the natural sciences and engineering research council of canada , in part by the fonds fcar of the qubec government , and in part by the us department of energy under grant number doe / de - fg02 - 88er-40388 .
one of us ( c. g. ) is happy to acknowledge discussions with , and the hospitality of , the nuclear theory group of the state university of new york at stony brook , where this work was completed .
we are also happy to acknowledge useful discussions with s. rudaz .
99 see , for example , hildegard meyer - ortmanns , rev .
mod . phys . * 68 * , 473 ( 1996 ) , and references therein .
john w. harris and berndt mller , ann .
nucl . part .
, * 46 * , 71 ( 1996 ) .
see , for example , proceedings of the 12@xmath84 international conference on ultra - relativistic nucleus - nucleus collisions , heidelberg , germany , 20 - 24 may , 1996 , nucl . phys . * a 610 * , 1c ( 1996 ) , and references therein .
a. drees , nucl .
phys . * a 610 * , 536c ( 1996 ) . charles gale , proceedings of the third conference on the physics and astrophysics of the quark - gluon plasma , jaipur , india , march 17 - 21 , 1997 , nucl - th/9706026 .
g. q. li , c. m. ko , and g. brown , phys .
lett . * 75 * , 4007 ( 1995 ) ; j. murray , w. bauer , and k. haglin , hep - ph/9611328 ; r. rapp , g. chanfray , and j. wambach , nucl . phys . * a 617 * , 472 ( 1997 ) ; james v. steele , hidenaga yamagishi , and ismail zahed , phys . lett . * b 384 * , 255 ( 1996 ) ;
see also ref . @xcite and references therein . c. song , c.m .
ko and c. gale , phys . rev . *
d 50 * , 1827 ( 1994 ) .
j. k. kim , p. ko , k. y. lee , and s. rudaz , phys .
rev.*d 53 * , 4787 ( 1996 ) .
k. haglin and c. gale , phys . rev . *
d 52 * , 6297 ( 1995 ) .
p. ko and s. rudaz , phys . rev . *
d 50 * , 6877(1994 ) .
b. a. li , phys . rev . * d 52 * , 5165 ( 1995 ) .
h. gomm , . kaymakcalan and j. schechter , phys . rev .
* d 30 * , 2345 ( 1984 ) .
l. xiong , e. shuryak and g. e. brown , phys . rev . *
d 46 * , 3798 ( 1992 ) .
g. janssen , k. holinde and j. speth , phys . rev .
* c 49 * , 2763 ( 1994 ) .
particle data group , phys . rev . * d 54 * , 1 ( 1996 ) .
oconnell , b.c .
pearce , a.w .
thomas , a.g .
williams , prog . part .
* 39 * , 201 ( 1997 ) .
kroll , t.d .
lee , and b. zumino , phys . rev . *
157 * , 1376 ( 1967 ) .
g. carter , p. j. ellis and s. rudaz , nucl . phys . * a 603 * , 367 ( 1996 ) .
c. song , phys
. rev . * c 47 * , 2861 ( 1993 ) .
see , for example , c. gale and p. lichard , phys .
d 49 * , 3338 ( 1994 ) .
z. huang , phys .
* b 361 * , 131 ( 1995 ) . g. penso and tran n. truong , phys . lett .
* b 95 * , 143 ( 1980 ) ; t. n. pham , c. roiesnel , and tran n. truong , _ ibid .
_ , * 80 * , 119 ( 1978 )
. g. ecker , j. gasser , h. leutwyler , a. pich , and e. de rafael , phys .
* b 223 * , 425 ( 1989 ) . c. song and c. m. ko , phys . rev .
* c 53 * , 2371 ( 1996 ) .
|
the production of dielectrons in reactions involving @xmath0 mesons and pions is studied .
we compare results obtained with different phenomenological lagrangians that have been used in connection with hadronic matter and finite nuclei .
we insist on the necessity for those interactions to satisfy known empirical properties of the strong interaction .
large off - shell effects in dielectron production are found and some consequences for the interpretation of heavy ion data are outlined .
we also compare with results obtained using experimentally - extracted spectral functions .
|
graphene , a one - atom thick single sheet of @xmath0-bonded carbon atoms in a honeycomb lattice , is a zero - gap semiconductor ( semi - metal ) that exhibits extraordinarily high electron mobility and shows considerable promise for applications in electronic and optical devices , high sensitivity gas detection , ultracapacitors and biodevices@xcite .
graphene can be separated from bulk graphene using a peeling technique known as mechanical exfoliation@xcite .
lithography techniques@xcite have recently been applied to fabricate patterned graphene nanoribbons ( gnr ) with widths below 100 ; such gnr were used to produce graphene field effect transistors with on - off ratios as high as @xmath1
. there has been concerted efforts to investigate the stability of graphene sheets @xcite .
since applications of graphene necessarily employ sheets of finite extent , it is also of interest to inquire as to the equilibrium shape of such finite sheets or flakes .
two distinct definitions of shape arise in the context of graphene sheets .
the first is associated with a finite , perfectly flat graphene sheet .
for such a sheet , the equilibrium shape is determined primarily by the energy of the graphene sheet edges .
edge energies are related to the thermodynamically stable shape in exactly the same way that surface energies are used to predict the equilibrium shape of a three dimensional crystal through the wulff construction @xcite .
the second is associated with equilibrium elastic distortions of the graphene sheet as a result of edge stresses ( analogous to surface stresses for a three - dimensional crystal ) .
several recent studies @xcite have shown that the two highest symmetry graphene edges have compressive edge stresses , leading to elastic distortions of the regions near these sheet edges ( edge warping ) . in this paper
, we employ density - functional theory ( dft ) to determine graphene edge energies and edge stresses as a function of edge orientation both in the absence and in the presence of hydrogen atoms .
we employ these results to investigate the equilibrium shape of a finite graphene sheet ; both for the flat sheet ( a two - dimensional intrinsic manifold ) and one for which elastic distortions are permitted ( a two - dimensional sheet embedded in three dimensions ) .
in order to determine the properties of the edges of graphene sheets , we employ density - functional theory ( dft)@xcite within the local density approximation ( lda ) using a self - consistent pseudopotential method , as implemented in the siesta code@xcite . a double-@xmath2 plus polarization basis set is used for the localized basis orbitals along with a large energy cutoff ( 400 ry ) for the real space integration .
an electronic temperature of 0.01 ev is used to smear the fermi - dirac distribution . in order to study graphene sheet edges , we focus on flat graphene nanoribbons ( gnr ) that are periodic in one direction and
are separated by 16 of vacuum in the other two directions . in the direction parallel to the ribbon axis
, we use a @xmath3-point sampling criterion such that the number of @xmath3 points @xmath4 in the periodic direction is determined by the smallest integer that fulfills @xmath5 , where @xmath6 is the length of the ribbon in the periodic direction .
this corresponds to a 32-@xmath3-point scheme for 4.28 , the dimension of the smallest unit cell of an armchair edge .
we minimize the energy with respect to ionic position in the gnr plane ; the relaxation is stopped when the force on each atom is less than 0.01 ev / .
we first determine the honeycomb lattice constant @xmath7 of the infinite graphene sheet , which is @xmath8 .
this corresponds to a c - c bond length of @xmath9 .
this agrees very well with the value of @xmath10 found in another density - functional calculation @xcite . between edge tangent vector @xmath11 and the armchair vector @xmath12 .
the two highest symmetry edges are shown : the @xmath13 ( 1,1 ) armchair edge and the @xmath14 ( 1,0 ) zigzag edge .
also show is the more general @xmath15 graphene edge , corresponding to @xmath16 . ]
a general graphene edge orientation can be described by a convention that is commonly used to describe the chirality of a single - wall carbon nanotube@xcite .
we use @xmath17 to describe the angle between the tangent to the edge @xmath18 and a second vector parallel to the armchair , as shown in fig .
[ fig : graphene - orientation.eps ] .
this angle is given by @xmath19}}.$ ] naturally , the armchair and zigzag edges correspond to @xmath13 and @xmath20 , respectively .
because of the symmetry of the honeycomb lattice , we only need to consider angles within the range of @xmath21 . in the present study ,
we study a discrete set of edge orientations in this @xmath17-range : @xmath22 , @xmath23 , @xmath24 , @xmath25 , @xmath26 , @xmath27 , and @xmath28 .
figure [ fig : ac-51-zz.eps]b shows a @xmath29 graphene nanoribbon , corresponding to @xmath30 ; each repeat period of this edge structure contains 5 zigzag units ( fig .
[ fig : ac-51-zz.eps]c ) and a kink that has the armchair structure ( fig . [ fig : ac-51-zz.eps]a ) .
( 1,1 ) , ( b ) @xmath31 ( 5,1 ) , and ( c ) zigzag @xmath14 ( 1,0 ) orientations . ] in the absence of hydrogen , the edge energy ( per unit length ) @xmath32 is @xmath33 , \label{eq : ee}\ ] ] where @xmath34 is the energy of an optimized gnr with edge orientation @xmath17 and width @xmath35 that contains @xmath36 carbon atoms , and
@xmath37 is the energy per carbon atom in the infinite , flat graphene sheet with the lattice parameter that minimizes the energy .
the factor of two in eq .
[ eq : ee ] accounts for the fact that a gnr has two edges . in this work
, we also consider the termination of all of the dangling bonds on the unreconstructed edges by hydrogen atoms . in this case
, the edge energy is @xmath38 \label{eq : edwithh}\ ] ] where @xmath39 is the energy of the graphene nanoribbon where the dangling bonds associated with carbon atoms on the edge are saturated with single h atoms and @xmath40 is the total number of such h atoms .
@xmath41 is the energy of the hydrogen molecule and the factor of @xmath42 in the last term accounts for the fact that each h@xmath43 molecule only contributes one h atom to each initially unsaturated bond .
note that eq .
[ eq : edwithh ] reduces to eq .
[ eq : ee ] in the absence of hydrogen @xmath44 . by analogy with the surface stress@xcite
, we define the edge stress @xmath45 as the work required to stretch a graphene edge by an infinitesimal strain @xmath46 ( parallel to the edge ) , @xmath47 , or in differential form by @xmath48 ( note : two definitions of surface stress @xmath49 have been widely employed in the literature : ( a ) @xmath50 and ( b ) @xmath51 , where @xmath52 is the surface energy .
we apply the approach used for surface stress definition ( a ) here , i.e. , @xmath53 .
this is consistent with the method employed by jun @xcite , but not by huang
_ et al . _
@xcite in their studies of graphene edges .
this definition yields the edge stress that produces elastic distortions of graphene nanoribbon edges @xcite . )
the derivative in eq .
[ eq : edgestress ] is determined using a method based upon the bulk stress calculation within the siesta code @xcite that assumes an affine transformation of the atomic coordinates according to the applied strain . throughout this work
, we adopt the convention that a negative ( positive ) edge stress corresponds to compression ( tension ) .
the first question we address , is how wide must a gnr be in order that the interactions between the opposite , parallel edges are negligible .
figure [ fig:05-width - vs - edge - energy - all - direction.eps ] shows the edge energy versus gnr width , without and with h - termination .
these figures clearly demonstrate that the edge energy converges to a nearly width - independent value for @xmath54 for all gnr orientations , both without and with h - termination .
we note that earlier first principles have both included @xcite and excluded@xcite spin polarization effects .
we have not included spin polarization effects here as they are expected to be more important for a proper determination of the band structure of zigzag edge graphene nanoribbons than for the overall edge thermodynamics .
unlike for all of the other edges , the @xmath55 ( 1,1 ) armchair edge energy exhibits a decaying oscillatory behavior with respect to gnr width @xmath35 , as seen in fig .
[ fig:05-width - vs - edge - energy - all - direction.eps ] .
this is consistent with earlier observations for the special case of an armchair gnr without h - termination@xcite .
et al._@xcite demonstrated that the energy bandgap of armchair gnrs can be grouped into 3 families according to the gnr width , @xmath35 .
these families correspond to @xmath56 , @xmath57 and @xmath58 , where @xmath59 is an integer and @xmath60 is the number of carbon rows parallel to the armchair gnr axis .
figures [ fig:04-agnr - three - family - edge - energy.eps ] a and b show the edge energy versus gnr width for the unterminated and h - terminated cases , where the data has been separated according to these families . with this division ,
the armchair edge energy versus @xmath35 largely converges monotonically .
( 1,1 ) armchair edge as a function of ribbon width @xmath35 grouped according to the division of the data into three families@xcite ( a ) without hydrogen termination and ( b ) with hydrogen termination . ] for the largest width gnrs in fig .
[ fig:05-width - vs - edge - energy - all - direction.eps ] .
the insets show the same data as in the main figure but over a narrower edge energy range .
the figure shows data for the unreconstructed edges both without and with h - termination .
also shown are data points for the reconstruction of the @xmath14 ( 1,0 ) zigzag edges [ zz(57 ) ] without and with h - termination .
( b ) the edge stress as a function of edge orientation @xmath17 for the same configurations as in ( a ) .
the hydrogenated zigzag(57 ) has too large of a tensile edge stress , 2.570 ev / to show on this plot . ]
the converged ( with respect to @xmath35 ) edge energies shown in fig .
[ fig:05-width - vs - edge - energy - all - direction.eps ] are summarized in fig .
[ fig : tau - gamma - angle - vs - edge - energy - and - edge - stress.eps]a as a function of edge type , @xmath17 . for the unreconstructed and unterminated edges ,
this data shows that the edge energy increases nearly linearly with @xmath17 ; the armchair ( zigzag ) edge has the lowest ( highest ) edge energy , @xmath61ev / ( @xmath62 ev / ) .
the energy ordering ( armchair versus zigzag ) can be understood by noting that the armchair edge reduces its energy by forming triple bonds in the armrest regions , as suggested by the fact that this bond length 1.258 is much smaller that the bulk graphene c - c bond length of 1.427 .
in contrast , the zigzag edge does not have the opportunity to form triple bonds ; the outer c - c bonds relax slightly to yield a bond length of 1.400 .
the edge energy difference between the armchair and zigzag edges is @xmath63 ev / without reconstruction or h - termination , in agreement with previous dft results@xcite ( 0.33 ev / ) .
the near linearity of the edge energy with respect to @xmath17 can be understood by noting that all of the unreconstructed edges are made up of a combination of armchair and zigzag units in proportion to the edge angle ; i.e. , steps on near armchair oriented edges are zigzag units and steps on near zigzag oriented edges are armchair units ( see the ( 5,1 ) edge in fig .
[ fig : ac-51-zz.eps ] ) .
the graphene edge stresses , with and without hydrogen termination , are presented in fig . [
fig : tau - gamma - angle - vs - edge - energy - and - edge - stress.eps]b . without hydrogen termination ,
the edge stresses are compressive ( negative ) for all @xmath17 . for the armchair [ @xmath55 , ( 1,1 ) ] edge @xmath64 ev / and for the zigzag [ @xmath14 , ( 1,0 ) ] edge @xmath65 ev / . these values are of the same order of magnitude as those obtained by jun@xcite ( i.e. , armchair edge - @xmath66
ev / and zigzag edge - @xmath67 ev / ) . calculations of the edge stresses using an adaptive intermolecular reactive empirical bond order ( airebo ) potential , significantly underestimate the magnitude of the armchair and zigzag edge stresses ( i.e. , they find that the edge stresses of the armchair and zigzag edges are @xmath68 ev / and @xmath69 ev / , respectively ) . while earlier studies only considered the edge stresses of the two highest symmetry edges , the results of fig .
[ fig : tau - gamma - angle - vs - edge - energy - and - edge - stress.eps]b show the edge energy for many intervening orientations . unlike for the edge energies ,
the edge stress is not a monotonic function of the edge orientation angle .
rather , the magnitude of the edge stress is a minimum at @xmath70 .
reconstruction of the zigzag edge to from the zigzag(57 ) structure considerably raises the zigzag edge stress to near zero ( slightly tensile ) .
the main conclusion from this part of the study is that in the absence of hydrogen , all edge orientations exhibit compressive edge stresses ( except for the zigzag edge upon reconstruction ) .
we can understand the sign of the edge stress by consideration of edge relaxation ( i.e. , like surface relaxation ) .
finnis and heine@xcite presented an argument in which surface atoms relax inward ( toward the bulk ) in order to smooth the electron density near the surface .
this lowers the kinetic energy of the electrons , while having a relatively weak effect on the pseudopotential contribution to the total energy .
the same thing happens at edges in the present two dimensional material .
when the edge atoms move inward , they create a force on the atoms below the edge that is parallel to the edge and tends to push these atoms apart . however , the atom spacing , parallel to the edge , is fixed by the constraint of the atoms in the bulk of the graphene sheet .
it is these forces that create the compressive edge stress .
when hydrogen adsorbs on the surface ( see below ) , the inward relaxation effect is strongly reduced ( due to near perfect saturation of the dangling bonds ) , thereby greatly reducing the magnitude of the compressive edge stresses .
when hydrogen terminates all of the dangling bonds of the ribbon edges , the unreconstructed edge energies drop by @xmath71 orders of magnitude , as seen in figs .
[ fig:05-width - vs - edge - energy - all - direction.eps]b and [ fig : tau - gamma - angle - vs - edge - energy - and - edge - stress.eps]a .
the edge energy in the presence of hydrogen is maximum for the zigzag [ @xmath14 , ( 1,0 ) ] edge ( @xmath72 ev / ) and minimum for the armchair [ @xmath73 , ( 1,1 ) ] edge ( @xmath74 ev / ) . while the magnitude of the edge energies drop dramatically upon hydrogen saturation , the ratio of the zigzag to armchair energies shows a much smaller change . saturating the gnr edges with hydrogen
significantly decreases the magnitude of the edge stresses ( see fig .
[ fig : tau - gamma - angle - vs - edge - energy - and - edge - stress.eps]b )
. interestingly , saturating the zigzag gnr edge with hydrogen changes the edge stress from highly compressive , @xmath75 ev / , to nearly zero , @xmath76 ev / .
this is consistent with the results of an earlier study@xcite that show a significant drop in the magnitude of the edge stress upon saturating the edges with hydrogen . in order to understand the effect of hydrogen on the gnr edges , we focus first on the hydrogen adsorption energy @xmath77 on the graphene edge : @xmath78 .
if we denote the number of h atoms adsorbed per unit length of the edge as @xmath79 , and the binding energy of the hydrogen molecule ( relative to the isolated hydrogen atoms ) as @xmath80 , we can rewrite @xmath81 ( see eq .
[ eq : edwithh ] ) as @xmath82 this change in edge energy ( per unit length ) upon saturation with hydrogen is simply the product of the number of h atoms adsorbed ( per unit length ) and the change in energy per hydrogen atom in going from h@xmath43 in the gas to h on the edge . for the unreconstructed edges of graphene
, @xmath83 has a very simple form : @xmath84 where @xmath7 is the honeycomb lattice constant .
@xmath83 is @xmath85 larger for the armchair ( @xmath13 ) than for the zigzag edge ( @xmath86 ) . .
the binding energy of a hydrogen atom in h@xmath43 , @xmath87 , is shown for reference . ]
the main effect of adsorbing hydrogen on the edges is to decrease the edge energy by @xmath88 .
figure [ fig : tau - gamma - angle - vs - edge - energy - and - edge - stress.eps]a shows that the energy of the zigzag edge decreases upon hydrogen adsorption more so than does the energy of the armchair edge .
this is contrary to what we should expect based upon the number of adsorbed hydrogen atoms per unit length , @xmath83 ( which decreases with increasing @xmath17 as @xmath89 ) .
however , the hydrogen adsorption energy @xmath77 increases with increasing @xmath17 ( as seen in fig .
[ fig:06-nm - angle - vs - h - adsorption - energy.eps ] ) more quickly than @xmath83 decreases .
this effect is associated with the fact that , before hydrogen termination , carbon atoms on the armchair edge have triple bonds to neighboring edge carbon atoms ( and no dangling bonds ) , while the zigzag edge has no triple bonds ( but does have dangling bonds)@xcite .
this suggests that hydrogen should bind more strongly to the zigzag edge atoms than to the armchair edge atoms .
this implies @xmath90 will be larger for the zigzag edge than for the armchair edge .
because the only term in @xmath77 that depends on edge type is @xmath90 , this implies higher h adsorption energy to the zigzag edge than to the armchair edge . the hydrogen adsorption energy on the zigzag(57 )
edge , 5.22 ev / atom , is significantly smaller than the adsorption energy on the unreconstructed zigzag edge ( see fig .
[ fig:06-nm - angle - vs - h - adsorption - energy.eps ] ) . this can be attributed to the fact that triple bonds formed in the heptagonal units on the zigzag(57 ) edge and to the lack of triple bonds on the unreconstructed zigzag edge . the difference in adsorption energy between the two zigzag edges
explains why the hydrogenated zigzag(57 ) edge energy is ( 0.249 ev / ) higher than the hydrogenated unreconstructed zigzag edge ( see fig .
[ fig : tau - gamma - angle - vs - edge - energy - and - edge - stress.eps]a ) .
it is important to focus upon h adsorption energy @xmath77 for the zigzag and hydrogenated zigzag edges ( presented above ) _ relative _ to the binding energy of hydrogen in molecular hydrogen @xmath91 ( on a _ per atom _ basis ) .
this comparison shows that while reconstruction reduces the tendency for h - passivation of the zigzag edge , h will still bind to this edge .
( this conclusion differs from the interpretation of the discussion in ref . . )
in this section , we investigate the equilibrium shape of a flat graphene sheet , following a procedure akin to that used for determining the equilibrium shape of a 3-dimensional crystal via the wulff construction ( e.g. , see ref . ) . the first step is to construct a polar plot of the edge energy @xmath32 as a function of the edge orientation angle @xmath17 .
this is shown in fig .
[ fig : noh - noreconstruction.eps]a for the case of pure , unreconstructed graphene edges using the data from fig .
[ fig : tau - gamma - angle - vs - edge - energy - and - edge - stress.eps]a .
next , we construct a ray from the origin at edge orientation angle @xmath17 on this polar plot .
then , we draw a line perpendicular to this ray at the point of its intersection with @xmath32 . this process is repeated for every point ( many points ) around the polar @xmath32 versus @xmath17 plot .
the inner envelope of these lines represents the equilibrium graphene flake shape .
this construction and the resultant equilibrium flat graphene flake shape ( i.e. , that which minimizes the edge energy at fixed flake area ) is shown in fig .
[ fig : noh - noreconstruction.eps]a for the case of graphene with no edge reconstruction and without hydrogen termination .
the equilibrium shape is a hexagon with flat ( 1,1 ) armchair edges . as a function of @xmath17 ( a ) for edges without hydrogen termination or reconstruction ,
( b ) with reconstruction but without hydrogen termination based upon a linear interpolation of the equilibrium armchair and zigzag(57 ) edge energies and ( c ) with hydrogen termination ( no reconstruction occurs ) .
the edge energy scales in ( a ) and ( b ) are identical , but are multiplied by a factor of 16.7 in ( c ) for clarity .
the red circles indicate data from our density functional calculations , the black lines are an interpolation of the data and the blue lines are drawn as perpendiculars to the radius @xmath32 vectors . the equilibrium , flat graphene flake shape is the inner envelope of the blue lines . ] the data presented in fig .
[ fig : tau - gamma - angle - vs - edge - energy - and - edge - stress.eps]a suggests that , in the absence of hydrogen , the zigzag edge reconstructs and that the reconstructed surface has lower energy than even the armchair edge ( which does not reconstruct ) . since we do not have data for the reconstruction of all edges with @xmath92 , we make the simplifying assumption that the energies of these edges can be represented by a linear interpolation between the energies of the unreconstructed @xmath55 armchair edge and that of the reconstructed @xmath14 zigzag(57 ) edge ( as holds for the unreconstructed edges - see fig .
[ fig : tau - gamma - angle - vs - edge - energy - and - edge - stress.eps]a ) .
the resultant polar @xmath32 versus @xmath17 plot is shown in fig .
[ fig : noh - noreconstruction.eps]b .
unlike the polar @xmath32 versus @xmath17 plot for the unreconstructed case , this one is relatively smooth and nearly circular , yet a six - fold symmetry is obvious . performing the wulff construction on this interpolated @xmath32 versus @xmath17 plot produces an equilibrium graphene flake shape that has six - fold symmetry , but no flat surfaces .
the absence of flat surfaces or facets can be traced to the fact that the polar @xmath32 versus @xmath17 plot here shows no sharp cusps ( discontinuity of slope ) .
more interesting , however , is that when we allow edge reconstruction to occur ( in the absence of hydrogen ) , the resultant six - fold equilibrium shape is rotated relative to that without edge reconstruction .
that is , the equilibrium graphene flake shape is dominated by surfaces at or near the zigzag edge orientation . when hydrogen is introduced into the system , such that all of the dangling bonds on the edges are saturated , the edge energy is greatly reduced , as discussed above .
the hydrogenation of the edges removes any reconstruction and the ( 1,1 ) armchair @xmath93 ( 1,1 ) edge has the lowest edge energy and the zigzag @xmath20 ( 1,0 ) edge has the highest . the resultant polar plot of @xmath32 versus @xmath17 is shown in fig .
[ fig : noh - noreconstruction.eps]c .
this plot shows strong cusps at the armchair orientations . performing the wulff construction
, we find that the equilibrium graphene flake shape is hexagonal with ( 1,1 ) armchair edges , as in the unreconstructed , unhydrogenated graphene edge case . the above analysis focused on the equilibrium shape of a flat , finite graphene sheet ( a two - dimensional intrinsic manifold ) .
however , a flat graphene flake is unstable against thermal rippling at finite temperature@xcite .
compressive edge stresses can also lead to elastic distortions of / near the graphene flake / ribbon edges , as discussed in the literature@xcite .
these distortions may be described as the buckling of the edges , associated with the compressive edge stress but constrained by the elasticity of the interior of the graphene flake .
the amplitude of the buckling @xmath94 has been described@xcite as @xmath95 where @xmath96 is the wavelength of the buckling , @xmath97 , @xmath98 and @xmath99 are the young s modulus and the poisson ratio of the graphene sheet , and where we have neglected the relatively minor effect of the elastic stiffness of the edge . assuming a typical wavelength of @xmath100 nm and @xmath101 ev / nm@xmath102 for graphene@xcite , we can use the data presented above to determine the buckling amplitude @xmath94 as a function of edge orientation @xmath17 and examine the effect of hydrogenation of the edge . using the data from fig .
[ fig : tau - gamma - angle - vs - edge - energy - and - edge - stress.eps]b . ] figure [ fig : bucklingamplitude ] shows the buckling amplitude from eq .
( [ eq : bucklingamplitude ] ) using the edge stress data from fig .
[ fig : tau - gamma - angle - vs - edge - energy - and - edge - stress.eps]b as a function of @xmath17 , with and without hydrogen on the edges .
the edge buckling amplitude found here is 3.05 and 3.11 for the unreconstructed armchair and zigzag edges without hydrogen , respectively , which is roughly consistent with the magnitude of the buckling amplitude reported in ref . .
the buckling amplitude is in the range of 2.70 to 3.11 over the entire range of @xmath17 for the unreconstructed edges without hydrogen .
interestingly , if the zigzag edge reconstructs to the ( 57 ) structure , the buckling amplitude , as per eq .
( [ eq : bucklingamplitude ] ) , is imaginary ( the edge stress is positive ) ; implying that the reconstructed zigzag edge does not buckle .
when the edges are saturated with hydrogen , the edge stress drops dramatically ( by a factor of @xmath103 7.54 to 425 for the armchair and zigzag edges , respectively ) and the buckling amplitude drops to 1.11 and 0.14 for the armchair and zigzag edges ( see fig .
[ fig : bucklingamplitude ] ) .
this represents a drop in edge buckling amplitude by a factor of @xmath103 2.3 to 22 .
in short , edge hydrogenation strongly reduces the edge stress driven graphene edge buckling to the point that it is , for most purposes , negligible .
we have employed density functional theory - based calculations of graphene nanoribbons to determine the edge orientation dependence of the edge energy and edge stress ( akin to surface stress in three - dimensional materials ) . in the absence of reconstruction ,
the edge energy is a nearly linear function of edge orientation angle ; increasing from the armchair orientation to the zigzag orientation .
reconstruction of the zigzag edge , however , lowers its energy to less than that of the armchair edge .
the edge stress for all edge orientations is compressive .
the magnitude of the edge stress is a minimum at an orientation nearly midway between the armchair and zigzag orientations .
reconstruction of the zigzag edge reduces the edge stress to near zero ( very slightly tensile ) .
hydrogen adsorption to graphene edges is favorable for all edge orientations , with a larger adsorption energy to the unreconstructed zigzag edge than to the armchair .
hydrogen adsorption dramatically lowers all edge energies and all edge stresses .
it also removes the reconstruction of the zigzag edge .
the thermodynamic properties of graphene edges play a key role in determining the equilibrium shape of a finite graphene sheet or flake . using the new edge energy data
, we determined that the equilibrium shape of a graphene sheet ( with unreconstructed edges ) is hexagonal with straight armchair edges .
this result is valid for both the hydrogen - free case with unreconstructed edges and in the presence of hydrogen .
however , if the zigzag edges reconstruct ( as thermodynamically favored ) , graphene flakes will have a six - fold symmetry , but with rounded ( rather than faceted ) edges .
interestingly , the six - fold symmetry is rotated with respect to the hydrogenated case and will be dominated by near zigzag edges .
the compressive edge stresses will lead to edge buckling ( out - of - the - plane of the graphene sheet ) for all edge orientations , in the absence of hydrogen .
the edge buckling amplitude is expected to be approximately 3 , but this value depends on buckling wavelength . exposing the graphene flake to hydrogen will dramatically decrease the buckling amplitude to the point that it may be too small to be of any significance .
the authors gratefully acknowledge useful discussions with paulo branicio , yong wei zhang , and nathaniel ng .
the basis sets and pseudopotentials were provided by young - woo son and su ying quek .
we are also grateful to julian gale for his input on the use of the siesta code .
|
we use density functional theory to determine the equilibrium shape of graphene flakes , through the calculation of the edge orientation dependence of the edge energy and edge stress of graphene nanoribbons .
the edge energy is a nearly linear function of edge orientation angle ; increasing from the armchair orientation to the zigzag orientation .
reconstruction of the zigzag edge lowers its energy to less than that of the armchair edge .
the edge stress for all edge orientations is compressive , however , reconstruction of the zigzag edge reduces this edge stress to near zero .
hydrogen adsorption is favorable for all edge orientations ; dramatically lowering all edge energies and all edge stresses .
it also removes the reconstruction of the zigzag edge . using the new edge energy data , we determine the equilibrium shape of a graphene sheet ( with unreconstructed edges ) to be hexagonal with straight armchair edges in the presence and absence of hydrogen . however , zigzag edge reconstruction produces graphene flakes with a six - fold symmetry , but with rounded edges .
this shape is dominated by near zigzag edges .
the compressive edge stresses will lead to edge buckling ( out - of - the - plane of the graphene sheet ) for all edge orientations , in the absence of hydrogen . exposing
the graphene flake to hydrogen dramatically decreases the buckling amplitude .
|
the major goal of colliding heavy - ions at the relativistic heavy ion collider ( rhic ) at brookhaven national laboratory and the large hadron collider ( lhc ) at cern is to study matter at extreme conditions of temperature and energy densities , where quarks and gluons , rather than mesons and baryons , define the relevant degrees of freedom @xcite .
this new phase of matter , the quark - gluon plasma ( qgp ) , is governed by the principles of quantum chromodynamics ( qcd ) and is the result of a phase transition from the normal nuclear matter @xcite .
experiments at rhic and lhc are on the quest to unearth the nature of the qcd phase transition and to get a glimpse of how matter behaves at extreme conditions .
the beam energy scan ( bes ) program at rhic has been initiated to explore the onset of phase transition by scanning the collision energy over a larger range and to locate the critical point in the qcd phase diagram .
the thermodynamic state of the qcd matter can be specified by the temperature @xmath4 and the chemical potentials @xmath5 , @xmath6 and @xmath7 corresponding to the conserved charges of qcd , namely baryon number ( @xmath8 ) , strangeness ( @xmath9 ) , and electric charge ( @xmath10 ) , respectively .
phase transitions are associated with the transformation of thermodynamic quantities such as pressure , entropy and energy density , as well as a set of response functions , like , specific heat , compressibility and susceptibility with change in @xmath4 , @xmath5 , @xmath7 and @xmath6 . in this article
, we discuss the specific heat ( @xmath0 ) of the system produced in heavy - ion collisions at relativistic energies and its behaviour as a function of collision energy .
specific heat is a thermodynamic quantity characterizing the equation of state of the system . for a system undergoing phase transition , @xmath0
is expected to diverge at the critical point .
temperature fluctuation of the system provides an estimation of @xmath0 . near the critical point
, the specific heat is normally expressed in terms of a power law , @xmath11 , where @xmath12 is the critical temperature and @xmath13 is critical exponent .
thus the variation of thermal fluctuations with temperature can be effectively used to probe the critical point .
hadron resonance gas ( hrg ) model analysis of the particle yields indicate the formation of a thermal source for the produced particles in heavy - ion collisions @xcite .
the production of large number of particles in each collision at the rhic and lhc energies makes it even possible to study several quantities on an event - by - event basis @xcite and hence measure their event to event fluctuations .
thus , with the measurement of @xmath4 on an event - by - event basis , it is possible to extract the @xmath0 of the hot and dense strongly interacting matter produced in heavy - ion collisions . assuming complete thermal equilibrium up to the surface of last scattering which is the kinetic freeze - out surface , @xmath0 is then expected to reveal the thermodynamic state of the matter at the moment of kinetic freeze - out .
the specific heat and its variation as a function of temperature have been studied extensively in different theoretical calculations .
statistical and hrg models have been used to obtain @xmath0 as a function of temperature in hadron gas and in quark - gluon matter @xcite . in ref .
@xcite , a parton and hadron cascade model has been used to investigate @xmath0 as a function of beam energy for the initial partonic stage using quark - gluon matter and the final stage using hadronic matter .
lattice qcd calculations @xcite provide estimations of @xmath0 for a wide range of temperatures . in ref .
@xcite , continuum limits of @xmath0 have been calculated in quenched qcd at temperatures of 2@xmath12 and 3@xmath12 , where @xmath12 is the transition temperature .
it is found that @xmath0 differs significantly from that of the ideal gas .
recent lattice calculations using ( 2 + 1)-flavor qcd with almost physical quark masses give the results of @xmath0 for a temperature range of 130 to 400 mev @xcite .
the low temperature ( hadron phase ) results agree well with hrg .
the specific heat has its origin in the event - by - event temperature fluctuations , which manifests through the fluctuations in the transverse momenta ( @xmath2 ) @xcite .
event - by - event fluctuations of @xmath1 have been reported by experiments at the cern super proton synchrotron ( sps ) @xcite and beam energy scan at rhic @xcite .
the values of @xmath0 extracted from the experimental results have large errors @xcite . the @xmath2 fluctuation data from ref .
@xcite yielded the value of @xmath0 to be @xmath14 at @xmath15 mev for sps energies .
the statistical fluctuations arising from the finite multiplicity distributions of charged particles may significantly affect the extracted thermodynamic fluctuations @xcite . in the present work , this is taken care of by subtracting the widths of the results of mixed events from the real data . since radial flow affects the estimation of temperature , its effect has also been considered .
finally , the values of @xmath0 have been calculated as a function of beam energy from published experimental data and compared to lattice and hrg calculations .
further predictions have been made for the lhc energies .
the paper is organized as follows . in section - ii , we present the methodology for extraction of @xmath0 from event - by - event average transverse momentum distributions and from theoretical calculations .
the hrg model calculations of specific heat are presented in section iii .
event - by - event distributions of mean transverse momenta have been reported by several experiments at cern - sps and rhic .
the @xmath1 distributions and corresponding values of effective temperatures are presented in section iv .
a compilation of the results of specific heat from the existing experimental data is presented in section v. estimates for the specific heat at lhc using the ampt model @xcite have been included in this section . a discussion on the results of @xmath0is given in section vi .
the paper is summarised with an outlook in section vii .
the heat capacity @xmath16 of a system is defined as @xcite : @xmath17 where @xmath4 , @xmath18 and @xmath19 are temperature , volume and energy of the system , respectively .
equivalently , @xmath16 of a system in thermal equilibrium to a bath at @xmath4 can be computed from the event - by - event fluctuations of @xmath19 : @xmath20 for a system in equilibrium , the event - by - event temperature fluctuation is controlled by the heat capacity : @xmath21 , \label{eqn3}\end{aligned}\ ] ] where @xmath22 is the mean temperature and @xmath23 is the variance in temperature .
this yields the expression for @xmath16 @xcite : @xmath24 heat capacity thus can be estimated from the fluctuations in energy or temperature . for a system in equilibrium , the mean values of @xmath4 and @xmath19 are related by an equation of state .
however , the fluctuations in energy and temperature have very different behaviour .
energy being an extensive quantity , its fluctuation has a volume dependent component .
so energy is not suited for obtaining the heat capacity . on the other hand ,
temperature fluctuations provide a good major for estimating the @xmath0 @xcite
. the temperature of the system can be obtained from the transverse momentum ( @xmath2 ) spectra of the emitted particles .
an exponential boltzmann - type fit to the @xmath2 spectra gives a measure of the temperature : @xmath25 where @xmath26 is a normalization factor and @xmath27 is the apparent or effective temperature of the system @xcite .
for obtaining the event - by - event fluctuation , the temperature needs to be estimated in every event .
the fitting is possible only for central heavy - ion collisions at the lhc energies when the number of particles is at least one thousand in every event . even in this case , the error associated with the fitting will be relatively large .
this can be overcome by making a connection of mean transverse momentum ( @xmath1 ) of particles in every event with the temperature . since the calculation of the mean value is more stable , this method of temperature estimation
can also be used for collisions at rhic energies .
the @xmath1 can be written as @xcite : @xmath28 where @xmath29 is the rest mass of the particle .
note that the integration for @xmath2 is from 0 to @xmath30 .
but in reality the @xmath2 window is finite . for a range of @xmath2 within @xmath31 to @xmath32
, we obtain : @xmath33 this equation links the value of @xmath1 within a specified range of @xmath2 to @xmath27 . in order to validate the relation between @xmath2 to @xmath27 ,
we have generated a large number of events using the ampt model @xcite for pb+pb collisions at @xmath3 = 2.76 tev .
the goal is to compare the values of @xmath27 obtained from event - by - event @xmath2 distribution and from @xmath1 distributions . for top central ( top 5% cross section ) collisions , @xmath2
distribution of pions has been constructed for each event within a rapidity range of -1.0 to 1.0 .
the distribution is fitted to an exponential function and the inverse slope parameter ( @xmath27 ) is extracted within fit range , @xmath34 gev .
[ fig_ampt ] shows the extracted event - by - event @xmath27 distribution ( as solid circles ) .
for the same set of events , the values of @xmath1 has been calculated within the same @xmath35 and @xmath2 ranges for each event . from the value of @xmath1 for each event , the @xmath27 is calculated using eqn .
( [ eqn7 ] ) . resulting @xmath27 distribution has been plotted as open squares in fig .
[ fig_ampt ] .
both the @xmath27 distributions are observed to be same .
this validates the relationship of @xmath1 and @xmath27 as given in eqn .
( [ eqn7 ] ) .
distributions of pions for central pb+pb collisions at @xmath3= 2.76 tev from the ampt model within rapidity range of -1.0 to 1.0 and @xmath36 gev .
@xmath27 distributions , obtained by fitting the @xmath2 distribution of each event and from the @xmath1 for are presented .
, scaledwidth=49.0% ] we note that the extracted temperature , @xmath27 , is a combination of kinetic freeze - out temperature ( @xmath37 ) and transverse flow velocity ( @xmath38 ) of the system : @xmath39 for pion , @xmath40 .
the event - by - event fluctuations of @xmath38 needs to be taken into account for calculating the fluctuation in kinetic temperature @xcite .
fluctuation in @xmath38 dominates over the fluctuation in @xmath37 for small systems ( e.g. pp ) @xcite , asymmetric ( e.g. , ppb ) @xcite and non - central collisions . for central au+au collisions at @xmath3 = 200 gev @xcite , @xmath41 and for central pb+pb collisions at @xmath3 = 2.76 tev @xcite , @xmath42 , which translate to @xmath37 as @xmath43 gev and @xmath44 gev , respectively by using blast - wave fit @xcite . for the present work ,
we consider 10% fluctuation in @xmath38 and calculate its effect on specific heat .
@xmath16 is calculated using the equation , @xmath45 the values of @xmath46 are obtained from the blast - wave fits to the @xmath2 distributions of identified particles . with this , we obtain the specific heat as the heat capacity per number of particles ( @xmath47 ) as ( @xmath0 = @xmath48 ) within the system .
let us put the specific heat calculated in the present scenario ( heat capacity per particle ) in perspective with quantities normally quoted in theoretical calculations .
for an ideal gas of particles of mass @xmath49 and degeneracy factor @xmath50 at temperature @xmath4 , zero chemical potential and volume @xmath18 , the number of particles @xmath51 can be expressed using boltzmann statistics : n & = & g + & = & g d^3x + & = & g d^3q + & = & g , where @xmath52 , @xmath53 $ ] and we have taken @xmath54 .
the energy @xmath55 is given by : e = g .
the heat capacity ( from eqn .
[ eqn1 ] ) can be written as , c&=&g d^3x ( ) + & = & g d^3q ( q^2 + ( ) ^2 ) + & = & g .
@xmath56 $ ] is a dimensionless quantity .
the specific heat is the heat capacity per unit phase space volume , c_v = c/ , where @xmath57 is an estimate of the phase space volume . in lattice calculations
one extracts the dimensionless quantity @xmath58 and investigate its temperature dependence @xcite , so in these calculations @xmath59 . however , in experiments it is simpler to measure the dimensionless quantity @xmath48 where @xmath47 is the charged particle multiplicity , and thus @xmath60 , where @xmath47 is taken as pseudorapidity ( @xmath35 ) density of charged particles at mid rapidity ( @xmath61 at @xmath62 ) .
we compare the experimental results to other model calculations for @xmath48 as in ref .
@xcite , where a parton and hadron cascade model , paciae has been used to compute @xmath48 .
we also compare with hrg where it is straightforward to obtain both , @xmath58 and @xmath48 .
the current continuum estimates of lattice qcd thermodynamics in the low temperature and density phase ( hadronic phase ) show good agreement with that of an ideal hadron resonance gas @xcite . assuming complete chemical equilibrium between all hadrons , the hadron chemical potential of the @xmath63 species @xmath64 can be written as _
i = b_i_b + q_i_q + s_i_s , [ eq.mui ] where @xmath65 , @xmath66 and @xmath67 are the baryon number , electric charge and strangeness quantum numbers of the @xmath63 hadron , respectively .
the hrg partition function @xmath68 in the grand canonical ensemble at @xmath69 can be expressed as : @xmath70 , \label{eq.pressurehrg}\end{aligned}\ ] ] where the sum runs over all hadrons and resonances up to mass @xmath712 gev as listed in the review of particle physics @xcite .
@xmath72 for mesons and @xmath73 for baryons . here
@xmath74 , @xmath75 , @xmath65 , @xmath66 and @xmath67 refer to the properties of the @xmath76th hadron species : its degeneracy factor , mass , baryon number , electric charge and strangeness respectively .
@xmath18 is the volume of the fireball under study .
@xmath77 is the modified bessel function of the second kind . from @xmath78 ,
all thermodynamic quantities could be computed . from @xmath78 , @xmath19 is obtained as follows : e & = & t^2 + _ i _ in_i[eq.energy ] where , n_i & = & t[eq.number ] .
from @xmath19 it is straightforward to compute @xmath16 using eq .
[ eqn1 ] .
the values of @xmath37 have been reported by experiments at rhic and lhc , as the final state particles give the information about @xmath37 from the particle spectra @xcite .
these are obtained by making combined fits to the identified particle spectra using the boltzmann - gibbs blast - wave model .
figure [ fig_tch_tkin ] gives the @xmath37 values for different beam energies and collision systems @xcite .
in addition , chemical freeze - out temperatures ( @xmath79 ) , extracted from the identified particle yield by using thermal model calculations @xcite , are also shown in the figure .
we find that the difference between @xmath79 and @xmath37 increases with the increase of beam energy .
lattice calculations indicate that @xmath0 is a monotonically increasing function of @xmath4 at zero @xmath5 .
thus we expect that the difference between the @xmath0 extracted at the chemical and kinetic freeze - out surfaces should also increase with beam energy following the trends of @xmath79 and @xmath37 .
we calculate @xmath0 from hrg model for two scenarios . in the first case ,
we compute at the chemical freeze - out surface using the extracted @xmath79 as well as @xmath5
. however , in the experiment one can only determine the @xmath0 at the kinetic freeze - out surface where the momentum exchange freezes .
the thermal conditions at the kinetic freeze - out surface are much different from that at the chemical freeze - out surface .
hence the fireball is expected to have different @xmath0 at the two surfaces . in the second case
, we try to estimate @xmath0 at the kinetic freeze - out surface using @xmath37and zero hadron chemical potentials . for both scenarios ,
we calculate @xmath58 and @xmath48 for a wide range of beam energy , from @xmath3 = 1.91 gev to 2.76 tev .
the results of @xmath0 are shown in the fig .
[ specific1 ] .
it is observed that the trend of @xmath0 as a function of @xmath3 is similar to the nature followed by chemical and kinetic freeze - out temperatures .
the value of @xmath48 corresponding to @xmath37 shows a sharp drop with increase of energy , and beyond @xmath3 = 62.4 gev , the rate of decrease is very slow .
a better estimate of @xmath0 in hrg could be made with realistic hadron chemical potentials taking into account the conservation of hadron number from the chemical to the kinetic freeze - out surfaces .
= 2.76 tev @xcite .
thermal model calculation @xcite to @xmath79 is also shown .
, scaledwidth=49.0% ] as a function of collision energy from the hrg model for two temperature settings , @xmath79 with finite @xmath80 , and @xmath37 , and two phase space volumes , @xmath81 and @xmath82 .
lattice calculation for @xmath0 at @xmath83 mev @xcite and stefan - boltzmann limit are indicated in the figure .
, scaledwidth=49.0% ] recent lattice calculations for @xmath0 have been reported @xcite as a function of temperature .
the lattice results are at zero baryonic potential , hence only relevant at the lhc and higher energies . the value of @xmath0 as indicated in the fig .
[ specific1 ] is for @xmath83 mev , corresponding to the qcd transition temperature .
it is seen that at the transition temperature and below , hrg results of @xmath58 agree well with lattice calculations @xcite . the steffan - boltzmann non - interacting gas limit ( @xmath0 @xmath84 66 ) is also shown in the figure .
experimental data for @xmath1 distributions have been reported by experiments at sps and rhic @xcite . in the left panel of fig .
[ fig_auau ] we present the @xmath1 distributions from the star experiment @xcite for the 5% most central au+au collisions at @xmath3 = 20 , 62.4 , 130 and 200 gev .
the results are shown for charged particle tracks within @xmath85 and @xmath36 gev .
the solid points are the event - by - event @xmath1 distributions from the experimental data , whereas the open circles are the corresponding results for mixed events .
the mixed events are created by randomly selecting charged particles from different events .
the mixed event distributions contain all the systematic effects arising from the detector effects , such as efficiency and acceptance , as well as include statistical fluctuations .
the non - statistical or dynamical fluctuations in @xmath1 can be extracted by subtracting the width of the mixed event distribution from that of the real data . it has been observed that the @xmath1 distributions are nicely described by using the gamma ( @xmath86 ) distribution @xcite : @xmath87 here @xmath88 represents the @xmath1 .
the mean ( @xmath89 ) and standard deviation ( @xmath90 ) of the distribution are related to the fit parameters ( @xmath13 and @xmath91 ) by @xmath92 and @xmath93 .
both the real and mixed event @xmath1 distributions are fitted with the @xmath86 function and the fits are shown by the solid and dashed lines , respectively , in the left panels of fig .
[ fig_auau ] .
the fitted distributions are used to generate a large number of @xmath1 values for which corresponding @xmath27 values are calculated from eqn .
the resulting histograms represent event - by - event @xmath27 distributions , which are shown in the right panels of fig .
[ fig_auau ] for both real data and mixed events .
these distributions are also fitted by the @xmath86 function as shown by the solid and dashed lines for data and mixed events , respectively .
table [ table1 ] lists the fit parameters for event - by - event @xmath27 distributions for data and mixed events . the system size dependence of @xmath1 and @xmath27 distributions
have been studied with the star experimental data of top 10% central cu+cu collisions at @xmath3 = 62.4 and 200 gev @xcite .
the results are presented in in fig .
[ fig_cucu ] . corresponding @xmath86 distribution fit parameters to the event - by - event @xmath27 distributions for top 10% central collisions are tabulated in table [ table2 ] . .
the event - by - event @xmath27 distributions for central ( top 5% ) au+au collisions are fitted by the gamma function .
table gives fhe fit parameters , @xmath13 and @xmath91 along with mean ( @xmath89 ) and standard deviation ( @xmath90 ) . [ cols="^,^,^,^,^,^ " , ] [ table2 ] from these two figures and the given tables for @xmath1 and @xmath27 distributions for au+au and cu+cu collisions at rhic energies , we can infer that : ( a ) the mean values of the event - by - event @xmath1 and @xmath27 consistently increase with the increase of beam energy , ( b ) the widths of the distributions decrease with the increase of beam energy .
in addition , the widths for cu+cu system are observed to be larger compared to the corresponding widths of the au+au system .
this may be because of the smaller system size for cu+cu compared to au+au system .
experimental data for event - by - event @xmath1 distributions are not available for pb+pb collisions at lhc energies @xcite .
the string melting mode of ampt model is used to generate central ( top 5% ) pb+pb collisions at @xmath3 = 2.76 tev .
the @xmath1 and @xmath27 distributions are constructed from these generated events as shown in fig 1 .
this distribution will be used to extract specific heat at the lhc energy .
the widths of the @xmath27 distributions are strongly affected by statistical fluctuations , which need to be subtracted as the heat capacity is related only to the dynamical part of the fluctuation .
the width contains two components : @xmath94 @xmath95 values are obtained by subtracting the widths of the @xmath27 distributions for mixed events from the real data . with this , eqn .
[ eqn9 ] is expressed as : @xmath96 the heat capacity @xmath16 is calculated from eqn .
[ eqnfinal ] by using the values of @xmath37 from fig .
[ fig_tch_tkin ] .
knowing the heat capacity , the specific heat , @xmath0 is obtained by dividing @xmath16 by number of charged particles in the system . since the experimental results presented here are at mid - rapidity , we have divided the value of @xmath16 by charged particle multiplicity at mid - rapidity @xcite to obtain the specific heat .
this is presented in fig .
[ specific1 ] for au+au and cu+cu collisions at rhic energies .
the estimated @xmath48 for the lhc energy from the ampt model using fig .
[ fig_ampt ] is also shown in the figure .
the errors in the data points are estimated mainly from the following sources : ( a ) error in extraction of @xmath37 using the blast - wave fits , ( b ) error in charge particle multiplicity density , and ( c ) error in @xmath1 as reported in the experimental data .
the error in @xmath37 takes into account the spread in the value of ( @xmath38 ) .
it is observed that @xmath48 has a sharp drop from @xmath3 = 20 gev to 62.4 gev , beyond which the decrease is rather slow up to the lhc energy . as a function of collision energy for central au+au and cu+cu collisions at rhic energies .
result from ampt model is given for the lhc energy at @xmath3 = 2.76 tev .
hrg calculations at @xmath37 are shown in the figure .
model calculations for three different scenarios from ref .
@xcite are superimposed on the experimental results .
, scaledwidth=49.0% ] hrg calculations for @xmath48 with @xmath37 are superimposed in fig .
[ specific2 ] .
these results follow the experimental data points quite well . in ref .
@xcite , specific heat for central ( top 5% ) au+au collisions at rhic energies are discussed using a parton and hadron cascade model , called paciae .
the results of the model calculations are presented for three cases : hadronic matter in the final state ( hm ) , quark - gluon matter in the partonic initial state ( qgm ) , and hadronic matter via quark - gluon matter ( hm via qgm ) .
these results for au+au collisions are also presented in fig .
[ specific2 ] .
the results of these models miss the experimental data point at @xmath3 = 20 gev , but can explain the data at higher energies .
the paciae is a transport model with lo pqcd cross sections considered between partons
. it might be possible that at lower beam energies , the lo pqcd description is not appropriate .
in an earlier publication by r. korus _ et al . _ ( ref .
@xcite ) , the experimental data of @xmath2 correlations from the na49 experiment @xcite for pb+pb collision at laboratory energy of 158 gev had used to calculate specific heat , which yielded a value of @xmath14 .
the large error bars of these results made the reported results insignificant .
one of the possible reasons for the large errors is the low particle multiplicity which gives a significant hindrance to the calculation of dynamic temperature fluctuations @xcite . at the sps energies , @xmath1 distributions
have been reported by na49 collaboration for laboratory energies of 20 , 30 , 40 80 and 158 gev @xcite and the ceres collaboration at 40 , 80 and 158 gev @xcite . in both of the data sets , the @xmath1 distributions for real data and mixed event are indistinguishable .
these are also prominent in the ratio plots of real and mixed events as shown in ref .
thus the extraction of dynamic fluctuation in temperature and so the specific heat is not possible . in the present work ,
we have probed much higher energy collisions , where the charged particle multiplicities in each event are large , allowing for the extraction of the dynamical part of the temperature fluctuation by overcoming the statistical fluctuations .
the results of specific heat for cu+cu collisions are close to that of au+au collisions .
this shows that although a large change of volume happens in going from cu+cu to au+au systems , the two systems are not very different thermodynamically .
several sources of uncertainty may affect the extraction of specific heat . in the context of the results presented in fig .
[ specific2 ] , three sources of uncertainty , _ viz .
_ , effect of finite particle multiplicity , spread of @xmath37from the fits of @xmath2distributions , and the radial flow fluctuations , have been discussed .
apart from these , it is worthwhile to point out some other uncertainties .
fluctuations in the impact parameter of the collision and thus the fluctuation in the number of participating nucleons gives an uncertainty to the event - by - event mean @xmath2distribution .
choice of narrow bins in centrality has been made to minimize this uncertainty .
another source of uncertainty may come from the choice of the @xmath2window .
the lower bound of the @xmath2window needs to be chosen properly in order to reduce the final state effects such as resonance decay and hadronic scattering .
similarly , the upper limit on pt needs to be chosen such that the effect from mini - jets and jets are minimized .
although it is implicit that radial flow fluctuations are minimum for central collisions because of inherent symmetry of the system , its detailed study can be made using an event - by - event hydrodynamic model .
these studies will help to pin down the errors of the results extracted in fig .
[ specific2 ] .
we have studied the excitation energy dependence of specific heat of hadronic matter formed in heavy - ion collisions corresponding to rhic and lhc energies . in the present work , dynamical component of the temperature fluctuation
is calculated from @xmath1 distributions . from this , the specific heat is obtained as heat capacity per charge particle .
we employ the hrg model to calculate heat capacity from the variation of energy of the system with temperature .
results of the hrg calculations are close to the data . with increase of collision energy ,
@xmath0 shows a sharp drop from low energy till @xmath3 = 62.4 gev , beyond which the rate of decrease is very slow . in this regard ,
we look forward to results of bes program of rhic , where the collision energy and centrality dependences of @xmath0 are expected to provide important signatures for the onset of the qgp phase transition . in order to probe the qcd critical point ,
we propose a finer scan of beam energies for the second phase of bes program ( bes - ii ) from 7.7 gev to 62.4 gev .
a sudden change in @xmath0 is expected at a particular beam energy within this range .
studies of heat capacity at high baryon density and lower temperatures accessible at facility for antiproton and ion research ( fair ) would be of high interest , unless it is critically challenged by statistical fluctuations
. predictions for @xmath0 at the lhc at @xmath3 = 2.76 tev are made using different models .
it will be interesting to obtain @xmath0 at the highest lhc energy of @xmath3 = 5.02 tev in order to make a direct comparison to lattice calculations . in literature
, it has been also proposed to calculate thermal conductivity from transverse energy ( @xmath97 ) fluctuations , which can be explored in future studies .
the excitation energy dependence of @xmath0 provides important information regarding the thermodynamic properties , such as , heat conductivity , speed of sound ( @xmath98 ) , compressibility ( @xmath99 ) , etc .
, which may reveal better understandings of the matter formed in relativistic nuclear collisions . * acknowledgement * sc acknowledges
centre for nuclear theory " [ pic xii - r@xmath100d - vec-5.02.0500 ] , variable energy cyclotron centre for support .
this research used resources of the lhc grid computing centre at the variable energy cyclotron centre , kolkata .
|
we report the excitation energy dependence of specific heat ( @xmath0 ) of hadronic matter at freeze - out in au+au and cu+cu collisions at the relativistic heavy ion collider energies by analyzing the published data on event - by - event mean transverse momentum ( @xmath1 ) distributions .
the @xmath1 distributions in finite @xmath2 ranges are converted to distributions of effective temperatures , and dynamical fluctuations in temperature are extracted by subtracting widths of the corresponding mixed event distributions . the heat capacity per particle at the kinetic freeze - out surface
is presented as a function of collision energy , which shows a sharp rise in @xmath0 below @xmath3 = 62.4 gev .
we employ the hadron resonance gas ( hrg ) model to estimate @xmath0 at the chemical and kinetic freeze - out surfaces .
the experimental results are compared to the hrg and other theoretical model calculations .
hrg results show good agreement with data .
model predictions for @xmath0 at the large hadron collider energy are presented .
|
in mechanics and physics exist two ways to describe and model a particulate , inhomogeneous material like powder , ceramics or concrete .
the first approach is based on continuum theory and relies on empirical assumptions about the macroscopic material behavior @xcite . for a summary of recent progress on the macroscopic modeling of sintering processes
see refs .
@xcite and the references therein .
this ansatz is complemented by a `` microscopic '' description of the material on the particle or grain level where the particles and their interactions are modeled one by one @xcite .
the former involves stress , strain and plastic yield conditions @xcite , whereas the latter deals with local force - deformation laws for each contact @xcite .
the macroscopic approach neglects the microstructure due to its nature and often isotropy is assumed @xcite .
in contrast to network models or models with fixed topology @xcite we will follow the path of microscopic , dynamic modeling , where no assumptions about either geometry , topology , homogeneoity , or isotropy of the powder packing are involved . however , we have to assume certain contact force - laws and , furthermore , restrict ourselves to two dimensions . in order to test
the approach , a model system in a two - dimensional box filled with cohesive - frictional disks of different sizes , see sec .
[ sec : model ] , is examined by means of a `` microscopic '' discrete element method ( dem ) .
the microscopic interaction model for plastic deformations , friction , and cohesion is discussed and time- , temperature- , and history - dependent behavior is introduced and rationalized in sec .
[ sec : demmodel ] .
the long time sintering of a block of material is simulated and the results are discussed in sec . [
sec : results ] .
one possibility to obtain information about the material behavior is to perform elementary tests in the laboratory .
however , because it is difficult to observe what is going on inside the material , an alternative way is the simulations with the discrete element model ( dem ) @xcite .
the numerical `` experiment '' chosen is a bi - axial box set - up , see fig .
[ fig : schemes ] , where the left and bottom walls are fixed , and a stress- or strain - controlled deformation is applied to the other walls . in the future ,
quantitative agreement between experiment and simulation has to be achieved , however , this issue is far from the scope of this paper .
stress control means that the wall is subject to a predefined external pressure that , in equilibrium , is cancelled by the stress , which the material exerts on the wall . in a typical `` experiment '' , the particles
are rapidly compressed with constant , isotropic pressure and then left alone for some time and at constant temperature , typically close below the melting point .
the stress - controlled motion of the walls is described by the differential equation @xmath0 where @xmath1 is the mass of the wall .
values of @xmath1 large as compared to the sample mass lead to slow adaption and vibrations , whereas small values allow for a rapid adaption to the actual situation .
three forces are active : ( i ) the force @xmath2 due to the bulk material , ( ii ) the force @xmath3 due to the external pressure , and ( iii ) a viscous force which damps the motion of the wall so that oscillations are reduced . inside the system , @xmath4 disks with radii @xmath5 ( @xmath6 , @xmath7 , @xmath4 ) and
height @xmath8 are placed .
the radii are drawn from a homogeneous distribution with mean @xmath9 and relative width @xmath10 so that @xmath11 $ ] . the particle - particle interactions and the parameters involved
are discussed in the next section .
the elementary units of granular materials are mesoscopic grains which deform under stress , possibly yield , change their properties with time , and can behave different for different temperatures .
since the realistic modeling of the deformations of the particles is much too complicated to allow for a subsequent many - particle simulation , we relate the interaction force to the overlap @xmath12 of two particles , see fig .
[ fig : force ] . in the absence of long - range forces
, an interaction takes place only if particles are in contact and thus @xmath13 . in that case , the forces are split into a normal and a tangential component denoted by @xmath14 and @xmath15 , respectively . if all forces @xmath16 acting on the particle @xmath17 , either from other particles , from boundaries or from external forces , are known , the problem is reduced to the integration of newton s equations of motion for the translational and rotational degrees of freedom @xmath18 with the mass @xmath19 of particle @xmath17 , its position @xmath20 the total force @xmath21 , its moment of inertia @xmath22 , its angular velocity @xmath23 , the total torque @xmath24 , and the center - contact vector @xmath25 .
the integration of the equations of motion is performed with a standard molecular dynamics verlet algorithm together with verlet - table neighborhood search @xcite . two particles @xmath17 and @xmath26 at positions @xmath20 and @xmath27 , with radii @xmath5 and @xmath28 ,
interact only if they are in contact so that their overlap @xmath29 is positive , @xmath30 , with the unit vector @xmath31 pointing from @xmath26 to @xmath17 .
the force on particle @xmath17 , from particle @xmath26 can be written as @xmath32 , with @xmath33 perpendicular to @xmath34 . in this subsection ,
the normal forces are discussed . as first step ,
we discuss the time- and temperature - independent behavior of the contact forces between a pair of particles . for this , we modify and extend the linear hysteretic spring model @xcite .
it is the simplest version of some more complicated nonlinear - hysteretic force laws @xcite , which reflects the fact that at the contact point , plastic ( permanent ) deformations may take place .
the repulsive ( hysteretic ) force can be written as @xmath35 with @xmath36 , see fig .
[ fig : force ] . during
the initial loading the force increases linearly with the overlap @xmath12 , until the maximum overlap @xmath37 is reached , which is kept in memory as a _ history parameter_. the line with slope @xmath38 thus defines the maximum force possible for a given @xmath12 . during unloading the force drops , on the line with slope @xmath39 , from its value at @xmath37 down to zero at the force - free overlap @xmath40 reloading at any instant leads to an increase of the force along this line , until the maximum force is reached ; for still increasing @xmath12 , the force follows again the line with slope @xmath38 and @xmath37 has to be adjusted accordingly .
unloading below @xmath41 leads to negative , attractive forces until at the overlap @xmath42 the minimum force @xmath43 , i.e. the maximum attractive force , is obtained as a function of the model parameters @xmath38 , @xmath39 , @xmath44 , and the history parameter @xmath37 .
further unloading leads to attractive forces on the branch @xmath45 . and
@xmath39 , is reached for @xmath46 , so that @xmath47 .
this would lead to a discontinuity at @xmath48 that is avoided by using finite @xmath44 . ]
the cone formed by the lines with slope @xmath38 and @xmath49 defines the range of possible force values .
if a force would fall outside the cone , it is forced to remain on the limit lines .
departure from these lines into the cone takes place in the case of unloading and reloading , respectively . between these two extremes , unloading and
reloading follow the same line with slope @xmath39 .
possible equilibrium states are indicated as circles in fig .
[ fig : force ] , where the upper and lower circle correspond to a pre - stressed and stress - free state , respectively . in the case of collisions of particles and for large deformations
, dissipation takes place due to the hysteretic nature of the force - law . for small displacements around some equilibrium state
, the model does not contain dissipation .
therefore , in order to allow for stronger dissipation and thus faster relaxation , a viscous , velocity dependent dissipative force in normal direction , @xmath50 is assumed with some damping coefficient @xmath51 .
the half - period of a vibration around the equilibrium position , see fig .
[ fig : force ] , can be computed for arbitrary values of @xmath38 and @xmath44 , as long as the overlap fulfills the condition @xmath52 . in this case
, @xmath39 determines the stiffness and one obtains a typical response time on the contact level @xcite , @xmath53 the eigenfrequency of the contact , the rescaled damping coefficient @xmath54 , and the reduced mass @xmath55 .
the time - step of the simulation @xmath56 has to be chosen such that @xmath57 for a proper integration of the equations of motion , so that we chose a typical time - step @xmath58 after the model parameters @xmath39 and @xmath51 are specified . in order to account for the fact that a larger contact surface leads to a larger contact stiffness , the coefficient @xmath39 is made dependent on the maximum overlap history parameter @xmath37 ( and thus on the force - free overlap @xmath41 ) , as long as the overlap is below the threshold @xmath59 that corresponds to the `` complete melting '' of the particles .
complete melting is here the limit of an incompressible liquid that is contained in the model , however , neither discussed in detail nor verified for reasons of brevity .
the overlap @xmath59 corresponding to the stress - free fluid , is computed such that the volume fraction in the system equals unity .
the volume fraction of a dense packing of rigid spheres is @xmath60 if all the material would be melted , the volume fraction is @xmath61 which leads to the ratio of fluid and solid volumes @xmath62 the minimum radius for the incompressible melt is @xmath63 which corresponds to the maximum overlap @xmath64 in two dimensions . in the following , we use @xmath65 in order to magnify the range of possible non - fluid overlaps .
the stiffness is maximal in the fluid limit for @xmath66 , which corresponds to @xmath67 , and varies between @xmath38 and @xmath39 for smaller overlaps , so that @xmath68 for large overlaps ( in the fluid regime ) , the stiffness and the force is thus only dependent on @xmath39 , independent of @xmath38 . for smaller overlaps
both @xmath38 and @xmath39 affect the force together with the history of this contact .
the stiffness in the incompressible fluid should diverge , however , for reasons of numerical stability , we have to limit the maximum stiffness to @xmath39 .
larger stiffness values would require smaller time - steps which would reduce the simulation efficiency .
therefore , the model does not really take the incompressibility of the liquid into account , see fig . [
fig : force2 ] . in the fluid no plastic deformation can take place so that @xmath41 is fixed and can not be shifted to larger force free overlaps in contrast to the hysteretic model described above . in summary ,
the hysteretic stiffness model takes into account an increasing stiffness with increasing overlap .
the first loading is plastic with low stiffness , and subsequent un- and reloading are stiffer because the material was initially compressed . as a consequence ,
also the maximum cohesive force depends on the maximum compression which was experienced by the contact during its history .
if the material is compressed so strong that the liquid density is reached , the force - free overlap is equal to the fluid equivalent overlap and the material behaves like a fluid . if a solid or a liquid ( we assume a simple material here - not like water - where the density dependence on the temperature is continuous through the phase transition ) is heated , in general , its volume increases so that its density decreases .
therefore , we assume a temperature dependent density of the single particles ( disks with radius @xmath69 and height @xmath8 ) : @xmath70 with the density change per unit temperature @xmath71 . this corresponds ( in linear approximation ) to a change of the particle radius @xmath72~ , \label{eq : at}\ ] ] with the relative change of the radius per unit temperature @xmath73 .
this approximation can be used if the range of temperatures is rather narrow and the changes per unit temperature are very small . in the following , we use @xmath74k@xmath75 , so that the particle radius is changed by 0.01 per - cent if the temperature is changed by one kelvin . in the interesting range of temperatures between a low temperature ( 80@xmath76c ) and some melting point ( 120@xmath76c ) , the radius changes by one per - cent , accordingly . for the temperature dependence , we focus on an inhomogeneous material with a melting temperature @xmath77 and assume that the material behaves static , as described above , if the temperature @xmath78 is much smaller than the melting temperature .
the behavior of the stiffness @xmath38 is schematically shown in fig .
[ fig : ktschem ] as a function of the temperature .
the material becomes softer when @xmath79 becomes small and will lose all stiffness in the limit @xmath80 . the temperature range in which the melting takes place is quantified by @xmath81 . in the transition regime @xmath82 ,
the particles are significantly softer than in the _ cold _ limit @xmath83 . in the _ hot _ and
the liquid regime , @xmath84 , one has @xmath85 and the particles lose their nature , however , the ` incompressibility ' is accounted for with a stiffness @xmath39 , as defined in eq .
( [ eq : k2 ] ) , and @xmath66 is fixed . when the temperature is increased to a rather large value , close to the melting point
, two particles under stress and in equilibrium due to compressive forces will lose stiffness and thus will deform stronger so that their overlap becomes larger .
therefore , we assume for the stiffness coefficient @xmath86 \nonumber \\ & = & \frac{k_1}{2 } \left [ 1 + \tanh ( \tau ) \right ] ~ , \label{eq : k1t}\end{aligned}\ ] ] with the dimensionless temperature difference @xmath87 , and the typical range of considerable temperature dependency @xmath81 , where the temperature has a strong effect on the stiffness of the particles , see fig . [
fig : ktschem ] .
if experimentally available in the future , the function @xmath88 can be replaced by any other function @xmath89 that decays from unity to zero at @xmath90 .
when @xmath38 is reduced due to an increase in temperature ( @xmath91 ) , we assume that @xmath92 remains constant , so that one obtains a larger force - free overlap @xmath93 \delta^+_{\rm max}$ ] .
thus the material volume shrinks due to sintering at the contact level . note that @xmath39 is not changed directly when @xmath94 is decreased , see the left panel in fig . [
fig : force3 ] or eq .
( [ eq : k2 ] ) . the cohesion in this model ,
however , is directly affected by a change of @xmath94 , see eq .
( [ eq : kc ] ) . in a pre - stressed situation ,
corresponding to a finite confining force at the contact , also @xmath95 is shifted in order to balance the confining force but only after @xmath96 became smaller than the confining force .
if , as an example , the material has a melting point @xmath97c with a range of softening of @xmath98c , for a temperature of @xmath99c , the stiffness @xmath38 is reduced to @xmath100 of its cold - limit value , for a temperature of @xmath101c the stiffness is @xmath102 , and for a temperature of @xmath103c the stiffness is only @xmath104 . for a temperature of @xmath105c only a stiffness of
@xmath106 remains .
the fact that there is a remaining stiffness above the melting temperature can be attributed to the inhomogeneity of the material , i.e. not all the material melts at the same temperature .
if later in time the temperature is decreased again , @xmath94 is adjusted according to eq .
( [ eq : k1 t ] ) , but since the melted ( sintered ) area around the contact point will not return to its previous state , we now assume that @xmath107 , so that the maximum overlap increases to the value @xmath108 $ ] , see the right panel in fig .
[ fig : force3 ] .
therefore , a temperature cycle involving a temperature close to the melting temperature leads to a contact situation similar to the one obtained through some larger maximum compressive force . the contact deformation / area and
thus the force - free overlap become larger due to the partial melting of the surface and also the stiffness is increased accordingly .
+ the temperature dependence thus can lead to changes of the stiffness and increases the overlap ( deformation ) of the particles .
it is effective only if the temperature is changed the time dependence will be introduced in the next subsection . in the material
there are several time dependent processes taking place .
since we are interested in the long time behavior of the material , we assume that heat conduction and equilibration take place instantaneously , as long as temperature changes are small and slow . the realistic simulation of heat - conduction in the sample is far from the scope of this study .
therefore , the particle size is adjusted directly with the temperature according to eq .
( [ eq : at ] ) .
in addition to the direct effect of a temperature change on the particle size and the stiffness , the material may change its internal , atomistic structure such that defects heal and disappear .
this effect will occur mainly in the regime of high temperatures close to the melting point .
therefore , in order to account for such slow microscopic processes , a time dependence is introduced that leads to a change of the material stiffness @xmath38 with time .
the change takes place extremely slowly with an algebraic time dependence , so that @xmath109 lags behind when varying from its actual value to the desired , final value @xmath94 , as defined in the above equation ( [ eq : k1 t ] ) .
when the temperature is increased from a small value @xmath110 to @xmath78 , then @xmath111 changes to @xmath94 following the law @xmath112 ~ , \label{eq : k1time}\ ] ] which corresponds to the rate of change @xmath113 ^ 2}{k_1(t ) \
, t_0 } ~ , \label{eq : dk1time}\ ] ] with the time scale @xmath114 on which a typical change takes place .
note that @xmath38 , @xmath94 , and @xmath109 are different , in general , and correspond to the maximum @xmath38 , the temperature dependent @xmath115 , and the time dependent @xmath109 that tends towards @xmath94 .
the sign in eq .
( [ eq : dk1time ] ) is chosen according to the sign of @xmath116 $ ] in eq .
( [ eq : k1time]).^2 = y|y|$ ] . for the numerical integration ,
we remark , that the value of the actual stiffness should be @xmath117 , otherwise we obtained a numerically instable behavior in some situations ] assume @xmath118c and @xmath99c , so that @xmath119 corresponds to @xmath120 .
the stiffness as a function of time is plotted in fig .
[ fig : k1time ] for the time constant @xmath121s .
note that , due to the factor @xmath94 in the denominator of eq .
( [ eq : dk1time ] ) , the change of stiffness is faster for higher temperatures . in the hot limit ,
changes take place very rapidly , whereas in the cold limit changes are extremely slow .
finally , we note that the adaption / relaxation of @xmath109 to the desired @xmath94 value follows the above equations in all cases except when the temperature is going down and the @xmath122 .
in that situation the contacts freeze rapidly and thus have to become strong as fast as the system cools down .
the cohesive properties of a particle contact depend on the temperature , in so far that a melted contact should have weak tensile and compressive strength .
therefore , we couple the cohesive parameter @xmath44 to the magnitude of @xmath109 , see eq .
( [ eq : k1time ] ) , which decreases with temperature increasing .
in addition , in order to take into account a reduced tensile strength of a soft contact with weak deformation and thus small overlap , the cohesion is directly related to the stiffness @xmath123 , see eq .
( [ eq : k2 ] ) .
thus , we propose @xmath124 this is an arbitrary choice for the cohesive force factor @xmath44 , but as long as no detailed experimental results are available , we stick to this empirical law , see fig . [
fig : force2 ] for a schematic picture .
the force in tangential direction is implemented in the spirit of cundall and strack @xcite who introduced a tangential spring in order to account for static friction .
various authors have used this idea and numerous variants were implemented , see @xcite for a summary and discussion .
since we combine cohesion and friction and introduce time and temperature dependencies , it is convenient to repeat the model and define the implementation .
the tangential force is coupled to the normal force via coulomb s law , i.e. @xmath125 , where for the limit case one has sliding friction and for the case of small forces @xmath126 , one has static coulomb friction .
the latter situation requires an elastic spring in order to allow for a restoring force and a non - zero remaining tangential force in static equilibrium due to activated coulomb friction . as a consequence of the cohesion force in normal direction , attractive forces are possible so that coulomb s law has to be modified @xmath127 with the minimum ( maximum attractive ) force @xmath128 the equal @xmath129 in eq .
( [ eq : ftc0 ] ) corresponds to ( fully ) activated , sliding friction , while the smaller @xmath130 corresponds to static friction .
thus the tangential force is related to the normal force relative to the point of cohesion - failure . for normal forces larger than @xmath131
, friction is always active and the amplitude is proportional to @xmath132 and @xmath133 .
this can lead to a stable equilibrium of the solid also at @xmath134 and activated tangential static friction .
note that @xmath131 tends towards zero for vanishing overlap , so that the difference from the original model also vanishes for vanishing overlap .
if @xmath135 , the tangential force is active , and we project the tangential spring @xmath136 into the actual tangential plane by subtracting a possible ( small ) normal component .. @xmath137 this action is relevant for an already existing spring , if the spring is new , the tangential force is zero anyway , however , the change of the spring - length defined below , is well defined .
next , the tangential velocity is computed @xmath138 with the total relative velocity @xmath139 in the next step we calculate the tangential test - force as the sum of the tangential spring and a tangential viscous force ( similar to the normal viscous force ) @xmath140 with the tangential spring stiffness @xmath141 , with a typical stiffness ratio @xmath142 , see @xcite , and the tangential dissipation parameter @xmath143 . if @xmath144 , with @xmath145 , one has static friction and , on the other hand , if @xmath146 , sliding friction is active . in the former case ,
the tangential test force is incremented @xmath147 to be used in the next iteration in eq .
( [ eq : chimap ] ) , and the force @xmath148 from eq .
( [ eq : f_static ] ) is used . in the latter case , the tangential spring is adjusted to a length which is consistent with coulomb s condition @xmath149 with the tangential unit vector , @xmath150 , defined by the tangential spring , and the coulomb force is used . inserting @xmath151 into eq .
( [ eq : f_static ] ) leads to @xmath152 in the next iteration . and @xmath153 are not necessarily parallel in three dimensions .
however , the mapping in eq .
( [ eq : xinew ] ) works always , rotating the new spring such that the direction of the frictional force is unchanged ] in short notation this reads @xmath154 note that the tangential force described above is identical to the classical cundall - strack spring only in the limit @xmath155 and @xmath156 . besides the combination of the cohesive and the frictional force , also the tangential dissipation is non - standard .
furthermore , we remark that the cohesion could also be coupled to friction in the sense that a broken contact looses its tensile strength when it is assumed brittle , so that @xmath157 , i.e. if one has a sliding contact with @xmath158 . would not be affected by the type of the contact being either sliding or sticking . in this study we assume , as an arbitrary , possibly inconsistent choice , @xmath159 , thus disregarding cohesion in the sliding situation .
the effect of this choice has to be examined in more detailed elsewhere . ] in parallel to the change of normal stiffness , the tangential stiffness is always kept in a constant ratio to @xmath39 so that @xmath160 since the stiffness in tangential direction is based on the same arguments as the material stiffness in normal direction .
the friction is coupled to the temperature dependent value of the stiffness @xmath109 , because friction should not be present in a liquid at large enough temperatures , so that @xmath161 thus friction is modified together with the changes in normal direction .
no new ideas are introduced for the tangential forces .
in this section , the sintering model is applied to the sintering process of a particulate material sample .
the material is initially a loose powder and first has to be prepared at low temperature from time @xmath114 to time @xmath162 , see fig.[fig : schemnt ] .
the preparation takes place with a system as described in section [ sec : model ] with isotropic external pressure @xmath163 or @xmath164 .
pressure is measured in units of nm@xmath75 due to the two - dimensional nature of the model , i.e. @xmath165 has the same units as the spring stiffness @xmath39 .
the correct units of the pressure can be obtained by division through the length of the cylinders ( particles ) .
since the length is rather arbitrary , we drop the unit of pressure in the following for the sake of simplicity . due to the linearity of the model , some stress @xmath166
can be rescaled / non - dimensionalized with the stiffness @xmath39 and has thus to be read as dimensionless quantity @xmath167 .
however , a detailed study of the scaling behavior of the stresses is far from the scope of this study .
note that the pressure is therefore related to the typical overlap of two particles : a pressure of @xmath168 corresponds to @xmath169 or @xmath170 , while @xmath171 corresponds to approximately @xmath172 .
this can also be interpreted as mean normal force @xmath173
. however , these are rough estimates only , since the contact forces and the overlaps are strongly varying in magnitude for one situation , as will be discussed in more detail below . the other system parameters are summarized in table [ tab : tabsys ] , where multiple numbers mean that a series with the corresponding values was performed .
.summary of the system properties and material parameters as used in the simulations [ cols="<,<,^",options="header " , ] the preparation of the sample consists of an initial relaxation period at constant temperature @xmath174c until time @xmath162 , when the system is heated up to @xmath175c between time @xmath162 and @xmath176 . during the sintering time @xmath177 , the system is allowed to sinter with a much shorter relaxation time @xmath178 .
this ` trick ' allows for a long time sintering simulation while keeping the simulation time small . during sintering ,
the time axis should be stretched by a factor of @xmath179 in order to obtain the real - time behavior . at the end of the sintering process , at time @xmath180
, the sample is slowly cooled down and , at time @xmath181 allowed to relax at constant temperature until time @xmath182 . with this finished sample
, tests will be performed later , but first we discuss the preparation process . for the preparation of the sample , we use the times @xmath183 , @xmath184 , @xmath185 , different sintering times @xmath186 , @xmath187 , and @xmath188 . note that the time unit is arbitrary , since a slower preparation procedure did not lead to noteably different results , i.e. the process takes place in the quasi - static limit .
only the long - time sintering is affected by a change of the relaxation time @xmath114 .
the variation of @xmath109 with time is plotted in fig . [
fig : k1_prepare ] , where the algebraic , slow decay of the stiffness with time becomes evident .
the number of particles was @xmath189 , the sidestress @xmath168 , and the other parameters are given in table [ tab : tabsys ] .
the longer the sintering time @xmath177 , the lower the value of @xmath38 gets . for short sintering time ,
the lowest values are never reached , because the system is cooled down before the sintering is finished . at the end of the sintering time , @xmath38 is increasing during the cooling process of the sample and reaches its inital value .
however , the melting and sintering of the contacts is _ not reversed _ , as becomes evident when plotting the volume fraction @xmath190 in fig .
[ fig : nu_prepare ] .
the initial preparation step leads to a rather low density of @xmath191 . at time @xmath162 ,
the relaxed sample is heated up and , at the same time , the wall friction is switched off .
the latter has an immediate effect , a slight increase in density is possible due to reorganizations .
the increased temperature becomes only effective at time @xmath176 , when the relaxation time @xmath114 is decreased in order to accelerate the evolution of the system .
note that the increase in density due to the sintering seemingly appears quite rapid due to the quenched time axis during sintering .
up to the beginning of the sintering process all densities are equal due to an identical preparation of the sample . for increasing duration of sintering , @xmath177 ,
the density successivly increases with @xmath177 .
this effect is strongest at the beginning , but continues also for longer times .
below , we will discuss different reasons for this increase in density in more detail . at the end of the sintering process ,
the system is cooled down , corresponding to the final , small increase in density , and then relaxes to the final configuration . as a control simulation , also
three particles were simulation and sintered in the same way as the larger samples with @xmath192 and @xmath189 above .
a sample with only three particles has almost no possibility for a rearrangement after the initial load is applied ( actually , the three particles were either arranged on a triangle or on a line and did not change their configuration for the cases observed ) .
this artificial test - case shows that the strong increase in density observed for larger samples is caused by re - arrangements of the packing and not only due to the force model .
more quantitatively , the sintering process leads to a densification of about 5% and 16% for @xmath171 and @xmath168 , respectively , whereas the densification due to the contact law , without rearrangements of the packing , can be estimated to the respective magnitudes of about 0.2% and 8% ( data not shown here ) .
thus , for low confining pressure , reorganizations are much more important than in the case of large external pressure at least for the parameters used here . in this spirit ,
network - models or dem simulations with fixed arrangements @xcite can not account for the reorganization of the packing , which is of eminent importance during the sintering process . another control simulation without thermal expansion of the particles , @xmath193 , gave no new insights and did not change the outcome of the simulations after cool - down , even though the values of the density were slightly different from the values obtained with a thermal expansion . therefore , we conclude that the qualitative results are not dependent on detailled parameter values and the sintering behavior , macroscopic as well as microscopic , is generic within the framework of this model . in summary
, we obtain that longer sintering and larger confining pressures lead to higher densities of the sintered sample . as an example , after the end of the longest sintering process ( dashed line ) at @xmath194 , a density of almost @xmath195 is reached for @xmath168 . only after cool - down , the maximum density of almost @xmath196 is reached due to negative thermal expansion . through test simulations with different parameters
, we verified that the increase in density is only partially due to the contact model , but also is caused by reorganizations in the sample .
the thermal expansion of the particles , on the other hand , does not affect the results in a drastic way . in order to understand the behavior of the material during sintering
, we take a look at some microscopic quantities of the system , like the coordination number , in fig .
[ fig : cont1 ] .
we note two major issues : first , the larger confining stress , @xmath168 , leads to a larger number of contacts per particle , whereas the samples with small confining stress , @xmath171 , approach a coordination number of @xmath197 , as can be expected for a frictionless , isostatic arrangement of disks .
second , the coordination number sometimes decays , whereas the density continuously increases .
we attribute this to the fact , that inside the sample , reorganizations can take place , which save space and thus increase the density , but at the same time can lead to a smaller @xmath198 . in order to deepen this insight , the coordination number
, the density and the measured wall - stress @xmath199 are plotted in fig .
[ fig : cont2 ] . the continuous increase in density
is sometimes accompanied by a decrease of the coordination number and large stress fluctuations .
large scale re - organizations of the packing thus may be responsible for a detectable fluctuation of the stress at the boundaries of the sample , possibly connected to sound emission .
a very specific microscopic property of the sintered sample is the statistics of the contacts , i.e. the probability distribution function ( pdf ) to find a certain overlap @xmath12 .
after the initial preparation of the powder sample at low temperature , the probability for larger overlaps decays rapidly no large deformations have taken place .
the overlap probabilities at the final stage of the sintering process are plotted in fig .
[ fig : overlap3 ] for @xmath189 , and both side pressures @xmath171 and @xmath168 .
the probability distributions show many contacts with small overlaps and a rapidly decaying probability for large overlaps at short sintering times .
the mean overlap increases , as expected , with the side stress and the sintering time .
the larger side stress leads to an underpopulation of small overlaps , because _ all _ particles are compressed quite strongly .
the probability distributions after the cool - down and the subsequent relaxation is plotted in fig .
[ fig : overlap5 ] for the same simulations .
the remarkable difference is not the width of the distribution that is only slightly wider .
the most striking difference is the shape , showing that after cool - down contacts with small overlap are rarefied , while contacts with a the typical overlap are overpopulated as compared to the situation before cool - down .
the probability for the largest overlaps is still decreasing rapidly . due to the advanced contact force law ,
as introduced and used in this stude , it is not possible to directly obtain the contact force from the overlap .
therefore , the pdf for the normal contact forces is plotted in figs .
[ fig : force10 ] and [ fig : force100 ] for small and large side - stresses @xmath171 and @xmath168 , respectively , and for different sintering times , @xmath177 as given in the inset .
comparing the pdf for the two situations , after sintering and after cool - down , the astonishing outcome of the simulation is the fact that the contact forces are mostly repulsive and rather small in the hot situation , just before cool - down .
this situation changes during cooling down : the cooling goes ahead with a broadening of the distribution towards both positive and negative forces .
this qualitative result is independent of the side pressure , however , the width of the distribution increases with increasing side stress and increasing sintering time .
another interesing and unexpected result is the observation that the force distribution
_ becomes narrower _ with sintering .
in other words , the extremely large forces are `` destroyed '' due to long time sintering and the distribution becomes more homogeneous in the sense that its width becomes smaller , see top panel in fig .
[ fig : force100 ] . from the tangential force distribution ( data not shown ) , there are less clear observations to be made .
the force distribution shows that during sintering the tangential forces become weaker as can be expected from the force model , see eq .
( [ eq : mu ] ) , since @xmath109 decays with increasing temperature and time .
thus longer sintering leads to weaker tangential forces .
however , after cool - down , the tangential force distribution is almost independent of the side stress and the sintering time , if both are sufficiently large .
with other words , only for very small side stresses and sintering times , the tangential forces are weakly activated , for larger @xmath165 or @xmath177 , the tangential forces reach a saturation distribution that is decaying almost exponentially for large @xmath200 .
the second possibility to look at the tangential forces is to measure the amount of friction that is activated , namely @xmath201 .
this quantity should not become larger than @xmath202 , as is consistently observed from the data .
contacts where @xmath203 are referred to as contacts with fully activated friction .
these become less probable for longer sintering times since , as mentioned before , the attractive forces become stronger after long sintering and the cooling down .
larger attractive forces correspond to a larger magnitude of @xmath131 , so that @xmath204 becomes smaller .
the samples prepared via the procedure described above are now tested via two test experiments , a compression test , where the vertical confining pressure is slowly increased , and a vibration test , where the confining stress is removed , and the sample is vibrated vertically on a flat bottom .
the compression test is performed for two sample sizes , namely with @xmath192 and @xmath189 , in order to judge the effect of a rather small particle number on the outcome of the simulations .
the compression test is a variant of the bi - axial compression frequently used in soil- or powder mechanics .
a given compressive stress in one direction ( @xmath205 ) is kept constant by moving the right , vertical wall , if neccessary .
the other confining stress @xmath206 is increased by moving the vertical wall down in a defined way see also section [ sec : model ] .
this vertical motion is thus strain controlled , where the vertical strain is defined as @xmath207 , with the vertical position of the top wall @xmath208 and @xmath209 .
note that a compressive vertical strain is defined positive for convenience here . for the compression test
, the previously prepared samples are used and the top wall is displaced slowly with the vertical strain @xmath210 .
the density of a sample with @xmath192 particles is plotted against the strain for various sintering times and for the two pressures @xmath171 and @xmath168 in fig .
[ fig : s1nu ] .
the results show again that samples , which sintered longer have a higher density .
furthermore , the compressive pressure also affects the sample density in magnitude and also in the variation , i.e. the higher compressive pressure leads to higher densities and also to a broader variation in densities between . during compression ,
the density slightly increases first and then decreases strongly .
the former is due to compression , the latter is due to dilatancy that is neccessary for the material to shear .
the longer the sample was sintered , the stronger is the change in density .
we relate this to the existence of the attractive contact forces . for longer sintering ,
stronger attraction is activated , holding together parts of the sample , so that the compression test leads to fragements with increasing size for increasing sintering duration . for the shortest sintering ,
the density change due to compression is negligible and the sample fragments into single grains .
the behavior of the material , as for instance its stiffness , is better described in terms of the vertical stress that the material can sustain under load , as plotted in fig .
[ fig : s1sxx ] . the vertical stress in the sample increases and it fails typically at some strain and at some magnitude of stress .
the failure stress increases with increasing sintering time and increasing external stress .
the material stiffness ( the slope in this representation ) is increased by a factor of about two when the confining stress is increased by a factor of ten .
moreover , the critical strain where the material fails increases with increasing confining pressure .
finally , a rather large jump in material strength is observed for sintering times between @xmath211 and @xmath212 for the small pressure , not paralleled by a similar outcome for the large stress .
thus the combination of sintering time , compressive pressure and test mode appears quite non - linear and not straightforward . for the compression test of the larger samples , also the prepared samples are used and the top wall is displaced slowly with the vertical strain . the density of a sample with @xmath189 particles is plotted against the strain for various sintering times and for the two pressures @xmath165 in fig . [
fig : s3nu ] .
the results are qualitatively similar , only that the larger sample shows somewhat larger densities and a weaker decay of density after failure .
the vertical stress is plotted against the strain in fig .
[ fig : s3sxx ] . with increasing strain , the vertical stress in the sample increases
and it fails typically at some slightly larger strain and stress as in the small sample .
the failure stress increases with increasing sintering time , increasing external stress , and increasing sample size .
however , the latter observation is not clearly a size - effect , since the sample size is so small that a non - negligible fraction of the particles is in contact with the walls and thus leads to different outcome .
the material stiffness ( dashed line in fig .
[ fig : s3sxx ] ) is again increased by about a factor of two when the confining stress is increased by a factor of ten . moreover ,
the critical strain where the material fails increases with increasing confining pressure , sintering time and system size .
one compression test for a long sintering time , @xmath189 and @xmath171 is presented in fig .
[ fig:300comp10 ] . during compression
( from top to bottom ) , the lines of strong attraction ( so to say the backbone of the sample ) are desctroyed and . at the same time
, more and more frictional contacts occur due to local shear .
moreover , gaps between the parts of the sample open and it fragments into pieces .
this is only a representative example for the compression test ; a more detailed study of the fracture behavior , sample - size- and sintering - time - dependence is far from the scope of this study .
the sintered samples can also be vibrated ( in the gravitational field ) in order to probe their stability .
the results of the previous tests are paralleled by a vibration test , see figs.[fig : vibrate10 ] and [ fig : vibrate100 ] .
short sintering times leads to unstable material samples , whereas the sample becomes more and more stable with increasing sintering time and confining pressure .
the sample with the longest sintering time @xmath213 is almost perfectly stable even under strong shaking
some corner- or boundary particles sometimes break off . for shorter sintering time
the sample is less stable and fragments into pieces .
for very short sintering time , the sample consists of single particles only .
in summary , a discrete model for the sintering of particulate materials was introduced and simple material samples were sintered for different times and confining pressures
. then they were tested with respect to their anisotropic load strength : longer sintering and stronger confining pressure systematically increases the density and the strength of the material . depending on the sintering duration
, either isolated particles , fragments or a single solid block of material could be produced . besides this macroscopic point of view , also the microscopic picture was examined .
the series of astonishing observations includes : ( i ) the coordination number may slightly decrease due to reorganizations while the density monotonously increases .
( ii ) sintering leads to a broadening overlap distribution , but to a narrowing force distributon , and thus to a homogenization of the sample .
( iii ) the normal forces become strongly attractive during the cooling down of the sample below the melting temperature . besides these facts , a lot of open questions concerning the sintering process remain , especially concerning the connection between the microscopic contact - model parameters and the macroscopic material parameters .
the research to be done is an accurate testing of the model via a comparison with experimental data .
since these are only available in three - dimensional systems , the 2d model presented here may be not helpful .
however , the model can easily be extended to three dimensions , where ` only ' more particles are needed .
note that the model presented here increases the amount of computation necessary for each contact by a large factor , so that the number of particles possible to simulate becomes rather small for a standard computer .
thus an extension to three dimensions requires a proper tuning of the implementation of the force - model and possibly the view of particles as blocks of material that can not fragment , rather than isolated particles .
the last missing ingredient in the model is a rolling resistance which accounts for a torque resistance of the contacts .
the authors thank m. ltzel , j. tomas , s. diebels , h. besserer , and g. a. d'addetta for discussions and acknowledge financial support by the deutsche forschungsgemeinschaft ( dfg ) and by the daimlerchrysler research division .
h. riedel and b. blug . a comprehensive model for solid state sintering and its application to silicon carbide .
in t .- j .
chuang and j. w. rudnicki , editors , _ multiscale deformation and fracture in materials and structures , the james r. rice 60th anniversary volume , solid mechanics and its application - vol . 84 _ , pages 4970 , dordrecht , 2001 .
kluwer academic publishers .
s. luding .
collisions & contacts between two particles . in h.
j. herrmann , j .-
hovi , and s. luding , editors , _ physics of dry granular media - nato asi series e350 _ , page 285 , dordrecht , 1998 .
kluwer academic publishers .
j. svoboda , h. riedel , and h. zipse .
equilibrium pore surfaces , sintering stresses and constitutive equations for the intermediate and late stages of sintering i. computation of equilibrium surfaces . , 42:435443 , 1994 .
h. riedel , h. zipse , and j. svoboda .
equilibrium pore surfaces , sintering stresses and constitutive equations for the intermediate and late stages of sintering ii .
diffusional densification and creep . , 42:445452 , 1994 .
l. brendel and s. dippel .
lasting contacts in molecular dynamics simulations . in h.
j. herrmann , j .-
hovi , and s. luding , editors , _ physics of dry granular media _
, page 313 , dordrecht , 1998 .
kluwer academic publishers .
|
a discrete model for the sintering of polydisperse , inhomogeneous arrays of cylinders is presented with empirical contact force - laws , taking into account plastic deformations , cohesion , temperature dependence ( melting ) , and long - time effects .
samples are prepared under constant isotropic load , and are sintered for different sintering times . increasing both external load and
sintering time leads to a stronger , stiffer sample after cooling down .
the material behavior is interpreted from both microscopic and macroscopic points of view . among the interesting results
is the observation , that the coordination number , even though it has the tendency to increase , sometimes slightly decreases , whereas the density continuously increases during sintering this is interpreted as an indicator of reorganization effects in the packing .
another result of this study is the finding , that strongly attractive contacts occur during cool - down of the sample and leave a sintered block of material with almost equally strong attractive and repulsive contact forces .
|
in the past most discussions of supersymmetric ( susy)@xcite phenomenology assumed r parity ( @xmath0 ) conservation where , r_p=(-1)^2j + 3b + l this implies that susy particles are pair produced , every susy particle decays into another susy particle and that there is a _
lsp _ that it is stable .
but this is just an _ ad hoc _ assumption without a deep justification . in this talk
we will review how @xmath0 can be broken , either spontaneously or explicitly , and discuss the most important features of these models @xcite .
we will also describe recent results @xcite on one loop generated masses and mixings in the context of a model that is a minimal extension of the minimal extension of the mssm
gut @xcite in which @xmath0 violation ( rpv ) is introduced via a bilinear term in the mssm superpotential @xcite .
in the original proposal @xcite the content was just the mssm and the breaking was induced by = v_l the problem with this model was that the majoron @xmath1 coupled to @xmath2 with gauge strength and therefore the decay @xmath3 contributed to the invisible @xmath4 width the equivalent of half a ( light ) neutrino family .
after lep i this was excluded . the way to avoid the previous difficulty is to enlarge the model and make @xmath1 mostly out of _ isosinglets_.
this was proposed by masiero and valle @xcite .
the content is the mssm plus a few isosinglet superfields that carry lepton number , ^c_i(1,0,-1 ) ; s_i ( 1,0,1 ) ; ( 1,0,0 ) the model is defined by the superpotential @xcite , w&=&h_u u^c q h_u + h_d d^c q h_d + h_e e^c l h_d & & + ( h_0 h_u h_d - ^2 ) & & + h _ ^c l h_u + h ^c s where the lepton number assignments are shown in .
.lepton number assignments .
[ cols="<,^,^,^,^,^ " , ] [ table0 ] the spontaneous breaking of r parity and lepton number is driven by @xcite v_r = v_s = v_l = the electroweak breaking and fermion masses arise from = v_u = v_d with @xmath5 fixed by the w mass .
the majoron is given by the imaginary part of ( v_u h_u - v_d h_d ) + - + where @xmath6 .
since the majoron is mainly an 1 singlet it does not contribute to the invisible @xmath2 decay width . the sbrp model has been extensively studied . the implications for accelerator @xcite and non
accelerator @xcite physics have been presented before and we will not discuss them here @xcite . as in this talk
we are concerned with the neutrino properties in the context of @xmath0 models we will only review here the neutrino results .
* _ neutrinos have mass _ + neutrinos are massless at lagrangian level but get mass from the mixing with neutralinos@xcite . in the sbrp model it is possible to have non zero masses for two neutrinos @xcite . * _ neutrinos mix _ +
the mixing is related to the the coupling matrix @xmath7 .
this matrix has to be non diagonal in generation space to allow _ _ + j and therefore evading @xcite the _ critical density argument _ against @xmath8 in the mev range . * _ avoiding bbn constraints on the @xmath9 _ + in the _ sm _ bbn arguments @xcite rule out @xmath10 masses in the range 0.5 mev < m _ _
< 35 mev we have shown @xcite that _ sbrp _ models can evade that constraint due to new annihilation channels _ _ j j
the most general superpotential @xmath11 with the particle content of the mssm is given by @xcite w= w_mssm + w _ where @xmath12 \label{wmssm}\end{aligned}\ ] ] and @xmath13\cr & & \cr & & + \lambda^{''}_{ijk } \widehat d_i
\widehat d_j\widehat u_k + \varepsilon_{ab}\ , \epsilon_i\widehat l_i^a\widehat h_u^b \label{wrpv}\end{aligned}\ ] ] where @xmath14 are generation indices , @xmath15 are @xmath16 indices .
the set of soft supersymmetry breaking terms are v^soft = v^soft _ mssm+v^soft _ @xmath17\cr & \hskip -3mm\hskip -3 mm & \vb{20 } + \varepsilon_{ab}\left [ a_u^{ij}\widetilde q^a_i\widetilde u_j h_u^b + a_d^{ij}\widetilde q^b_i\widetilde d_j h_d^a \right.\cr & \hskip -3mm\hskip -3 mm & \vb{18 } \left .
+ a_e^{ij}\widetilde l^b_i\widetilde r_j h_d^a -b\mu h_d^a h_u^b + h.c.\right ] \ , , \label{softmssm}\end{aligned}\ ] ] and @xmath18 \cr & \hskip -3mm\hskip -3 mm & \vb{18 } + a_{\lambda^{''}}^{ij}\widetilde d_i\widetilde d_j \widetilde u_k + \varepsilon_{ab}\ , b_i\epsilon_i\widetilde l^a_i h_u^b + h.c . \label{softrpv}\end{aligned}\ ] ] the bilinear @xmath0 violating term _ can not _ be eliminated by superfield redefinition as sometimes it is claimed .
to show this we consider the case @xcite where all the trilinear couplings in are zero and for simplicity we take @xmath19 . then the superpotential is @xmath20\end{aligned}\ ] ] consider now the rotation h_d=h_d-_3l_3,l_3=_3h_d+l_3 in the new basis @xmath21 where ^2=^2+_3 ^ 2 but the soft terms , @xmath22\cr & & \vb{18 } + \cdots\end{aligned}\ ] ] in the rotated basis is @xmath23 + \cdots\end{aligned}\ ] ] the last two terms violate @xmath0 and induce a non - zero vev for the @xmath24 sneutrino field in the rotated basis @xmath25 , where v_3-_3 ( v_1m^2+v_2b ) and @xmath26
the superpotential @xmath11 for the bilinear @xmath0 violation model is given by @xcite @xmath27\end{aligned}\ ] ] where @xmath14 are generation indices , @xmath15 are @xmath16 indices .
the set of soft supersymmetry breaking terms are @xmath28\cr \vb{18 } & \hskip -6 mm & + \varepsilon_{ab}\left [ a_u^{ij}\widetilde q^a_i\widetilde u_j h_u^b + a_d^{ij}\widetilde q^b_i\widetilde d_j h_d^a\right.\cr \vb{18 } & \hskip -6 mm & \left .
+ a_e^{ij}\widetilde l^b_i\widetilde r_j h_d^a \!-\!b\mu h_d^a h_u^b+b_i\epsilon_i\widetilde l^a_i h_u^b\right ] \,.\cr \vb{12 } & \hskip -6mm&\end{aligned}\ ] ] the electroweak symmetry is broken when the vevs of the two higgs doublets @xmath29 and @xmath30 , and the sneutrinos .
@xmath31}\choose { h^-_d } } \\
h_u&=&{{h^+_u}\choose{{1\over{\sqrt{2}}}[\chi^0_u+v_u+ i\varphi^0_u]}}\\ l_i&=&{{{1\over{\sqrt{2 } } } [ \tilde\nu^r_{i}+v_i+i\tilde\nu^i_{i}]}\choose{\tilde\ell^{i}}}\end{aligned}\ ] ] the gauge bosons @xmath11 and @xmath4 acquire masses m_w^2=g^2v^2 ; m_z^2=(g^2+g^2)v^2 where v^2v_d^2+v_u^2+v_1 ^ 2+v_2 ^ 2+v_3 ^ 2=(246 ) ^2 we introduce the following notation in spherical coordinates : @xmath32 which preserves the mssm definition @xmath33 .
the angles @xmath34 are equal to @xmath35 in the mssm limit . the full scalar potential may be written as v_total = _ i | w z_i |^2 + v_d + v_soft + v_rc where @xmath36 denotes any one of the scalar fields in the theory , @xmath37 are the usual @xmath38-terms , @xmath39 the susy soft breaking terms , and @xmath40 are the one - loop radiative corrections . in writing @xmath40 we use the diagrammatic method and find the minimization conditions by correcting to one loop the tadpole equations .
this method has advantages with respect to the effective potential when we calculate the one loop corrected scalar masses .
the scalar potential contains linear terms v_linear = t_d^0_d+t_u^0_u+t_i^r_i t_^0 _ , where we have introduced the notation ^0_=(^0_d,^0_u,^r_1,^r_2,^r_3 ) and @xmath41 . the one loop tadpoles are @xmath42 where @xmath43 are the finite one loop tadpoles .
the @xmath44model is a one ( three ) parameter(s ) generalization of the mssm .
it can be thought as an effective model showing the more important features of the sbrp model @xcite at the weak scale .
the mass matrices , charged and neutral currents , are similar to the sbrp model if we identify v_r h _ the @xmath0 violating parameters @xmath45 and @xmath46 violate lepton number , inducing a non - zero mass for only one neutrino , which could be considered to be the the @xmath10 .
the @xmath47 and @xmath48 remain massless in first approximation .
as we will explain below , they acquire masses from supersymmetric loops @xcite that are typically smaller than the tree level mass .
( 0,8 ) ( 0,0 ) in the @xmath44model and in the mssm as a function of @xmath49.,title="fig:",width=264 ] the model has the mssm as a limit .
this can be illustrated in figure [ fig1 ] where we show the ratio of the lightest cp - even higgs boson mass @xmath50 in the @xmath44model and in the mssm as a function of @xmath49 .
many other results concerning this model and the implications for physics at the accelerators can be found in ref .
at @xmath51 we assume the standard minimal supergravity unifications assumptions , & & a_t = a_b = a _ a , & & b = b_2=a-1 , & & m_h_d^2 = m_h_u^2 = m_l^2 = m_r^2 = m_0 ^ 2 , & & m_q^2 = m_u^2 = m_d^2 = m_0 ^ 2 , & & m_3 = m_2 = m_1 = m_1/2 in order to determine the values of the yukawa couplings and of the soft breaking scalar masses at low energies we first run the rge s from the unification scale @xmath52 gev down to the weak scale .
we randomly give values at the unification scale for the parameters of the theory .
ccccc 10 ^ -2 & & h^2_t_gut / 4 & & 1 10 ^ -5 & & h^2_b_gut / 4 & & 1 -3&&a / m_0&&3 0&&^2_gut / m_0 ^ 2&&10 0&&m_1/2/m_0&&5 10
^ -2 & & ^2_i_gut / m_0
^ 2 & & 10 the values of @xmath53 are defined in such a way that we get the charged lepton masses correctly . as the charginos mix with the leptons , through a mass matrix given by _
c = where @xmath54 is the usual mssm chargino mass matrix , m_c=@xmath55 is the lepton mass matrix , that we consider diagonal , m_l = and @xmath56 and @xmath57 are matrices that are non zero due to the violation of @xmath0 and are given by a^t=-3ptb=-3ptwe used an iterative procedure to accomplish that the three lightest eigenvalues of @xmath58 are in agreement with the experimental masses of the leptons . after running the rge
we have a complete set of parameters , yukawa couplings and soft - breaking masses @xmath59 to study the minimization .
this is done by the following method : we solve the minimization equations for the soft masses squared .
this is easy because those equations are linear on the soft masses squared .
the values obtained in this way , that we call @xmath60 are not equal to the values @xmath59 that we got via rge . to achieve equality
we define a function = max ( , ) i with the obvious property that 1 then we adjust the parameters to minimize @xmath61 . before we end this section
let us discuss the counting of free parameters in this model and in the minimal n=1 supergravity unified version of the mssm .
in we show this counting for the mssm and in for the @xmath44model .
finally , we note that in either case , the sign of the mixing parameter @xmath62 is physical and has to be taken into account .
cccparameters -9pt&-9pt conditions -9pt&-9pt free parameters @xmath63 , @xmath64 , @xmath65 -9pt&-9pt @xmath66 , @xmath67 -9pt&-9pt @xmath68 @xmath69 , @xmath70,@xmath71 -9pt&-9pt @xmath72 , @xmath73 -9pt&-9pt 2 extra @xmath74 , @xmath56 , @xmath62 -9pt&-9pt @xmath75 , @xmath76 -9pt&-9pt ( _ e.g. _ @xmath50 , @xmath77)total = 9 -9pt&-9pt total = 6 -9pt&-9pt total = 3 [ table1 ] cccparameters -9pt&-9pt conditions -9pt&-9pt free parameters @xmath63 , @xmath64 , @xmath65 -9pt&-9pt @xmath66 , @xmath67 -9pt&-9pt @xmath68 , @xmath45 @xmath69 , @xmath70 , @xmath71 -9pt&-9pt @xmath72 , @xmath73 -9pt&-9pt @xmath74,@xmath56 , @xmath62 -9pt&-9pt @xmath75 -9pt&-9pt 2 extra @xmath46 , @xmath45 -9pt&-9pt ( @xmath78 ) -9pt&-9pt ( _ e.g. _ @xmath50 , @xmath77)total = 15 -9pt&-9pt total = 9 -9pt&-9pt total = 6 [ table2 ] there is a strong motivation to consider gut theories where _ both _ gauge and yukawa unification can achieved .
this is because besides achieving gauge coupling unification , gut theories also reduce the number of free parameters in the yukawa sector and this is normally a desirable feature .
the situation with respect to the mssm can be summarized as follows : * in @xmath79 models , @xmath80 at @xmath81 .
the predicted ratio @xmath82 at @xmath83 agrees with experiments . * a relation between @xmath84 and @xmath68
is predicted .
two solutions are possible : low and high @xmath68 . * in @xmath85 and @xmath86 models
@xmath87 at @xmath81 . in this case , only the large @xmath68 solution survives .
we have shown @xcite that the @xmath44model allows @xmath88 yukawa unification for any value of @xmath68 and satisfying perturbativity of the couplings .
we also find the @xmath89 yukawa unification easier to achieve than in the mssm , occurring in a wider high @xmath68 region .
in the basis ^0t= ( -i,-i^3,_d^1,_u^2 , _ e , _ , _ ) the neutral fermions mass terms in the lagrangian are given by _
m=-12(^0)^tm_n^0+h.c . where the neutralino / neutrino mass matrix is _
n = with _ ^0 - 2pt=-4pt m= the mass matrix @xmath90 is diagonalized by ^*m_nn^-1=diag(m_^0_i , m__j ) [ eq : neumdiag ] where @xmath91 for the neutralinos , and @xmath92 for the neutrinos .
if the parameters are small it is convenient to define @xcite the matrix = m _ ^0 ^ -1 if the elements of this matrix satisfy _ ij 1 then one can find an approximate solution for mixing matrix @xmath93 .
explicitly we have @xmath94 where _
i = v_i + v_d _ i [ lambdai ] from and one can see that @xmath95 in the mssm limit where @xmath96 , @xmath97 . in leading order in @xmath98 the mixing matrix @xmath93 is given by , ^ * -2pt = -2pt ( -1pt cc n^ * & 0 + 0 & v_^t -1pt ) ( -1pt cc 1 -1 2 ^ & ^ + - & 1 -1 2 ^ -1pt ) the second matrix above block - diagonalizes @xmath90 approximately to the form diag(@xmath99 ) m_eff & -3mm=-3 mm & - m _ ^0 ^ -1 m^t & -3mm=-3 mm & ( -2 mm ccc _
e^2 -1pt&-1pt _ e _ -1pt&-1pt _ e _ + _ e _ -1pt&-1pt _ ^2 -1pt&-1pt _ _ + _ e _ -1pt&-1pt _ _ -1pt&-1pt _ ^2 -3 mm ) the sub - matrices @xmath100 and @xmath101 in eq .
diagonalize @xmath102 and @xmath103 n^*m_^0 n^ = diag(m_^0_i ) , v_^t m_eff v _ = diag(0,0,m
_ ) , where m _
= tr(m_eff ) = ||^2 . for @xmath101
we have ( we can rotate away one angle ) v_&= & ( ccc 1 & 0 & 0 + 0 & _ 23 & -_23 + 0 & _ 23 & _ 23 ) & & ( ccc _ 13 & 0 & -_13 + 0 & 1 & 0 + _ 13 & 0 & _ 13 ) , where the mixing angles can be expressed in terms of the _ alignment vector _ @xmath104 as follows : _ 13 = - , _ 23 = .
the self energy for the neutralino / neutrino is then m^pole_ij= m^_ij(_r ) + m_ij where m_ij&-3mm=-3 mm & _ = 0 where ^v=(^l+^r ) ^v=(^l+^r ) and = -_e + 4 ( 0,4.5 ) ( 0,2.5 ) ( 0,4.5 ) ( 3.5,2.5 ) ( 0,3 ) ( 0,-0.45 ) ( 0,3 ) ( 3.5,0 ) in are shown the classes of diagrams contributing to the self energy at one loop .
these diagrams can be calculated in a straightforward way .
for instance the @xmath11 diagram in the @xmath105 gauge gives ^v_ij & -5pt=-5pt & - _
k=1 ^ 5 2 ( o^ncw_l jk o^cnw_l ki + o^ncw_r jk o^cnw_r ki ) & -15pt -15pt & b_1(p^2,m^2_k , m^2_w ) ^v_ij & -5pt=-5pt & - _
k=1 ^ 5 ( -4 ) ( o^ncw_l jk o^cnw_r ki + o^ncw_r jk o^cnw_l ki ) & -15pt -15pt & m_k b_0(p^2,m^2_k , m^2_w ) where @xmath106 and @xmath107 are the passarino - veltman functions , and the coupling matrices appear in the vertices in the following way i ^ ( o^cnw_l ji p_l + o^cnw_r ji p_r ) ( 0,0 ) ( 0,-1 ) ( 0,2.5 ) ( 0,0 ) ( 0,0 ) ( 2.5,0 ) ( 0,0 ) ( 5,0.2 ) ( 0,2.5 ) ( 1,0 ) ( 0,0 ) ( 4,0 ) ( 0,2.5 ) ( 0.5,0 ) ( 0,0 ) ( 2.4,0 ) ( 0,0 ) ( 4.5,0 ) when calculating the self energies the question of gauge invariance arises . in the @xmath108 gauge
the sets of diagrams of figs .
[ fig3]-[fig5 ] depend on @xmath98 .
we have shown that the gauge dependence cancels among the diagrams in each set .
so in the actual calculations we considered the tadpoles needed in those sets to ensure gauge invariance .
the other tadpoles were included in the minimization procedure of .
this is a gauge invariant splitting .
the one loop corrected mass matrix is given by m^1l= m^0l_diag + m^1l where m^0l_diag= n m_n n^t now we diagonalize the 1loop mass matrix m^1l_diag = n m^1l n^t then the mass eigenstates are related to the weak basis states by _ 0^mass= n^1l_i _ 0^weak with ^1l = n n the usual convention in neutrino physics _
= u_k _ k is recovered in our notation as u_k= n^1l_4+k , 4 + assuming hierarchy in the masses @xmath109 and @xmath110 the survival probabilities for the solar and atmospheric neutrinos are @xmath111 as @xmath112 has to be small we neglect it and write the usual two neutrino mixing angle as ^2 ( 2 _ 12 ) = 4 u_e1 ^ 2 u_e2 ^ 2 and ^2 ( 2 _ 13 ) = 4 u_3 ^ 2 ( 1 -u_3 ^ 2 ) we have found @xcite , that if @xmath113 then the approximate formulas hold @xmath114 then if we take _ 1 _ 2 _ 3 we immediately get maximal mixing for the atmospheric neutrinos .
this is shown in figure [ atm05 ] where we see that maximality of the mixing is only possible for @xmath115 .
@xmath116 ( 0,4 ) ( 0.5,0 ) for @xmath117 and @xmath118 .
@xmath119 has an upper cut of @xmath120 in this plot , since larger values lead to larger scatter for very small @xmath121,title="fig:",width=264 ] @xmath122 to get bimaximality we have to fix the solar angle .
we have discovered that if @xmath123 and @xmath124 we get bimaximality if the following sign condition applies ( _ / _ ) ( _ / _ ) 0 this is illustrated in figure [ solar04 ] . in practice
we do not need perfect maximality .
we took @xmath125 ( 0,4 ) ( 0.5,0 ) for @xmath126 and @xmath115 applying the sign condition.,title="fig:",width=264 ] @xmath127 4/5 |_/_| 5/4 |_e/_| 1 [ atmangle ] to fix the atmospheric angle and ( _ /_)(_/
_ ) 0 0.6 _ e/ _ 1.2 [ solangle ] to fix the solar angle .
next we have to fix the masses to solve the atmospheric and solar neutrino problems .
we found @xcite that the range 0.03 gev^2 || 0.25 gev^2 [ atmmass ] fixes the tree level mass to reproduce the atmospheric neutrino problem .
this is illustrated in figure [ fig8 ] where , besides the conditions in eqs.([atmangle ] ) and ( [ solangle ] ) to fix the angles and condition eq .
( [ atmmass ] ) , all the other parameters were chosen randomly .
consistency of the parameters was required in the sense that minimization the scalar potential including the tadpoles was performed as well as the matching with the rge solutions with universality at gut scale .
@xmath128 @xmath129 $ ] ( 0,4 ) ( 0.5,0 ) versus @xmath116 .
all points obey eqs .
( [ atmangle ] ) , ( [ solangle ] and ( [ atmmass ] ) no further cut applied except that @xmath130.,title="fig:",width=264 ] @xmath131 @xmath132 @xmath129 $ ] ( 0,4 ) ( 0.5,0 ) as a function of @xmath133 .
a box is drawn to guide the eye.,title="fig:",width=264 ] @xmath134 @xmath132 @xmath129 $ ] ( 0,4 ) ( 0.5,0 ) versus @xmath125 for those points which have @xmath128 and @xmath116 correct and at the same time fit either the vacuum or la - msw solutions.,title="fig:",width=264 ] @xmath135 finally we have to check if it is possible to have masses in the range to solve the solar neutrino problem .
we found @xcite that the relevant parameter for this purpose is ( _ 1 ^ 2+_1 ^ 2+_1 ^ 2)/|| ^2/|| depending on this quantity in the range @xmath136 the solar neutrino problem can be solved : low values give _ just - so _ solutions , high values tend to give large angle msw ( la - msw ) .
this is illustrated in figure [ fig9 ] and in figure [ fig10 ] .
another question of relevance that we addressed was the study of the decay length of the lightest neutralino .
this is important , because if the decay length is greater than the detector , then in practice it will be invisible like in the mssm .
as we can see in figure [ fig12 ] , that is not the case , the neutralino decays well inside the detector leading to novel signatures .
@xmath137 @xmath138 ( 0,7.5 ) ( 0.5,0 ) -8 mm @xmath139 @xmath140 -8 mm
there is a viable model for sbrp that leads to a very rich phenomenology , both at laboratory experiments , and at present ( lep ) and future ( lhc , lnc ) accelerators .
we have shown that the radiative breaking of both the gauge symmetry and @xmath0 can be achieved . in these type of models neutrinos have mass and can decay thus avoiding the critical density argument .
they also can evade the bbn limits on a @xmath10 on the mev scale .
most of these phenomenology can be described by an effective model with bilinear explicit @xmath0 violation .
we have calculated the _
one loop _ corrected masses and mixings for the neutrinos in a completely consistent way , including the rg equations and correctly minimizing the potential .
we have shown that it is possible to get bimaximal solutions for both the atmospheric and solar neutrino problems .
we emphasize that the lightest neutralino decays inside the detectors , thus leading to a very different phenomenology than the mssm .
daz , j.c .
romo , and j.w.f .
valle , ; m.a .
daz , talk given at _
international europhysics conference on high energy physics _ , jerusalem , israel , 19 - 26
aug 1997 , ; j.c .
romo , talk given at _
international workshop on physics beyond the standard model : from theory to experiment _ ( valencia 97 ) , valencia , spain , 13 - 17 oct 1997 , ; j.w.f .
valle , review talk given at the _ workshop on physics beyond the standard model : beyond the desert : accelerator and nonaccelerator approaches _ , tegernsee , germany , 8 - 14 jun 1997 , .
f. de campos , m.a .
garca - jareo , a.s .
joshipura , j. rosiek , and j. w. f. valle , ; t. banks , y. grossman , e. nardi , and y. nir , ; a. s. joshipura and m.nowakowski , ; r. hempfling , ; f. vissani and a.yu .
smirnov , ; h. p. nilles and n. polonsky , ; b. de carlos , p. l. white , ; s. roy and b. mukhopadhyaya , .
a lopez - fernandez , j. romo , f. de campos and j. w. f. valle , ; _ ibidem _ , proceedings of moriond 94 .
pag.81 - 86 , edited by j. tran thanh van , ditions frontires , 1994 .
j. c. romo , f. de campos , m. a. garcia - jareno , m. b. magro and j. w. f. valle , .
|
we review models where r parity is broken , either spontaneously or explicitly . in this last case
we consider the situation where r parity is broken via bilinear terms in the superpotential .
we show that although at tree level only one neutrino gets mass , at one loop level all three neutrinos became massive .
we study the conditions under which bimaximal mixing can be achieved and show that the masses can be in the correct ranges needed for solving the atmospheric and solar neutrino problems .
|
the source coding problem for correlated sources has been regarded as an important research area in information theory , and various types of coding problems were studied so far ( e.g. @xcite ) . in particular , our focus in this paper is the lossy coding problem posed by heegard and berger @xcite . in the problem , there are one encoder and multiple decoders ( see fig .
[ fig : scenario ] ) . in this paper , we only treat the case with two decoders . the encoder send an encoded version of principal source @xmath0 , and decoders reproduce the principal source within prescribed distortion levels by the help of side - information @xmath1 and @xmath2 respectively . in this
setting , heegard and berger showed an upper bound on the rate distortion function .
they also showed that the upper bound is tight if the sources are degraded , i.e. , @xmath0 , @xmath1 , and @xmath2 form a markov chain in this order .
although some researchers investigated variants of heegard and berger
s problem @xcite , there is no conclusive result , i.e. , an upper bound and a lower bound coincide , without the degraded assumption , and whether heegard and berger s upper bound is tight or not for non - degraded cases has been a long - standing open problem . in order to provide some insight to this problem
, we investigate a special case of heegard and berger s problem in this paper .
specifically , we consider the case such that the correlated sources @xmath3 is a direct product of two components correlated sources @xmath4 and @xmath5 and the components are independent each other ( see fig . [
fig : scenario ] ) .
furthermore , we exclusively consider the case such that each component is degraded , i.e. , either eq . ( [ eq : matched - degraded ] ) or eq .
( [ eq : unmatched - degraded ] ) is satisfied .
although the problem setting treated in this paper seems too restrictive at first glance , this is not the case .
the rate distortion function of product sources can be strictly smaller than the summation of the component - wise rate distortion functions even though the components are independent each other , an intuitive example of which is illustrated in fig .
[ fig : example ] .
therefore , the rate distortion function of product sources is not trivial , and it is interesting to characterize the rate distortion function . when eq .
( [ eq : matched - degraded ] ) is satisfied , the joint sources @xmath3 are degraded .
thus , heegard and berger s result suggests that their upper bound is tight . on the otherhand ,
when eq .
( [ eq : unmatched - degraded ] ) is satisfied , the joint sources are not degraded .
thus , whether heegard and berger s upper bound is tight or not is unclear so far . in this paper , we show that the upper bound is tight , i.e. , we characterize the rate - distortion function . as long as the author s knowledge
, this is the first example such that the rate - distortion function is characterized without the degraded assumption .
the rest of the paper is organized as follows . in section [ sec : preliminaries ] , we explain the problem setting treated in this paper , and also explain known results . in section [ sec : main ] , we show our main result and its proof . in section [ sec : gaussian ] , we show the gaussian example . ) for the product source is strictly smaller than the summation of the component - wise rate distortion functions . @xmath6 and @xmath7 are independent uniform binary random variables , and @xmath8 represent a constant random variable .
when two components are encoded and decoded separately , @xmath9 bit must be sent for each components , which means @xmath10 bits must be sent to reproduce @xmath11 at both decoders . on the otherhand ,
if the encoder sends @xmath12 , then both decoders can reproduce @xmath11 as in the network coding @xcite .
thus , when the components are encoded and decoded jointly , @xmath9 bit suffice for the decoders to reproduce @xmath11 .
this example can be also regarded as a special case of the complementary delivery network @xcite , and the relation between the complementary delivery network and the network coding was pointed out in @xcite . ]
in this section , we explain the problem formulation treated in this paper , which is a special case of the problem formulation treated by heegard and berger in @xcite .
then , we explain some known results .
let @xmath13 be product of correlated sources , i.e. , components @xmath4 and @xmath5 are independent each other .
the alphabet of the sources are denoted by @xmath14 , @xmath15 , and @xmath16 respectively .
let @xmath17 be independently identical distributed copies of @xmath3 .
let @xmath18 , @xmath19 , @xmath20 , and @xmath21 be reproduction alphabets , and for @xmath22 let @xmath23 be distortion measures .
the coding system treated in this paper consists of one encoder @xmath24 and two decoders @xmath25 and @xmath26 for quadruplet @xmath27 , rate @xmath28 is said to be @xmath29-achievable if , for any @xmath30 , there exists a sequence of codes @xmath31 such that @xmath32 and @xmath33 & \le & \hat{d}_i + \gamma , \\ \frac{1}{n } \sum_{t=1}^n \mathbb{e}[\tilde{d}_i(x_{it},\tilde{x}_{it } ) ] & \le & \tilde{d}_i + \gamma\end{aligned}\ ] ] for @xmath22 are satisfied for sufficiently large @xmath34 , where @xmath35 and @xmath36 .
then , the rate - distortion function is defined as @xmath37 in @xcite , heegard and berger showed an upper bound on the rate - distortion function .
[ proposition : heegard - direct ] ( ( * ? ? ?
* theorem 2 ) ) let @xmath38 be auxiliary random variables satisfying 1 . @xmath39 .
there exist functions @xmath40 and @xmath41 such that @xmath42 \le \hat{d}_i$ ] and @xmath43 \le \tilde{d}_i$ ] for @xmath44 .
then , we have @xmath45 in @xcite , heegard and berger also showed an upper bound on the rate - distortion function for more than three decoders .
however , timo _ et .
pointed out that the statement of ( * ? ? ?
* theorem 2 ) for more than three decoders is invalid , and only the statement for two decoders is valid @xcite . in @xcite , they also showed a corrected upper bound on the rate - distortion function for more than three decoders .
when the component sources are degraded in matched order , i.e. , @xmath46 are satisfied , then the joint sources @xmath3 are degraded , i.e. , @xmath47 for the degraded sources , heegard and berger @xcite showed that the upper bound in proposition [ proposition : heegard - direct ] is tight .
in particular for product of two sources .
we have the following statement .
[ proposition - matched - case ] ( @xcite ) if the components sources are degraded in matched order , i.e. , eq .
( [ eq : matched - degraded ] ) is satisfied , then we have @xmath48,\end{aligned}\ ] ] where the minimization is taken over all auxiliary random variables @xmath49 satisfying the followings : 1 .
@xmath50 for @xmath22 .
2 . @xmath51 and @xmath52 are independent each other .
there exist functions @xmath53 and @xmath54 such that @xmath55 \le \hat{d}_i$ ] and @xmath56 \le \tilde{d}_i$ ] for @xmath44 .
4 . @xmath57 and @xmath58 for @xmath22 , where @xmath59 and @xmath60 are ranges of @xmath61 and @xmath62 respectively .
technically , the result in @xcite does not directly imply proposition [ proposition - matched - case ] , because proposition [ proposition - matched - case ] states the stronger condition on the auxirially random variables , i.e. , @xmath63 and @xmath64 are independent .
we give a proof of proposition [ proposition - matched - case ] in appendix [ appendix : proof - of - prposition - matched ] for readers convenience .
note that @xmath65 is nothing but the summation of the component - wise rate distortion functions , i.e. , @xmath66 where @xmath67\end{aligned}\ ] ] and the minimization in eq .
( [ component - rate - distortion - function ] ) is taken over all @xmath68 satisfying the conditions 1 , 3 , and 4 in proposition [ proposition - matched - case ] .
this fact implies that the optimal scheme for the degraded product sources is to combine the component - wise optimal scheme .
when sources @xmath3 are not necessarily degraded , whether the upper bound in proposition [ proposition : heegard - direct ] is tight or not has been an open problem for a long time . in the next section , we will show that the upper bound is tight if the component sources satisfy eq .
( [ eq : unmatched - degraded ] ) .
in this section , we consider the case in which the component sources are degraded in mismatched order , i.e. , @xmath69 are satisfied . in this case , the joint sources @xmath3 are not degraded , and the rate - distortion function @xmath70 has not been clarified .
the following is our main result .
[ theorem - main ] suppose that eq .
( [ eq : unmatched - degraded ] ) is satisfied .
then , we have @xmath71,\end{aligned}\ ] ] where the minimization is taken over all auxiliary random variables @xmath49 satisfying the followings : 1 .
@xmath50 for @xmath22 .
2 . @xmath51 and @xmath52 are independent each other .
3 . there exist functions @xmath72 ,
@xmath73 , @xmath74 , and @xmath75 such that @xmath76 \le \hat{d}_i\end{aligned}\ ] ] and @xmath77 \le \tilde{d}_i\end{aligned}\ ] ] for @xmath44 .
4 . @xmath57 and @xmath58 for @xmath22 , where @xmath59 and @xmath60 are ranges of @xmath61 and @xmath62 respectively .
when the distortion levels are all @xmath78 , we have the following corollary , which can be also derived as a straightforward consequence of sgarro s result @xcite .
when @xmath79 , we have @xmath80 it should be noted that @xmath81 in general .
this fact implies that the combination of the component - wise optimal scheme is not necessarily optimal even though the components are independent each other .
this phenomenon also appears for a lossy case , which will be mentioned in section [ sec : gaussian ] by a gaussian example .
[ remark : exor - construction ] in @xcite , tian and diggavi proposed a coding scheme that is different from @xcite .
although joint encoding and decoding is required to achieve the rate - distortion function given in theorem [ theorem - main ] , we can construct a code that achieve the rate - distortion function from component - wise coding scheme of @xcite in a similar manner as the example of fig . [ fig : example ] .
when we apply the coding scheme of @xcite to the first component source @xmath4 , the source @xmath82 is quantized into the common description @xmath83 and the private description @xmath84 .
then , the common description @xmath83 is first bin coded at rate @xmath85 and then @xmath83 is bin coded again at extra rate @xmath86 by using the first bin index @xmath87 , the first decoder ( with @xmath88 ) can reconstruct the common description @xmath83 . by using both the first bin index @xmath87 and the extra bin index
@xmath89 , the second decoder ( with @xmath90 ) can reconstruct @xmath83 .
after that the private description @xmath84 is transmitted to the first decoder at rate @xmath91 similarly , when we apply the coding scheme of @xcite to the second component source @xmath5 , the source @xmath92 is quantized into the common description @xmath93 and the private description @xmath94 .
then , the common description @xmath93 is bin coded into @xmath95 and @xmath96 at rate @xmath97 and @xmath98 respectively so that the first decoder ( with @xmath99 ) can reconstruct @xmath93 from both @xmath95 and @xmath96 and the second decoder ( with @xmath100 ) can reconstruct @xmath93 from @xmath95 .
the private description @xmath94 is also transmitted to the second decoder at rate @xmath101 by using the above two component - wise coding scheme , we can construct a joint encoding and decoding scheme as follows .
first , the encoder sends @xmath102 .
this requires the rate @xmath103.\end{aligned}\ ] ] note that the first ( second ) decoder can obtain @xmath96 ( @xmath89 ) by first reconstructing @xmath83 ( @xmath93 ) and then subtracting @xmath89 ( @xmath96 ) from @xmath104 .
the encoder also sends the private descriptions @xmath84 and @xmath94 at rates @xmath105 and @xmath106 respectively .
consequently , the total rate coincide with the rate - distortion function given in theorem [ theorem - main ] .
the direct part is a straightforward consequence of proposition [ proposition : heegard - direct ] . for any auxiliary random variables
@xmath107 satisfying the conditions in theorem [ theorem - main ] , let @xmath108 then , proposition [ proposition : heegard - direct ] implies theorem [ theorem - main ] .
@xmath109 as we have mentioned in section [ sec : preliminaries ] , heegard and berger showed the converse coding theorem for degraded case . in the course of the proof
, they essentially showed the following lemma , which can be shown only for the degraded case .
although our purpose is to show the converse coding theorem for the non - degraded case , we need the following lemma in our converse proof of theorem [ theorem - main ] .
a proof of lemma [ lemma : degraded ] is given in appendix [ appendix : proof - of - lemma - degraded ] .
[ lemma : degraded ] let @xmath110 be product of correlated sources such that @xmath111 for both @xmath112 and @xmath113 .
let @xmath114 be independently identical distributed copies of @xmath115 .
then , for any function @xmath116 , we have @xmath117,\end{aligned}\ ] ] where we use the notations @xmath118 , @xmath119 , and etc .. it should be noted that the bounds in lemma [ lemma : degraded ] do not involve the decoders , which is important when we use lemma [ lemma : degraded ] in the converse proof below
. _ proof of converse part ) _
suppose that the rate @xmath28 is @xmath29-achievable .
then , for any @xmath30 there exists a code @xmath120 such that @xmath121 and @xmath122 for @xmath22 are satisfied , where we set @xmath123 , @xmath124 $ ] and @xmath125 $ ] for @xmath126 and @xmath127@xmath128 .
the key idea of the proof is to derive two lower bound on @xmath129 by using lemma [ lemma : degraded ] as follows .
first , let @xmath130 and @xmath131 . then , since @xmath115 satisfies eq .
( [ eq : condition - lemma ] ) , we can use lemma [ lemma : degraded ] , and we have @xmath132 \nonumber \\ & = & \frac{1}{n } \sum_{t=1}^n \left [ i(s_n , y_{1t}^-,y_2^n , z_{1t}^-,z_{1t}^+,z_2^n ; x_{1 t } | z_{1 t } ) \right .
\nonumber \\ & & + i(y_{1t}^+ ; x_{1 t } | y_{1 t } , s_n , y_{1t}^-,y_2^n , z_{1t}^-,z_{1t}^+,z_2^n ) \nonumber \\ & & \left .
+ i(s_n , y_1^n , y_{2t}^- , y_{2t}^+ , z_1^n , z_{2t}^-,z_{2t}^+ ; x_{2 t } | z_{2 t } ) \right ] \label{eq : proof-1 } \\
& = & \frac{1}{n } \sum_{t=1}^n \left [ i(w_{1 t } ; x_{1 t } | z_{1 t } ) + i(u_{1 t } ; x_{1 t } | y_{1 t } , w_{1 t } ) \right . \nonumber \\ & & \left .
+ i(w_{2 t } , u_{2 t } ; x_{2 t } | z_{2 t } ) \right ] \nonumber \\ & = & i(w_{1 t } ; x_{1 t } | z_{1 t } , t ) + i(u_{1 t } ; x_{1 t } | y_{1 t } , w_{1 t } , t ) \nonumber \\ & & + i(w_{2 t } , u_{2 t } ; x_{2 t } | z_{2 t } , t ) \nonumber \\ & = & i(w_{1 t } , t ; x_{1 t } | z_{1 t } ) + i(u_{1 t } ; x_{1 t } | y_{1 t } , w_{1 t } , t ) \nonumber
\\ & & + i(w_{2 t } , t , u_{2 t } ; x_{2 t } | z_{2 t } ) \label{eq : proof-2}\end{aligned}\ ] ] where we used the fact that @xmath133 is degraded version of @xmath134 in eq .
( [ eq : proof-1 ] ) , and we set @xmath135 and @xmath136 is the uniform random numbers on @xmath137 that are independent of the other random variables .
note that @xmath138 satisfy @xmath139 for @xmath22 .
similarly , let @xmath140 and @xmath141 . then , since @xmath115 satisfies eq .
( [ eq : condition - lemma ] ) , we can use lemma [ lemma : degraded ] , and we have @xmath142 \nonumber \\ & = & \frac{1}{n } \sum_{t=1}^n \left [ i(s_n , y_1^n , y_{2t}^-,y_{2t}^+,z_1^n , z_{2t}^- ; x_{2t}| y_{2 t } ) \right .
\nonumber \\ & & + i(z_{2t}^+ ; x_{2 t } | z_{2 t } , s_n , y_1^n , y_{2t}^-,y_{2t}^+,z_1^n , z_{2t}^- ) \nonumber \\ & & \left .
+ i(s_n , y_{1t}^-,y_{1t}^+ , y_2^n , z_{1t}^- , z_{1t}^+ , z_2^n ; x_{1t}| y_{1 t } ) \right ] \label{eq : proof-3 } \\
& = & \frac{1}{n } \sum_{t=1}^n \left [ i(w_{2 t } ; x_{2 t } | y_{2 t } ) + i(u_{2 t } ; x_{2 t } | z_{2 t } , w_{2 t } ) \right .
\nonumber \\ & & \left .
+ i(w_{1 t } , u_{1 t } ; x_{1 t } | y_{1 t } ) \right ] , \nonumber \\ & = & i(w_{2 t } ; x_{2 t } | y_{2 t } , t ) \nonumber \\ & & + i(u_{2 t } ; x_{2 t } | z_{2 t } , w_{2 t } , t ) \nonumber \\ & & + i(w_{1 t } , u_{1 t } ; x_{1 t } | y_{1 t } , t ) \nonumber \\ & = & i(w_{2 t } , t ; x_{2 t } | y_{2 t } ) \nonumber \\ & & + i(u_{2 t } ; x_{2 t } | z_{2 t } , w_{2 t } , t ) \nonumber \\ & & + i(w_{1 t } , t , u_{1 t } ; x_{1 t } | y_{1 t } ) , \label{eq : proof-4}\end{aligned}\ ] ] where we used the fact that @xmath143 is degraded version of @xmath144 in eq .
( [ eq : proof-3 ] ) .
since @xmath145 includes @xmath146 , there exists a function @xmath147 satisfying @xmath148 = \hat{d}_{1t}$ ] .
since @xmath149 includes @xmath150 , there exists a function @xmath151 satisfying @xmath152 = \hat{d}_{2t}$ ] .
since @xmath153 includes @xmath154 , there exists a function @xmath155 satisfying @xmath156= \tilde{d}_{1t}$ ] . since @xmath157 includes @xmath158 , there exists a function @xmath159 satisfying @xmath160 = \tilde{d}_{2t}$ ] .
thus , there exist functions @xmath161 , @xmath162 , @xmath163 , and @xmath164 satisfying @xmath165 & \le & \hat{d}_i+ \gamma \label{eq : distortion - condition-1 } \\
\mathbb{e}[\tilde{d}(x_i , \tilde{x}_i ) ] & \le & \tilde{d}_i+ \gamma \label{eq : distortion - condition-2}\end{aligned}\ ] ] for @xmath22 .
thus , by combining eqs .
( [ eq : proof-0 ] ) , ( [ eq : proof-2 ] ) , and ( [ eq : proof-4 ] ) , and by taking @xmath166 , @xmath167 , @xmath168 , and @xmath169 , we have that there exist @xmath49 satisfying eqs .
( [ eq : distortion - condition-1 ] ) and ( [ eq : distortion - condition-2 ] ) and @xmath170 although the auxirially random variables @xmath51 and @xmath52 chosen above are not independent each other , they never appear in any term simultaneously
. thus we can take @xmath51 and @xmath52 to be independent each other .
since @xmath30 is arbitrary , by the continuity of @xmath171 with respect to @xmath29 , and by using the support lemma @xcite , we have the statement of the theorem . @xmath109
to illustrate our main result , we show an explicit form of the rate distortion function for the gaussian example .
we consider jointly gaussian sources @xmath172 given by @xmath173 and @xmath174 . where @xmath175 and @xmath176 are gaussian noises with variances @xmath177 and @xmath178 such that @xmath179 and @xmath180 respectively .
the conditional variance of @xmath181 given @xmath182 is denoted by @xmath183 etc .. to avoid tedious exceptional cases , we assume that @xmath184 and @xmath185 for @xmath22 . in the above setting , the rate - distortion function is given by the following theorem .
we have @xmath186 \\ & & + \frac{1}{2}\log \frac{(b_1^ * + \sigma_{1,n_y}^{-1})}{(\tilde{d}_1^{-1 } - \sigma_{1,n_z}^{-1 } + \sigma_{1,n_y}^{-1 } ) } \\ & & + \frac{1}{2}\log \frac{(b_2^ * + \sigma_{2,n_z}^{-1})}{(\hat{d}_2^{-1 } - \sigma_{2,n_y}^{-1 } + \sigma_{2,n_z}^{-1})},\end{aligned}\ ] ] where @xmath187,\\ b_2^ * & = & \max [ \hat{d}_2^{-1 } - \sigma_{2,n_y } , \tilde{d}_2^{-1 } - \sigma_{2,n_z}^{-1}].\end{aligned}\ ] ] note that the component - wise rate - distortion functions are given by @xmath188 by noting @xmath179 and @xmath189 , we can find @xmath190 which implies that the combination of the component - wise optimal scheme is suboptimal for gaussian product sources .
in this paper , we investigated the lossy coding problem for a product of two sources with two decoders , and characterized the rate - distortion function .
it should be noted that the present work is motivated by the results on product of two broadcast channels @xcite .
the technique used in our converse proof can be regarded as the enhance technique introduced by weingarten _ et .
our future research agenda is to extend our result to vector gaussian sources in which there exist correlation between component sources .
the author would like to thank prof .
tomohiko uyematsu and prof .
hirosuke yamamoto for valuable comments .
he also would like to thank prof .
yasutada oohama for valuable discussions .
he also appreciate the comments from anonymous reviewers of isit 2011 , especially for pointing out remark [ remark : exor - construction ] .
since the direct part directly follows from proposition [ proposition : heegard - direct ] , we only prove the converse part .
suppose that @xmath28 is @xmath29-achievable .
then , for any @xmath30 , there exists a code @xmath120 satisfying eqs .
( [ eq : proof-0])([eq : distortion - proof-2 ] ) , where we use the same notation as in section [ subsec : proof - converse ] .
we will lower bound @xmath129 by using lemma [ lemma : degraded ] .
let @xmath130 and @xmath191 .
then , from lemma [ lemma : degraded ] , we have @xmath192 \\ & = & \frac{1}{n}\sum_{t=1}^n \left [ i(w_{1 t } ; x_{1t}|z_{1 t } ) + i(u_{1t};x_{1t}|y_{1t},w_{1 t } ) \right .
\\ & & \left .
+ i(w_{2t};x_{2t}|z_{2 t } ) + i(u_{2t};x_{2t}|y_{2t},w_{2 t } ) \right ] \\ & = & i(w_{1t};x_{1t}|z_{1t},t ) + i(u_{1t};x_{1t}|y_{1t},w_{1t},t ) \\ & & + i(w_{2t};x_{2t}|z_{2t},t ) + i(u_{2t};x_{2t}|y_{2t},w_{2t},t ) \\ & = & i(w_{1t},t ; x_{1t}|z_{1 t } ) + i(u_{1t};x_{1t}|y_{1t},w_{1t},t ) \\ & & + i(w_{2t},t;x_{2t}|z_{1 t } ) + i(u_{2t};x_{2t}|y_{2t},w_{2t},t ) , \end{aligned}\ ] ] where we set @xmath193 and @xmath136 is the uniform random number on @xmath137 that are independent of the other random variables .
note that @xmath138 satisfy @xmath139 for @xmath22 . in a similar reason as in section [ subsec : proof - converse ] , there exist functions @xmath194 and @xmath195 satisfying eqs .
( [ eq : distortion - condition-1 ] ) and ( [ eq : distortion - condition-2 ] ) for @xmath22 . thus , by taking @xmath196 , @xmath197 , @xmath168 , and @xmath198 , we have that there exist @xmath49 satisfying eqs .
( [ eq : distortion - condition-1 ] ) and ( [ eq : distortion - condition-2 ] ) and @xmath199 although the auxirially random variables @xmath51 and @xmath52 chosen above are not independent each other , they never appear in any term simultaneously .
thus we can take @xmath51 and @xmath52 to be independent each other .
since @xmath30 is arbitrary , by the continuity of @xmath65 with respect to @xmath29 , and by using the support lemma @xcite , we have the statement of the theorem .
@xmath109 first , by chain rules , we have @xmath200.\end{aligned}\ ] ] since @xmath201 and @xmath202 are independent , we have @xmath203 similarly , since @xmath204 and @xmath205 are independent , we have @xmath206 furthermore , since the markov chains @xmath207 and @xmath208 hold , we have @xmath209 and @xmath210 thus , we have @xmath211 .
\label{eq : lemma - proof-1}\end{aligned}\ ] ] by chain rules , we have @xmath212 similarly , we have @xmath213 from eq .
( [ eq : matched - degraded ] ) , we have @xmath214 thus , we have @xmath215 similarly , from eq .
( [ eq : matched - degraded ] ) , we have @xmath216 finally , by substituting eqs .
( [ eq : lemma - proof-2])([eq : lemma - proof-5 ] ) into eq .
( [ eq : lemma - proof-1 ] ) , we have @xmath117.\end{aligned}\ ] ] @xmath109 a. kimura , t. uyematsu , s. kuzuoka , and s. watanabe , `` universal source coding over generalized complementary delivery networks , '' _ ieee trans .
inform . theory _
55 , no . 3 , pp .
13601373 , march 2009 .
|
this paper investigates a lossy source coding problem in which two decoders can access their side - information respectively .
the correlated sources are a product of two component correlated sources , and we exclusively investigate the case such that each component is degraded .
we show the rate - distortion function for that case , and give the following observations .
when the components are degraded in matched order , the rate distortion function of the product sources is equal to the sum of the component - wise rate distortion functions . on the otherhand ,
the former is strictly smaller than the latter when the component sources are degraded in mismatched order .
|
the response of ice as a function of polarization ( `` birefringence '' ) is characterized by differences in either wavespeed or absorption along linear ( generally orthogonal ) axes . over a km - scale pathlength , in the absence of any preferred in - ice direction , one might expect any time propagation asymmetry at the single - crystal level to be macroscopically mitigated by the randomness of the corresponding single - crystal orientation . in such a case , over a total pathlength @xmath2 consisting of @xmath3 unit steps , each characterized by an asymmetry @xmath4 , the average propagation time along each polarization axis should have a gaussian distribution , centered at @xmath5 , with width @xmath6 .
the asymmetry distribution would therefore be a gaussian of width @xmath7 , centered at zero . for 1% birefringence ( @xmath8 ) , @xmath2=1000 m , and step sizes corresponding to typical grain sizes ( @xmath9 m , or n=@xmath10 ) , we expect @xmath110.1 ns . if , however , bulk flow of the ice sheet results in a preferred in - ice direction much longer than typical grain sizes , the propagation time asymmetry can be @xmath12(10 ns ) .
efforts are underway to use the antarctic ice sheet as an ultra - high energy neutrino target@xcite .
neutrino - ice collisions result in the production of charged particles which emanate from the interaction point with velocities approaching the speed - of - light _ in vacuo _ : @xmath13 . in a medium with index - of - refraction @xmath14 , detection of the resulting cherenkov radiation in either the near - uv ( tev - scale neutrinos ) or radio wavelength regime ( pev - scale neutrinos ) by a suite of sensors can be used to reconstruct the kinematics of the initial neutrino , provided the absorption and refraction of the original electromagnetic signal due to the intervening ice can be reliably estimated .
the rice experiment@xcite demonstrated the feasibility of the radio - detection approach over the last decade .
the askaryan radio array ( ara ) collaboration@xcite seeks to substantially enlarge the current rice footprint at south pole by instrumenting an 80 @xmath15 area over the period 2010 - 2015 .
ice birefringence could result in an initially ns - scale rf pulse being resolved into two components , with a time stagger comparable to the signal duration itself , requiring a trigger system with a correspondingly long signal integration time .
complete characterization of the ice permittivity , as a function of depth , and also polarization is therefore essential in obtaining a reliable estimate of the neutrino detection efficiency .
dielectric studies of ice have been ongoing for nearly a century@xcite , with radio frequency studies well over half a century old@xcite .
the study of ice birefringence is itself a discipline which is also several decades old@xcite . at visible wavelengths ( 590 nm )
, ice exhibits a birefringent asymmetry of @xmath00.3% .
early rf studies focused on single - crystal measurements in the lab , with complementary samples of data collected _ in situ_. even prior to the advent of ns - scale signal capture , the field was already quite mature , with estimates of birefringent asymmetries of order 1%@xcite . to explain the asymmetry observed in those early _ in situ _ data , the possible effects of geometrically distorted air bubbles and/or other sources of in - ice density contrast , internal layering , and the inherent single - crystal anisotropy , macroscopically manifest as a preferred axis of the crystal orientation fabric ( cof ) , were all considered .
the physical mechanism for cof alignment , however , was not fully explained experimental data up to that point contradicted the hypothesis that internal strain due to ice flow is exclusively responsible for the observed birefringence .
briefly , crystal orientation correlates with depth as follows@xcite : i ) in deeper antarctic ice , crystals align with the horizontal glide plane ; near the surface , there is no preferred orientation , ii ) complete cof alignment results in an order - of - magnitude faster bulk flow than randomly oriented cof .
note the interplay between ice flow and cof alignment - as shear increases , cof alignment increases , which , in turn , facilitates ice flow .
the effect can be particularly dramatic at ice divides@xcite , given the large internal strains expected at those locales .
the correlation between cof alignment and ice flow was verified by an analysis of airborne radar ( 60 mhz ) data taken in the vicinity of vostok station , where the derived east antarctic cof alignment was found to corroborate direct vostok ice core data@xcite .
that study produced many results relevant to our investigation most notably the conclusion that the variation of echo strength with increasing depth indicates scattering dominated by density , acidity , and cof effects , in the top , middle , and bottom third of the ice sheet at vostok . to the extent that birefringence results from cof
, we would therefore expect that non - zero birefringence would be most noticeable for deeper returns .
similarly , airborne - sounding data taken at dronning maud land , some 3000 km west of vostok , were also compared against ice - core data@xcite .
150 mhz pulses of 60 ns and 600 ns duration revealed the presence of an horizontally extended ( @xmath05 km ) plane , at a depth of 20252045 m , or approximately 800 m above the bed . that deep reflector was found to be inconsistent with acidity scattering , and more compatible with a discontinuity in the cof .
direct comparison of vostok , or dronning maud data to south pole , however , must properly take into account the variation in ice flow due to the different underlying stratigraphy and the fact that much of the ice sheet at vostok lies in proximity to a nearly resistance - less lake . over a length scale of tens of kilometers ,
south polar ice is observed to flow at the surface with a velocity of approximately 910 m / yr@xcite in a direction which corresponds to about 153@xmath16 in the coordinate system used for these measurements ( corresponding to the 40 degree west longitude line ) .
the velocity profile , as a function of depth , has been determined by price _
et al._@xcite , and is reproduced in figure [ fig : iceflow - v - depth.eps ] . to the extent that ice flow shear implies cof - alignment , and cof - alignment is singly responsible for birefringence , we would therefore expect very little birefringence to a depth of @xmath02 km , with increasing birefringence at greater depths , if the ice flow at present is the same as the ice flow through the history of the ice sheet formation .
there have been several more recent@xcite _ in situ _ measurements of antarctic ice@xcite which have also been interpreted as evidence for cof - induced birefringence . in a study of internal layer reflections conducted at dome fuji@xcite , co -
polarized ( transmitter [ tx ] and receiver [ rx ] antenna polarizations parallel , projected onto the horizontal plane ) and cross - polarized ( tx and rx perpendicular ) measurements were made over the full azimuth .
signals broadcast at a given frequency using three - element yagis were detected after internal scattering from within the ice ; both @xmath17 and @xmath18 modulations of the received amplitude were observed . under the assumption that the ordinary and extra - ordinary birefringent axes are orthogonal ,
the latter was interpreted as due to birefringent - induced interference effects in the time - domain . in particular , as the transmitter - receiver systems were rotated in azimuth ,
signal drops were observed at angles @xmath19 ( with m an integer ) relative to the presumed horizontal cof - alignment , which were attributed to destructive interference between the ordinary vs. extra - ordinary signals at the broadcast wavelength .
those data@xcite imply @xmath20 variation over a depth of 1500 meters , which ( given that @xmath213 m for 179 mhz in - ice broadcasts ) imply a birefringent asymmetry of order 0.1% .
the authors note that their measured birefringence is somewhat weaker at 60 mhz , although an authoritative recent study of pure laboratory ice obtained an asymmetry @xmath22 of [email protected]%@xcite at both 1 mhz and 39 ghz . in a comprehensive attempt to model the dome fuji data , and neglecting any possible tilting of the reflecting internal layers ( which must occur coherently over an aeral scale of order the fresnel zone to be significant ) ,
the dependence of radar scattering on density , cof , and acidity effects were assessed@xcite . in principle , the type of scattering can be elucidated on the basis of signal strength : cof and acidity - based layers typically reflect -30 -60 db of the incident power ; density scattering , integrated through a vertical chord is typically a factor 10 larger ( in power ) .
to the extent that acidity scattering is a pure conductivity effect , we would expect it to vary as the inverse of frequency .
to the extent that density and cof scattering is due to a variation of @xmath24 , we would expect such scattering to show much weaker frequency dependence , depending on the proximity of debye resonances . in order to simplify the interpretation of data
, the authors made the assumption that one scattering effect was dominant , and that although all types of scattering can result in large cross - polarized signals , only cof produces anisotropic scattering which also gave the observed azimuthal interference patterns .
in 2006 , our group previously used time - domain bedrock reflections observed at a site near taylor dome@xcite to estimate a birefringent asymmetry of 0.12% , projected onto the vertical @xmath1-axis ( perpendicular to the surface ) .
a lack of conclusive ice flow data at taylor dome prevented a correlation with the local ice flow direction from being established . a follow - up study in 2008@xcite at the south pole searched for , but found no detectable azimuthal variation in echo return times through the upper 1600 m of the ice sheet .
table [ tab : birefsum ] summarizes some recent birefringence measurements .
.summary of recent birefringence measurements .
[ tab : birefsum ] [ cols="^,^,^,^",options="header " , ] = 0 corresponds to the reference polarization orientation for tem horns , chosen to correspond to the long axis of the mapo building for convenience .
surface ice flow direction at south pole is shown in red , corresponding to @xmath25=27@xmath16 ( @xmath26 + 153@xmath16 ) .
magnitude of surface ice flow is measured to be @xmath09 m / yr ( fig . [
fig : iceflow - v - depth.eps ] ) , and uniform to a depth of @xmath02 km .
( data courtesy of kurtis skoog.),width=264 ] = 0 corresponds to the reference polarization orientation for tem horns , chosen to correspond to the long axis of the mapo building for convenience .
surface ice flow direction at south pole is shown in red , corresponding to @xmath25=27@xmath16 ( @xmath26 + 153@xmath16 ) .
magnitude of surface ice flow is measured to be @xmath09 m / yr ( fig . [
fig : iceflow - v - depth.eps ] ) , and uniform to a depth of @xmath02 km .
( data courtesy of kurtis skoog.),width=302 ]
the martin a. pomerantz observatory ( mapo ) building , located at the south pole , is used to house the signal generator and data acquisition system used for our measurements .
two 1.25-cm thick lmr-500 coaxial cables , each approximately 40 m long , were fed from within mapo through a conduit at the bottom of the building and out onto the snow , and connect to the transmitter and receiver horn antennas ( separated by 50 m , and on opposite sides of the mapo building ) , respectively .
complete details of this measurement , including calibration of antennas , amplifiers , filters , cables , and verification of both the thermal noise floor , as well as the determination of the polarization dependence of the output signal , are provided in our 2008 study , probing the upper half of the ice sheet@xcite .
our experimental geometry , relative to the surface ice flow axis , is identical to those previous measurements , and shown in figure [ fig : iceflow ] .
as before , a lecroy 950 waverunner digital oscilloscope is used for data capture .
the primary difference is that , for the measurements described herein , higher - power pulsers are used , with a maximum output voltage of @xmath03000 v. these higher - power pulsers now permit observation of the bottom reflection itself . to ensure consistent results
, we employed two pulsers ( grant corporation hyps , and fid model fpg6 - 1pnk ) , two sets of antennas ( a pair of tem horn antennas built by the institute of nuclear resarch [ inr ] , moscow , russia and two seavey co. quad - ridge , dual - polarization antennas of the same type used by the anita experiment ) , and also two amplifier gains ( 36 db , or 52 db gain ) , with a variety of high - pass filters to allow multiple comparisons of signal shape and amplitude .
the output signal shape from the fid pulser is shown in figure [ fid - output ] .
the combination of the fid pulser plus the inr horn antennas , at 36 db gain , with 100 mhz high - pass filtering were found to yield the best signal - to - noise , without amplifier saturation .
this configuration provides the bulk of the results herein , and is defined as the ` run 1 ' configuration . to improve signal : noise , waveforms are typically averaged over @xmath010000 captures .
each averaged waveform is designated by the angle of the transmission polarization axis relative to our zero - degree reference ( fig .
[ fig : iceflow ] ) , as well as the corresponding angle for the receiver axis .
we attempted to detect birefringent effects using the same approach as that used in our previous taylor dome analysis@xcite ; namely , we search for a measurable time difference in received signals , as a function of the orientation of the long axis ( transmitted signal polarization axis ) of the horn antennas .
although the received spectral power is expected to be mostly determined by the @xmath27 dependence of the horn antenna effective area , we have examined the signal strength as a function of frequency ( transforming 32-ns [ 64-sample ] time segments ) , as well , presuming that the frequency characteristics of the bedrock reflection may be differentiable from internal layer scattering , density scattering , etc . and thereby help elucidate distinct echoes .
the obtained voltages bracketing the time for the bedrock reflection ( @xmath034 @xmath28s ) , as a function of rotation angle , are shown in figure [ fig : rt090 ] .
as we sweep in azimuth , we note an apparent shift in the bedrock reflection time , between the zero degree and + 30 degree orientation data sets .
figures [ fig : sprefl09 ] and [ fig : sprefl09_1 ] present the data for the -30 degree vs. + 60 degree orientations , zooming in around the expected bedrock reflection time to illustrate the @xmath050 ns time shift , and also showing the fourier transform ( ft ) .
.,width=491 ] .,width=491 ] by comparison , the echo returns observed at earlier times are shown in figure [ fig : rt090_0 ] and zoomed in figure [ fig : rt090_1 ] .
based on the time stagger in the bedrock reflection , and assuming that the time delay is linear with ice depth , we would expect to observe a corresponding 20 ns delay at an echo time of @xmath014 @xmath28s .
no such obvious shift is observed , consistent with the results obtained in our previous study ( see fig . [ fig:13.9us - all ] , taken from our 2008 south polar data analysis ) .
our data therefore suggest that the birefringent asymmetry is primarily generated in the lower half of the ice sheet .
ideally , we would have sufficient resolution to observe discrete scattering layers near the bed , and thereby map the magnitude of the birefringent delay as a function of depth .
although a visual inspection of the waveforms reveals the presence of enhancements in the received voltage at echo times of @xmath024 and @xmath026 microseconds , they are not sufficiently convincing to allow an extraction of a birefringent asymmetry .
previous res measurements have , in fact , similarly observed a lack of evident echoes in deep ice for @xmath010 microseconds preceding the bedrock echo ( the so - called `` echo - free zone '' ( efz)@xcite ) the efz is illustrated by the distribution of recorded voltages , corrected for averaging and amplification , for echo times greater than 25 microseconds , as shown in figure [ hv_fit_log.eps ] . the expected rms voltage , assuming a 220 k environment and comparable system noise , and using @xmath29 with @xmath30 boltzmann s constant and @xmath31 the bandwidth of our receiver system ( @xmath01 ghz ) is approximately 20 microvolts . figures [ fig : sprefl09 ] and [ fig : sprefl09_1 ] display an apparently extended signal shape , including some enhanced power apparently preceding the primary bedrock reflection .
this extended signal shape , persisting for @xmath0200300 ns , was similarly observed in our studies at taylor dome , and is also present in the bedrock reflection data taken by ground - penetrating radar measurements from , e.g , the center for remote sensing of ice sheets ( cresis)@xcite . the primary reflecting area on the bedrock surface is expected to be roughly one fresnel zone wide , or approximately @xmath32 in radius . with @xmath331 m , and @xmath345.6 km , the time difference between a normally incident ray and a ray reflecting at the periphery of the first fresnel zone should be @xmath35 , or approximately 20 ns , considerably smaller than our observations .
given the typical antenna beamwidth of @xmath030 degrees in air ( @xmath020 degrees in ice ) , incoherent scattering extending across several fresnel zones is therefore likely responsible for much of the signal extension observed .
the assumption of incoherent scattering from macroscopic rubble near the bed is also consistent with our observation that signals observed with the seavey antennas , which are responsive for frequencies greater than 250 mhz , are conspicuously more extended than the signals observed using the inr horns , which have good vswr down to 100 mhz ; longer - wavelength components should be less susceptible to decimeter - scaled bedrock scatterers if the bottom 10 meters of the ice sheet also contains rubble , presumably resulting from glacial motion across the bed , the resulting scattering would also explain the apparent ` pre - signal ' observed in several of the waveforms .
the mixing of rock with basal water / ice was noted over 40 years ago after the first core to bedrock was extracted at byrd station@xcite
. 50 ns.,width=491 ] measurements were conducted using a variety of gains , as well as antennas . due to varying locations of transmitter / receiver , as well as different instrumental delays , the absolute time of the measured bottom echo can shift .
nevertheless , for each separate ` run ' ( defined as one stable hardware configuration ) , when selecting the maximum voltage signal observed in the interval 3300035000 ns return time , a time delay of approximately 53 ns was observed in the signal arrival times between the two modes ( presumably corresponding to propagation along the major and minor birefringence axes ) , as we rotate antennas in azimuth , as illustrated in figure [ fig : v1v2 ] . when broadcasting along one of the two birefringent axes , we expect to observe only one signal ( assuming perfect isolation between the vpol and hpol terminals of our antennas ) ; when broadcasting at 45 degrees relative to the two birefringent axes , we expect to observe equal amplitudes for the ` early ' vs. ` late ' received signals .
we have summarized the relative amplitudes of the two signals in figure [ fig : v1v2_1 ] ; a value greater than 1 indicates that the ` early ' signal was measured to have larger voltage ( and vice versa ) .
we note that this method of determining the optical axis orientation by the observed elliptical polarization is very similar to the formalism outlined over three decades ago by hargreaves@xcite . interpreted as due exclusively to birefringence
, our data indicates that the birefringent axes are oriented at approximately 40 and 50 degrees , respectively , relative to the zero degree axis , or within 25 degrees of the local ice flow direction .
previous measurements of birefringence ( @xcite ) found a lack of clear correlation between the optical axis and the local ice flow direction ; those prior measurements attributed the misalignment to a multi - step process leading to the in - ice crystal orientation .
we can not , of course , exclude the possibility that an asymmetry in the imaginary portion of the dielectric constant is responsible for at least some of the amplitude variation we are observing ; fujita _
et al_@xcite claimed an extreme difference of 15% in the loss tangent for broadcasts parallel vs. perpendicular to the optical axis . in the absence of birefringence , if we broadcast along one polarization , using antennas with excellent cross - polarization rejection , we would expect to observe little or no received signal in a receiver with perpendicular alignment . according to the manufacturer s specifications for the seavey antennas ,
the cross - polarization rejection between is approximately 10:1 in voltage ( better than 20 db rejection in power ) ; we similarly measure better than 14 db rejection in the lab for the inr antennas .
nevertheless , we note considerable cross - polarized power measured at our receivers , as shown in figure [ fig : xpol.eps ] .
this is , in fact , not the first observation of such co - polarization power at the pole a reported rotation of the basal echo by 90 degrees was reported over 40 years ago@xcite , consistent with our 2004 bedrock reflection data from south pole@xcite . for transmission at an angle @xmath36 relative to the ordinary axis , the signal amplitudes measured for the `` fast '' signal in the co - polarization vs. cross - polarization orientations should be @xmath37 and @xmath38 , respectively .
conversely , the signal amplitudes , projected onto the extraordinary axis , in the co - polarization vs. cross - polarization orientations should be @xmath39 and @xmath40 , respectively .
if birefringence were solely responsible for the observed cross - polarized signal , we would therefore expect equal magnitudes of cross - polarized signal observed in all orientations , at variance with the observation in figure [ fig : xpol.eps ] .
this suggests an additional mechanism is at least partly responsible for the amplitude variation observed for the early vs. late bedrock echoes .
our 2008 analysis at south pole@xcite also showed considerable observed cross - polarized signal strength for internal reflections , in a regime where the average birefringent effect was found to be consistent with zero ; our measurements at taylor dome@xcite displayed somewhat less cross - polarized power in the bedrock reflection , as shown in figure [ fig : td - hh - vv-1.eps ] .
hh rotate '' refers to the tx and rx orientation for which the vv axes of both tx and rx have been rotated into the initial hh orientation . in this figure ,
the three uppermost signals have been vertically offset to enhance visual clarity.,width=491 ] echoes provided by polarization - grating - like reflectors , as well as the mechanism responsible for the observed extended bedrock echo signal shape may interfere with birefringent asymmetries . the fact that each of these has an associated , and generally unknown preferred axis complicates an unambiguous association of the amplitude data shown in figure [ fig : v1v2_1 ] with the birefringent asymmetry axis alone . if the reflecting plane is not horizontal , then there will also be an asymmetry generated upon reflection , for the electric field components perpendicular , and parallel to the reflecting surface , as well . in principle
, faraday rotation can also lead to considerable cross - polarized power ; however , a recent laboratory study of cold ice disfavors that possibility@xcite .
to determine the ice thickness at our measurement point , we use the average time delay observed for run 1 ( 33972 ns ) , corrected for the total cable delay ( 160.3 ns ; figure [ fid - output ] ) and use a temperature - weighted average em wave velocity of @xmath41 m/@xmath28s below the firn@xcite . through the firn ,
we use direct measurements of radio propagation wavespeed@xcite , implying a one - way transit time of 1102 ns through the upper 200 m of the ice sheet at south pole . taking the errors from @xcite
gives an implied depth of @xmath42 m. however , field measurements of radio frequency wave velocity have shown variations in the permittivity of up to 1% ; uncertainties in impurity levels can , in principle , contribute an approximately equivalent systematic error . taken together , these imply a systematic depth error of @xmath030 m.
knowing the total cable length ( @xmath0125 m ) , the cable loss per unit length ( @xmath05 db/100 m at 500 mhz ) , the maximum signal amplitude at the output of the pulser ( 2.5 kv ) , the measured reflected signal amplitude @xmath44 , the bandpass of the antennas ( @xmath01 ghz ) , and the net gain of the amplifiers + filters in the system , we can determine `` absolutely ''
the average attenuation length to the bed by direct application of the friis equation : @xmath45 , with @xmath46 the bedrock reflectivity ( taken to be 0.3 ) , @xmath47 and @xmath48 the measured signal voltages , @xmath49 the antenna gain of transmitter and receiver antenna ( taken to be 12 dbi given the expected in - ice focusing , and consistent with values used for our previous 2004 analysis with the same antennas@xcite ) , @xmath50 the total round - trip echo path ( 5.7 km ) , and @xmath51 the broadcast wavelength . since we are using sharp time - domain signals , rather than the limited - duration
tone s used for our previous , dedicated attenuation length study , we do not expect the same precision as attained in that previous study , and correspondingly use @xmath51=1 meter as sufficient for this coarse check of the received signal amplitudes .
averaged over the entire vertical chord , we obtain an implied average attenuation length @xmath52627 m , entirely consistent with those 2004 measurements at pole . by performing a fourier
transform on the received waveforms , we can investigate the dispersive characteristics of ice over the frequency range 200 mhz
figure [ fig : dispersion ] displays the run 1 data at a tx / rx angle of 5 degrees . for this plot , the amplitude at high frequencies has been multiplied by frequency ( squared ) to compensate for the expected wavelength dependence of both receiver and transmitter effective height . from this figure , we estimate synchronicity of received power to within 4 ns for all frequencies considered , implying a variation in the real part of the dielectric constant less than 0.012% over the range of frequencies considered in this analysis . this asymmetry is consistent with the naive expectation that the variation in dielectric constant with frequency should be relatively small at radio frequencies , far from debye resonances for ice .
by contrast , in their extensive review of radio ice sheet sounding , dowdeswell and evans@xcite suggest a variation in the dielectric constant of approximately @xmath53 over the interval 1100 mhz , corresponding to a wave speed variation of about 1.25% .
we observe approximate synchronicity at all frequencies.,width=491 ]
an in - ice neutrino interaction can produce either a single particle ( muon ) propagating over km - scale distances , or a compact electromagnetic or hadronic shower , limited to @xmath010 m in extent .
the former is best detected with suitably located photomultiplier tubes sensitive to uv radiation .
the latter is best detected with englacial radio receivers sensitive to the expected @xmath0ghz bandwidth , 1-ns duration `` askaryan '' signal impulse .
since the measured birefringent asymmetry is considerably larger in magnitude than 1 ns , a 1 ghz bandwidth receiver will generally observe ` split ' signals , with relative amplitudes depending on the geometry and actual orientation of the optical axis relative to the signal propagation direction .
monte carlo simulations indicate that , without compensating trigger electronics , the birefringent asymmetry measured here would result in an approximately 5% reduction in equivalent neutrino detection volume .
the smallness of this effect is primarily due to the fact that a ) the measured asymmetry is limited to deeper , poorer quality ( from the standpoint of rf transmission ) polar ice , b ) the measured asymmetry is restricted to the vertical axis , while neutrino signals are predominantly observed along near - horizontal incident angles .
we have measured a birefringent effect , in the lower half of the ice sheet at south pole .
the magnitude of the observed effect ( @xmath00.3% ) is approximately consistent with other measurements of the radio frequency properties of cold polar ice
. it would be desirable to correlate our measurement with direct physical measurements of south polar ice down to the bed .
unfortunately , despite having perhaps the most developed infrastructural base in the antarctic interior , an ice core of the type taken at siple dome , dome c , monning draud or vostok has not yet been extracted .
such a core would provide incontrovertible , and direct evidence for the cof correlation only suggested by our current measurements .
if such an association could be verified , the application of birefringent and/or cof measurements to perform glacial tomography could potentially develop as a discipline unto its own@xcite .
the fact that our measurements for the upper half of the ice sheet are in direct contradiction to those made at dome fuji imply that the ice characteristics are considerably different between those two locales ; the fact that the interior is often regarded as one monolithic glacial mass notwithstanding .
our results are , however , in agreement with the analysis of data taken in the vicinity of vostok , which found evidence for cof scattering dominating only in the lowest third of the ice sheet , in agreement with direct core data@xcite . a follow - up measurement would accumulate more cross - polarization data , with yet a higher - power pulser , and attempt to further elucidate the birefringent dependence on depth .
as hargreaves@xcite points out , rotation of transmitter / receiver antennas through the full azimuth in the cross - polarization orientation vs. rotation of tx / rx in the co - polarization orientation allows one to separate out effects due to birefringence vs. effects due to asymmetric reflecting planes .
assuming different reflection coefficients in the x- vs. y - directions ( @xmath54 and @xmath55 ) , the maximum signal measured with transmitter perpendicular to the receiver , compared to the minimum signal measured in the same configuration allows an extraction of @xmath54 vs. @xmath55 . the matrix - based approach of fujita _
et al._@xcite also allows a separation of the various contributions to anisotropic scattering based on a comprehensive analysis of the elliptical polarization characteristics over a broad range of geometries and depths .
time limitations , however , prevented us from making the requisite suite of measurements needed to fully exploit that formalism .
in addition to accumulating a more complete data sample , a successor experiment would also take advantage of the ability to broadcast , and study the reflections from , elliptically polarized signal of arbitrary ellipticity using the dual polarization ports of the seavey horn antennas .
also particularly interesting would be data which might further reveal the frequency dependence of echo returns .
figure [ fig : fullfft ] displays the fft , as a function of return time , for the @xmath56 orientation ( chosen at random ) . orientation , as a function of depth .
voltages have been scaled by @xmath57 to compensate for transmitter / receiver frequency response.,width=491 ] although inconclusive , we observe considerable high - frequency power associated with , e.g. , the return at @xmath010000 ns , consistent with expectations from density / cof scattering .
a pure conducting plane reflector would be expected to be : a ) uniform as a function of rotation in azimuth , and b ) have a power spectrum falling inversely with frequency , at variance with our observations . with enough resolution
, one might hope to observe the sequence of frequency - independent , frequency - dependent , and again frequency - independent returns ( and over the full azimuth ) that would be expected in a model where the scattering is dominated by density , acid , and cof effects , respectively .
additional data at higher power , and using a short - duration ` tone ' would be very useful in this respect . in the process of performing this measurement ,
one conspicuous pattern that emerges is the similarity of waveforms delineated by the optical axes we identify in this analysis .
i.e. , beyond 15000 ns , sharp scattering lines consistently appear at the same depth for data taken at angles @xmath58 ; a different set of reflectors are evident for data taken at angles @xmath59 ( and consistently within that angular interval ) .
similarly , signal shapes ( sharp or extended ) of internally reflected radio waves are repeatable within the two distinct angular ranges ( figs [ fig:25800 - 26600_0.eps ] , [ fig:25800 - 26600_1.eps ] , [ fig:25800 - 26600_2.eps ] , [ fig:25800 - 26600_3.eps ] ) ; this is perhaps an indication that preferred crystal orientation is also accompanied by an asymmetry in absorption .
finally , we re - emphasize that we have sensitivity only to the projection of the optical axis onto the vertical the true birefringent asymmetry could , in principle , be much larger than the value reported here .
the ara experiment is expected to have the capability to broadcast in - ice signal horizontally over several km and should therefore provide essential vertical polarization information at radio frequencies .
the authors particularly thank chris allen ( u. of kansas ) and kenichi matsuoka ( u. of washington ) for very helpful discussions , as well as our colleagues on the rice and anita experiments .
we also thank andy bricker of lawrence high school ( lawrence , ks ) for his assistance working with the lawrence students .
this work was supported by the national science foundation s office of polar programs ( grant opp-0826747 ) and quarknet programs .
any opinions , findings , and conclusions or recommendations expressed in this material are those of the author(s ) and do not necessarily reflect the views of the national science foundation . 99 abassi , r. and 217 others , 2010 .
`` icecube collaboration contributions to the 2009 international cosmic ray conference '' , arxiv:1004.2093 .
kravchenko , i. and 20 others , 2006 .
`` rice limits on the diffuse ultrahigh energy neutrino flux '' , _ phys .
* 73 * ( 082002 ) .
gorham , p.w . and 40 others , 2009 .
`` new limits on the ultrahigh energy cosmic neutrino flux from the anita experiment '' , _ phys .
* 103 * ( 051103 ) .
gow , a. , h. ueda and d. garfield , 1968 .
`` antarctic ice sheet : preliminary results of first core hole to bedrock '' , _ science _ , * 161*(3845 ) , 1011 .
ara : the askaryan radio array , approved by the national science foundation ( april , 2010 ) .
ehringhaus , a , 1917 .
_ neus jb .
miner geol .
palont _ , * b41 * ( 342 - 419 ) .
auty , r. and r. cole , 1952 .
`` dielectric properties of ice and solid @xmath60 . '' _ j. chem .
_ * 20*(8 ) , 1309 .
hargreaves , n.d . , 1977 .
`` the polarization of radio signals in the radio echo sounding of ice sheets . ''
_ j. phys .
d , appl . phys _ , * 10*(9 ) , 1285 - 1304 .
hargreaves , n.d . , 1978 , `` the radio frequency birefringence of polar ice . '' _
j. glac . _
* 21 * , 301 - 313 .
riffenburgh , b. , 2008 . _ encyclopedia of the antarctic_. siegert , m.j . and r. kwok , 2000 .
`` ice - sheet radar layering and the development of preferred crystal orientation fabrics between lake vostok and ridge b , central east antarctica . '' _ earth and planetary science letters _ , * 179*(2 ) , 227 - 235 . price , p.b . and 9 others , 2002 .
temperature profile for glacial ice at the south pole : implications for life in a nearby subglacial lake .
_ , * 99*(12 ) , 78447877 .
cresis , 2008 .
https://www.cresis.ku.edu/research/dataproductsandmodeling.html fujita , s. , h. maeno , and k. matsuoka , 2006 .
`` radio - wave depolarization and scattering within ice sheets : a matrix - based model to link radar and ice - core measurements and its application '' , _ j. glac . _ , * 52 * , 178 .
matsuoka , k. , t. furukawa , s. fujita , n. maeno n , s. uratsuka , r. naruse , o. watanabe , 2003 .
`` crystal orientation fabrics within the antarctic ice sheet revealed by a multipolarization plane and dual - frequency radar survey . ''
j. geophys .
* 108 * ( b10 ) .
matsuoka , t. , s. fujita , s. morishima and s. mae , 1997 .
`` precise measurement of dielectric anisotropy in ice ih at 39 ghz . '' _ j. appl .
phys . _ * 81*(5 ) , 2344 .
matsuoka , k. , s. uratsuka , s. fujita , f. nishio , 2004 .
`` ice - flow - induced scattering zone within the antarctic ice sheet revealed by high - frequency airborne radar . ''
_ j. glac . _ * 50*(170 ) , 382 - 388 .
besson , d.z . and 42 others , 2008 .
`` in situ radioglaciological measurements near taylor dome , antarctica and implications for uhe neutrino astronomy . '' _ astropart .
phys . _ * 29*(2 ) .
besson , d. , r. keast and r. velasco , 2009 .
`` in situ and laboratory studies of radiofrequency propagation through ice and implications for siting a large - scale antarctic neutrino detector . '' _ astropart .
phys . _ , * 31*(5 ) , 348 - 358 .
doake , c. , h. corr , and a. jenkins , 2002 .
`` polarization of radio waves transmitted through .
antarctic ice shelves . '' _ ann .
glac . _ * 34*(1 ) , 165 - 170 .
doake , c. , h. corr , a. jenkins , k. nichols , c. stewart , 2003 .
`` applications of sar polarimetry and polarimetric interferometry . '' _ euro .
space agency _ * 529 * , 313 - 320 .
fujita , s. , t. matsuoka , s.morishima , s. mae , 1993 .
`` the measurement on the dielectric properties of ice at hf , vhf and microwave frequencies . '' _ ieee igarss 93_. fujita , s. and s. mae , 1993 .
`` relation between ice sheet internal radio - echo reflections at mizuho station , antarctica . '' _ ann .
* 17 * , 269 - 75 .
doake , c.s.m . , 1981 .
`` polarization of radio waves propagated through george vi ice shelf . ''
_ british antarctic survey bulletin_. woodruff , a.h . and c.s.m .
doake , 1979 .
`` depolarization of radio waves can distinguish between floating and grounded ice sheets . '' _
_ * 23 * , 223 .
fujita , s. , k. matsuoka , h. maeno , t. furukawa , 2003 .
`` scattering of vhf radio waves from within an ice sheet containing the vertical - girdle - type ice fabric and anisotropic reflection boundaries . '' _ ann .
glac . _ * 37 * , 305 .
robin , g. de q. , drewry , d. j. , and meldrum , d. t. , 1977 .
`` international studies of ice sheet and bedrock . ''
_ philosophical transactions of the royal society of london , series b , biological sciences _ 279(963 ) , 185196 .
fujita , s. , maeno .
h. , and uratsuka , s. , 1999 , `` nature of radio echo layering in the antarctic ice sheet detected by a two - frequency experiment . ''
_ journal of geophysical research _ 104 , * b6 * , 1301313024 .
matsuoka , t. , fujita , s. , morishima , s. , and mae , s. , 1997 , `` precise measurement of dielectric anisotropy in ice ih at 39 ghz '' _ journal of applied physics .
_ 81(5 ) , 2344 2348
. jiracek g. r. , 1967 .
_ university of wisconsin res .
67 - 1 _ barwick , s. , d. besson , p. gorham , d. saltzberg , 2005 .
`` south polar _ in situ _ radio frequency ice attenuation . '' _
_ * 04j067 * dowdeswell , j.a .
and s. evans , 2004 .
`` investigations of the form and flow of ice sheets and glaciers using radio - echo sounding . '' _ rep .
_ , * 67 * kravchenko , i. , d. besson and j. meyers , 2005 .
`` _ in situ _ index of refraction measurements of the south polar firn with the rice detector '' , _ j. glac . _ 03j061 .
eisen , o. , i. hamann , s. kipfstuhl , d. steinhage , and f. wilhelms , `` direct evidence for continuous radar reflector originating from changes in crystal - orientation fabric '' , _ the cryosphere _ , 1 , 2007 , pp 1 - 10 .
http://www.the-cryosphere.net/1/1/2007/ eisen , o. , f. wilhelms , d. steinhage and j. schwander , 2006 .
`` improved method to determine res - reflector depths from ice - core profiles of permittivity and conductivity '' , _
_ * 52 * , 177 ( 299 - 310 ) martaan , c. , g.h.gudmandsson , h.pritchard and o.gagliardini , 2008
. `` ice flow and crystal fabric near ice divides . '' _ geophys .
res . abstracts _ , * 10*.
|
using a bistatic radar echo sounding ( res ) system developed for calibration of the rice particle astrophysics experiment at the south pole , we have studied radio frequency ( rf ) reflections off the bedrock .
the total propagation time of @xmath0ns - duration , vertically ( @xmath1 ) broadcast radio signals , as a function of polarization orientation in the horizontal plane , provides a direct probe of the geometry - dependence of the ice permittivity to a depth of 2.8 km .
we observe clear birefringent asymmetries along @xmath1 in the lowest half of the ice sheet , at a fractional level @xmath00.3% .
this result is in contrast to expectations based on measurements at dome fuji , for which birefringence was observed in the upper 1.5 km of the ice sheet .
this effect , combined with the increased radio frequency attenuation expected near the bedrock , renders the lower half thickness of south polar ice less favorable than the upper half of the ice sheet in terms of its ultra - high energy neutrino detection potential .
radio frequency birefringence in south polar ice and implications for neutrino reconstruction i. kravchenko + _ university of nebraska , department of physics and astronomy , lincoln , ne , 68588 - 0111 _ + d. besson , a. ramos , j. remmers + _ university of kansas , dept .
of physics and astronomy , lawrence , ks , 66045 - 7582 _ +
|
most of the experimentally observed hadrons can be interpreted as @xmath2/@xmath3 states and accommodated in the quark model @xcite . up to
now there has accumulated some evidence of the exotic state with @xmath4 @xcite .
such a quantum number is not accessible for a pair of quark and anti - quark .
it is sometimes labelled as an exotic hybrid meson with the particle contents @xmath5 .
recently we have investigated the @xmath6 state using the tetraquark currents @xcite .
the extracted mass and characteristic decay pattern are quite similar to those expected for the exotic hybrid meson .
such a result is expected .
since the gluon field creates a pair of @xmath7 easily , the hybrid operator @xmath8 transforms into a tetraquark interpolating operator with the same exotic quantum number . in quantum field theory different operators with the same quantum number mix and tend to couple to the same physical state . using the same tetraquark formalism developed in the study of the low - lying scalar mesons @xcite and the exotic @xmath6 mesons @xcite
, we study the possible @xmath9 states composed of light quarks . for a neutral quark model state @xmath7 , we know that @xmath10 ensures @xmath11 hence @xmath12 . in other words , states with @xmath13 are strictly forbidden .
on the other hand , the gauge invariant scalar and pseudoscalar operators composed of a pair of the gluon field are @xmath14 and @xmath15 , both of which carry the even c - parity .
we construct all the local tetraquark currents with @xmath16 .
there are two kinds of constructions : @xmath17 and @xmath18 .
they can be related to each other by using the fierz transformation . as usual
, we use the first set @xcite .
their flavor structure can be @xmath19 , @xmath20 , and @xmath21 ( @xmath17 ) . with all these independent currents
, we perform the qcd sum rule analysis . as a byproduct
, we notice that there does not exist any tetraquark interpolating operator without derivative for the @xmath22 case .
this paper is organized as follows . in sec .
[ sec : current ] , we construct the tetraquark currents with @xmath23 using the diquark ( @xmath24 ) and antidiquark ( @xmath25 ) fields .
the tetraquark currents constructed with the quark - antiquark ( @xmath26 ) pairs are shown in appendix.[app ] .
we present the spectral density in sec .
[ sec : ope ] and perform the numerical analysis in sec . [
sec : num ] . for comparison , we present the finite energy sum rule analysis in the appendix.[sec : fesr ] . the last section is a short summary .
in this section , we construct the tetraquark interpolating currents with @xmath27 using diquark and antidiquark fields .
such a quantum number can not be accessed by a @xmath7 pair .
the currents can be similarly constructed by using the quark - antiquark pairs . however , as shown in appendix.[app ] , these two constructions are equivalent as we have shown several times in our previous studies @xcite .
the pseudoscalar tetraquark currents can be constructed using five independent diquark fields , which are constructed by five independent @xmath28-matrices @xmath29 where @xmath30 represents the @xmath31 , @xmath32 and @xmath33 quarks , and @xmath34 are the color indices . to compose a color singlet pseudoscalar tetraquark current
, the diquark and antidiquark should have the same color and spin symmetries .
so the color structure of the tetraquark is either @xmath35 or @xmath36 , which is denoted by labels @xmath37 and @xmath38 respectively .
therefore , considering both the color and lorentz structures , there are altogether ten terms of products @xmath39 we list them as follows @xmath40 in the above expressions , the flavor structure is fixed at the same time due to the pauli principle .
the currents @xmath41 , @xmath42 , @xmath43 belong to the symmetric flavor representation @xmath44 where both diquark and antidiquark fields have the symmetric flavor structure .
the currents @xmath45 , @xmath46 , @xmath47 belong to the antisymmetric flavor representation @xmath48 , where both diquark and antidiquark fields have the antisymmetric flavor structure .
@xmath49 , @xmath50 for @xmath51 and @xmath52 , @xmath53 for @xmath54 , where @xmath55 represents the mixed flavor symmetry .
the isovector states with charges can be observed in the experiments more easily , therefore in this paper we concentrate on the isovector currents which was shown in the @xmath56 tetraquark weight diagram in fig 1 @xcite .
we have : @xmath57 we do not differentiate @xmath31 and @xmath32 quarks and denote them by @xmath58 .
we are only interested in those neutral components .
the other states do not carry definite c - parity .
it turns out that the neutral isovector and isoscalar states have the same qcd sum rules .
our following discussions are valid for both of them . taking the charge - conjugation transformation
, we get @xmath59 @xmath47 and @xmath43 have even charge - conjugation parity .
we conclude that the currents with @xmath27 are : @xmath60 adding different quauk contents as shown in eq .
( [ content ] ) , there are altogether seven independent currents as shown : 1 . for @xmath61 : @xmath62 where @xmath63 belongs to the @xmath64 representation and contains up and down quarks only while @xmath65 belongs to the @xmath66 representation and contains one @xmath67 quark pair .
+ 2 . for @xmath68
: @xmath69 where @xmath70 and @xmath71 belong to the @xmath72 representation and contain only u , d quarks while @xmath73 and @xmath74 belong to the @xmath66 representation and contain one @xmath67 quark pair .
+ 3 . for @xmath75 : @xmath76 where @xmath77 belongs to the @xmath66 and contains one @xmath67 quark pair .
it is understood that there exists the other part @xmath78 $ ] in the expressions of @xmath79 .
now we move on to the @xmath80 case .
there are also ten independent scalar tetraquark currents without derivative : @xmath81 the flavor structure is again fixed due to the pauli principle . to have a charge - conjugation parity
, we fix the quark contents to be : @xmath82 and @xmath83 ( or @xmath84 and @xmath85 ) . after performing the charge - conjugation transformation , we find that they all have an even charge - conjugation parity , for example : @xmath86 therefore , the @xmath80 tetraquark interpolating currents without derivatives do not exist .
we consider the two - point correlation function in the framework of qcd sum rule @xcite : @xmath87 where @xmath88 is an interpolating current
. we can calculate @xmath89 at the quark gluon level using the propagator : @xmath90|0\rangle \\ \nonumber & = & \frac{i\delta^{ab}}{2\pi^2x^4}\hat{x } + \frac{i}{32\pi^2}\frac{\lambda^n_{ab}}{2}gg_{\mu\nu}^n\frac{1}{x^2}(\sigma^{\mu\nu}\hat{x}+\hat{x}\sigma^{\mu\nu } ) -\frac{\delta^{ab}}{12}\langle\bar{q}q\rangle \\ & & + \frac{\delta^{ab}x^2}{192}\langle g_s \bar{q}\sigma gq\rangle -\frac{m_q\delta^{ab}}{4\pi^2x^2}+\frac{i\delta^{ab}m_q\langle\bar{q}q\rangle}{48}\hat{x } + \frac{i\delta^{ab}m_q^2\langle\bar{q}q\rangle}{8\pi^2x^2}\hat{x},\end{aligned}\ ] ] where @xmath91 . with the dispersion relation @xmath89
is related to the observable at the hadron level @xmath92 where @xmath93 here , the usual pole plus continuum parametrization of the hadronic spectral density is adopted . up to dimension 12
, the spectral density @xmath94 at the quark and gluon level reads : @xmath95s \\ & & - ( \frac{3 m_q^2 \langle\bar{q}q\rangle^2}{2\pi^2}+\frac{\langle g_s^2gg\rangle m_q \langle\bar{q}q\rangle}{192\pi^4})+(\frac{16}{9}m_q \langle\bar{q}q\rangle^3-\frac{1}{\pi^2}m_q^2 \langle\bar{q}q\rangle \langle g_s \bar{q}\sigma gq\rangle)\delta(s ) \ , , \end{aligned}\ ] ] @xmath96s \\ & & + ( \frac{m_s^2 \langle\bar{s}s\rangle^2}{12\pi^2 } - \frac{m_s^2 \langle\bar{q}q\rangle^2}{3\pi^2}-\frac{m_s \langle\bar{s}s\rangle \langle g_s^2gg\rangle}{384\pi^4})-(\frac{m_s^2 \langle\bar{u}u\rangle \langle g_s \bar{q}\sigma gq\rangle}{6\pi^2}-\frac{8}{9}m_s \langle\bar{s}s\rangle \langle\bar{q}q\rangle^2)\delta(s ) \ , , \end{aligned}\ ] ] @xmath97s \\ & & - ( \frac{6 m_q^2 \langle\bar{q}q\rangle^2}{\pi^2}-\frac{5 \langle g_s^2gg\rangle m_q \langle\bar{q}q\rangle}{96\pi^4 } ) + ( \frac{64}{9}m_q \langle\bar{q}q\rangle^3-\frac{4}{\pi^2}m_q^2 \langle\bar{q}q\rangle \langle g_s \bar{q}\sigma gq\rangle)\delta(s ) \
, , \end{aligned}\ ] ] @xmath98s \\ & & + ( \frac{m_s^2 \langle\bar{s}s\rangle^2}{3\pi^2 } -\frac{4m_s^2 \langle\bar{q}q\rangle^2}{3\pi^2}+ \frac{5m_s \langle\bar{s}s\rangle \langle g_s^2gg\rangle}{192\pi^4})-(\frac{2m_s^2 \langle\bar{u}u\rangle \langle g_s \bar{q}\sigma gq\rangle}{3\pi^2}-\frac{32}{9}m_s \langle\bar{s}s\rangle \langle\bar{q}q\rangle^2)\delta(s ) \ , , \end{aligned}\ ] ] @xmath99s \\ & & - ( \frac{3 m_q^2 \langle\bar{q}q\rangle^2}{\pi^2}-\frac { \langle g_s^2gg\rangle m_q \langle\bar{q}q\rangle}{96\pi^4 } ) + ( \frac{32}{9}m_q \langle\bar{q}q\rangle^3-\frac{2}{\pi^2}m_q^2 \langle\bar{q}q\rangle \langle g_s \bar{q}\sigma gq\rangle)\delta(s ) \ , , \end{aligned}\ ] ] @xmath100s \\ & & + ( \frac{m_s^2 \langle\bar{s}s\rangle^2}{6\pi^2 } - \frac{2m_s^2 \langle\bar{q}q\rangle^2}{3\pi^2}+\frac{m_s \langle\bar{s}s\rangle \langle g_s^2gg\rangle}{192\pi^4 } ) -(\frac{m_s^2 \langle\bar{u}u\rangle \langle g_s \bar{q}\sigma gq\rangle}{3\pi^2}-\frac{16}{9}m_s \langle\bar{s}s\rangle \langle\bar{q}q\rangle^2)\delta(s ) \ , , \label{pi6}\end{aligned}\ ] ] @xmath101s \\ & & + ( \frac{m_s^2 \langle\bar{s}s\rangle^2}{24\pi^2}- \frac{m_s^2 \langle\bar{q}q\rangle^2}{6\pi^2}+ \frac{m_s \langle\bar{s}s\rangle \langle g_s^2gg\rangle}{384\pi^4})-(\frac{m_s^2 \langle\bar{u}u\rangle \langle g_s \bar{q}\sigma gq\rangle}{12\pi^2}-\frac{4}{9}m_s \langle\bar{s}s\rangle \langle\bar{q}q\rangle^2)\delta(s ) \ , .\end{aligned}\ ] ] it is interesting to note several important features of the above spectral densities : * first the special lorentz structure of the @xmath27 interpolating currents forbids the appearance of the four - quark type of condensates @xmath102 , @xmath103 @xmath104 and @xmath105 .
usually these terms play an important role in the multiquark sum rules .
the feynman diagrams for the dimension 10 condensate @xmath106 are shown in fig .
[ fig : mixed - condensate ] . + [ cols= " < , < , > " , ]
|
in order to explore the possible existence of the exotic @xmath0 state , we have constructed the tetraquark interpolating operators systematically . as a byproduct
, we notice the @xmath1 tetraquark operators without derivatives do not exist .
the special lorentz structure of the @xmath0 currents forbids the four - quark type of corrections to the spectral density .
now the gluon condensates are the dominant power corrections .
none of the seven interpolating currents supports a resonant signal .
therefore we conclude that the exotic @xmath0 state does not exist below 2 gev , which is consistent with the current experimental observations .
|
the electromagnetic production of kaons provides an important source of information on hadronic physics with strangeness . due to the rather weak coupling at the electromagnetic vertex ,
the reaction operator is quite simple and , therefore , the use a of single channel analysis may be quite sufficient to describe the process . this process can be used to simultaneously study kaon - hyperon - nucleon coupling constants , isospin symmetry , hadronic form factors , baryon and meson resonances , and contributions of kaon - hyperon final states to the magnetic moment of the nucleon . in this report
we will focus only on the two last topics .
for a detailed discussion of the elementary operator we refer to ref.@xcite .
the background part of the operator consists of the standard born terms along with the @xmath3(892 ) and @xmath4(1270 ) vector meson poles in the @xmath5-channel .
the resonance part of the @xmath6 operator includes the @xmath7(1650 ) , @xmath8(1710 ) , and @xmath1 resonances .
we also include hadronic form factors in hadronic vertices by employing the gauge method of haberzettl @xcite which leads to an excellent agreement between experimental data and model calculations .
a brief inspection of the particle data table reveals that less than 40% of the nucleon resonances predicted by constituent quark models are observed in @xmath9 scattering experiments .
quark model studies have suggested that those `` missing '' resonances may couple to other channels , such as the @xmath10 and @xmath11 channels .
stimulated by the new @xmath12 saphir total cross section data @xcite , which suggest a structure around @xmath13 mev , we start the investigation by using our isobar model . as shown in fig.[fig : missing ] , our previous model can not reproduce the total cross section . the constituent quark model of capstick and roberts
@xcite predicts many new states around 1900 mev .
however , only a few of them have been calculated to have a significant @xmath10 decay width @xcite .
we have performed fits for each of the possible states , allowing the fit to determine the mass , width and coupling constants of the resonance .
we found that only in the case of the @xmath14(1895 ) state couplings are obtained that are in remarkable agreement @xcite with recent quark model predictions .
the result is shown in fig.[fig : missing ] , where without this resonance the model shows only one peak near threshold , while inclusion of the new resonance leads to a second peak at @xmath15 slightly below 1900 mev , in accordance with the new saphir data . in the case of @xmath16 photoproduction
we find that a better agreement with experimental data can be achieved by including the @xmath1 resonance . as shown in fig.[fig : p13role ] the inclusion of this state does not influence the @xmath17 channel appreciably , in contrast to the other three @xmath18 channels , where the prominent effect is found in the @xmath16 channel .
this is due both to the isospin coefficient which enhances the @xmath19 decay over the @xmath20 coupling and the very different structure of the born terms in @xmath21 photoproduction .
the extracted fractional decay width is found to be consistent with the prediction of the quark model , whose magnitude is almost one order smaller than the value given by particle data group @xcite .
the gerasimov - drell - hearn ( gdh ) sum rule @xcite relates the anomalous magnetic moment of the nucleon @xmath22 to its excitation spectrum in the resonance region by an integral .
previous analyses on pion photoproduction multipole amplitudes show a discrepancy between theoretical prediction and indirect experimental investigations of the gdh sum rule @xcite . here
we calculate the contribution of kaon photoproduction processes to the gdh sum rule by using our isobar model . in our previous calculation @xcite
, we have fixed the upper limit of integration at 2.2 gev .
exact formulation was obtained by integrating the structure function @xmath23 , and an approximation of the upper bound of integral is achieved by calculating total cross section @xmath24 .
we obtained a value of @xmath25 and @xmath26 .
note that although this is relatively small , the result is consistent with the karliner s work @xcite . in order to improve our gdh calculation
we increase the upper limit of integration to higher energies . for the present
, we use the regge model given in ref.@xcite to calculate the cross sections in the energy region between 5 and 15 gev .
the result is shown in fig.[fig : glv ] , where it is clear that contributions from higher energies are very small .
nevertheless , the important message here is the convergence of @xmath23 which indicates the convergence of the integral in the case of kaon . in the future
, we will improve our model by reggeizing the @xmath27channel intermediate state .
some preliminary results have been reported in ref .
. 9 f. x. lee , t. mart , c. bennhold , and l. e. wright , nucl - th/9907119 ( 1999 ) ; + t. mart , c. bennhold , and c. e. hyde - wright , phys . rev .
c 51 ( 1995 ) r1074 .
h. haberzettl , c. bennhold , t. mart , and t. feuster , phys .
c 58 ( 1998 ) r40 .
s. capstick and w. roberts , phys .
d 58 ( 1998 ) 074011 .
t. mart and c. bennhold , phys .
c 61 ( 2000 ) 012201(r ) .
m. q. tran _
et al_. , phys .
b 445 ( 1998 ) 20 . s. goers _ et al_. , phys .
b 464 ( 1999 ) 331 .
abbhhm collaboration , phys .
188 ( 1969 ) 2060 .
t. mart , phys .
c 62 ( 2000 ) 038201 .
gerasimov , yad .
fiz . 2 ( 1965 ) 598 ; sov .
phys . 2 ( 1966 ) 430
drell and a.c .
hearn , phys .
( 1966 ) 908 .
i. karliner , phys .
d 7 ( 1973 ) 2717 .
s. sumowidagdo and t. mart , phys .
c 60 ( 1999 ) 028201 .
m. guidal , j .-
m . laget , and m. vanderhaeghen , nucl .
a627 ( 1997 ) 645 .
t. mart , s. sumowidagdo , c. bennhold , and h. haberzettl , nucl - th/0002036 ( 2000 ) .
|
some applications of the elementary kaon photoproduction process are discussed : the investigation of missing resonances in the @xmath0 channel , the role of the @xmath1 state in the @xmath2 channel , and the calculation of the gerasimov - drell - hearn sum rule . for the latter , we present an extension of our previous study to higher energies .
|
transport properties of electron wave packets underlie the functioning of electronic devices and are an important factor in photoemission spectroscopies and electron diffraction techniques . in a solid , as opposed to a single atom , special role is played by inelastic scattering giving rise to a limited electron lifetime and mean free path .
the inelastic scattering is phenomenologically described as electron absorption @xcite , and its implications for stationary processes are well understood @xcite .
a much less studied question is how the wave packet evolves and propagates on a time scale comparable to its lifetime .
experimental access to such ultrafast processes is given by modern time - resolved photoelectron spectroscopies , in particular by the streaking method @xcite , in which a subfemtosecond time resolution is achieved by mapping time to energy using a strong laser field : the electron wave packet created by an ultrashort pulse of extreme ultraviolet radiation ( xuv ) is accelerated by the superimposed laser field , and the energy by which its spectrum shifts up or down depends on the electron release time @xmath1 relative to the temporal profile of the laser field @xmath2 .
various models have been developed that treated the problem by the time - dependent schrdinger equation ( tdse ) @xcite or by applying a classical model @xcite .
the majority of the studies consider a solid in which the low - frequency laser field is strongly damped by the dielectric response @xcite , so the electron needs to travel some distance before it gets exposed to the streaking field .
dependence of optical potential @xmath3 on energy .
small circles are the values empirically derived from _ ab initio _ analysis of arpes spectra ( full circles ) and vleed spectra ( open circles ) .
large circles are the _ ab initio _ @xmath4 calculation for w(100 ) ( fleur - spex ) . the dependence @xmath5 is well fitted by a function @xmath6 with @xmath7 ev and @xmath8 ev for w(100 ) , @xmath9 ev for al(111 ) , and @xmath10 ev for tite@xmath11 .
, scaledwidth=48.0% ]
assuming an instantaneous xuv excitation , the time of flight from the depth where the photoelectron is created to the surface is determined by the group velocity of the wave packet . because of the inelastic scattering the photoelectrons excited deeper in the crystal have smaller probability to reach the detector , which is assumed to scale with the depth @xmath12 as @xmath13 .
then the average depth traversed by the electrons is just the mean free path ( mfp ) @xmath14 , and the time to reach the surface can be estimated as @xmath15 , where @xmath16 is the group velocity of the emitted wave packet @xcite .
often it is sufficient to use the values of @xmath14 averaged over an energy interval of several evs that can be inferred from the well - known universal @xmath17 curve .
this is , however , not sufficient for arpes because mfp may strongly vary over a few ev interval as a result of the specific band structure . at the same time , experience with arpes and vleed ( very low energy electron diffraction )
@xcite suggests that the photoelectron final - state lifetime changes rather smoothly with energy , see fig .
[ univi](a ) .
thus , it is reasonable to rely on the average values for the lifetime , and to derive the rapidly varying mfp as @xmath18 , where @xmath3 is the inverse lifetime ( optical potential ) , which enters the hamiltonian through the imaginary term @xmath19 .
for a nearly free electron , the wave vector @xmath20 acquires an imaginary part @xmath21 , which yields @xmath22 , and the time to reach the surface is then @xmath23 note that the group velocity cancels , so @xmath24 depends on energy through the function @xmath5 .
vast empirical data suggest that the @xmath5 curves are similar for different materials : @xmath3 grows with energy with similar rate , see fig . [
univi](a ) . because the empirically derived data may be affected by various extrinsic factors and by the imperfection of the apparatus it is important to theoretically corroborate this observation .
microscopically , optical potential is associated with the imaginary part of the self - energy .
figure [ univi](a ) shows an _ ab initio _
@xmath4 calculation for a high - energy conducting band of w(110 ) .
it is seen to agree well with the empirical results , which supports our understanding of the electron absorption as coming predominantly from the electron - electron interaction and corroborates the empirical estimate of its growth with energy .
thus , within this simplest approach the arrival delay of photoelectrons with final energy @xmath25 relative to those with energy @xmath26 behaves roughly as @xmath27 , see fig .
[ univi](b ) : the electron with a lower energy comes later , and the delay @xmath28 decreases as the final state energy grows .
the recent measurements by neppl @xcite on w(110 ) of the phase shift between the spectrograms of valence 5@xmath29 and semi - core 4@xmath30 states qualitatively follow this trend [ open circles in fig . [ univi](b ) ] , but the absolute values in the experiment are about three times larger . at the same time ,
the value of @xmath31 as measured on mg(0001 ) at @xmath32 ev is 4 times smaller than expected from the simple mfp - based theory .
such drastic disagreement in absolute values calls for the search of mechanisms that , in addition to the finite lifetime , affect the transport of the photoelectron to the surface .
a number of fundamental questions arise : what is the role of the finite duration of the pump pulse , i.e. , at what point does the wave packet start its motion with the group velocity ? does the electron have the group velocity dictated by the band structure if it travels only the distance of a few atomic layers ?
how does the packet propagate when its energy falls into a forbidden gap , where the group velocity can not be defined as @xmath0 ? can the electron excited at the outermost atomic plane be thought of as not feeling the band structure because it moves away from the surface , towards vacuum ?
apart from that , an important theoretical question arises of whether the absorbing potential @xmath19 can be used in the time - dependent schrdinger equation to adequately describe both the photocurrent attenuation and the implications for the transport : inelastic scattering is a stochastic process , so the photocurrent is an incoherent overlap of randomly dephased contributions , whereas the optical - potential - damped wave packet remains fully coherent , which may have implications both for its transport and for its acceleration by the laser field .
to resolve these questions , we need a model that does not incorporate any assumptions about the answers and , at the same time , can be treated numerically exactly . for the present proof - of - concept calculation we use a one - dimensional model of a crystal , which treats photoexcitation , transport , and streaking fully quantum - mechanically without resorting to a perturbational treatment of any of the terms in the hamiltonian or to a heuristic separation of the xuv and the laser pulses .
we employ a particle - in - the - box method : the box comprises a thick crystal slab lying on a structureless substrate [ the piecewise constant potential @xmath33 in fig .
[ procedure](a ) ] and the vacuum half - space . in ref .
@xcite this model was applied to photoelectron streaking by a spatially uniform laser field in the absence of inelastic scattering .
the present model allows for a rapid decay of the laser field into the solid and includes the inelastic scattering in two alternative ways : via the optical potential and by a straightforward ensemble averaging over random configurations .
we perform a series of numerical experiments , in which the system is excited by an xuv pulse of duration @xmath34 fs and is simultaneously acted upon by the laser pulse of duration @xmath35 fs , fig .
[ procedure](b ) . both light fields are given by the dipole operator @xmath12 , and the laser field
is screened by multiplying it by a smooth step function @xmath36 .
the temporal envelopes of both pulses are of the form @xmath37 .
the laser pulse is @xmath38 $ ] , with photon energy @xmath39 ev and amplitude @xmath40 v / cm .
the tdse is solved with the split - operator technique in matrix form in terms of exact eigenfunctions ( discrete spectrum ) of the unperturbed hamiltonian @xmath41 , so the crystal potential is fully taken into account for both initial and final states .
the spectrum of @xmath42 is truncated at 200 ev above the vacuum level , which ensures the convergence of all the results . apart from @xmath43 and the two external fields the hamiltonian includes inelastic scattering , which is either a static absorbing potential @xmath44 $ ] or a stochastic perturbation @xmath45 $ ] , where @xmath46 and @xmath47 are random functions of @xmath48 and @xmath12 , respectively . here
@xmath46 is fixed and @xmath49 is the @xmath50th sample in a random ensemble .
the ultimate spectrum is then the average over @xmath51 random walks , fig .
[ procedure](c ) .
the peak displacement @xmath52 from its laser - free position as a function of the time shift @xmath53 between the xuv and the laser pulse gives the key to the temporal information : by fitting the measured @xmath54 points with the momentum transfer function @xmath55 , we infer the arrival time of the electron in vacuum @xmath24 , see fig .
[ procedure](e ) for the motion in the random potential and fig .
[ procedure](f ) for the motion in the absorbing potential .
for a detailed study of the photoelectron transport to the surface it is convenient to exactly know its initial position in space .
this is achieved by introducing a small defect at one of the layers and considering the photoemission from the localized state at the defect . figures [ procedure](e ) and [ procedure](f ) show streaking curves for the initial state localized at the third layer for @xmath56 and 90 ev .
note that the results by the phenomenological absorbing potential ( f ) and by the microscopic random collisions ( e ) perfectly agree .
counterintuitively , the electron at the higher energy arrives 40 as later than at the lower energy : @xmath57 as at 90 ev and @xmath58 as at 78 ev .
figure [ procedure](g ) shows the arrival time of the electron initially at the 2nd , 3rd , and 4th layer for photon energies from 65 to 159 ev .
the most striking are the two minima at @xmath59 and at 143 ev ( @xmath60 and at 102 ev ) , at which the arrival time even shows negative values .
the minima @xmath61 and @xmath62 are located at the lower edges of the band gaps , and minimum @xmath63 occurs at the energy where the ionization cross - section vanishes , see fig .
[ procedure](d ) . there
the @xmath64 curve strongly deviates from that derived from the group velocity , @xmath65 [ dashed curve in fig .
[ procedure](g ) ] . to prove that the discrepancy is not due to the rather indirect streaking method to measure @xmath24
, we employ an alternative method : we switch off the laser and follow the propagation of the wave packet up to a distance of 500 a.u . away from the surface . by measuring the center of gravity of the packet in a number of time points
we determine its equation of motion in vacuum @xmath66 and obtain the time point @xmath67 at which it has arrived at the surface .
the @xmath68 and @xmath64 curves are seen to agree well , fig .
[ procedure](h ) , which means that the clock provided by the streaking technique is accurate , and that the fast delivery of the photoelectron to the surface is effected during the excitation process .
figure [ evolution ] shows the time evolution of the spectrum at the @xmath64 minimum , @xmath59 ev , and at @xmath69 ev , where @xmath64 well agrees with the instantaneous approximation @xmath70 .
the spectral evolution maps in the two cases are qualitatively different : while at 100 ev the spectrum rapidly concentrates around @xmath71 , at 80 ev the evolution is much slower : at @xmath72 as the intensity is still spread over a 40 ev wide spectral interval .
clearly , while the spectral coefficients @xmath73 of the packet @xmath74 keep changing , the velocity of its gravity center may deviate from the weighted group velocity of the stationary waves @xmath75 , and during the xuv pulse the wave packet may travel a distance comparable with the mean free path .
such unusual behavior happens every time when the central energy of the wave packet approaches an intensity minimum , be it a vanishing matrix element ( cooper minimum ) or a spectral gap . for the emission from the first layer , at its cooper minimum ( @xmath76 ev ) the xuv pulse , on the contrary , detains the packet in the solid causing a delay of @xmath77 as .
all the other features of the @xmath64 curve are similar to those of the emission from the deeper layers .
figure [ procedure](h ) shows that the inelastic scattering plays an important role in the formation of the spatial shape and transport of the packet : the displacement effect is more pronounced at larger @xmath3 . in this phenomenological approach the dephasing of the wave packet
is neglected , and the question arises if this may bring about any artifacts .
this simplification turns out to be not crucial : figure [ procedure](g ) shows the @xmath64 data by statistical averaging over 144 configurations ( large circles ) for emission from the 2nd and 4th layer and fig .
[ procedure](i ) for the 1st layer .
the parameters of the random perturbation are chosen such that it gives the same mfp as @xmath78 ev . the dephasing is seen to cause a broadening of the @xmath64 features , but otherwise the behavior of the incoherent ensemble is in perfect accord with that of the coherent packet moving in the absorbing medium . that such two very different systems agree
to within fine details suggests that this advancement phenomenon is robust , and it should be expected to occur in real solids .
the present calculation has revealed a peculiar dynamics of photoelectrons excited by an ultra - short pulse .
the interesting features arise from the interplay between the elastic scattering from the crystal lattice and inelastic scattering that causes the damping of the electron wave .
three aspects can be distinguished : first , the effect of inelastic scattering is to increase the average velocity of the wave packet , which is natural because it is the slower components that are damped stronger , as they spend more time in the solid .
the range of the advancement effect depends on the interaction with the crystal lattice : at the special points , and around the gaps it may exceed 200 as , whereas in the free - electron - like region it reduces to 20 as , which is still comparable with the measured relative delays , see fig .
[ univi](b ) .
this effect can be understood in terms of the group velocity of bloch electrons . at certain photon energies , however , owing to a complicated non - free - electron - like structure of the final states , the wave packet performs a complicated motion , and it may approach the surface much faster than an instantaneously created packet . this may also lead to an interesting effect that the electron starting from the outermost layer is overtaken by the electron coming from the depth of the crystal . close to the spectral gaps the temporal shift of the spectrogram does not necessarily agree with the equation of motion of the center of gravity of the wave packet , but apart from the special points discussed above the discrepancy of the two methods does not exceed 20 as .
|
laser - assisted photoemission from a solid is considered within a numerically exactly solvable one - dimensional model of a crystal .
the effect of the inelastic scattering and of the finite duration of the pump pulse on the photoelectron dynamics is elucidated .
the phenomenological result that the photoexcited wave packet moves with the group velocity @xmath0 and traverses on average the distance equal to the mean free path is found to hold for energies far from the spectral gaps of the final state band structure . on the contrary , close to a spectral gap the photoelectron
is found to move during the excitation by the pump pulse with a velocity higher than the group velocity .
|
the dynamics of language has recently been identified as a rich field for interdisciplinary application of statistical techniques traditionally associated with mathematics and physics .
a host of analytical and numerical models have been proposed , aimed at reproducing more or less quantitatively several aspects of language as a dynamical sociocultural entity.@xcite among them , models of language evolution focus on the joint processes of mutation of linguistic features and of language acquisition , switching , and adoption by human populations.@xcite bit - string models of language evolution , prototyped by schulze s model,@xcite conceive that a language is completely characterized by a sequence of dichotomic properties .
each of them represents , for instance , whether the language in question possesses certain grammatical property or not .
the sequence is naturally represented as a string of binary variables ( bits ) , whose length is the number of yes / no questions which fully identify a language .
emphasis in the study of bit - string language models has been put on agent - based numerical simulations,@xcite addressing possible explanations for the distribution of population among languages and the abundance of language families .
the language spoken by an individual can mutate at a certain rate , and each individual can also abandon his / her language and adopt a different one copied from a randomly chosen member of the population . upon variation of the mutation rate ,
simulations show a sharp transition between a state of dominance , where most of the population speaks the same language , and a situation where the population is homogeneously distributed among languages.@xcite in this paper , i introduce a formulation of bit - string language models based on differential equations for the population fraction speaking each language .
the evolution turns out to be a combination of replicator dynamics , representing language switching , and diffusion in the bit - string ( hypercubic ) space , representing mutation .
while the stability of the homogeneous state can be exactly analyzed , the study of the dominance state requires a mean - field - like approximation .
the results , nevertheless , are in very good agreement with numerical resolution of the differential equations and with agent - based simulations .
they predict a region of bistability , where dominance and homogeneity coexist as possible asymptotic states of the system , and the disappearance of the dominance state through a tangent bifurcation .
the present formulation provides an alternative to simulations to trace language progress and regression in the framework of bit - string models .
as advanced in the introduction , in bit - string models an individual language is represented by a string of @xmath0 binary variables ( bits ) , each of them adopting one of two possible values , say , @xmath1 and @xmath2 .
the total number of possible languages is @xmath3 .
evolution is driven by two dynamical rules .
first , the language of an individual can mutate with a certain probability per time unit .
each mutation event consists of a change in a single , randomly selected bit of the individual s language , from @xmath1 to @xmath2 or _
vice versa_. second , an individual can give up his / her language and adopt a new one . to account for the preference for more widespread languages , the new one is chosen by selecting an individual at random from the whole population and adopting his / her language . in this way
, the probability of switching to a given language ( say , to language @xmath4 ) equals the fraction of the population speaking @xmath4 , denoted by @xmath5 .
moreover , the probability of abandoning the original language ( say , language @xmath6 ) is weighted by a monotonically decreasing function @xmath7 of the fraction of the population speaking @xmath6 .
a language is thus less likely to be given up if it is spoken by a large fraction of the total population .
previous numerical analyses of bit - string models have considered the cases @xmath8 and @xmath9 , both of which satisfy @xmath10 and @xmath11.@xcite bit - string models are added with population dynamics , for instance , asexual reproduction with births at a given rate , and deaths with a mortality rate proportional to the population size .
this insures that after a transient , and up to small fluctuations the total population remains constant in time .
a newborn inherits the language of the parent , except possibly for mutation or switching to a new language , following essentially the same rules as described above .
the effect of births on language evolution , thus , is effectively the same as mutation and switching by `` adult '' individuals . in its original formulation,@xcite schulze s model
considers that the features of a given ( `` superior '' ) language may be preferred to those of any other language . during mutation events , with a certain probability , a bit copies the value of the homologous bit in the `` superior '' language , enhancing the possibility that this language is eventually spoken by most of the population .
here , i disregard this mechanism and assume that all languages are intrinsically equivalent . the dynamical rules of bit - string language models can be readily expressed as differential equations for the fractions @xmath12 ( @xmath13 ) of the population speaking each language , which satisfy the normalization condition @xmath14 at all times .
for the sake of simplicity , i will assume that the total population remains exactly constant in time , incorporating newborn language s mutation and switching to the respective overall mechanisms .
the contribution of mutation to the time variation of @xmath5 consists of a gain term proportional to @xmath15 , where @xmath16 if languages @xmath4 and @xmath6 differ by just one bit , and @xmath17 otherwise .
the sum runs over all the languages .
the gain term represents the fraction of the population speaking any language @xmath6 which mutates to language @xmath4 in a time unit .
the factor @xmath18 accounts for the fact that a mutation event can lead , with identical probability , from each language @xmath6 to @xmath0 different languages .
correspondingly , there is a loss term proportional to @xmath19 , which represents mutations of language @xmath4 to other languages . in turn
, the gain term associated to the mechanism of language switching is proportional to @xmath20 , because the probability of abandoning @xmath6 is @xmath7 , and the probability of adopting @xmath4 is @xmath5 .
the corresponding loss term is proportional to @xmath21 . if @xmath22 and @xmath23 are , respectively , the rates of mutation and switching events , the evolution of the fraction @xmath5 ( @xmath24 ) is given by @xmath25 summing these equations over the index @xmath4 yields @xmath26 , which insures that eq .
( [ norm ] ) holds if the initial fraction satisfy normalization . the functions @xmath27 make the equations nonlinear , so that little can be expected from trying to solve them exactly .
it is worthwhile mentioning that a joint dynamical description of mutation and population growth here corresponding to switching to widespread languages has been proposed for the process of language learning and the evolution of universal grammar.@xcite several exact results are known for the relevant equations,@xcite though they depend crucially on the fact that the functions which play a role analogous to that of @xmath28 are linear on the population fractions . also , it is important to realize that the representation of languages as bit strings is essentially irrelevant to the form of the evolution equation ( [ fund ] ) .
as long as each language is interpreted as a kind of _ state _ to which a certain fraction of the population is assigned at each time , the specific way in which such states are individualized does not play a role in the equation . at most ,
it is necessary that those states are suitably labeled , in such a way as to discern the languages between which population can be transferred due to mutations , i.e. as to fix de coefficients @xmath29 .
otherwise , the mathematical form of the equations will be the same irrespectively of the individual characterization of languages .
though , for clarity , i discuss eq .
( [ fund ] ) with reference to bit - string models , it is important to bear in mind that most results will also hold for other models driven by the mechanisms of mutation and switching .
the contributions of mutation and switching to the evolution of @xmath5 have opposite , competing effects .
while mutation spreads individuals over different languages , switching tends to concentrate the population on languages with an already large number of speakers .
it is useful to begin with a separate analysis of the two mechanisms , which are closely related to well - understood processes in other areas of science .
then , one can proceed to show that the combination of the two competing mechanisms gives rise to a critical transition between language diversity and dominance , as effectively observed to occur in numerical simulations of bit - string models.@xcite
in order to isolate from each other the effects of mutation and switching , let us first disregard the latter , setting @xmath30 in eq .
( [ fund ] ) . taking into account that @xmath31 , the dynamics without switching is governed by the equations @xmath32 @xmath33 .
this expression emphasizes the fact that mutation is a form of linear diffusion , with population transfer between languages at a rate proportional to the population difference @xmath34 .
the diffusion process takes place on the set of bit strings of length @xmath0 , which can be assimilated to the vertices of the @xmath0-dimensional hypercube of unitary side .
mutation events transfer population between languages which differ in a single bit , i.e. between hypercube vertices at hamming distance @xmath35 .
each vertex has @xmath0 neighbour vertices at @xmath35 .
since @xmath16 if the hamming distance between @xmath4 and @xmath6 equals one , and @xmath17 otherwise , one has @xmath36 .
therefore , introducing the hypercube adjacency matrix @xmath37 , eq .
( [ dif1 ] ) can be rewritten as @xmath38 with @xmath39 and @xmath40 , where @xmath41 is the identity matrix .
the solution to the linear equation ( [ dif2 ] ) can be found by standard methods , in terms of the eigenvalues @xmath42 and eigenvectors @xmath43 ( @xmath44 ) of the matrix @xmath45 .
one of the eigenvalues of @xmath45 , say @xmath46 , equals zero , and all the remaining eigenvalues are negative .
consequently , the eigenvector corresponding to @xmath46 , @xmath47 , yields the equilibrium solution which , due to the normalization condition ( [ norm ] ) , is @xmath48 for all @xmath4 . in other words , starting from any initial condition , the system asymptotically reaches a homogeneous state . in the context of the original problem , in the absence of language switching
, mutation leads for long times to a state where the population is homogeneously distributed among all possible languages . if , now , @xmath49 so that mutation is absent , eq .
( [ fund ] ) becomes a member of a well - studied class of nonlinear equations , generally written as @xmath50 @xmath33 , with @xmath51 .
equations ( [ replic ] ) define the so - called replicator dynamics.@xcite they describe the evolution of a system of interacting species in terms of the fractions @xmath52 , where @xmath53 is the number of individuals of species @xmath4 and @xmath54 is the total population in the system . the function @xmath55 is the reproductive rate , or ( fisherian @xcite ) fitness , of species @xmath4 .
it is assumed to depend on the population fractions of all other species .
the replicator equations are at the basis of the mathematical description of evolutionary dynamics.@xcite the first term in the right - hand side of eq .
( [ fund ] ) is obtained from ( [ replic ] ) for @xmath56 . in bit - string language models , therefore , the `` fitness '' to be assigned to each language is a given function of the corresponding population fraction , the same for all languages , and does not depend on the populations of other languages .
the negative proportionality between @xmath57 and @xmath58 points out that the function @xmath28 defines a kind of `` unfitness . '' in fact , it measures the probability with which a language is abandoned for adoption of another one .
though the general solution to the replicator equations is not known , some generic mathematical properties make it possible to characterize aspects of the evolution and of its asymptotic state .
these properties can be straightforwardly translated to eq .
( [ fund ] ) when @xmath49 .
first , the ratio of the populations fractions of any two languages @xmath4 and @xmath6 satisfy the `` quotient rule '' @xcite @xmath59,\ ] ] which depends on @xmath5 and @xmath60 only .
it can be formally integrated , yielding @xmath61 dt ' \right ) .\ ] ] suppose now that the initial population of language @xmath4 is larger than that of @xmath6 , i.e. @xmath62 . since @xmath28 is a decreasing function of @xmath63 , the initial value of @xmath64 is positive .
this implies that the ratio @xmath65 will grow , making the difference between the two populations larger .
this kind of feedback effect will be enhanced as time elapses . if , on the other hand , @xmath66 , the ratio @xmath65 will monotonically decrease along the evolution .
since the normalization condition ( [ norm ] ) holds , this result suggests that , generically , the language which initially has the largest population fraction , say language @xmath4 , will asymptotically accumulate the whole population , @xmath67 as @xmath68 , while the fractions of all the other languages will asymptotically vanish .
the only exception to this behaviour happens if two or more languages have exactly the same initial population . for two of these languages , in fact ,
@xmath69 , and @xmath5 and @xmath60 remain identical at all times .
if their initial populations are larger than that of any other language in the system , they will asymptotically share the whole population in equal parts , and the remaining populations will vanish .
the same results arise from a global stability analysis of eq .
( [ fund ] ) , based on the fact that , for @xmath49 , it admits a lyapunov functional,@xcite @xmath70 its time derivative is @xmath71 with @xmath72 and @xmath73 . due to the definiteness of the sign of @xmath74 , lyapunov s theorem holds and , in particular
, the asymptotic state of the differential system is in the manifold where @xmath75 , i.e. where @xmath76 .
this condition , along with eq .
( [ norm ] ) , is met if @xmath77 among the @xmath78 population fractions are equal to @xmath79 ( @xmath80 ) , while the remaining @xmath81 fractions are equal to zero . as advanced from the analysis of eq .
( [ quot ] ) , the asymptotic state consists of a number of languages with identical non - vanishing populations , while all the other languages are absent . note that the homogeneous stationary state where @xmath82 for all @xmath4 , obtained in section [ diff ] for the case without switching , is also found among the equilibrium solutions of the system without mutation . here
, however , this equilibrium solution corresponds to an extremely special initial condition , where all the initial populations are identical . as a matter of fact , all the equilibria with @xmath83 , where more than one language survives asymptotically , are rather special , since they require that two or more populations are initially identical .
these equilibria would not be robust under the effects of fluctuations in the initial condition , or of noise during the evolution . in the most generic case ,
on the other hand , there is a single maximal population .
the above analysis shows that this population will grow to the expense of the others . in the absence of mutation , thus , the system generically approaches a state where only one language survives .
this dominant language accumulates the whole population .
under the combined effects of mutation and switching , it is expected that the system approaches a state where the population is distributed among languages in a way that interpolates between the stationary solutions discussed separately for each mechanism in sections [ diff ] and [ repl ] . with
the notation introduced there , eq .
( [ fund ] ) can be rewritten as @xmath84 + \frac{\mu}{l } \sum_{j=1}^n d_{ij } x_j .\ ] ] stationary solutions are given by equating the right - hand side to zero for every @xmath33 , and solving for @xmath5 .
it should be clear by now that a particular solution is the homogeneous distribution @xmath85 for all @xmath4 .
numerical simulations of agent - based bit - string language models @xcite confirm the presence of this stationary state of homogeneity for large mutation rates . for small @xmath22 , on the other hand , they suggest that there is a stationary solution with a dominant language , which accumulates most of the population , together with several less populated languages . in terms of the results of sections [ diff ] and [ repl ] ,
this solution should be interpreted as the consequence of the interplay between the replicator dynamics of switching , which concentrates population in a single language , and mutation , which redistributes part of the population among `` dialects '' around the dominant language . generally , a joint lyapunov functional for replicator dynamics and diffusion does not exist , so that a global stability analysis of eq .
( [ joint ] ) is not possible along the lines used in section [ repl ] . to my knowledge , the only exception is the case of linear unfitness , @xmath86 , where the functional @xmath87 satisfies @xmath88 , so that @xmath89 . in any case , the local stability of the homogeneous state @xmath85 can be analyzed by standard linearization of eq .
( [ joint ] ) . from this analysis
it turns out that the homogeneous state is a stable equilibrium above the critical mutation rate @xmath90 where @xmath91 is the derivative of the unfitness function .
note that , unless @xmath28 has a singularity at @xmath92 , the critical mutation rate is very small if the number of languages @xmath78 is large .
this seems to strongly disagree with numerical results:@xcite with @xmath93 , @xmath94 ( @xmath95 ) , and @xmath8 , agent - based simulations indicate that homogeneity is asymptotically approached for @xmath96 , while the above equation predicts @xmath97 . as i show below , however , this discrepancy is fallacious . performing a linear stability analysis for the equilibrium state where the population is concentrated around a dominant language , requires first to explicitly find the corresponding stationary solution @xmath98 , which turns out not to be trivial at all .
the equilibrium equations derived from eq .
( [ joint ] ) couple nonlinearly , through the average @xmath99 , all the population fractions @xmath98 .
an approximate solution can however be found in the limit of small mutation rates , @xmath100 , when essentially all the population speaks the dominant language .
assuming that the equilibrium population fraction @xmath101 of a language at hamming distance @xmath102 from the dominant one is of order @xmath103 , and keeping only the most significant terms in powers of @xmath22 , yields @xmath104 the population fraction of the dominant language , @xmath105 , can be obtained from the normalization condition and , within the same approximation order , is @xmath106 .
note that this solution for small @xmath22 is independent of the form of @xmath28 .
this is in agreement with numerical simulations for very small mutation rates , which have found no sensible dependence on the unfitness function.@xcite note also that the relevant small parameter in this approximation is the ratio @xmath107 .
linear stability analysis of the solution ( [ peak ] ) shows that , as expected , it is stable for small @xmath22 .
an instability is predicted for @xmath108 , but this critical value of the mutation rate is outside the validity range of the approximation . comparing with eq .
( [ mu1 ] ) , however , this result suggests that there may be an interval in @xmath22 , just above @xmath109 , where _ both _ the homogeneous state and the dominance state are stable
. such behaviour would be consistent with known features in models of language learning with mutations.@xcite in the following section , i support this conjecture from a different approach , and validate the result by means of numerical resolution of eq .
( [ joint ] ) .
in order to progress beyond the limit of small mutation rates , the solution to eq .
( [ joint ] ) must be approached from a different perspective .
i thus focus the attention on the evolution of the maximal population fraction all over the system , @xmath110 , which i assume is reached at a single language . furthermore , invoking a kind of mean - field approximation , i suppose that the population fractions of the remaining @xmath111 languages are mutually identical . due to normalization ,
their value is @xmath112 . replacing in eq .
( [ joint ] ) yields an evolution equation for @xmath110 : @xmath113-\mu\frac{n x_{\max}-1}{n-1 } .\ ] ] encouragingly , @xmath114 is an equilibrium solution of eq .
( [ xmax ] ) , for any form of @xmath28 . for this solution , the population fractions of all the other languages
are also @xmath115 , so that eq .
( [ xmax ] ) correctly predicts the existence of a stationary homogeneous state . on the other hand ,
its stability threshold is predicted by linearization of eq .
( [ xmax ] ) at @xmath116 which , for large @xmath78 , differs from the critical mutation rate given in eq .
( [ mu1 ] ) by a factor of two .
this is however understandable , taking into account that not any deviation from homogeneity is compatible with the present approximation , so that not all eigenvalues of the original problem are at work in eq .
( [ xmax ] ) to bring the system to equilibrium
. nevertheless , the dependence of @xmath117 and @xmath109 on @xmath78 is the same and , in particular , @xmath117 tends to zero for large @xmath78 .
note also that , for @xmath49 , @xmath118 is always an equilibrium solution .
it turns out to be stable for any decreasing function @xmath28 .
this solution corresponds to the dominance state in the absence of mutation , where all the population speaks the same language .
the existence and stability of other equilibrium solutions for eq .
( [ xmax ] ) depend on the form of the unfitness @xmath28 . for the sake of concreteness ,
i discuss the problem in the limit of large @xmath78 , where the equation for the equilibria of eq .
( [ xmax ] ) takes the simpler form @xmath119-\mu x_{\max}^*.\ ] ] here , i have assumed that @xmath10 . the homogeneous state is now given by the trivial solution @xmath120 .
it can be easily realized that , if @xmath28 decreases with @xmath63 and @xmath22 is positive but not too large , two additional solutions exist in the interval @xmath121 .
the larger solution is a stable equilibrium of eq .
( [ xmax ] ) , and tends to @xmath118 for @xmath122 , thus corresponding to the dominance state .
the lower solution is unstable , and approaches zero in the same limit .
as the mutation rate grows , these two solutions approach each other , and eventually collide and disappear through a tangent bifurcation at a critical mutation rate @xmath123 .
the scenario is illustrated in fig .
[ f1 ] for three forms of the unfitness @xmath28 .
the curves show @xmath124 as a function of the ( normalized ) mutation rate @xmath107 , corresponding to the dominance state ( full line ) and the unstable solution ( dotted line ) . for these low - degree polynomial forms of @xmath28
the critical point of the tangent bifurcation can be exactly calculated . for @xmath9 , @xmath125 , and @xmath126 ,
the respective critical mutation rates are @xmath127 , @xmath128 , and @xmath129 .
the first value is in very good agreement with the critical point reported from agent - based numerical simulations ( @xmath130 for @xmath93 and @xmath95),@xcite and makes it possible to identify the transition detected numerically as the tangent bifurcation at @xmath123 . as advanced at the end of section [ stati ] ,
the present results predict that , for mutation rates satisfying @xmath131 , both dominance and homogeneity are stable states for the distribution of population among languages . in other words , two stable solutions coexist and can be asymptotically approached during the evolution .
the asymptotic state is selected by the initial condition : in eq .
( [ xmax ] ) , the attraction basins of the two solutions are separated by the intermediate unstable state . this dependence on the initial condition has also been noticed in agent - based simulations.@xcite in spite of the agreement with simulations , eq .
( [ xmax ] ) remains the outcome of a rather rough assumption on the population distribution over languages .
it is therefore worthwhile to compare its predictions with results from the numerical resolution of eq .
( [ joint ] ) .
this is also an opportunity to consider relatively small values of @xmath78 , which have been disregarded in eq .
( [ xmaxeq ] ) .
curves in fig .
[ f2 ] show the equilibria of eq .
( [ xmax ] ) as functions of the mutation rate @xmath22 , for @xmath132 ( @xmath133 ) , @xmath93 , and two forms of the unfitness @xmath28 .
dots stand for the maximal population fraction obtained , at long times , from the numerical resolution of eq .
( [ joint ] ) with each function @xmath28 .
two kinds of initial conditions have been considered . in one of them ,
the initial population fractions are chosen at random , all of them close to the homogeneous state , and satisfying the normalization ( [ norm ] ) . in the other initial condition ,
one of the population fractions equals unity , and all the other are zero .
as expected , the former are found to asymptotically approach the value @xmath134 . when the initial population is concentrated in a single language , on the other hand
, @xmath124 is relatively large for mutation rates below the tangent bifurcation , and drops to @xmath115 above it .
the agreement between the results from eqs .
( [ joint ] ) and ( [ xmax ] ) is excellent for small values of @xmath22 and , perhaps not unexpectedly , worsens as the critical point @xmath123 is approached .
however , the overall pictures are qualitatively identical .
in this paper , i have presented a formulation of bit - string models of language evolution based on differential equations for the fraction of the population speaking each language .
the formulation highlights the fact that these models conceive language evolution as combining a diffusion mechanism , given by mutation between similar languages , with replicator dynamics for language switching , when an individual adopts the language of a randomly selected member of the population . the combination of replicator dynamics and diffusion is not new in the literature of biological evolution models .
in contrast with bit - string language models , however , these applications to biological evolution often assume that the fitness of individuals of each species is independent of the population , though it varies between species.@xcite in bit - string models , on the other hand , the dependence of the unfitness on the population fraction is essential to represent preference for more widespread languages .
the decrease of the unfitness of a language when its population grows is the key ingredient which shapes the behaviour of bit - string models . in particular , the occurrence of a tangent bifurcation where the dominance state with most of the population speaking the same language disappears , and the existence of a parameter region where dominance and homogeneity coexist and are stable , are direct consequences of such dependence .
this bistability opens the possibility that , in a system of languages divided into weakly - interacting domains , where inter - domain switching is much less likely than intra - domain transitions , some domains converge to the dominance state while others approach homogeneity , even when the evolution parameters are identical all over the system .
it is worth remarking that a dominance - homogeneity transition is known to happen in genetic space within replication - mutation models of molecular evolution , specifically , in eigen s model.@xcite though the involved critical phenomenon is mathematically different from the tangent bifurcation disclosed here , they are qualitatively much the same from the viewpoint of the competing balance between the basic mechanisms at work .
it would be interesting , by inspiration from mathematical studies of biological evolution , to add bit - string models with some non - uniformity in the individual properties of languages , for instance , assigning different parameters to the unfitness function of each language .
in fact , it may well be that the preference for a very widespread language is inhibited by intrinsic difficulties to acquire it .
_ vice versa _ , a language whose acquisition is perceivably easier could be chosen as a common communication means by populations speaking more widespread but more intricate languages .
do you imagine the whole of mankind eventually deciding to speak mandarin and its dialects?@xcite think also of mark twain s `` the horrors of the german language.''@xcite let me finally point out that comparison of numerical results from bit - string models and empirical data of language statistics specifically , the distribution of language sizes , and the abundance within language families has been focused on transient stages of the evolutionary process . in agreement with this , though not analyzed in detail here , the present study shows that asymptotic states in bit - string models are not a good representation of empirical observations .
transient effects in the present formulation could be addressed by numerical resolution of eqs .
( [ fund ] ) , but such analysis would deserve a separate presentation .
|
a formulation of bit - string models of language evolution , based on differential equations for the population speaking each language , is introduced and preliminarily studied .
connections with replicator dynamics and diffusion processes are pointed out .
the stability of the dominance state , where most of the population speaks a single language , is analyzed within a mean - field - like approximation , while the homogeneous state , where the population is evenly distributed among languages , can be exactly studied .
this analysis discloses the existence of a bistability region , where dominance coexists with homogeneity as possible asymptotic states
. numerical resolution of the differential system validates these findings .
|
shot noise , i.e. , current fluctuations due to the discrete nature of electrons , describes the correlation in electron transport through mesoscopic systems , such as quantum dots ( qds ) or molecular devices .
recently , shot noise measurements have provided additional information that is not available in conventional conductance measurements ( for reviews , see refs .
@xcite and @xcite ) .
in contrast to poissonian noise often observed in classical transport , quantum transport of electrons is usually described by binomial statistics @xcite , as has been used to explain the suppression of noise in noninteracting conductors @xcite . for quantum transport in many mesoscopic systems , the coulomb repulsion allows an electron to enter a region only after the departure of another electron .
this effect induces a negative electron correlation and , therefore , leads to a sub - poissonian shot noise @xcite . instead of coulomb blockade , an electron blockade due to the pauli exclusion principle also has a similar effect @xcite .
recently , it was found that the interplay between the coulomb repulsion and pauli - exclusion principle can lead to electron bunching , i.e. , a super - poissonian shot noise @xcite .
there have been numerous theoretical studies on electron correlations @xcite .
electron correlations in a single qd connected to two terminals have been studied theoretically in both sequential @xcite and coherent tunneling regimes @xcite .
a sub - poissonian fano factor of electrons transporting through two tunnel junctions was found to be due to a coulomb blockade @xcite . in contrast , a super - poissonian fano factor has recently been predicted based on a dynamic channel blockade @xcite which occurs when a qd is connected to spin polarized leads @xcite or is coupled to a localized level @xcite at low temperature . in a three - terminal qd @xcite , positive cross correlations have been obtained by lifting the spin degeneracy @xcite , comparing to the predicted intrinsic negative cross correlation when the level spacing of the dot is much smaller than thermal fluctuations @xcite .
also , a super - poissonian fano factor has been found in the two - terminal case in the cotunneling regime @xcite .
noise measurements have also been performed in many experimental realizations @xcite .
the specific conditions under which super - poissonian shot noise occur was investigated .
positive noise correlations due to the effect of interacting localized states @xcite was observed in metal - semiconductor field transistors @xcite , mesoscopic tunnel barriers @xcite , and self - assembled stacked coupled qds @xcite .
dynamic channel blockade induced super - poissonian shot noise has also been verified in both the weak tunneling @xcite and the quantum hall regimes@xcite in gaas / algaas qds . in addition , positive cross correlations were also recently observed in an electronic beam splitter @xcite , as well as in capacitively coupled qds @xcite .
recently , snchez _ et al _
@xcite . investigated the noise properties of electrons transporting through a qd containing two orbital levels .
only the case with at most one electron in the qd was considered .
they found a super - poissonian shot noise due to dynamic channel blockade .
after applying an ac field to drive transitions between the two levels , the dynamic channel blockade is suppressed and the electron shot noise becomes sub - poissonian at a large field strength . also , djuric _ et al _ @xcite .
considered the spin dependent transport of electrons through a qd with a single orbital level .
their results show that when the bias voltage is beyond that corresponding to the coulomb blockade regime , the spin blockade is lifted and the spin - up and spin - down electrons can both tunnel through the qd independently . as a result
, only a sub - poissonian shot noise of electrons was obtained . in the present work
, we investigate the shot noise of electron transport through a single qd under a magnetic field .
the qd is connected to two non - magnetic leads and can take a single - electron spin - up , single - electron spin - down or two - electron singlet states . in the nonequilibrium case ,
a bias voltage is applied and this leads to different chemical potentials in the two leads . following a setup used to detect single - spin decoherence @xcite , there is a spin - dependent energy cost for adding a second electron into the qd , due to the effects of both coulomb interactions and zeeman splitting .
if the left - lead chemical potential can supply the extra energy for only one of the spin orientations , the coulomb blockade is partially lifted . in this regime ,
the transport mechanisms for the spin - up and spin - down electrons are different . here
, we also found a super - poissonian shot noise , similar to that in @xcite , but it is caused by the effects of both interacting localized states and dynamic channel blockade is obtained .
moreover , we show that these effects can be suppressed by applying an electron spin resonance ( esr ) driving magnetic field . when the esr driving field is strong enough , the electron shot noise will change from super - poissonian to sub - poissonian . the paper is organized as follows . in sec .
ii , we introduce the theoretical model . the master equation and generating function
are explained in sec .
electron shot noise in the presence of an esr magnetic field and couplings to the environment in various bias voltages are discussed in sec .
section v is a brief summary .
we consider a single qd connected to two electron reservoirs by tunneling barriers , as schematically shown in fig . 1 .
electron transport can occur when there are discrete energy levels of the qd within the bias window defined by the chemical potentials @xmath0 and @xmath1 of the two leads .
the full hamiltonian reads @xmath2 with @xmath3 where @xmath4 ( @xmath5 ) is the creation ( annihilation ) operator of an electron with momentum @xmath6 and spin @xmath7 in lead @xmath8 @xmath9 .
@xmath10 represents the hamiltonian of an isolated qd and is given by @xmath11 where @xmath12 describes the orbital energy for the electron spin @xmath7 and @xmath13 is the coulomb interaction between electrons in the dot .
we have defined @xmath14 ( @xmath15 ) as the creation ( annihilation ) operator for electrons with spin @xmath7 , and @xmath16 is the number operator .
external magnetic fields @xmath17 and @xmath18 are applied in the @xmath19 and @xmath20 directions , respectively .
they result in an interaction @xmath21 given by @xmath22 , \label{hspin}\end{aligned}\ ] ] where @xmath23 and @xmath24 with @xmath25 being the electron g factor , @xmath26 the bohr magnetron and @xmath27 the pauli operator . the first term in eq .
( [ hspin ] ) generates a zeeman splitting between two different spin states .
the energy levels of these two spin states are @xmath28 . the second term in eq .
( [ hspin ] ) is an esr driving field with frequency @xmath29 , which produces spin flips when it is resonant with the zeeman splitting .
the tunneling coupling between the qd and the leads is given by @xmath30 where @xmath31 characterizes the coupling strength between the qd and the left ( right ) lead .
in the present discussion , only a single orbital level in the qd is considered . with one electron in the dot
, it can be at either the spin - up ground state , @xmath32 , or the spin - down excited state , @xmath33 .
when an additional electron enters the dot , we only need to consider the singlet ground state with two opposite spins , i.e. , @xmath34 .
this is because triplet states have symmetric spin wavefunctions .
one of the electrons must then occupy a higher orbital level to attain an antisymmetric orbital wavefunction .
in general , the energy difference between two orbital levels in a qd is much larger than the exchange energy between the spins .
the triplet states can hence be neglected @xcite .
when the qd is occupied by one spin - up ( down ) electron , the energy cost @xmath35 to add a spin - down ( up ) electron to form a singlet state is @xmath36 @xcite .
we first consider the case @xmath37 , as shown in figs .
[ fig1](a ) and [ fig1](b ) . from an initial spin - down state
, a spin - up electron can enter the qd since @xmath38 .
moreover , for an initial spin - up state , the condition @xmath39 implies that a second electron can not enter the dot and the tunneling through the qd is blocked at low temperature .
however , the qd becomes unblocked after exciting the electron spin via an esr driving field ( spin flips ) , but only a spin - up electron can tunnel into the qd from the left lead to form a singlet state .
then , either a spin - up or spin - down electron hops onto the right lead from this singlet state , leaving the qd in a single electron state again . in the intermediate regime with @xmath40
[ see figs .
[ fig1](c ) and [ fig1](d ) ] , the transport mechanism is similar to that in the previous regime .
the only difference is that a single spin - down electron in the dot can now hop onto the right lead , because @xmath41 .
alternatively , when the right - lead chemical potential is further tuned down to the third regime with @xmath42 , tunneling is always ensured , as shown in figs .
[ fig1](e ) and [ fig1](f ) .
we have derived a particle number resolved master equation to describe the time evolution of the reduced density matrix , @xmath43 , of the qd @xcite .
the diagonal element @xmath44 @xmath45 gives the occupation probability of state @xmath46 assuming that @xmath47 electrons have arrived at the right lead at time @xmath48 .
the off - diagonal element @xmath49 describes the coherence of two spin states . in the born - markov approximation , the master equation @xcite is given , after some algebra , by @xmath50 where @xmath51 .
the time dependence of the coefficients in eq .
( [ mastereq ] ) has been eliminated by applying a rotating - wave approximation at the resonant condition @xmath52 .
we now explain the the transition rate @xmath53 from state @xmath54 to another state @xmath46 , i.e. , @xmath55 , where @xmath56,@xmath57,@xmath58,@xmath59 .
we have defined @xmath60 and @xmath61 $ ] with the fermi distribution @xmath62}\}^{-1}$ ] .
also , @xmath63 is the half - width of the dot level due to coupling to the electrodes , while @xmath64 and @xmath65 denote , respectively , the density of states and transition amplitude at lead @xmath8 . here , the transition rates are spin - independent since the leads are non - magnetic .
this is in contrast to the spin - dependent transition rates for ferromagnetic leads ( e.g. , ref .
similar definitions are also applied to @xmath66 , @xmath67 , @xmath68 , and @xmath69 .
effects of the environment , such as phonons , nuclear spins , etc . , are effectively taken into account by introducing an additional relaxation rate @xmath70 . here
, we neglect longitudinal relaxation , i.e. , @xmath71 @xcite . to calculate the electron correlations
, one can use the generating function @xcite , @xmath72 .
the equation of motion for the generating function reads , @xmath73 where @xmath74 is the transition matrix depending on the counting variable @xmath75 .
the matrix @xmath74 can be obtained from the master equation , eq .
( [ mastereq ] ) .
one can obtain the correlations from the derivatives of the generating function w.r.t .
the electron counting variable , @xmath76 in particular , the number @xmath47 of electrons reaching the right lead has a mean @xmath77 and a variance @xmath78 laplace transforming eq .
( [ eq - gf ] ) , we get @xmath79 the long time behavior can be extracted from the pole , @xmath80 , closest to zero . from the taylor expansion of the pole @xmath81 , one obtains @xmath82 , which yields@xcite @xmath83 in the long time limit , full statistical information about the electron transport can be obtained from the coefficients @xmath84 . in particular ,
the shot noise fano factor is given by @xmath85 .
@xmath86 @xmath87 indicates super(sub)-poissonian noise while the classical poissonian noise corresponds to @xmath88 .
we have considered above electron transport through a single qd in the regime @xmath89 .
thus , at low temperature that we are interested in with @xmath90 , @xmath91 , and @xmath92 , where @xmath93 , or @xmath94 .
this leads to @xmath95 , @xmath96 , and @xmath97 . by varying the chemical potential @xmath1 of the right lead ,
one arrives at the following three different tunneling regimes : ( i ) @xmath98 , ( ii ) @xmath40 , and ( iii ) @xmath99 , in which the transport mechanisms are different . for simplicity , we define @xmath100 , @xmath101 , and @xmath102 .
let us first consider the case without a driving field @xmath103 . in the first regime ,
i.e. , @xmath104 , a spin - up or spin - down electron enters the initially empty dot .
if it is a spin - down electron , an additional spin - up electron can enter the qd and occupies the energy level , @xmath105 , as shown in fig .
[ fig1](b ) .
this electron can later tunnel out to the right lead , leaving a spin - down electron in the qd .
the whole process then repeats and a continuous current results .
however , if the qd is initially occupied by a spin - up electron , transport will be completely suppressed due to the coulomb blockade , i.e. , @xmath106 , in the low temperature limit . in this case
, no current can be detected . as a result
, a current can be detected only with a probability of @xmath107 when @xmath98 .
for this reason , we will only further study the transport properties for @xmath108 , i.e. , @xmath109 , in this subsection .
the fano factor is obtained as @xmath110 \nonumber\\ & -2\pi_2\!~\!(\gamma+\gamma_l+\gamma_r)(\gamma_l+2\gamma_r)(1-\chi_1)\bigg\},\end{aligned}\ ] ] where @xmath111 in the regime @xmath40 ( @xmath112 ) , the spin - up level is occupied most of the time since it is well below the chemical potentials of the two leads .
its occupation blocks further transport through others levels of the qd .
this mechanism is termed as _ dynamic channel blockade _ , and it leads to enhanced shot noise @xcite .
once this spin - up electron tunnels out to the right lead due to thermal fluctuations , subsequent electron with either spin up or spin down tunnels into the qd . for a spin - up
electron , we are back to the previous situation and just one electron is transported in this cycle .
in contrast , for a spin - down electron , it can quickly tunnel to the right lead , or an additional spin - up electron enters the qd . a super - poissonian shot noise due to the _ dynamic channel blockade _ results and the fano factor is given by @xmath113 where @xmath114 in figure [ fig2 ] , the fano factor is plotted as a function of the relaxation rate @xmath115 .
it can be seen that the fano factor decreases with the relaxation rate @xmath115 in this regime .
this is because the relaxation process increases the occupation of the spin - up state , which blocks further tunneling event and makes the current noise poissonian . in the large bias limit ,
i.e. , @xmath116 ( @xmath117 ) , tunneling for both spin orientations are allowed .
the _ dynamic channel blockade _ is suppressed since the spin - up electron can now quickly tunnel into the right lead .
then , the current noise becomes sub - poissonian with a negative correlation between electron transport events@xcite .
we have @xmath118 ^ 2},\label{noac-2}\end{aligned}\ ] ] where @xmath119.\end{aligned}\ ] ] our results explained here are different from those for a two - level - system case considered in ref .
@xcite , where the current noise is independent of the coupling to outside environment in the large bias regime .
the transport mechanisms involving either of the two orbital levels discussed there are the same . as a result , the relaxation effect will not affect the current noise . for our system ,
the electron transport process depends on the precise state in the qd involved . with an initial spin - up electron
, it will tunnel to the right lead .
starting from a spin - down state , it may tunnel to the right lead or an additional electron enters the qd to form a singlet .
even though the environment - induced relaxation have no direct effect on the singlet state , the noise shows a dependence on the relaxation effect [ see eq .
( [ noac-2 ] ) ] . from fig .
[ fig2 ] , however , one notes that the effect of the relaxation does not alter the super- or sub - poissonian characteristics of the shot noise in both regimes discussed above . in this subsection , we discuss the effects of an esr driving field on the current shot noise .
the esr driving field that induces spin flips between two opposite spin states will affect the transport through the qd differently in the three regimes considered above . for @xmath98
, we have @xmath120 , and @xmath121 .
the physical mechanism in this regime is similar to that for two interacting localized states discussed in ref .
@xcite , in which two impurity levels @xmath122 and @xmath123 were considered .
when the transition rate through the lower impurity level @xmath122 is much smaller than that through the upper impurity level @xmath123 , contribution of @xmath122 to the current is negligible . due to coulomb interaction between these two states , transport through @xmath123 is strongly modulated by the occupancy at @xmath122 . as a result ,
the current jumps randomly between zero and a non - zero value .
this modulation leads to an enhanced current shot noise @xcite .
when the right - lead chemical potential @xmath1 is tuned to the intermediate regime , @xmath40 , with @xmath112 , and @xmath109 , the _ dynamic channel blockade _ will induce a super - poissonian shot noise similar to the _ non - driven _ case in explained subsection 4.1 .
furthermore , if the spin flips generated by the driving field are very frequent , the _ dynamic channel blockade _ will be suppressed .
we then get @xmath127 -6\omega^2\xi_2(\gamma_l+\gamma_r)\ } , \end{aligned}\ ] ] where @xmath128 from fig .
[ fig3 ] , one can see that the fano factor decreases gradually with increasing driving field strength .
this indicates that the current shot noise changes from super - poissoian to sub - poissonian . in the large bias regime , @xmath116 ,
the current shot noise is always sub - poissonian , as shown in fig .
the expression for the fano factor is too lengthy and is not shown here . due to the interplay with the singlet state ,
our results are distinct from those without a driving field at a small relaxation rate .
when the spin flips generated by the driving field is not able to compete with a large relaxation rate , however , the noise tends to be the same in the two cases , as shown in the inset of fig .
in conclusion , we have investigated current noise in a single qd under the influence of both external magnetic fields and the outside environment .
when the dot is initially been occupied by an electron , the zeeman splitting causes a spin - dependent energy cost for adding an additional electron , i.e. , @xmath129 . if the left - lead chemical potential @xmath0 lies within @xmath105 and @xmath130 , the coulomb blockade will be partially lifted . in this setup ,
the transport mechanisms for spin - up and spin - down electrons are different @xcite . in previous studies
@xcite , only a sub - poissonian shot noise was found when the coulomb blockade was completely lifted by applying a large bias voltage , i.e. , @xmath131 .
however , in the full coulomb blockade regime , a super - poissonian shot noise was obtained in a qd containing two orbital levels due to dynamic channel blockade @xcite . in the more general case considered here
, we find a super - poissonian electron shot noise induced by the effects of both interacting localized states and dynamic channel blockade via tuning the chemical potential of the right lead @xmath1 .
moreover , we show that these mechanisms are suppressed by rabi oscillations between the two spin states ( generated by an esr driving field ) . as a result , the current shot noise will change to sub - poissonian at a large rabi frequency .
this work was supported by the srfdp , the nfrpc grant no . 2006cb921205 and the national natural science foundation of china grant nos .
10534060 and 10625416 .
99 blanter y m and bttiker m 2000 _ phys . rep _ * 336 * 1 lesovik g b 1989 _ jetp lett . _ * 49 * 592 ; bttiker m 1990 _ phys . rev . lett . _ * 65 * 2901 safonov s s , savchenko a k , bagrets d a , jouravlev o n , nazarov y v , linfield e h and ritchie d a 2003 _ phys
lett . _ * 91 * 136801
|
we study electron transport through a quantum dot , connected to non - magnetic leads , in a magnetic field . a super - poissonian electron noise due to the effects of both interacting localized states and dynamic channel blockade
is found when the coulomb blockade is partially lifted .
this is sharp contrast to the sub - poissonian shot noise found in the previous studies for a large bias voltage , where the coulomb blockade is completely lifted .
moreover , we show that the super - poissonian shot noise can be suppressed by applying an electron spin resonance ( esr ) driving field . for a sufficiently strong esr driving field strength ,
the super - poissonian shot noise will change to be sub - poissonian .
|
in recent years , much research effort has been dedicated to study the electronic transport through magnetic molecules and single atoms in order to combine molecular electronics with spintronics@xcite .
experiments and theoretical works have demonstrated that molecular and atomic spin states can be inferred and sometimes switched through an electrical current@xcite . among the many interesting phenomena arising in devices comprising magnetic molecules
, there is the kondo effect@xcite .
below the so called kondo temperature , @xmath0 , characteristic of each system , the coupling between the electrons from the electrodes and the spin of the molecule promotes the formation of a many - body state with a fully or partially quenched magnetic moment .
this results in a new resonant transport channel at the electrodes fermi level . to date
the kondo effect has been studied in a number of molecular devices @xcite , and exotic manifestations , such as the orbital kondo effect , have also been reported@xcite .
_ ab - initio _ computational studies play a prominent role in molecular electronics and spintronics .
in particular density functional theory ( dft ) combined with the non - equilibrium green s functions ( negf ) formalism@xcite , known as dft+negf for short , has become the dominant method to address electronic transport@xcite . as initial step in a typical dft+negf study of a molecular device ,
one optimizes the atomic configuration of the device active region , usually called `` scattering region '' or `` extended molecule '' .
then the scattering region is joint to two semi - infinite left- and right - hand side leads ( electrodes ) , whose effect on the states in the scattering region is taken into account by the so - called leads self - energies .
this approach allows to treat the extended molecule as an open system , with the leads self - energies that provide quantitative estimates of the molecule - electrode hybridization and of its dynamic character@xcite . finally , the device transmission function and conductance are calculated within the landauer framework@xcite , by considering the kohn - sham ( ks ) eigenvalues as the single - particle excitations .
although this assumption is formally not correct , since even in exact dft only the highest occupied molecular orbital ( homo ) can be rigorously associated to the negative of the ionization potential@xcite , it practically works well for metallic point - contacts@xcite , nanowires and nanotubes @xcite , quasi two - dimensional systems@xcite , and tunnel junctions@xcite .
in contrast , the calculation of transport properties via the ks eigenvalues encounters some drastic limitations in case of molecules .
the fundamental gap between the homo and the lowest unoccupied molecular orbital ( lumo ) is often severely underestimated by the ks gap computed with standard ( semi-)local exchange - correlation density functionals .
furthermore , non - local correlation effects , such as the dynamical response of the electronic system to the addition of an electron or hole@xcite , are not captured .
these shortcomings hinder the ability of dft+negf to predict the correct energy level alignment between a molecule and the electrodes , which often results in overestimated values for the conductance . consequently , different improvements have been proposed , such as corrections for self - interaction error@xcite , scissor operator schemes@xcite and constrained - dft@xcite .
furthermore , in the recent years , there have been several attempts to move beyond the dft+negf method by using the @xmath1 approximation of the many - body perturbation theory@xcite .
the description of the kondo effect , even at the qualitative level , still represents a challenge . on the one hand , electron correlations leading to the kondo effect are beyond the @xmath1 perturbative scheme . on the other hand , the dft ks spectrum fails to display any kondo - related feature , and so does the conductance computed via the landauer approach .
we note that in principle dft is able to capture the kondo effect in one - orbital lattice models@xcite if the exchange - correlation potential has the correct derivative discontinuity at integer number of electrons , and if the conductance is computed from the density through the friedel sum rule@xcite and not from the ks states . in order to describe the kondo effect in real molecular systems and to overcome
the limitations of dft+negf and @xmath1 , recent studies have proposed to combine dft with model calculations , thus extending to molecular electronics theoretical schemes originally proposed for the study of strongly correlated solid state materials .
these schemes include the dynamical mean - field theory ( dmft)@xcite . the combination of dft and models is typically achieved by partitioning the system of interest in two coupled subsystems , a weakly correlated one , whose electronic structure is well accounted for by dft , and a strongly correlated one .
mathematically , this means that part of the system of interest is projected onto the correlated sub - system , and that the rest is integrated out as an effective bath . in case of molecular devices
, this approach ultimately leads to the reformulation of the electronic structure and transport problem in terms of an effective anderson impurity model ( aim ) , which then has to be solved either exactly or within some approximations .
the potential of this approach has been firstly demonstrated by comparing to photoemission experiments@xcite the computed spectral properties of single magnetic atoms@xcite and molecules@xcite on metallic surfaces .
then , the linear - response ( i.e. zero - bias ) transport properties have been addressed for example by smogunov , tosatti and co - workers@xcite , and jacob and co - workers@xcite .
we have recently contributed to further develop the dft+negf scheme including dmft to study spin transport in solid state devices such as multilayered heterostructures@xcite . in this article , we present our scheme to project out from dft+negf an aim , which is then solved numerically using continuous time quantum monte carlo ( ctqmc)@xcite . the developed method allows us to evaluate the temperature - dependent linear - response transport properties of magnetic molecules on metal surfaces . an important class of such molecules is formed by the stable organic radicals , which are are paramagnetic compounds presenting an unpaired electron in a singly occupied molecular orbital ( somo)@xcite . in particular , here we consider the 1,3,5-triphenyl-6-oxoverdazyl ( tov , fig .
[ fig.tov ] ) .
this is the first organic radical for which the kondo effect was experimentally observed in a scanning tunneling microscope ( stm ) setup with a gold substrate , and the kondo temperature was reported to be about 37 k@xcite . since the system is well characterized by stm , the comparison of the calculated density of states and kondo temperature with the experiment serves as a stringent test for the theory .
additionally , we demonstrate how the different contributions to the conductance that originate from elastic , non - coherent and quantum interference effects can be disentangled .
an open issue concerns the calculation of the electron - electron interaction energy for the anderson impurity .
here we suggest that a partially screened value should be used in order to reproduce the experimental kondo temperature .
the article is separated in two main parts . in the first part ( sec .
[ sec : method ] ) we describe the theoretical methods , while in the second ( secs .
[ sec : comput ] and [ sec : results ] ) we present the results for the tov molecule on au . in the first part
we initially summarize the projection scheme and outline how the different contributions to the transmission are computed ( secs .
[ sec : negf_setup ] to [ sec : teq ] ) , and then we describe the employed ctqmc algorithm ( sec .
[ sec : ctqmc ] ) and the analytic continuation of the ctqmc data ( sec .
[ sec : analyticcont ] ) . in the second part , we start by listing the computational details of the dft and negf calculations ( sec . [
sec : comput ] ) .
we present the dft results for the gas phase tov , for the tov on gold ( sec .
[ sec : geo ] ) and the dft+negf results for the transport properties ( sec .
[ sec : teq ] ) .
we then introduce the aim and discuss its solution within the mean - field approximation ( sec .
[ sec_mf ] ) and by ctqmc ( sec . [ sec_kondo ] ) . finally , we present the transport properties computed via dft+negf+ctqmc ( sec . [
sec : tctqmc ] ) and we conclude ( sec . [ sec : conclusions ] ) .
the first step of the method is the separation of the anderson impurity sub - system from the full system calculated in the dft+negf setup .
this is done by an appropriate projection scheme , outlined in subsections [ sec : negf_setup ] , [ sec : projection ] , and [ sec : hybridization ] , and schematically summarized in appendix [ sec : appendixsum ] .
then , the aim with an effective interaction term is introduced and solved .
this means that the many - body green s function and self - energy are computed ( subsection [ sec : interaction ] ) so that the transport properties can be obtained ( subsection [ sec : teq ] ) .
the method can , in principle , address both zero- and finite - bias transport problems provided the availability of a computationally efficient out - of - equilibrium solver for the aim .
nevertheless , in this work we only consider the zero - bias case , and therefore the solution of the aim is achieved by using ctqmc for quantum systems in equilibrium at finite temperatures , as outlined in subsection [ sec : ctqmc ] .
the use of ctqmc requires a scheme to carry out the analytic continuation of the many - body green s function from the discrete imaginary matsubara frequencies onto real energies this is presented in subsection [ sec : analyticcont ] . a schematic summary of the whole method is shown in fig .
[ fig : schematicmethod ] .
the method has been implemented in the negf code _ _
smeagol__@xcite , which uses a linear combination of atomic orbitals ( lcao ) basis set @xmath2 , and obtains the ks hamiltonian from the dft package siesta@xcite .
note however that the method is general and can be readily used for any code based on the lcao approach .
each basis orbital @xmath3 in _ smeagol _ is characterized by its integer index @xmath4 , which is a collective index that includes the atom , @xmath5 , the orbital @xmath6 , and the angular momentum @xmath7 indices . the orbital index @xmath6 can run over different radial functions corresponding to the same angular momentum , according to the multiple - zetas scheme@xcite .
any operator @xmath8 can be expressed in this basis by using its matrix form @xmath9 , with the matrix elements given by @xmath10 as a matter of notation we remark that in general upper case symbols represent matrices , and to distinguish operators from matrices we explicitly add a hat on top of the symbol for operators .
the setup used in the calculations is schematically illustrated in fig .
[ fig : p_schematic ] , where the basis set described above is used to expand the charge density@xcite .
like in standard electron transport simulations , the system is first split into a semi - infinite left lead , a so - called scattering region ( or _ extended molecule _
( em ) ) , and the semi - infinite right lead@xcite .
we denote the number of basis orbitals within the em by @xmath11
. we can then introduce the hamiltonian matrix of the em as @xmath12 , where the basis orbitals span all orbitals within the em , and where @xmath13 is the hamiltonian operator . since in general
the basis set is non - orthogonal , we also need to introduce the overlap matrix of the em , @xmath14 , given by @xmath15 we note here that we shift all energies of leads and em in such a way to set the fermi energy , @xmath16 , equal to 0
. such a global shift of the spectrum does not affect the properties of the system .
the em is then further subdivided in a total of 3 subsystems .
a set of pre - determined wave - functions @xmath17 ( @xmath18 $ ] ) , defines the _ anderson impurity _ ( ai ) , with @xmath19 being to the number of interacting states .
the _ interacting region _ ( ir ) includes all basis orbitals inside the em that contribute to the ai , which corresponds to the collection of those basis orbitals of the em where any of the @xmath20 is non - zero .
the number of basis orbitals in the ir , @xmath21 , is usually much larger than @xmath19 .
we require that the overlap and hamiltonian matrix elements between orbitals within the ir and the orbitals of the leads outside the em is zero , so that the em has to be chosen large enough to ensure this . as last subspace we introduce the _ extended interacting region _ ( er ) , which includes all basis orbitals inside the em that have a finite overlap or hamiltonian matrix element with the basis orbitals within the ir .
we denote the number of orbitals within the er as @xmath22 , with @xmath23 . ordered by decreasing size , the 4 subsystems are then : em , er , ir , and ai .
the ai is a subsystem of the ir , the ir is a subsystem of the er , and the er is a subsystem of the em ( see fig .
[ fig : p_schematic ] ) .
( @xmath24 ) , the one of the em as @xmath25 , and the coupling hamiltonian matrix as @xmath26 ( @xmath27 ) .
the em is further subdivided in the _ interacting region _
( ir ) and the _ extended interacting region _ ( er ) .
the ir includes a set of interacting molecular orbitals @xmath28 , with @xmath18 $ ] ( @xmath19 is the number of interacting molecular orbitals ; in the schematic figure @xmath29 ) , forming the _ anderson impurity _ ( ai ) . the er consists of all orbitals in the em that have finite overlap with the ir orbitals , so that the ir orbitals are always also part of the er . in total the system is therefore subdivided into 4 subspaces of decreasing size : em , er , ir , and ai .
, width=321 ] the first step is therefore to define the set of molecular orbitals , @xmath28 , inside the em that constitute the ai .
the method outlined here is applicable for any arbitrary set of wave - functions , although in general the choice of the @xmath28 is based on physical intuition . in the simplest case
one can simply take a combination of @xmath30 or @xmath31 orbitals for correlated magnetic atoms in the system , or else the wave - functions of the homo or lumo of a molecule attached to metal electrodes .
a practical way to construct such a set of interacting molecular orbitals is to extract from the full em hamiltonian , @xmath25 , and overlap , @xmath14 , matrices sub - blocks with basis orbitals on the molecule , which we denote as @xmath32 and @xmath33 , respectively .
we can then calculate the eigenvalues @xmath34 and eigenvectors @xmath17 of these sub - systems by solving @xmath35 .
note that each of the @xmath17 is a vector of dimension @xmath21 .
once the set of wave - functions that defines the ai as @xmath36 is chosen ( here we have added the subscript `` ir '' to indicate explicitly that the wave - functions extend over the ir ) , the projection matrix onto the ai inside the ir , @xmath37 , is then defined as @xmath38 which is of dimension @xmath39 .
we then construct the er by adding to the ir all basis orbitals within the em that have finite overlap with the ir basis orbitals , and construct the set of wave - functions @xmath40 .
each @xmath41 is a vector of length @xmath22 , and is equal to @xmath42 for its elements within the ir , and 0 for the others .
these vectors therefore define the ai in the er , and lead to the er projection matrix onto the ai , @xmath43 , given by @xmath44 which is of dimension @xmath45 .
note that the simulation setup in _ smeagol _ requires that one leads unit cell is included at both the left and right ends of the em .
for the projection onto the ai we then further require that the orbitals of those cells can not be part of the ir , while they are allowed to be part of the er . if the basis orbitals indices are approximately ordered from left to right in the em , then the overlap matrix of the em@xcite can be written in the following general block - matrix structure @xmath46 where @xmath47 is a @xmath48 matrix .
@xmath49 ( @xmath50 is a square matrix that includes all the @xmath51 ( @xmath52 ) orbitals within the em on the left ( right ) of the er , so that @xmath53 .
the dimensions of the off - diagonal matrix blocks are determined from the ones of the diagonal blocks , and are therefore not explicitly given .
we subdivide the hamiltonian matrix in an analogous way .
note that by construction the following important relations are satisfied : @xmath54 , @xmath55 , @xmath56 , and @xmath57 .
the green s function ( gf ) matrix of the em is then given by the standard form@xcite @xmath58^{-1 } , \label{eq : gf}\ ] ] which has the block - matrix structure analogous to the one of the @xmath25 and @xmath14 matrices in eq .
( [ eq : ssmall ] ) ; @xmath59 is an arbitrary complex number , and @xmath60 and @xmath61 are the left and right leads self - energies that describe the coupling of the em to the leads .
these are computed according the algorithm in ref . .
importantly , @xmath62 can be either the retarded gf if @xmath63 , with @xmath64 the real energy and @xmath65 a vanishingly small positive real number , or the matsubara gf if @xmath66 , with @xmath67 , for @xmath68 and @xmath69 the inverse temperature ( @xmath70 is the boltzmann constant , @xmath71 is the temperature ) .
spin indices are omitted in above equations and in the following subsections to simplify the notation , but they will be explicitly re - introduced when required to emphasize spin - dependent relations . here
we outline how to separate explicitly the ai sub - system from the rest of the system , which we refer to as the `` bath '' , and that includes the orthogonal subspace to the ai within the em as well as the semi - infinite electrodes . to this aim
we introduce a basis transformation matrix , @xmath72 , which transforms the overlap , hamiltonian , and self - energy matrices as @xmath73 where we denote the transformed matrices with a bar on top of the symbol .
this transformation is required to bring any general matrix extending over the orbitals of the em , @xmath74 , into a transformed form , @xmath75 , in which the top left corner describes the ai , the bottom right corner describes the part of the bath included in the em ( b ) , and the off - diagonal blocks describe the connection terms .
we note that the full bath includes orbitals from both the em and the semi - infinite electrodes , while the @xmath76 matrix @xmath77 contains only the @xmath78 bath orbitals within the em . in mathematical terms
we therefore require @xmath75 to have the following block - matrix structure : @xmath79 we also require the projection to lead to a zero overlap between the ai orbitals and the bath orbitals .
the form of @xmath72 to achieve this for the transport setup is derived in appendix [ sec : appendixafp ] , and is given by @xmath80 here and in the following we denote an identity matrix of dimensions @xmath81 as @xmath82 . we have introduced @xmath83 which is equal to the projection matrix @xmath84 [ eq .
( [ eq : uer ] ) ] , multiplied by the @xmath85 matrix @xmath86 , which in general is constructed in such a way to orthogonalize both the overlap and hamiltonian matrices of the ai itself . in principle this second transformation is arbitrary and can also be omitted , depending on how the ai problem is solved .
the @xmath87 matrix @xmath88 spans the orthogonal space to the ai inside the er , so that @xmath89 , and therefore projects onto the non - interacting part of the er . we denote the number of non - interacting states within the er as @xmath90 , so that @xmath91 .
there is some freedom in the construction of the matrices @xmath86 and @xmath88 , since they are not uniquely defined .
we construct them in such a way to leave the non - interacting part close to the original system , and the detailed relations to construct @xmath86 and @xmath88 are given in appendix [ sec : appendixafp ] , eqs .
( [ eq : w2ai ] ) and ( [ eq : wni ] ) . the explicit form of the final transformed overlap matrix is then evaluated to @xmath92 which as required has zero overlap between ai and bath orbitals , and where @xmath93 the final general form of the projected hamiltonian matrix is @xmath94 where @xmath95 the general structure of the resulting matrices has the required shape given in eq .
( [ eq : aibath ] ) , with @xmath96 and @xmath97 note that the transformed self - energy matrices extend only over the block of the bath inside the em , and are zero for the other matrix blocks .
this is ensured by the requirement that the orbitals of the left and right leads unit cells included at the boundaries of the em are not part of the ir .
the transformation for the gf is given by @xmath98 and can also be evaluated directly in the transformed system as @xmath99^{-1}. \label{eq : gfbar}\ ] ] it is possible to evaluate the required inverse of @xmath72 by blocks ( appendix [ sec : appendixaip ] ) , and the result is @xmath100 where @xmath101 with @xmath102 the form of the @xmath103 block - matrix @xmath104 is given in appendix [ sec : appendixaip ] , eq .
( [ eq : winvni ] ) . by removing the ai orbitals from the system
we can introduce the gf of only the bath orbitals within the em , @xmath105 , as @xmath106^{-1 } , \label{eq : bargb}\ ] ] which can be written in a block - matrix structure analogous to the one of @xmath107 [ eq .
( [ eq : sb ] ) ] as @xmath108 by using also eqs .
( [ eq : gfbar ] ) , ( [ eq : smatexp ] ) and ( [ eq : hmatexp ] ) the gf on the ai is then obtained as @xmath109^{-1}. \label{eq : gbar2}\end{aligned}\ ] ] here we have introduced the so - called hybridization function , @xmath110 , which is given by @xmath111 we remark that in general @xmath110 is a dense @xmath85 matrix .
therefore , optionally , as alternative possibility one can modify the transformation matrix @xmath86 to an energy - dependent form that diagonalizes @xmath110 rather than @xmath112 . in appendix
[ sec : appendixhyb ] we derive an equivalent , commonly used expression for the hybridization function .
furthermore , in appendix [ sec : appendixlimitdelta ] we show that @xmath110 exhibits the correct physical decay for large @xmath59 .
once @xmath113 and @xmath105 are known , then the remaining block matrices of the gf within the em [ see eq .
( [ eq : aibath ] ) for the general matrix structure ] can be evaluated to @xmath114 the ai is fully characterized at the ks - level through the `` on - site '' upper - block @xmath115 of the hamiltonian matrix in eq .
( [ eq : hmatexp ] ) , which is coupled to the bath via the off - diagonal block @xmath116 .
the aim hamiltonian operator can be rewritten in its standard form as an operator in second quantization @xmath117 , with @xmath118.\end{aligned}\ ] ] here @xmath119 is the annihilation ( creation ) operator for an electron of spin @xmath120 in the orbital @xmath121 of the ai , while @xmath122 is the annihilation ( creation ) operator for an electron of spin @xmath120 in an orbital @xmath123 of the total bath , whose hamiltonian matrix is formally written as @xmath124 .
we note that @xmath124 includes the elements of @xmath125 obtained from the projection of the em hamiltonian with the eq .
( [ projection ] ) , as well as the elements of the semi - infinite electrodes .
the effect of the leads is accounted for in the bath green s function [ eq . ( [ eq : bargb ] ) ] , and therefore in the hybridization function [ eq . [ eq : bardelta2 ] ] through the projected self - energies @xmath126 and @xmath127 , so that the full @xmath124 never needs to be explicitly computed . in order to account for many - body correlation effects , we supplement the ai with an effective coulomb interaction , expressed by the operator @xmath128 then the interacting aim hamiltonian operator is defined as @xmath129 @xmath130 is typically chosen to have the form of a generalized hubbard - like interaction , while @xmath131 is the double - counting correction@xcite . this double - counting correction is required in order to subtract the correlation effects in the ai that are already included at the ks level .
the exact form of @xmath131 is unfortunately not known , and several approximations have been introduced , the most common one being the so - called `` fully localized limit''@xcite .
this is the one that we also employ . for a single impurity one - orbital anderson model ( siam ) one
has @xmath132 with the hubbard @xmath133 being a real number , and @xmath134 ( @xmath135 ) the occupation operator for up - spin ( down - spin ) electrons on the ai .
the double - counting correction in the fully localized limit has the simple expression @xmath136 where @xmath6 is the dft total occupation of the impurity .
+ the solution of the interacting aim leads to the many - body gf on the ai , @xmath137 .
one can then define the many - body self - energy on the ai , @xmath138 , by means of the dyson equation @xmath139 since we are considering the case of a single interacting region , the full many - body self - energy matrix of the em is zero for all elements , except for those of the ai @xmath140 here we use the notation @xmath141 to denote a @xmath81 matrix with all zeroes ; the sizes of the off - diagonal blocks are determined by the ones of the diagonal blocks . if @xmath142 is known , it can be used to directly calculate the many - body gf as @xmath143^{-1}. \label{eq : gmbbar}\ ] ] in cases when the many - body self - energy is required in the original basis , this can be obtained by applying the inverse transformation as @xmath144 by using eqs .
( [ eq : winvsimple ] ) and ( [ eq : sigmamb ] ) the explicit structure of @xmath145 becomes @xmath146 the non - zero block of @xmath147 extends over the er , and is given by @xmath148 with @xmath149 given in eq .
( [ eq : wiabc ] ) . after calculating @xmath150 one
can then obtain the full many - body gf in the original basis as @xmath151^{-1}. \label{eq : gmb}\ ] ] the current flowing out of the em into the right electrode , @xmath152 , can be written as sum of a component transmitted into the right electrode from the left electrode , @xmath153 , and a component flowing from the ai into the right electrode , @xmath154,@xcite @xmath155 a similar expression holds for the current flowing from the left electrode into the em , @xmath156 , as outlined in appendix [ sec : appendixcurrent ] . at steady state
the current conservation condition implies that @xmath157 .
+ the value of @xmath153 can be evaluated using the left to right energy dependent many - body transmission coefficient , @xmath158 , which we will simply denote as transmission , @xmath159 , from now on in order to simplify the notation .
@xmath153 is then given by @xmath160 where @xmath161 is the electron charge , @xmath162 is the planck constant , @xmath163 ( @xmath164 ) is the fermi dirac distribution at the chemical potential of the left ( right ) electrode , and @xmath165\nonumber\\ & = & \mathrm{tr}\left[\bar{\gamma}_\mathrm{l}(e ) { \bar{g}^{\mathrm{mb}\dagger}}(e ) \bar{\gamma}_\mathrm{r}(e ) \bar{g}^\mathrm{mb}(e)\right].\label{eq : tre}\end{aligned}\ ] ] @xmath166 includes all interference effects between the possible transport channels across the system , including the effects of interactions on the ai .
the so called coupling matrices @xmath167 are defined as @xmath168,\label{eq : gammal}\\ \gamma_\mathrm{r}(e)&= & i\left[\sigma_\mathrm{r}(e)-\sigma_\mathrm{r}^\dagger(e)\right],\label{eq : gammar}\end{aligned}\ ] ] and @xmath169 .
note that if the matsubara gf is computed by using eqs .
( [ eq : gf ] ) or ( [ eq : gmbbar ] ) with @xmath66 , one has to perform the analytic continuation from the complex matsubara energies to the real axis in order to obtain the retarded gf prior to the calculation of the transmission .
the procedure that we have chosen to carry out such operation is described in sec.[sec : analytic ] .
note also that we can equally express the transmission and current in terms of the quantities in the original basis and in the transformed one . in what follows we work in the transformed basis .
+ the transport properties of a molecular device in dft+negf are usually expressed in terms of the transmission coefficient for the ks system@xcite , denoted as @xmath170 , which is given by@xcite @xmath171.\label{eq : tr0}\end{aligned}\ ] ] it has the analogous structure to @xmath159 , but with the many - body gf replaced by the ks one . therefore @xmath159 accounts for the renormalization of the coherent transport properties via many - body effects not included at the ks dft level , and in the following , we refer to @xmath153 as the coherent , or elastic , component of the current .
in contrast , @xmath154 represents the incoherent component of the current .
+ the incoherent component of the current flowing from the ai to the right electrode is given by ( see appendix [ sec : appendixcurrent ] ) @xmath172 , \label{eq : irai}\end{aligned}\ ] ] where we have introduced the occupation matrix of the ai , @xmath173,@xcite and @xmath174,\ ] ] which is defined in a similar way to the coupling matrices in eqs .
( [ eq : gammal ] ) and ( [ eq : gammar ] ) , but with the many - body self - energy replacing the the leads self - energies . the general structure of the occupation matrix is analogous to the one of @xmath175 in eq .
( [ eq : sigmamb ] ) , and can be written as @xmath176 the quantity inside the trace of eq .
( [ eq : irai ] ) can be interpreted as a transmission matrix , given by @xmath177 , times the difference in the distribution matrices , given by @xmath178 .
it therefore has the analogous structure to @xmath153 in eq .
( [ eq : ielastic ] ) , with the only difference that since @xmath179 is a matrix it can not be moved outside the trace .
the analogous equations for the current from the left lead into the em , @xmath156 , are given in appendix [ sec : appendixcurrent ] .
if we assume that the matrices @xmath180 and @xmath181 are zero , or more generally that their values are small enough that they can be neglected , and that @xmath182 ( @xmath183 ) extends only over the region @xmath184 ( @xmath185 ) , then we can obtain the transmission function from the gf block matrix of the ni as @xmath186,\ ] ] where @xmath187 $ ] and @xmath188 $ ] .
the matrices @xmath189 are given by @xmath190^{-1}\bar{k}_{\alpha,\mathrm{ni}},\\ \bar{\sigma}_\mathrm{r , ni}(e)&=\bar{k}_{\mathrm{ni},\beta}\left[\bar{k}_{\beta\beta}- \left(\bar{\sigma}_{\mathrm{r}}\right)_{\beta\beta}(e)\right]^{-1}\bar{k}_{\beta,\mathrm{ni}},\end{aligned}\ ] ] with @xmath191 having the analogous block - matrix structure to @xmath192 in eq .
( [ eq : hmatexp ] ) , and for real energies @xmath193 and @xmath194 .
note that @xmath195 , @xmath196 , and with the assumptions made in this paragraph we have @xmath197 and @xmath198 . + with eq .
( [ eq : gfexpand ] ) the gf block matrix of the ni can be obtained from the one of the ai as @xmath199 with @xmath200 the transmission can then be decomposed in three components @xmath201 where we have introduced the background or bath transmission @xmath202 , \label{eq : tbath}\ ] ] the transmission through the ai @xmath203 , \label{eq : tai1}\ ] ] and the interference term of the transmission @xmath204\nonumber\\ & + & \mathrm{tr}\left[\bar{\gamma}_\mathrm{l , ni}(e ) { \bar{g}}_\mathrm{ni}^{\dagger}(e ) \bar{\gamma}_\mathrm{r , ni } ( e ) \delta\bar{g}_\mathrm{ni}^\mathrm{mb}(e)\right ] .
\label{eq : tinterference}\end{aligned}\ ] ] often the background and ai transmission do not interfere significantly , so that it is useful to give estimates for their separate values . in that case typically
the transmission is composed of a background value , onto which the peaks due to transport through the ai are added . the transmission through the ai can be rewritten as @xmath205 , \label{eq : tai2}\ ] ] which depends only on the gf of the ai , and on the hybridization matrices of the ai , given by @xmath206 we can perform the analogous transformations also for @xmath154 from eq .
( [ eq : irai ] ) , and obtain @xmath207,\end{aligned}\ ] ] where we have defined @xmath208 .
\label{eq : gammamb}\ ] ] in general therefore the calculation of the current requires the knowledge of the impurity occupation matrix @xmath209 , which in turn requires the solution of the non - equilibrium problem . in refs .
@xcite a specific shape is implicitly assumed for this matrix@xcite , which allows simplified estimates of the currents .
the approach proposed in those references requires the inversion of the coupling matrices @xmath210 and @xmath211 , which is however not defined in the general case @xcite .
we now consider the special case where @xmath212 , with @xmath213 a constant@xcite .
this rather strict condition can usually only be fulfilled for a siam , so that in the remainder of this section we consider only the siam . in this case
it is possible to avoid the calculation of @xmath214 by using the current conservation condition @xmath157@xcite , and one obtains @xmath215 with @xmath216 where we have used the fact that for the siam all quantities in the equation are just numbers instead of matrices . using eqs .
( [ eq : ielastic ] ) and ( [ eq : iraispecial ] ) for the special case of a siam , and @xmath212 ,
we can write the total current as @xmath217 with the total effective transmission , which includes elastic and incoherent terms , given by @xmath218 we can collect the terms that describe the effective total transmission across the ai , @xmath219 , as@xcite @xmath220 . \label{eq : ttai}\end{aligned}\ ] ] eqs .
( [ eq : itotal]-[eq : ttai ] ) extend the results of ref .
@xcite to the more general case including background transmission and interference terms , as typically found for stm experiments of molecules on surfaces .
+ in this article we only consider applications to linear - response transport , so that the conductance @xmath221 is given by @xmath222 this equation generalizes the landauer formula used in dft+negf to the case of an interacting em , with the only difference that @xmath170 ( used in the landauer formula ) is replaced by @xmath219 . therefore ,
following the standard practice used in dft+negf that consists in analyzing the zero - bias transport properties by means of the transmission at equilibrium @xmath170 , here we will present the results for zero - bias transport in presence of many - body effects by plotting @xmath219 .
[ sec : ctqmc ] in the present work the aim is solved by using ctqmc for quantum systems in thermodynamic equilibrium@xcite , since we only address linear - response transport . in this case
the method is well - established .
we note however that recently there have been a number of developments towards the extension of ctqmc to out - of - equilibrium problems , both in steady - state and time - dependent frameworks@xcite .
+ in this work , two different algorithms have been considered : the weak - coupling approach , called continuous - time auxiliary field ( ct - aux)@xcite , and the hybridization expansion , strong coupling approach ( ct - hyb)@xcite .
ct - aux scales as the product of the interaction @xmath133 and of the inverse temperature , so that its application turns out too computationally demanding for kondo systems at temperatures of the order of only a few kelvin .
in contrast , we found ct - hyb able to provide quite accurate results at a reasonable computational cost for the specific au / tov system considered in the following . since this system is described as a siam , here we present the method only for this case .
however , we remark that our implementation can treat multi - orbital systems as well , although only for density - density interaction terms .
ctqmc is restricted to finite - temperatures , where the partition function is given by @xmath223 $ ] .
the starting step , which is common to all algorithms , is to separate the interacting aim hamiltonian [ eq .
( [ iaim ] ) ] in two parts , a reference hamiltonian @xmath224 and a perturbation hamiltonian @xmath225 , so that @xmath226 .
each ctqmc algorithm differs in the exact definition of @xmath224 and @xmath225@xcite . while in ct - aux and other weak - coupling approaches @xmath225
is set equal to the effective coulomb interaction term @xmath227 [ eq . ( [ hubbard ] ) ] , in ct - hyb one imposes@xcite @xmath228 with @xmath229 , @xmath230 , @xmath231 and @xmath232 defined in subsection [ sec : interaction ] . therefore the perturbation hamiltonian in ct - hyb is represented by the bath - ai coupling , while the reference hamiltonian is that of the decoupled atomic - like correlated ai and of the isolated bath .
having divided the hamiltonian in two parts , one is able to introduce the interaction picture , where an operator @xmath8 depends on the imaginary time @xmath233 as @xmath234 , with @xmath235 , and the partition function is written as the standard time - ordered exponential @xmath236\nonumber\\ & = \sum_{n=0}^\infty\int_0^\beta d\tau_1 ... \int_{\tau_{n-1}}^\beta d{\tau}_{n}\ , w_n , \label{part_f}\end{aligned}\ ] ] with @xmath237 the time - ordering operator , and @xmath238.\label{w_part_f}\ ] ] the partition function has the form of an integral over a configuration space . in such space , any particular configuration is specified by the expansion order @xmath6 , the times @xmath239 and a set of discrete variables , for instance the spin , and it is characterized by the probability distribution @xmath240 .
it is this integral , which is ultimately evaluated by monte carlo techniques . + by using the definitions of @xmath224 and @xmath225 in eqs ( [ h1 ] ) and ( [ h2 ] ) , @xmath241 in eq .
( [ w_part_f ] ) becomes@xcite @xmath242\\ \times\prod_\sigma \det{d_\sigma^{-1}}(\tau_{1}^\sigma, ... ,\tau_{n_\sigma}^\sigma;{\tau'}_{1}^\sigma, ... ,{\tau'}_{n_\sigma}^\sigma).\label{wc}\end{gathered}\ ] ] here @xmath243 is the bath partition function , @xmath244 , @xmath245 and the matrix @xmath246 has elements @xmath247 which are the fourier transforms of the hybridization function @xmath248 evaluated for the imaginary - time interval @xmath249 , which separate pairs of operators @xmath250 and @xmath251 in eq .
( [ wc ] ) . in the function
@xmath241 the trace accounts for the impurity that fluctuates between different quantum states as electrons jump in and out , while the determinants resum the bath evolutions which are compatible with the sequence of quantum fluctuations in the impurity .
note that here we have explicitly re - introduced the spin - index that was dropped from the equations in the previous subsections .
however , if the bath is non - magnetic like in most cases , the hybridization function is the same for spin - up and down , and can therefore be obtained from non spin - polarized calculations . in the specific case of a one - orbital hubbard interaction , ct - hyb can be efficiently implemented by using the so - called `` segment representation''@xcite .
this means that each configuration is depicted by segments , which represent time intervals @xmath252 during which an electron of a given spin resides on the impurity .
notably , with such representation the trace in @xmath241 can be evaluated in polynomial time .
new configurations are then obtained by either adding or removing segments .
this is enough to ensure ergodicity , although other operations ( such as shifting the segments end - points ) and global updates must be implemented to ensure an efficient sampling .
each update is accepted or rejected according to the metropolis algorithm .
the acceptance probability is efficiently computed with standard fast matrix update methods@xcite . as seen in eq .
( [ delta_tau ] ) , ct - hyb requires the fourier transform of the hybridization function , which in principle is calculated through a summation over an infinite number of frequencies .
however , in practice only a finite number of frequencies smaller than a certain cutoff @xmath253 can be inevitably summed up , although the high - frequency limit of @xmath254 determines @xmath255 close to @xmath256 .
therefore , in order to accurately calculate @xmath257 for any arbitrary @xmath233 , we use a standard approach that consists in adding and removing from eq .
( [ fourier ] ) the high frequency limit of @xmath258 .
this limit is given by @xmath259 , with @xmath260 given in eq .
( [ eq : m1 ] ) of the appendix [ sec : appendixlimitdelta ] .
hence we evaluate @xmath257 as @xmath261+\frac { m_1}2,\end{gathered}\ ] ] where the summation is restricted to positive frequencies , since @xmath262 and @xmath263 .
+ during monte carlo sampling the many - body matsubara gf can be directly estimated . however , following boehnke _
@xcite , we use an expansion of the @xmath264 in terms of legendre polynomials , @xmath265 $ ] , @xmath266 and we estimate the expansion coefficients @xmath267 . the advantage of this choice is twofold .
first , the transformation ( [ gl ] ) can be written as a unitary transformation .
second , only few legendre coefficients are needed to express the matsubara gf .
this is because the legendre coefficients for a matsubara gf decay faster than any power of the legendre expansion index @xmath268@xcite .
a valuable effect of this is that the statistical noise is filtered out due to the cutoff at a certain expansion order .
after extensive tests for the specific system studied in this work , we found that an appropriate choice for the cutoff of the legendre polynomials is around 100 .
however , such cutoff must generally be determined for each specific case .
+ the many - body self - energy is obtained either from the dyson eq .
( [ eq : sai ] ) , or by using the expression@xcite @xmath269 where @xmath270 is the matsubara representation of the correlation function @xmath271 , which can be easily computed in the legendre polynomial basis at no extra cost@xcite .
this last approach usually provides much more accurate results than the calculation through the dyson equation . the inversion of the gf in eq .
( [ eq : sai ] ) amplifies the statistical noise , in particular at high matsubara frequencies , when the difference between interacting and non - interacting gf is very small .
in contrast , such problem is not present when using eq .
( [ fg ] ) .
+ finally , we note that , although in this subsection we have made explicit the dependence of the gf on the spin - index for notation completeness , the spin up and down gfs are equal as long as the substrate is non - magnetic , and there is no zeeman - like term in the aim hamiltonian .
since this is the case for the application presented in this work , we will once again drop this index in the following .
[ sec : analyticcont ] in order to compute the transport properties , as outlined in subsection [ sec : teq ] , we need the retarded many - body gf on the ai , @xmath272 , whose imaginary part defines the spectral function @xmath273 normalized as @xmath274 and which is related to the real part via the kramers - kronig relation @xmath275 however , ctqmc returns the matsubara gf @xmath276 and not @xmath272 .
the relation between the spectral function , eq .
( [ a_mb ] ) , and the matsubara gf is determined by the hilbert transformation @xmath277 which must therefore be inverted to find the unknown @xmath278 from @xmath276 .
unfortunately , the inversion of the hilbert transformation belongs to the class of ill - posed problems , for which there is no unique solution in a mathematical sense . despite that ,
several numerical methods have been proposed during the last few decades to deal with this problem in the context of quantum monte carlo methods@xcite . here
we employ the stochastic optimization ( so ) , and our implementation is based on the original proposal by mishchenko _
et al._@xcite , with some modifications .
we have verified that there are only negligible quantitative differences between the results obtained with our code and the original version by mishchenko@xcite .
other methods , such as the pade approximation@xcite and the maximum entropy@xcite methods , which are routinely employed by the dmft practitioners , have been tested as well .
however , the results are found to depend critically on the precision of the monte carlo data , so that they result in overall unsatisfactory performances .
the so relies on the parametrization of the spectral function @xmath279 as a sum of @xmath280 rectangles determined by the height @xmath281 , the length @xmath282 and the center @xmath283 , and which is normalized as in eq .
( [ normalization ] ) .
this means that @xmath284 with @xmath285\\ 0 \mbox { otherwise}\end{cases}.\ ] ] accordingly , the gf obtained by using this @xmath286 in eq .
[ analytic ] reads @xmath287 .
\label{gt_analytic}\ ] ] the optimization algorithm proceeds through a series of global stochastic updates that change the number , the size and the location of the rectangles in order to minimize a given deviation function between @xmath288 and the original set of gfs obtained by ctqmc .
importantly , the fact that @xmath289 has an analytic expression [ eq . ( [ gt_analytic ] ) ] largely improves the performances of the method .
once the many - body retarded gf is computed for real energies , the many - body self - energy on the real energy axis can be obtained by using the dyson eq .
( [ eq : sai ] ) .
alternatively , the analytic continuation of the self - energy @xmath290 can be directly carried out @xcite .
the dft calculations for the tov molecule in gas phase are performed by using the siesta code@xcite .
norm - conserving troullier - martin pseudopotentials are used together with a double-@xmath291 plus polarization quality basis set .
the perdew - burke - ernzerhof ( pbe ) generalized gradient approximation ( gga)@xcite for the exchange - correlation density functional is considered .
additional all - electron calculations are carried out with the fhi - aims package@xcite . in particular , fhi - aims is used to compare the results of pbe with those obtained with the pbe0 hybrid functional@xcite . furthermore , the g@xmath292w@xmath292 method@xcite of the many - body perturbation theory is also employed .
g@xmath292w@xmath292 calculations are carried out by using the ( generalized)-ks orbitals and eigenvalues obtained by either pbe or pbe0 ( the notation g@xmath292w@xmath292@pbe and g@xmath292w@xmath292@pbe0 is used in order to distinguish between the two cases ) .
the computational details are the same as those described in ref .
+ fhi - aims is also employed to optimize the geometry of the tov molecule on au within a supercell approach .
the au(111 ) surface is experimentally found to present fcc and hcp domains separated by ridge regions , and with the molecules that are adsorbed on both domains@xcite .
here we consider the fcc surface , which is modeled as a 4-layer slab with a ( 5x5 ) square unit cell , and each slab is separated by 60 of vacuum from its periodic image . only the molecule and the first two au layers are allowed to relax until forces are smaller than @xmath293 ev/@xmath294 .
the pbe+vdw@xmath295 method@xcite is employed in order to include the effect of the van der waals ( vdw ) interactions .
pbe+vdw@xmath295 has been extensively tested for molecules both physisorbed and chemisorbed on metallic substrates , and the results are generally more accurate than those obtained with other common vdw - corrected gga functionals@xcite .
the values of the screened @xmath296 coefficient , the vdw radius and the polarizability for au are provided by ruiz _
_ in the original article about pbe+vdw@xmath295@xcite .
the standard numerical atom - centered orbitals basis set `` tier 1 '' and `` tier 2 '' are considered for au and h , c , n , o , respectively .
the used @xmath70-points mesh is equal to @xmath297 .
all dft calculations for the gas phase molecule and for the molecule on au are spin - polarized in order to account for the magnetism .
_ smeagol _ electron transport calculations are performed with the relaxed geometries described in the previous subsection , where further au layers are added below the molecule , and the top electrode with the au stm tip is included [ fig .
[ fig : structtransport](a ) ] .
the real space mesh is set by an equivalent energy cutoff of 300 ry , and we use the local density approximation ( lda ) for the exchange correlation potential .
we verified that the transport properties are essentially the same when using the lda or gga functionals .
we employ a double-@xmath291 plus polarization basis set for the all atoms of the molecule , a double-@xmath291 basis for the au atoms , and a double-@xmath291 plus polarization basis with extended orbital cutoffs for the au atoms of the tip .
this ensures that there is an appropriate electronic coupling between tip and molecule also at extended tip - molecule separations .
we verified that due to the large size of the supercell in the plane we can evaluate the electronic structure at the @xmath298-point in the brillouin zone perpendicular to the transport direction .
all calculations of the dft+negf transport properties without the inclusion of the many - body self - energy are spin - polarized . in contrast , non - spin - polarized calculations are used to extract the hybridization function @xmath110 and on - site hamiltonian @xmath115 of the aim , since the magnetism of the tov is accounted for within the aim .
spin - polarized calculations of the hybridization function are only required if the substrate is magnetic .
( left ) and cfg@xmath299 ( right).,scaledwidth=50.0% ] .somo and sumo energies , @xmath300 and @xmath301 , and somo - sumo energy gap @xmath302 computed with several methods : pbe , pbe0 , @xmath303scf , and the g@xmath292w@xmath292 perturbative correction to the ( generalized-)ks spectrum computed with pbe(0 ) .
@xmath303scf provides the somo and sumo energies for an isolated molecule through finite total energy differences @xmath304 and @xmath305 , where @xmath306 is the pbe energy of the neutral molecule and @xmath307 [ @xmath308 is the energy of the corresponding cation ( anion)@xcite . [
cols="<,^,^,^,^",options="header " , ] the tov molecule has spin 1/2 due to an unpaired electron , delocalized mainly over the nitrogen @xmath309 orbitals of the planar verdazyl heterocycle ( fig .
[ fig.ldos ] ) . the atoms n2 and n4 carry slightly more spin than n1 and n5 , and the difference of the mulliken populations for spin up and down is equal to @xmath310 for n2 and n4 , and to @xmath311 for n1 and n5 . the spin delocalization over the verdazyl heterocycle reflects the character of the somo , and of the corresponding singly unoccupied molecular orbital ( sumo ) ( fig .
[ fig.ldos ] , left panel ) .
in contrast , the highest doubly - occupied molecular orbital ( homo ) and the lowest unoccupied molecular orbital ( lumo ) are mainly located on the phenyl rings .
the results are overall consistent with the established picture for the tov electronic structure@xcite .
+ although the four n atoms in the verdazyl heterocycle are in the same plane , the molecule does not have a flat conformation .
in fact , we find that the the two phenyl groups attached to n1 and n5 are twisted out of the verdazyl heterocycle plane , either in the same direction by about @xmath312 degrees ( geometry 1 in right panel of fig .
[ fig.ldos ] ) , or in opposite directions by about @xmath313 and @xmath314 degrees ( geometry 2 in the right panel of fig .
[ fig.ldos ] ) .
only the latter conformation ( geometry 2 ) has been previously reported in the literature@xcite , but we find that that the energy difference with the geometry 1 is below @xmath315 mev , which is the numerical accuracy of our calculations set by the used basis set@xcite .
in contrast , the energy difference with the geometry obtained by constraining all three phenyl groups in the same plane as the verdazyl heterocycle is about @xmath316 ev .
+ the molecule s spin can be fully compensated either by charging or by attaching a hydrogen to the oxygen atom , thus forming the h - tov molecule . in both cases
, we find a considerable geometrical rearrangement involving also the n atoms , so that the verdazyl core appears largely distorted , and the phenyl groups are further moved out - of - plane in an asymmetric way .
the optimized geometries for the negatively - charged tov and h - tov are displayed in fig .
[ fig.ldos ] , right panel .
negligible differences are found between the results obtained with pbe and pbe0 , and with siesta and fhi - aims .
+ the somo and the sumo are well separated in energy from all other molecular states , and the somo - sumo gaps obtained with different methods are reported in tab .
[ somo_sumo ] .
several works have demonstrated that g@xmath292w@xmath292 yields excellent results for gaps when the generalized ks orbitals from hybrid dft are used as a starting point for the perturbative calculation@xcite .
for this reason , and due to the lack of experimental data , in the following we consider the g@xmath292w@xmath292@pbe0 estimate of @xmath317 ev as reference value .
we note that that no difference is found within our numerical accuracy for the computed somo - sumo gap in case of the two different tov conformations ( geometries 1 and 2 in in the right panel of fig .
[ fig.ldos ] ) , so that the results in tab .
[ somo_sumo ] apply to both .
+ the geometry optimization of the tov / au system reveals that the molecule can be adsorbed in two different configurations , labelled cfg@xmath318 and cfg@xmath299 . in configuration cfg@xmath318 ,
tov is physisorbed and assumes an almost flat conformation on the surface .
the distances of n2/n4 and n1/n3 from the top au layer are @xmath319 and @xmath320 , respectively .
the average adsorption height of the phenyl rings attached to n1 and n3 is @xmath321 , while that of the phenyl ring attached to c3 is @xmath322 .
the adsorption energy is equal to @xmath323 ev , a value that is comparable to that of typical molecules of similar size ( such as perylenetetracarboxylic dianhydride , ptcda ) physisorbed on au@xcite . by analyzing the mulliken populations
, we note that a small charge transfer of about @xmath324 electrons from the au to the molecule occurs .
this leads to an overall decrease of the somo - sumo exchange splitting and to a partial compensation of the molecule magnetic moment , which reduces to @xmath325 .
configuration cfg@xmath299 corresponds to a qualitatively different physisorbed geometry , which has an adsorption energy of @xmath326 ev . for cfg@xmath299 one full electron
is transferred from the surface to the tov , so that the spin on the molecule completely disappears .
although the adsorption energy for cfg@xmath318 and cfg@xmath299 differs by only about @xmath327 mev , the energy barrier separating the two states is quite large ( @xmath328-@xmath329 ev ) .
indeed we observe that the increased charge transfer is accompanied by a substantial distortion of the tov verdazyl heterocycle , with one of the n atoms ( n4 ) being displaced out of the molecular plane towards the au surface ( see fig .
[ fig.barrier ] ) .
the distance between n4 and the surface decreases to @xmath330 , while the distance between n2 and the surface increases to @xmath331 .
this distortion of the verdazyl core is similar to that observed in the gas phase for the negatively - charged molecule ( fig .
[ fig.ldos ] ) , taking into account that now the phenyl groups are kept parallel to the surface because of the vdw interaction .
we note that our calculations for physisorbed h - tov reveal a similar deformation of the molecule .
we point out that the charge transfer and the consequent spin reduction or compensation are likely overestimated by the calculations , because pbe underestimates the somo - sumo gap .
this problem is commonly found in dft - based electronic transport simulations , and it has been cured in practice by using scissor operator schemes@xcite ( see the following section ) .
the overestimated charge transfer may have important implications for the geometry optimization , since it may artificially stabilize the configuration cfg@xmath299 over the configuration cfg@xmath318 .
the calculation of the correct charge transfer for structural relaxations can currently not be addressed on a fully quantitative level even with state - of - art methods .
dft energy functionals that increase the somo - sumo gap , such as hybrid functionals , do not provide a good description of the metal electronic structure . for configuration
cfg@xmath318 the charge transfer is overall rather small , and in cases of small or negligible charge - transfer pbe+vdw@xmath295 has been shown to perform remarkably well when compared to experimental data@xcite .
we therefore expect that the final geometry for cfg@xmath318 , and in particular the molecule - surface average distance , is accurately predicted . in experiments
two types of molecules are observed on the au surface , denoted as type - a and type - b@xcite .
while type - b molecules display a kondo resonance in the low bias conductance , such resonance is absent in type - a molecules .
type - a molecules have been identified as h - tov , while type - b molecules correspond to configuration cfg@xmath318@xcite .
we note that based on our results the type - a molecules could also correspond to molecules adsorbed with geometry cfg@xmath299 .
however , as discussed in the previous paragraph , it is likely that the calculations overestimate the stability of cfg@xmath299 , and in experiments most type - a molecules may indeed be h - tov . in the remaining part of the manuscript we will address only type - b molecules , where the kondo state has been measured experimentally .
the spin - polarized projected density of states ( pdos ) on the molecule in the transport setup for configuration cfg@xmath318 is shown in fig .
[ fig : structtransport](b ) .
after adsorption , the former gas - phase somo and sumo appear as lorentzian - like peaks that extend around @xmath16 .
the combined effects of the hybridization of the somo and sumo states with the au substrate , and of the small surface - to - molecule charge transfer , induce a reduction of the exchange splitting compared to the gas phase , and consequently of the magnetic moment .
the full width at half maximum of the somo and sumo lorentzian peaks gives their electronic coupling to the substrate , @xmath332 , which we calculate to be @xmath333 290 mev .
this system therefore presents a rather strong electronic coupling to the substrate . in contrast , the electronic coupling to the tip is usually much smaller , and it can be neglected compared to @xmath298 .
the analysis of the transport properties is carried out by looking at the transmission coefficient , @xmath166 ( sec .
[ sec : teq ] ) .
this decays exponentially with the tip height and depends sensitively on the in - plane position of the stm tip .
the transmission through the somo and sumo is largest when the tip is on top of the n atoms of the molecule , so that we place it above atom n2 in the subsequent evaluation of the transport properties , with a tip height of about 5.7 above the plane of the molecule .
the spin - polarized transmission as function of energy is shown in fig .
[ fig : tdistance](a ) , and it can be seen that , like the pdos , also @xmath166 is only slightly spin - split .
we investigate the change of @xmath166 for reduced coupling by rigidly shifting the molecule off from the surface by a distance , @xmath334 , of 0.5 , and 1.0 (fig .
[ fig : tdistance](b - c ) ) .
the results demonstrate that if the coupling is reduced by increasing the molecule - surface separation , then the peak of the radical state is less broad , and the spin splitting increases . for @xmath334 of 0.5
the radical is fully spin - split , while it is only partly spin - split for @xmath335 0.0 .
this shows that in quantitative predictions of hybridizations of molecules to surfaces the correct evaluation of the contact geometry is of central importance .
experimentally such a change of coupling might be achievable by systematically varying the groups attached to the tov molecule .
.,scaledwidth=43.0% ] as discussed in the previous section , the lda gas phase ks somo - sumo gap is underestimated when compared to experiment . on a surface
however the gap is reduced due to the image charge effect , which is not captured by the lda ks eigenvalues@xcite .
a fortuitous error cancellation between the gap underestimation and the neglect of image - charge effects may sometimes happen , but this is not generally the case .
here we evaluate the effect that a larger gap has on the transport properties by using a scissor operator ( sco ) approach@xcite , where we increase the somo - sumo gap by about 1 ev to illustrate the general trend ( fig .
[ fig : t_dft ] ) .
we apply the sco correction non - self - consistently as a postprocessing step on top of the lda converged charge density as well as self - consistently .
the opening of the somo - sumo gap leads to a fully spin - split state , where one can clearly identify somo and sumo peaks despite their large broadening .
self - consistency does not qualitatively change the results , demonstrating that applying the correction as a post - processing step to the converged lda solution is a good approximation .
the exact value of the somo - sumo gap , which corresponds to the interaction energy @xmath133 [ eq .
( [ hubbard ] ) ] , for molecules on surfaces can only be estimated .
we will address this issue for the tov molecule deposited on au in sec .
[ sec_kondo ] .
finally , we note that the dft+negf pdos and transmission are spin - split and a magnetic moment appears due to the unequal occupation of spin up and down states .
this is a characteristic broken - symmetry picture , which is only valid for applied magnetic fields@xcite . in absence of magnetic field the symmetry is not broken , and the dft symmetry breaking is an artifact caused by the effective exchange - correlation magnetic field . , and with a sco applied selfconsistently [ sco ( scf)].,scaledwidth=43.0% ] ( eq .
( [ eq : bardelta2])).,scaledwidth=43.0% ] the tov adsorbed on au is represented as a siam through the projection described in sec .
[ sec : method ] .
we remind that here non - spin - polarized dft calculations are used , since magnetism is accounted for in the siam calculation .
the hybridization function , calculated with eq .
( [ eq : bardelta2 ] ) , is shown in fig .
[ fig.delta ] , and mainly reflects the au dos . both the real and imaginary parts are rather featureless over a wide energy range , where only the @xmath336 states contribute to the au dos , while very sharp peaks are present at energies below about @xmath337 ev , where the fully filled @xmath338 bands are located . and mean - field local magnetic moment @xmath339 as function of @xmath133 .
bottom panel : mf and ctqmc squared local magnetic moment @xmath340 and the corresponding square local magnetic moment in the limit of no correlations @xmath341 .
results are for @xmath342
. error bars in the ctqmc estimates are smaller than the symbols . , scaledwidth=40.0% ] , @xmath343 , @xmath315 and @xmath344 ev . for @xmath345
ev the spectral function for spin up and down are superimposed as the systems is spin - degenerate.,scaledwidth=40.0% ] before solving the siam by using ctqmc , we address it within the mean - field approximation , where the hamiltonian @xmath346 in eq .
( [ hubbard ] ) is replaced by @xmath347 , and the double - counting correction is the same as in ctqmc [ eq .
( [ hdc ] ) ] .
the mean - field solution can be directly compared to the broken - symmetry dft+negf results , where @xmath348 plays the same role as the ks potential including the sco , and where @xmath133 determines the somo - sumo gap . at @xmath345 ,
the total ai occupation @xmath349 is equal to the lda occupation of the ai , obtained with the projection described in sec .
[ sec : method ] .
this is slightly larger than @xmath315 , reflecting the small metal - to - molecule charge transfer predicted by dft , and discussed in sec .
[ sec : geo ] .
the magnetic moment @xmath350 is equal to zero , because the aim is mapped from a non - spin - polarized calculation and @xmath351 . for @xmath352
ev , we see in fig.[fig.n_m ] ( top panel ) that the total occupation @xmath353 initially increases as function of @xmath133 , until the system undergoes a transition to the magnetic state . at this point ( between @xmath354 and @xmath355
ev ) the formation of the magnetic moment is accompanied by the reduction of @xmath356 , which then keeps decreasing monotonically for increasing @xmath133 towards the half - filling limit .
this behavior is qualitatively the same that we find when we apply the sco in the dft+negf ( sec .
[ sec : transportdft ] ) .
the spectral function of the ai is a lorentzian at @xmath345 ev ( fig .
[ fig.dos ] ) , and gradually splits into two ( almost ) lorentzian functions as @xmath133 is increased , one for spin up and one for spin down .
importantly , due to the small charge transfer from the au substrate , these functions are not symmetric around @xmath16 .
moreover , the spectral function for spin down is slightly sharper than that for spin up , which is due to the decrease of absolute value of the imaginary part of the hybridization function with increasing energy ( fig .
[ fig.delta ] ) .
this is overall similar to the dft+negf transmission peaks displayed in fig .
[ fig : t_dft ] . to conclude this section
, we take advantage of the known exact mean - field spectral function for real energies in order to assess the accuracy of the analytic continuation ( sec .
[ sec : analytic ] ) .
this is shown in fig .
[ fig.amf ] , which compares the exact spectral function evaluated directly for real energies , to the one `` analytically continued '' from complex matsubara frequencies to real energies , and which should in principle be identical to one another .
the accuracy is very good for energies around @xmath16 , where the two spectral functions are almost indistinguishable , while the agreement becomes progressively worse for much larger ( in absolute values ) energies .
in particular , the analytically continued spectral function appears broader than the exact one the further one moves away from @xmath16 .
this additional broadening is an artifact of the analytic continuation , and it is caused by the limited accuracy of the so ( sec . [ sec : analytic ] ) .
k ( green line ) . , scaledwidth=40.0% ] and at @xmath357 k. the noise is a well - know drawback of the so method@xcite . in principle
, it can be greatly reduced by averaging the results of many independent optimizations for the same data set , in practice however it can not be completely removed .
the presented results are obtained by averaging over 250 independent runs.,scaledwidth=43.0% ] ev and at several different temperatures , @xmath71 .
the inset shows a zoom for the @xmath358 ev - range around the fermi energy .
, scaledwidth=43.0% ] the ctqmc impurity occupation @xmath359 and squared local magnetic moment @xmath360 for several @xmath133-values at @xmath357 k are presented in fig .
[ fig.n_m ] .
we remark that @xmath361 and not @xmath362 is the appropriate quantity to look at in the analysis of the magnetization for the ctqmc results , since the occupation for spin up and down is equal in absence of any zeeman - like term in the hamiltonian , even in the local moment regime .
this shows that the exact ctqmc solution does not have the unphysical symmetry breaking found for the dft and mean field solutions .
the ctqmc @xmath360 gradually increases as function of @xmath133 , and there is no evidence of an abrupt magnetic transition at @xmath363 ev . nevertheless , when mean - field predicts a stable local magnetic moment for @xmath364 ev , the ctqmc @xmath360 is smaller than the corresponding mean - field value , while @xmath365 is slightly larger . in this region of parameters
the effect of electron correlation is therefore to favor the ai double occupancy compared to the mean - field picture , so that the magnetic moment is partly quenched .
this is a typical behavior expected for a kondo system . for large @xmath133 , after the transition from the kondo screened to the local magnetic moment regime , the mean - field and the ctqmc @xmath366 converge to the same value , and @xmath367 approaches the half - filling limit .
+ important insights into the dominant physics at different energy scales are provided by the ai spectral function @xmath368 . in fig .
[ fig.spectral_u ] we observe that the lorentzian peak found for @xmath345 ev gradually shrinks into a narrow and sharp peak for increasing values of @xmath133 , thus forming the kondo resonance centered around @xmath16 . at the same time , two broad satellite peaks , separated approximately by @xmath133 , develop , and they correspond to the somo and sumo states found in the dft+negf and in the mean - field dos .
note that , unlike in the mean - field case of fig
. [ fig.dos ] , here the somo and sumo are not formed by breaking the spin - symmetry through an unequal spin - occupation .
furthermore , they are much broader . even though the broadening is partly an artifact caused by the analytic continuation ( see sec .
[ sec_mf ] ) , we can still appreciate the real intrinsic broadening due to electron - electron interaction by comparing fig .
[ fig.spectral_u ] to fig .
[ fig.amf ] . + in fig .
[ fig.spectral_t ] the spectral function is plotted at fixed @xmath369 ev , but at different temperatures , ranging from @xmath370 k to @xmath371 k. the somo and sumo - related peaks do not change significantly as function of @xmath71 within the stochastic and numerical precision of the results , since @xmath133 is constant .
the kondo peak on the other hand progressively shrinks with increasing @xmath71 , in analogy to what has been reported in experiments .
we note that the energy scale for the kondo effect is set by the ratio between @xmath133 and the electronic coupling of the molecule to the substrate @xmath298 ( see below ) .
therefore , for a fixed @xmath298 , increasing @xmath71 at fixed @xmath133 has the analogous effect on the kondo feature as fixing @xmath71 and increasing @xmath133 .
+ the quality of the computed spectral function at @xmath372 can be checked by means of the generalized friedel sum rule@xcite @xmath373 we remind that in this work we set @xmath372 to @xmath374 as explained in subsection [ sec : negf_setup ] .
although in principle the generalized friedel sum rule is valid only at @xmath375 k , we find that it is satisfied within @xmath376 whenever the temperature is low enough that the system can be considered deep in the kondo regime .
this turns out to be the case for @xmath377 ev at @xmath357 k. deviations for larger @xmath133 are mostly due to the finite temperature , since the system gets closer to its kondo temperature ( see below ) .
+ we have also assessed the large energy behaviour of the spectral functions by computing their moments . the normalization condition defined in eq .
( [ normalization ] ) fixes the value of the zero moment , and is enforced in the analytic continuation ( sec .
[ sec : analyticcont ] ) .
the sum rule involving the first moment@xcite @xmath378 is fulfilled within @xmath379 to @xmath380 for every considered value of @xmath133 and @xmath71 [ @xmath381 is the contribution to the on - site energy deriving from the double - counting correction in eq .
( [ hdc ] ) ] .
much larger errors are found for higher moments .
overall these results and the verification of the generalized friedel sum rule confirm the observations in the previous sub - section about the expected accuracy of the computed spectral functions at different energies .
while the accuracy is good for energies around @xmath16 , there is a loss of accuracy at high energies , reflected by the error in the moments beyond the first . + in order to quantitatively relate our simulations to the experiment@xcite we compute the kondo temperature @xmath0 .
this is defined as@xcite @xmath382 where @xmath383 is the quasiparticle weight , also known as the quasiparticle mass - renormalization factor , @xmath384^{-1}.\label{z_real}\ ] ] since ctqmc simulations yield the self - energy at the matsubara frequencies , @xmath383 is approximated in the discrete form@xcite @xmath385^{-1}.\label{z_mats}\ ] ] with @xmath386 .
we note that the definition of @xmath0 in eq .
[ t_k1 ] was originally derived for the symmetric aim in the kondo limit , where the occupation is @xmath387 , and where the charge susceptibility is zero and the magnetic susceptibility becomes equal to that of the @xmath388 model , also called kondo model , with @xmath389 for @xmath390@xcite . despite that ,
in recent works@xcite it has been employed to study general aims with the parameters obtained from dft , and it was shown able to provide reliable estimates of @xmath0 , so that its use has become justified in practice . moreover , in our case the symmetric aim is a good approximation to the studied aim , since the investigated system is close to half - filling and its hybridization function appears rather featureless over a wide energy range around the fermi energy ( fig . [ fig.delta ] ) . in experiments@xcite @xmath0
is extracted from the measurement of the full width at half maximum of the kondo peak , @xmath391 , as a function of temperature by means of the approximate relation@xcite @xmath392 therefore , additionally to eq .
[ t_k1 ] , we have also considered this equation , with @xmath71 equal to our simulation temperature , and @xmath393 as the kondo - peak width of the spectral functions in fig .
[ fig.spectral_u ] . + for @xmath357 k. the ctqmc results obtained with eqs .
( [ t_k1 ] ) ( blue circles ) and ( [ t_k2 ] ) ( orange squares ) are compared to the experimental value ( horizontal green dashed line ) .
the black curve represents the fit of the ctqmc data from eq .
( [ t_k1 ] ) in the range of @xmath377 by using the model eq .
( [ gamma_u_t_asym ] ) for the asymmetric siam ( with @xmath394 ) .
this fit allows the extrapolation of the calculated @xmath0 to larger @xmath133 . for comparison
, the pink dash - dotted line shows the expected @xmath0 for the symmetric siam ( @xmath395).,scaledwidth=45.0% ] the estimated @xmath0 as a function of @xmath133 for simulations at @xmath357 k is plotted in fig .
[ fig.kondo_t ] on a logarithmic scale .
the results obtained by using the two alternative definitions in eqs .
( [ t_k1 ] ) and ( [ t_k2 ] ) are in good agreement , showing that both methods are largely equivalent for the evaluation of @xmath0 .
note that the numerical and stochastic precision of the computed @xmath383 , and consequently of @xmath0 through eq .
[ t_k1 ] , degrade for @xmath133 larger than 2 ev as @xmath383 becomes small , although its error bar stays of the order of 0.1 . in this case
the system is approaching the transition point between the kondo and the local moment regime , where @xmath383 is not a well - defined quantity anymore .
a precise evaluation of the kondo temperature via eq .
[ t_k1 ] for @xmath396 would require much lower simulation temperatures than @xmath357 k , where the system would be deep in the kondo regime , but these simulations turned out too computationally demanding .
the results can then be compared to the kondo temperature for the asymmetric siam , given by ( see page 168 in ref . [ ] ) @xmath397 which is valid for @xmath398 .
here @xmath399 , and @xmath400 is the on - site energy of the impurity .
the asymmetric siam reduces to the symmetric siam for @xmath395 .
[ gamma_u_t_asym ] shows that for large @xmath133 the value of @xmath0 decreases exponentially with @xmath133 . in our calculations
@xmath400 is used a parameter that is obtained by fitting with eq .
[ gamma_u_t_asym ] the @xmath0 , which is calculated via eq .
( [ t_k1 ] ) in fig .
[ fig.kondo_t ] , for small @xmath133 ( @xmath401 ev ) .
we then obtain @xmath394 .
this value deviates from the symmetric value @xmath395 because , as we pointed out above , there is a small substrate - to - molecule charge - transfer .
the fitted @xmath400 approximately agrees with @xmath402 ( @xmath403 for @xmath401 ) , and accordingly also to the position of the somo peak with respect to @xmath16 in the spectral functions in figs .
[ fig.spectral_u ] and [ fig.spectral_t ] .
this agreement confirms the applicability of the model , and shows that for @xmath377 ev our dft+negf+ctqmc calculations provide reliable estimates of the kondo temperature .
then , by using eq .
( [ gamma_u_t_asym ] ) with the fitted @xmath400 , we can extrapolate the calculated values of @xmath0 to larger values of @xmath133 , where the @xmath0 computed directly from the ctqmc data becomes inaccurate .
we note that for the special case of the symmetric siam with @xmath395 in eq .
( [ gamma_u_t_asym ] ) , we obtain a steeper decay of the predicted kondo temperature as function @xmath133 than for the fitted asymmetric case ( see fig .
[ fig.kondo_t ] ) .
the tov / au system has an experimental @xmath0 of @xmath404 k@xcite , and the estimated value of @xmath133 that gives such @xmath0 in our calculation for the fitted asymmetric siam equation is @xmath405 ev ( fig .
[ fig.kondo_t ] ) .
consistently , for this @xmath133 the value of @xmath383 is calculated to be approximately zero , and the squared local magnetic moment @xmath360 approaches the mean - field value ( see fig [ fig.n_m ] ) .
the estimated @xmath133 that reproduces the experimental kondo temperature can then be compared to the somo - sumo gap obtained for the gas phase molecule ( tab .
[ somo_sumo ] ) , which is computed to be @xmath317 with g@xmath292w@xmath292@pbe0 . when the molecule is put on the au surface at a height of about 3.1 ( sec .
[ sec : geo ] ) , metal - induced image - charge corrections reduce the static gap@xcite . in order to estimate this static gap reduction ,
a classical image - charge model can be used , with an image plane placed at 1 above the au surface @xcite . with this model the reduction of the gap results to be about 3.4 ev , which then leads to an approximate static @xmath133 for the molecule on the surface that is @xmath329 ev .
this static @xmath133 corresponds to the lower limit for @xmath133 , since it implies full screening of any charge fluctuation on the molecule by the au surface .
the value of @xmath405 ev that best matches experiment lies somewhat above this lower limit , but well below the upper limit of the gas phase molecule values .
we explain this result by observing that , since the fluctuations responsible for the kondo physics happen on a rather fast time - scale , they will generally not be fully screened .
while much effort has been dedicated to compute fully - screened @xmath133 values in both molecular and solid state systems with several alternative methods , such as constrained random phase approximation ( crpa)@xcite , constrained dft ( cdft)@xcite , screened hybrid functionals@xcite and ( self - consistent ) linear response@xcite , the intermediate situation that we suggest here has not been analyzed yet .
this represents therefore an important aspect for future theoretical studies about kondo physics .
we point out that an accurate estimate of the kondo temperature is a great challenge . as seen in eq .
( [ gamma_u_t_asym ] ) , the kondo temperature is an exponential function of the ratio between @xmath133 and @xmath298 .
this latter quantity is determined by the molecule adsorption distance and position , which means that the quantitative modeling of kondo systems critically depends on the accuracy of the dft structural relaxation for molecules on surfaces .
this in turn requires a correct description of both covalent and vdw interactions , which is an open direction of research in itself@xcite , despite notable recent advancements , such as the introduction of the pbe+vdw@xmath295 functional@xcite that we use here .
the total elastic transmission in presence of many - body correlations , @xmath166 , of the system is calculated with eq .
( [ eq : tre ] ) .
importantly , we have verified that the same result is obtained by using the original or the projected matrices , so that the evaluation of the inverse projection of the many - body self - energy [ eq .
( [ eq : sigmambinv ] ) ] is in principle not required for the calculation of the transmission .
the results for the setup outlined in sec . [ sec : tdft ] , and for three different values of @xmath133 , are given in fig .
[ fig : t_mb](a ) .
the non - spin - polarized transmission without correlations , @xmath406 , corresponds to @xmath166 for @xmath345 , and has a broad lorentzian peak around @xmath16 .
when increasing @xmath133 this peak gets narrower and becomes the kondo peak , with two broad satellite peaks separated by about @xmath133 , following the evolution of the dos in fig .
[ fig.spectral_u ] .
the peak below @xmath16 is less pronounced than the one above @xmath16 , which indicates an energy dependent coupling to the electrodes , and is in qualitative agreement with the results obtained using a scissor operator ( fig .
[ fig : t_dft ] ) . the breakdown of @xmath166 into its components , as derived in sec .
[ sec : teq ] , is shown in figs .
[ fig : t_mb](b - d ) . by construction the background transmission in fig .
[ fig : t_mb](b ) [ eq .
( [ eq : tbath ] ) ] is independent of @xmath133 , and reflects the dos and coupling of the tip to the substrate mediated by non - interacting states of the molecule . for the considered values of @xmath133 the transmission of the ai itself , defined in eq .
( [ eq : tai2 ] ) and shown in fig .
[ fig : t_mb](d ) , decays to zero for energies @xmath344 ev below and above @xmath16 , where the ai dos vanishes ( fig .
[ fig.spectral_u ] ) .
the interference terms are shown in fig .
[ fig : t_mb](c ) [ eq .
( [ eq : tinterference ] ) ] , and are rather small , so that in this case @xmath166 largely corresponds to the sum of @xmath407 and @xmath408 . in general however @xmath409 can be large and lead to a change of shape of the kondo peak from a symmetric lorentzian - type to an asymmetric fano - type , and our method allows to evaluate the contributions individually .
for the contribution of the ai we evaluate also the total transmission including incoherent components , @xmath410 , and the results are shown in fig .
[ fig : t_mb](d ) . qualitatively @xmath410 and @xmath408 are similar , but @xmath408 becomes significantly smaller than @xmath410 as @xmath133 increases , since it neglects incoherent components of the electron transport .
note that at zero temperature the many - body self - energy at @xmath16 vanishes in the kondo regime , so that one should have @xmath411 and @xmath412 .
however , at finite temperature the many - body self - energy is small , but not exactly 0 , which results in the kondo transmission peak shrinking when compared to the non - interacting case and to @xmath413 for @xmath414
. we can now evaluate how the transmission depends on the tip shape .
since the tip - molecule coupling is rather weak at our considered tip height , we can assume that the tip does not significantly influence the electron correlation effects in the molecule , so that we use the same many - body self - energy for the different tip shapes . to evaluate the transmission for a blunt tip , we remove the single - tip au atom , so that we have a tip with 3 au atoms in the bottom - most plane closest to the molecule .
we then shift the height of these atoms from the molecule to be equal to the one of the single - atom tip considered in fig .
[ fig : t_mb ] .
the comparison of the transmission for the two tip shapes , and for @xmath415 ev , is shown in fig .
[ fig : t_mb_bt ] .
the main difference is that the 3-atom tip has a stronger coupling to the molecule , which increases all contributions to the transmission .
the background transmission becomes less featured , since the increased dos around @xmath16 for the 1-atom tip is generally broadened for a 3-atom tip @xcite . with the used value of @xmath415 ev , which is close to the value obtained by comparing the kondo temperature to experiment , our transport results are in good agreement with experimental conductance measurements in ref .
@xcite , which show a kondo resonance at small bias on top of a background conductance , which is approximately constant in the considered low bias range . in general the relative height of background transmission to kondo peak depends on tip shape , size and position , and can therefore be one of the tools to infer a nanoscale junction geometry .
we have presented a rigorous way to extract an effective aim from dft+negf calculations for molecules adsorbed on a substrate .
this effective aim is then solved at finite - temperature with our implementation of ctqmc . with the developed method we can describe the linear - response conductance of systems in the kondo regime , and evaluate the separate contributions to the overall conductance originating from the transmission through the molecule itself , from the background transmission , and from interference terms . furthermore , for the siam we can compare the elastic contributions to the total conductance including incoherent effects .
the results for the specific case of the tov radical adsorbed on au demonstrate that the method can potentially return accurate values for the kondo temperature , provided that dft gives reliable adsorption geometries , and that one can estimate the value of @xmath133 for the effective electron - electron interaction .
we propose that a partially screened @xmath133 has to be considered to describe the kondo physics for such molecules on metallic surfaces . +
our results show that the developed method is a promising tool to evaluate the linear - response conductive properties of strongly correlated nanoscale systems .
we note that the extension beyond linear response to finite - bias problems is in principle straightforward , and the current can be calculated as outlined in sec .
[ sec : teq ] .
the main developments required for such finite bias calculations are accurate and computationally efficient solvers for the out - of - equilibrium aim in the steady - state .
such developments are still an open research direction , and important progresses have been recently made@xcite .
the authors are grateful to l. chioncel , a.r .
rocha and s. kurth for useful discussions .
a.d thanks i. de marco for some interactions about the problem of the analytic continuation .
the work was sponsored by the european union through the project 618082 acmol .
ad received additional support from the `` ministerio de economia y competitividad '' ( mineco ) of spain .
as first step in the projection scheme we introduce a new basis set @xmath416 , which is identical to the original set of @xmath417 , with the exception that the @xmath19 basis orbitals , which have largest overlap with the set @xmath418 are replaced with the corresponding @xmath41 of the ai .
note that by construction the replaced basis orbitals are always part of the ir , since @xmath41 has non - zero elements only within ir basis orbitals .
this transformation is achieved with the projection matrix @xmath419 the matrix @xmath420 has the dimension @xmath87 , and is an identity matrix with the @xmath19 columns removed , which correspond to the ones of the replaced basis functions .
we collect all the unity vectors removed when forming @xmath420 into a @xmath45 matrix @xmath421 , so that @xmath422 and @xmath423 .
the overlap and hamiltonian matrices in the new basis , @xmath424 and @xmath425 , are then obtained as @xmath426 the general structure of the resulting @xmath424 is @xmath427 where @xmath428 is a @xmath85 matrix , @xmath429 is an @xmath430 matrix , and the dimensions of the off - diagonal blocks are determined from the ones of the diagonal blocks .
importantly , the sub - blocks for the overlap between the ai and the @xmath184 and @xmath185 subsystem are 0 , since the er is chosen to be large enough to guarantee this by construction .
the basis functions of the ai are generally not orthogonal to the other er orbitals , so that @xmath432 . if the anderson impurity solver can deal with such non - orthogonal systems , then one can pass this matrix directly to such a solver .
most ai solvers however require orthogonality between the ai and all other orbitals in the system .
we therefore perform a second transformation , which orthogonalizes all the orbitals to the ones of the ai .
importantly , we require such transformation to keep the orbitals of the ai , and of the @xmath184 and @xmath185 regions , unchanged
. one can achieve this with the following transformation matrix : @xmath433 with @xmath434 this leads to the transformed hamiltonian , @xmath435 , and overlap , @xmath436 , matrices .
@xmath437 has the new structure @xmath438 with @xmath439 we therefore now have the ai orthogonal to all bath basis orbitals . as a further optional step we can diagonalize @xmath428 and @xmath440 . to this aim
we first orthogonalize the orbitals in the ai via a lwdin transformation by calculating the square root of the inverse of the ai overlap matrix , @xmath441 , and then obtain the eigenvalues and the matrix of all eigenvectors , @xmath442 , of the resulting eigenvalue system @xmath443 . here
@xmath442 is a @xmath85 matrix , with each column equal to an eigenvector of the eigenvalues system , and @xmath444 is a @xmath85 diagonal matrix , where the diagonal elements are the corresponding eigenvalues . the resulting transformation matrix to the new basis acts only on the orbitals of the ai , and is @xmath445 with @xmath446 the transformed matrices are @xmath447 , and @xmath448 , and keep the non - zero structure of @xmath449 and @xmath437 , except that the top - left @xmath85 elements become diagonal matrices .
the total transformation matrix , @xmath72 , then is @xmath450 which when multiplied out gives the result in eq .
( [ eq : wshort ] ) , repeated in the following for convenience @xmath451 here we have introduced @xmath452 with @xmath453 in this section we evaluate the inverse of @xmath72 [ eq .
( [ eq : wshort ] ) ] . by using eq .
( [ eq : w123 ] ) we obtain its general form as @xmath454 the inverse of @xmath455 is @xmath456 where @xmath457 is a @xmath458 matrix , and @xmath459 is a @xmath103 matrix .
they are obtained by evaluating @xmath460 if we recollect that all the unity vectors removed when forming @xmath420 are combined into a matrix @xmath421 ( @xmath461 ) , then we can give analytic expressions for @xmath457 and @xmath459 as @xmath462 note that since for large systems the matrices @xmath420 and @xmath421 consist mainly of zeroes , and the matrix @xmath463 is only of dimensions @xmath464 , the calculation of the non - zero elements of @xmath465 by using eq .
( [ eq : abu ] ) is a fast order(@xmath11 ) operation .
we write the dense @xmath471 matrix @xmath472 in the block matrix form analogous to the one of the overlap matrix [ eq .
( [ eq : ssmall ] ) ] as @xmath473 by using eqs .
( [ eq : gtran ] ) and ( [ eq : winvsimple ] ) we obtain the following expression for the green s function projected on the ai : @xmath474 which is an equivalent expression to the one given in eq .
( [ eq : gbar2 ] ) . if the task is to calculate @xmath475 , one can therefore also first use eq .
( [ eq : gbar1 ] ) to obtain @xmath476 , and then eq . ( [ eq : gbar2 ] )
to evaluate @xmath475 as @xmath477 however , especially for large @xmath59 , this indirect way of obtaining @xmath478 is significantly less accurate than the direct way through eq .
( [ eq : bardelta2 ] ) .
the reason is that for large @xmath59 the quantities @xmath59 and @xmath479 in eq .
( [ eq : bardelta1 ] ) are both very large , while the resulting difference is small , with a leading term proportional to @xmath480 ( see sec .
[ sec : appendixlimitdelta ] ) . in a practical calculation
we therefore always use eq .
( [ eq : bardelta2 ] ) to calculate @xmath481 . by means of eq .
( [ eq : bardelta2 ] ) we can evaluate the limit of @xmath481 as @xmath59 goes to infinity . in eq .
( [ eq : bardelta2 ] ) @xmath481 is given by a multiplication between energy independent matrices , @xmath482 and @xmath483 , with the energy dependent matrix @xmath484 .
we therefore need to evaluate the asymptotic limit of @xmath484 , defined in eq .
( [ eq : bargb ] ) , for large @xmath59 . to this aim
we first evaluate the limit for large @xmath59 of @xmath60 and @xmath61 .
the left self - energy satisfies the following recursive relation@xcite @xmath485 where @xmath486 ( @xmath487 ) is the left lead s onsite overlap ( hamiltonian ) matrix , and @xmath488 ( @xmath489 ) is the left lead s coupling overlap ( hamiltonian ) matrix between nearest neighbor cells .
if we expand this relation for large @xmath59 , we can see that there is a dominant term proportional to @xmath59 , a constant term , plus terms proportional to @xmath480 and smaller . for large
@xmath59 we can therefore write @xmath490 with @xmath491 and @xmath492 constant matrices to be evaluated . if we insert this relation in eq .
( [ eq : senerecursive ] ) and evaluate it for large energies , we obtain the recursive relation @xmath493 which has the analogous form of eq .
( [ eq : senerecursive ] ) when the hamiltonian is set to 0 , multiplied by @xmath480 .
@xmath494 can therefore be evaluated by using the standard method of ref . with the hamiltonian set to 0 , and can be interpreted as the `` self - energy '' matrix in the calculation of the inverse of the bare left semi - infinite overlap matrix .
when we perform the perturbation expansion of eq .
( [ eq : senerecursive])to the next lower order we obtain the following recursive relation for @xmath492 : @xmath495 since the coupling matrices @xmath489 and @xmath488 are usually small compared to the onsite terms , we can keep the result only to lowest order in these matrices , so that we can write @xmath496 however this remaining term is still of second order in the coupling matrices , so that it is usually small , and we can therefore neglect it altogether and approximate it as @xmath497 .
we can perform an analogous expansion for the right self - energy , for which we can then write @xmath498 , with @xmath499 . with this expansion for the leads self - energies , we can now expand @xmath500 in eq .
( [ eq : bargb ] ) to second order in @xmath480 as @xmath501^{-1}\nonumber\\ & \approx&\frac{1}{z } m_1 + \frac{1}{z^2 } m_2 + o\left(\frac{1}{z^3}\right ) , \label{eq : limitsg}\end{aligned}\ ] ] with @xmath502 the asymptotic limit of the hybridization function for large @xmath59 can then be evaluated by inserting eq .
( [ eq : limitsg ] ) into eq .
( [ eq : bardelta2 ] ) .
if the hamiltonian and overlap matrices are real , then also @xmath260 and @xmath503 are real . in this case , if the energy @xmath59 is purely imaginary , @xmath66 ( with @xmath504 a matsubara frequency or any other real number ) , then the leading term of the imaginary part of @xmath481 is proportional to @xmath260 and decays as @xmath505 , while the real part is proportional to @xmath503 and decays as @xmath506 . on the other hand , in the case of a very large real energy with a small imaginary part @xmath65 added , @xmath507 ,
the real part of the leading term of @xmath481 is proportional to @xmath260 and decays as @xmath508 , while the leading imaginary part is also proportional to @xmath260 and decays as @xmath509 .
the current flowing from the left electrode into the em , @xmath156 , is given by@xcite @xmath510.\end{aligned}\ ] ] the integral runs over real energies , @xmath64 , so that @xmath511 and @xmath512 are the retarded and advanced green s functions , respectively , and @xmath513 is the lesser green s function@xcite .
@xmath514 is defined in eq .
( [ eq : gammal ] ) . in an analogous way the current flowing from the em into the right electrode , @xmath152 , is @xmath515 , \label{eq : ir1}\end{aligned}\ ] ] where @xmath164 is the fermi dirac distribution at the chemical potential of the right electrode , and @xmath516 is defined in eq .
( [ eq : gammar ] ) . with @xmath517 defined in eq .
( [ eq : gammamb ] ) , the following relation is fulfilled @xmath518\bar{g}^{\mathrm{mb}{\raisebox{0.5ex}{\footnotesize $ \dagger$}}}(e)\nonumber\\ & = & -i\bar{g}^{\mathrm{mb}{\raisebox{0.5ex}{\footnotesize $ \dagger$}}}(e)\left[\bar{\gamma}_\mathrm{l}(e)+\bar{\gamma}_\mathrm{r}(e)+\bar{\gamma}^\mathrm{mb}(e)\right]\bar{g}^\mathrm{mb}(e),\nonumber\end{aligned}\ ] ] and the lesser gf is given by @xmath519\bar{g}^{\mathrm{mb}{\raisebox{0.5ex}{\footnotesize $ \dagger$}}}(e ) , \label{eq : glesser}\end{aligned}\ ] ] where we have introduced the occupation matrix of the ai , @xmath173@xcite . using relations ( [ eq : grelations ] ) and ( [ eq : glesser ] ) in eq .
( [ eq : ir1 ] ) , one obtains the expression for the current in eqs .
( [ eq : ir],[eq : ielastic],[eq : irai ] ) . in an analogous way one obtains the current @xmath156 as sum of a component transmitted from the left electrode into the right electrode , @xmath520 , and a component flowing from left electrode into the ai , @xmath521 ,
@xmath522 the value of @xmath520 can be evaluated as @xmath523 with @xmath524.\label{eq : trel}\end{aligned}\ ] ] 1 .
define a set of interacting orbitals , @xmath28 .
2 . determine the extensions of em , er , ir , and ai ; with this information subdivide the total @xmath14 and @xmath25 matrices according to eq .
( [ eq : ssmall ] ) , and set the basic projection matrices @xmath43 , @xmath420 , and @xmath421 .
3 . calculate @xmath526 and @xmath527 .
4 . calculate @xmath441 , and with it @xmath442 by solving the eigenvalue system @xmath443 .
then evaluate @xmath528 .
5 . with the stored quantities
calculate @xmath529 and @xmath88 , by using eqs .
( [ eq : wai ] ) and ( [ eq : wni ] ) ; with these quantities the total projection matrix , @xmath72 , is known through eq .
( [ eq : wshort ] ) . 6 .
the inverse projection matrix , @xmath530 , is obtained from eq .
( [ eq : winvsimple ] ) . 7 . required blocks of the transformed matrices , @xmath531 and @xmath192 , can then be calculated with the relations ( [ eq : smatexp ] ) and ( [ eq : hmatexp ] ) . 8 .
the hybridization function of the ai , @xmath481 , is obtained by using eq .
( [ eq : bardelta2 ] ) , and its green s function is obtained from eq .
( [ eq : gbar2 ] ) if needed
optionally , the asymptotic limits of @xmath481 for large energies are obtained from eqs .
( [ eq : limitsg][eq : m2 ] ) and eq .
( [ eq : bardelta2 ] ) . 10 .
once the many - body self - energy of the ai , @xmath532 , is calculated , its inverse projection is obtained by using eq .
( [ eq : sigmambinv ] ) ; this then allows to evaluate the full interacting gf , @xmath533 , of the em in the original basis by using eq .
( [ eq : gmb ] ) .
note however that to evaluate the transmission this inverse projection is not required .
j.j.parks , a.r .
champagne , t.a .
costi , w.w .
shum , a.n .
pasupathy , e. neuscamman , s. flores - torres , p.s .
cornaglia , a.a .
aligia , c.a .
balseiro , g.k .- l .
chan , h.d .
abruna , d.c .
ralph , science * 328 * , 1370 ( 2010 ) .
j. ferrer , c.j .
lambert , v. m garcia - suarez , d.zs .
manrique , d. visontai , l. oroszlany , r. rodriguez - ferradas , i. grace , s.w.d .
bailey , k. gillemot , hatef sadeghi and l.a .
algharagholy , new j. phys .
* 16 * , 093029 ( 2014 ) . n. blmer in _ the lda+dmft approach to strongly correlated materials _ , edited by e. pavarini , e. koch , d. vollhardt and a. lichtenstein , reihe modeling and simulation , vol . 1 ( verlag des forschungszentrum jlich , 2011 ) , open access at http://www.cond - mat.de / events / correl11/manuscripts/.
|
we present a computational method to quantitatively describe the linear - response conductance of nanoscale devices in the kondo regime .
this method relies on a projection scheme to extract an anderson impurity model from the results of density functional theory and non - equilibrium green s functions calculations .
the anderson impurity model is then solved by continuous time quantum monte carlo .
the developed formalism allows us to separate the different contributions to the transport , including coherent or non - coherent transport channels , and also the quantum interference between impurity and background transmission .
we apply the method to a scanning tunneling microscope setup for the 1,3,5-triphenyl-6-oxoverdazyl ( tov ) stable radical molecule adsorbed on gold .
the tov molecule has one unpaired electron , which when brought in contact with metal electrodes behaves like a prototypical single anderson impurity .
we evaluate the kondo temperature , the finite temperature spectral function and transport properties , finding good agreement with published experimental results .
|
in the past two years several narrow hadronic resonances have been discovered : new bottomonium and charmonium states , the @xmath0 and at least two new @xmath1 mesons . it is more than twenty years ago that new states with widths @xmath2 mev have been seen last , in the @xmath3 system .
this is an exciting situation as only a rather small number of discovered hadrons , including most notably quarkonia , the hydrogen of qcd , are narrow .
what makes this new era of hadron spectroscopy even more interesting is that many of these previously overlooked states appear to require more qcd than is allowed within simple quark - model @xmath4 mesons or @xmath5 baryons : at least some of the new states constitute the anti - thesis to the `` hydrogen of qcd '' .
if confirmed in a high statistics experiment , the @xmath6 pentaquark baryon and possibly other exotic baryons will add even more to this excitement .
lattice qcd is ideally positioned to compute the spectrum of reasonably stable hadrons as well as of non - perturbative properties relating to their internal structure .
the above described experimental discoveries were paralleled by significant advances in lattice methods , enabling computer simulations of qcd to become a precision predictive tool . in many areas
there are also lessons to be learned from combining lattice studies with effective field theory ( eft ) methods and/or model assumptions .
we have also witnessed experimental progress in the study of the spectrum of excited baryons and of electromagnetic transition form factors . generalised parton distributions ( gpds ) of the proton are now being studied intensively . as a stable particle
the nucleon lends itself to lattice studies . in the case of spin - independent structure functions
it will be hard for lattice simulations to compete with the experimental precision .
however , in addition to the theoretical satisfaction of verifying experimental measurements , this provides an ideal test ground for the methods and approximations employed in lattice studies . once spin and transversity
are included into the description of the nucleon , the experimental situation is far less clean and here there is real potential for lattice prediction rather than postdiction .
i will describe the present state of the field , interpret recent pentaquark and baryon mass calculations and briefly survey studies on the form and structure of baryons .
qcd can be regularized by introducing a space - time lattice cut - off @xmath7 .
the qcd coupling and @xmath8 quark masses whose values are not predicted by qcd should then be matched to reproduce @xmath9 experimental measurements of hadronic properties , for instance hadron masses .
everything else is a prediction and in this sense lattice qcd is a _ first principles _ approach .
the confinement of colour implies that finite size effects are usually tiny , as long as the spatial box extent @xmath10 .
we are fortunate to find that lattice spacings @xmath11 4 gev are sufficiently small to allow for controlled continuum limit extrapolations , @xmath12 .
this means that @xmath13 is sufficient , which makes qcd tractable on computers .
system with sea quarks @xcite.,title="fig : " ] + system with sea quarks @xcite.,title="fig : " ] + on a lattice with @xmath14 points , the lattice dirac operator is a huge matrix of dimension @xmath15 .
the inversion of this operator represents the major computational task of lattice qcd and this makes simulations incorporating sea quarks expensive . the algorithmic cost explodes with small @xmath16 masses , @xmath17 .
a smaller @xmath18 also requires a larger spatial lattice volume and the scaling behaviour , keeping @xmath19 fixed , is even worse : @xmath20 .
this is the main reason why many simulations are performed in the quenched approximation including only valence quarks and neglecting the polarization of the qcd vacuum due to sea quarks . while this approximation violates unitarity and does not even qualify as a quantum field theory , light hadron masses seem to agree within 10 % with experiment @xcite , indicating that the main effect of quark loops can be absorbed into redefinitions of the bare parameters of the theory .
the quenched approximation however goes terribly wrong at least in the scalar and pseudoscalar flavour - singlet sectors . in figure
[ fig : spin ] we display recent @xmath21 qcd results obtained by the lhp and sesam collaborations @xcite on the quark contribution @xmath22 to the proton spin , in the @xmath23 scheme at a scale @xmath24 gev .
the normalization is such that @xmath25 , where @xmath26 is the contribution from the quark angular momentum and @xmath27 from the gluons .
similar results have been obtained by the qcdsf collaboration @xcite . in these simulations , @xmath28 mev .
obviously , for infinite quark masses we expect @xmath29 .
it is therefore not surprising that the experimental value is overestimated and it is clear that smaller quark masses are absolutely essential to allow for a meaningful chiral extrapolation .
fortunately , with the advent of new fermion formulations @xcite that respect an exact lattice chiral symmetry , a reduction of the quark mass towards values @xmath30 mev has become possible @xcite , albeit so far only in the quenched approximation . in many lattice calculations
it is sufficient to calculate quark propagators that originate from a fixed source point . in these cases
only one column of the inverse dirac matrix needs to be calculated , navely reducing the effort by a factor @xmath31 in some cases diagrams with disconnected quark lines are needed .
examples are the physics of flavour singlet mesons , strong decays as well as parton distributions . in the latter case the complication can be avoided by assuming @xmath32 isospin symmetry and only calculating differences between @xmath33 and @xmath34 quark distributions .
disconnected quark lines require all - to - all propagators and hence a complete inversion of the dirac matrix appears necessary .
this turns out to be prohibitively expensive in terms of memory and computer time .
fortunately , sophisticated noise reduced stochastic estimator techniques have been developed over the past few years and as a result tremendous progress was achieved .
one such benchmark is the qcd string breaking problem , @xmath35 , where @xmath36 and @xmath37 is a static quark .
this represents one of the cleanest examples of a strong decay . within both ,
the transition matrix element as well as the @xmath38 state all - to - all propagators are required . in figure
[ fig : break ] we display the result of a recent sesam collaboration study @xcite with @xmath21 , @xmath39 and @xmath40 fm .
an extrapolation to physical light quark masses yields a string breaking distance @xmath41 fm .
the gap between the two states in the string breaking region is @xmath42 mev and we are able to resolve this with a resolution of 10 standard deviations ! in conclusion , all the long standing `` killers '' @xmath43 , @xmath44 and all - to - all propagators have been successfully tackled .
however , we are still a few years away from overcoming combinations of two of these simultaneously and possibly up to a decade separates us from precision simulations of flavour singlet diagrams with realistically light sea quarks .
not all these ingredients are always required at the same time . while ever bigger computers are an absolute necessity , most of the recent progress would have been impossible without novel methods .
the gain factor from faster computers was almost 5,000 over the past 15 years .
the factor from theoretical and algorithmic advances is harder to quantify .
qcd goes beyond the quark model and hence hadronic states that do not fit into a nave quark model of @xmath4 mesons and @xmath5 baryons are of particular interest .
many of the observed hadrons will contain considerable higher fock state components .
obviously , quantum numbers that are incomprehensible with a quark model meson or baryon interpretation provide us with the most clean - cut distinction . such examples do exist in the review of particle properties , namely the @xmath45 mesons @xmath46 and @xmath47 .
the minimal configuration required to obtain a vector state with positive charge either consists of two quarks and two antiquarks ( tetraquark / molecule ) or of quark , antiquark and a gluonic excitation ( hybrid meson ) .
these resonances are rather broad with widths @xmath48 300 mev . however , the ratio @xmath49 is very much the same as for the established @xmath50 vector meson .
another clear indication of an @xmath51 nature would for instance be a baryonic state with strangeness @xmath52 .
the minimal quark configuration in this case consists of five quarks ( pentaquark ) : @xmath53 . over the past two years several experiments
have presented evidence of a very narrow @xmath54 resonance @xcite , with decay @xmath55 and @xmath56 .
the parity has not yet been established .
however the mass is about 100 mev above the @xmath57 threshold and for @xmath58 an @xmath59-wave decay is possible , which might be difficult to reconcile with a width @xmath60 mev .
for @xmath61 a @xmath62-wave is required , still a bit puzzling but less so .
as the main decay channel does not require quark - antiquark pair creation one might hope to gain some insight from quenched lattice simulations and several attempts have been made @xcite .
two groups @xcite also investigated charmed pentaquarks and two studies @xcite incorporated the @xmath63 sector , in addition to @xmath64 .
two groups @xcite employed chiral overlap fermions while the others used conventional wilson - type lattice quarks .
only the kentucky group @xcite went down to @xmath30 mev while all other @xmath16 masses were larger than 400 mev .
the budapest - wuppertal group @xcite varied the lattice spacing and attempted a continuum - limit extrapolation . in all studies the negative parity mass came out to be lighter than the positive parity one , which is expected in the heavy quark limit .
there are two crucial questions to be asked : what happens when realistically light quark masses are approached ?
do we see resonant or scattering states ? resolving a resonance sitting on top of a tower of @xmath57 scattering states with different relative momenta appears rather hopeless at first .
however , there are two discovery tools available : variation of the lattice volume and of the creation operator . by varying the volume ( and the boundary conditions @xcite )
one will change the spectrum of @xmath57 scattering states as well as the coupling of a given operator to @xmath57 ( the spectral weight ) .
for the @xmath61 state which can only decay into a @xmath62-wave , the mass of the scattering state will depend on the lattice size since the smallest possible non - vanishing lattice momentum is @xmath65 . for @xmath66
the volume dependence of the lowest scattering state mass will be weak , however , the scaling of the spectral weight with the volume provides us with additional information .
mass dependence of @xmath64 @xmath53 `` pentaquark '' masses @xcite.,title="fig : " ] mass dependence of @xmath64 @xmath53 `` pentaquark '' masses @xcite.,title="fig : " ] + it turns out that the situation on the lattice is at least as ambiguous as the one encountered in experiment . to demonstrate this we display some kentucky - washington results @xcite in figure [ fig : penta ] .
it appears that the @xmath66 state dominantly couples to an @xmath59-wave @xmath57 .
the @xmath61 displays the qualitative volume dependence of a @xmath62-wave , however , it does not share its non - interacting mass .
it is most likely a @xmath62-wave scattering state .
this interpretation is supported by the observed volume dependence of the spectral weight . at very small
@xmath18 the situation becomes further complicated by the fact that there is no axial anomaly in the quenched approximation .
hence the flavour singlet @xmath67 is degenerate with the @xmath16 .
as this contribution comes in with a negative spectral weight , it is sometimes labelled a `` ghost '' .
the @xmath61 state can contain such a @xmath68 @xmath59-wave ( dashed lines ) .
the @xmath66 state becomes indistinguishable from a @xmath57 @xmath59-wave as the quark mass is reduced .
this does not conclusively exclude the possible existence of a nearby resonant state which might only couple very weakly to the creation operator used in this particular study . in order to draw more definite conclusions a variation of the creation operator as well as of the volume appears necessary , which is a very ambitious project @xcite .
if the pentaquark really was such a narrow resonance as some experiments suggest then maybe a lattice operator can be constructed that has a large overlap with this state but only a very small coupling to @xmath57 .
lattice studies of diquark interactions in a simplified , more controlled environment represent an alternative strategy to the brute force simulation of unstable states .
a baryon with one static and two light quarks constitutes one such arena .
one can of course also investigate multiquark interactions in the nonrelativistic limit of infinitely heavy quark masses .
such tetra- and penta - quark potentials have been studied recently by two groups @xcite and the results should provide model builders with some insight .
however , it is not clear how to relate these findings to the light quark limit in which chiral symmetry appears to play a bigger rle than instantaneous confining forces .
there exist quite a few narrow resonances very close to strong decay thresholds like the @xmath69 , the recently discovered @xmath70 charmonium state and the @xmath71 system .
it is very conceivable that such states contain a sizable multiquark component .
the question then arises if these are would - be quark model states or if these are true molecules / multiquark - states , that appear _ in addition _ to the quark model spectrum .
a fantastic arena to address this was provided by the recently discovered ( probably scalar ) @xmath72 and ( probably axialvector ) @xmath73 states .
first lattice studies @xcite have been performed , with somewhat contradictory interpretations of very compatible results .
one might hope that a similar lattice effort will also be dedicated on the comparatively cleaner and easier question of tetraquarks as has been on pentaquarks .
the spectrum of baryons has attracted renewed experimental and theoretical interest in recent years .
there is the question if the states can be understood in terms of quark models and if so by what sort of interaction and assumptions .
do gluonic excitations or pentaquark components play a rle for instance in the roper resonance ? what can we learn about quark - quark interactions within bound states ?
quark model predictions are somewhat obscured as the corresponding decay widths set a limit on the precision that can be expected for the resulting masses .
are missing states really `` missing '' or are they just obliterated due to the presence of many very broad , overlapping resonances ?
strongly decaying hadrons also pose problems in lattice simulations . at present almost all calculations of baryonic resonances
have been performed within the quenched approximation in which these are stable and hence the problem is circumvented .
this limitation can also be viewed as a virtue since most models suffer from an omission of quark pair creation effects too .
comparison with similar lattice results then allows to establish the validity range and applicability of a particular phenomenological model .
one strength of lattice methods is that simulations are not limited to the quark mass parameters found in nature .
investigating the quark mass dependence of results is a powerful tool . in the limit of large quark masses
one would expect to make contact with non - relativistic quark models while as @xmath74 overlap with chiral perturbation theory ( @xmath75pt ) predictions should be verifiable .
based on the assumptions that qcd bound states are mesons and baryons , that there is a mass gap and spontaneous chiral symmetry breaking at zero quark mass , an effective low energy chiral effective field theory ( @xmath75eft ) can be derived in the spirit of the born - oppenheimer approximation .
this will , to leading order , describe interactions between the ( fast moving ) massless nambu - goldstone pions and other hadrons . in nature quarks and
thus pions are not massless and the leading mass corrections are formally of order @xmath76 where @xmath77 gev .
the number of terms explodes at higher orders and predictive power is eventually lost , unless @xmath18 is sufficiently small to allow for an early truncation .
lattice simulations with sea quarks have so far been limited to unrealistically heavy pions , heavier than about 400 mev .
only recently masses as low as 180 mev have become possible @xcite . to allow for a controlled extrapolation of lattice results to the physical region
it is mandatory to establish an overlap between simulation data and @xmath75pt expectations . in general the size of this window
will depend on the observable in question .
chiral lattice fermion actions will make such a comparison with @xmath75pt cleaner . only in this case
an exact version of chiral symmetry can be formulated at finite lattice spacing . with other fermion discretizations , strictly speaking , a comparison should only be attempted after extrapolating lattice results to the continuum limit . while even @xmath78 mev @xmath79 ( modulo the ambiguity of `` @xmath80 '' _ vs. _ `` @xmath81 '' ) such a pion is still doomed to `` see '' some of the internal structure of the proton , with an inverse charge radius of about 250 mev .
hence it is doubtful that the quark and gluon nature of qcd can completely be ignored with such a `` hard '' pion probe .
navely , hadron masses are a polynomial in the quark mass , @xmath82 . however , pion loops give rise to a non - analytic functional dependence on the quark mass .
for instance the nucleon mass is given by , @xmath83m_{\pi}^4+a_5m_{\pi}^5+\cdots,\ ] ] with a renormalization scale @xmath84 .
the coefficients of the non - analytic terms can be related to phenomenological low energy constants .
for instance @xmath85 . apart from such constants , @xmath86 and @xmath87 contain terms @xmath88 and @xmath87 a contribution @xmath89 . in the quenched approximation ,
the leading non - analytic term is not proportional to @xmath90 but to @xmath91 , due to the @xmath67 becoming an additional goldstone pion . in figure
[ fig : meinulf ] we show a comparison @xcite between @xmath21 lattice data of the nucleon mass and the relativistic @xmath75pt expectation eq . ( [ eq : fit ] ) @xcite .
the lattice results were obtained by the cp - pacs and jlqcd collaborations @xcite as well as by the ukqcd and qcdsf collaborations @xcite , with ( non - chiral ) wilson - type fermions , on relatively fine lattices , @xmath92 fm and large volumes , @xmath93 .
the low energy parameters were fixed to phenomenological values and the fit comprises only of @xmath94 and @xmath95 . the quantitative agreement between curve and data for @xmath96 is accidental @xcite . a nave polynomial fit to the simulation data results in a nucleon mass much larger than the experimental value : lower order @xmath75pt only becomes applicable at smaller quark masses .
efts are based on the separation of scales .
if the @xmath75eft however is regulated in dimensional regularization then @xmath16 loop - integrals can receive significant contributions from momenta @xmath97 . to enhance the convergence the adelaide group @xcite suggested a `` finite range regularization '' approach which amounts to introducing a momentum cut - off which is then varied to achieve `` model independence '' . a hard cut - off
can also be provided by lattice regularization of the chiral expansion @xcite .
needless to say that all cut - off and scheme dependence will disappear at sufficiently high orders in the @xmath98-expansion . during the past four years we witnessed many lattice publications on the spectrum of excited baryons @xcite .
while the extraction of a ground state mass is relatively straight forward , radial excitations either require high statistics and some confidence in the fitting procedures or the design of sophisticated , non - local creation operators @xcite .
all but one study @xcite have been performed in the quenched approximation . only the lhpc - ukqcd - qcdsf collaborations @xcite , using the wilson - clover action , attempted a continuum limit extrapolation .
other strategies were implementations of improved wilson - type actions like the @xmath99 action by lee
_ et al . _
@xcite or the flic action by melnitchouk _
et al . _
@xcite as well as a recent study with wilson - clover fermions @xcite .
the bgr collaboration employed chirally improved fermions @xcite , the riken - bnl group used chiral domain wall fermions @xcite and the taiwan @xcite as well as the kentucky - washington groups @xcite made use of overlap fermions .
early articles shared the observation of the positive parity state being much heavier than the roper @xmath100 resonance while the negative parity ground state was compatible with the orbital excitation , @xmath101 .
one explanation would be that the resonance observed in nature might have little overlap with the dominantly @xmath5 state created on the lattice ; pentaquark or gluon components might be necessary . alternatively , maybe one should not take the exact position of a resonance with a width of @xmath102 overly seriously . , @xmath103 gev.,title="fig : " ] + , @xmath103 gev.,title="fig : " ] + new results obtained by the kentucky - washingon group @xcite and the bgr collaboration @xcite at lighter pion masses , @xmath104 mev and @xmath105 mev , respectively , are however compatible with experiment .
we display these in figures [ fig : dong ] and [ fig : lang ] .
rather unsurprisingly the creation operator requires a node in its spatial wave function to produce a significant overlap with the radial excitation .
this complication was circumvented in ref .
@xcite by a sophisticated fitting procedure while in ref .
@xcite such an adequate operator has been constructed , using a variational principle . making contact with
the light quark regime seems hopeless for @xmath106 mev .
however , the authors of two recent studies @xcite manage to extrapolate their results to the experimental values from such pion masses as well . note that the bgr collaboration only sees a clear signal of the radial excitation for @xmath107 mev while the bayesian fitting procedure of the kentucky group yields results at any quark mass .
the taiwan group @xcite in addition predicts the spectrum of doubly charmed baryons , with findings roughly compatible with earlier studies @xcite as well as with the selex candidate(s ) .
one might hope that in the near future the @xmath108 state can cleanly be disentangled from a @xmath62-wave @xmath109 scattering state or , at very small masses , from @xmath59-wave @xmath110/@xmath111 states . by studying the volume dependence and spectral weights
the kentucky group has taken steps in this direction .
most of the lattice studies include additional baryonic resonances where similar problems need to be addressed . in conclusion , we are close to an understanding of the transition between the heavy and light quark limits in the quenched approximation , an information invaluable to model builders .
quite a few results on moments of gpds , most notably from the qcdsf @xcite , sesam and lhp collaborations @xcite , exist .
a nice review of the state - of - the - art concerning spin - independent parton distributions and the axial charge can be found in ref .
the main problem here is an overestimation by about 60 % of @xmath112 if linearly extrapolated in @xmath113 , relative to experiment .
it is not clear whether this difference will reduce as the quark mass is decreased or if this has to do with the non - chiral fermion formulation used .
this issue will be clarified in the near future @xcite .
there has also been progress in resolving the momentum dependence of electromagnetic @xmath114 transition form factors in a quenched study @xcite , in the region @xmath115 .
the magnetic dipole form factor is significantly overestimated at large @xmath116 , due to the unrealistically small charge radii of @xmath117 and @xmath118 .
one might hope that such effects cancel in part from form factor ratios .
@xmath119 is fairly constant at -0.02(1 ) , once extrapolated to the chiral limit .
in contrast , @xmath120 decreases monotonously from -0.01(1 ) at 0.1 gev@xmath121 down to -0.09(3 ) at @xmath122 .
this behaviour is in good agreement with @xmath123 clas data while it is hard to reconcile with the oops point @xmath124\,\%$ ] at @xmath125 .
many lattice studies now include sea quarks . within the quenched approximation
, light quark masses close to the physical limit have been realised and the lattice provides a powerful tool for exploring the validity range of chiral expansions .
lattice pentaquark studies still yield ambiguous results .
a systematic study using a large set of creation operators , of lattice volumes , spacings and quark masses is possible but ambitious .
latest lattice data suggest that the mass of the roper resonance can be reproduced in the quenched approximation , thus indicating a non - exotic leading fock component .
a lot of progress has been made in understanding the structure of the nucleon and there is a strong push towards reducing the lattice quark masses , closer to the physical limit .
i thank thomas hemmert for useful comments . this work is part of the ec hadron physics i3 contract no .
rii3-ct-2004 - 506078 .
gb is supported by a pparc advanced fellowship ( grant ppa / a / s/2000/00271 ) as well as by pparc grant ppa / g/0/2002/0463 .
99 c. gattringer _
et al . _ [ bgr collab . ] , nucl .
b 677 ( 2004 ) 3 .
j. w. negele _ et al .
_ , nucl .
phys . proc .
suppl . 128 ( 2004 ) 170 .
g. s. bali , t. dssel , t. lippert , h. neff , z. prkacin and k. schilling [ sesam collab .
] , arxiv : hep - lat/0409137 and in preparation .
m. gckeler _ et al .
_ [ qcdsf collab . ] , phys .
lett . 92 ( 2004 ) 042002
. h. neuberger , phys .
b 417 ( 1998 ) 141 and references therein .
n. mathur _ et al .
_ , phys .
b 605 ( 2005 ) 137 .
see e.g. t. nakano , these proceedings ; m.v .
polyakov , these proceedings .
f. csikor , z. fodor , s. d. katz and t. g. kovacs , jhep 0311 ( 2003 ) 070 .
s. sasaki , phys .
lett . 93 ( 2004 ) 152001
. t. w. chiu and t. h. hsieh , arxiv : hep - ph/0403020 and arxiv : hep - ph/0501227 . n. mathur _ et al .
_ , phys .
d 70 ( 2004 ) 074508 .
n. ishii _ et al .
_ , phys .
d 71 ( 2005 ) 034001 ; arxiv : hep - lat/0501022 . c. alexandrou , g. koutsou and a. tsapalis , arxiv : hep - lat/0409065 .
g. t. fleming , arxiv : hep - lat/0501011 . c. alexandrou and g. koutsou , phys .
d 71 ( 2005 ) 014504 .
f. okiharu , h. suganuma and t. t. takahashi , arxiv : hep - lat/0407001 .
g. s. bali , phys .
d 68 ( 2003 ) 071501 .
a. dougall _ et al .
_ , phys .
b 569 ( 2003 ) 41 .
m. procura , t. hemmert and w. weise , phys .
d 69 ( 2004 ) 034505 .
m. gckeler , arxiv : hep - lat/0412013 .
a. ali khan _ et al . _
[ cp - pacs collab . ] , phys .
d 65 ( 2002 ) 054505 ; _ erratum ibid _
d 67 ( 2003 ) 059901 ; s. aoki _ et al .
_ [ jlqcd collab . ] , phys .
d 68 ( 2003 ) 054502 . c. r. allton _ et al .
_ [ ukqcd collab . ] , phys . rev .
d 65 ( 2002 ) 054502 ; a. ali khan _ et al .
_ [ qcdsf and ukqcd collabs . ] , nucl .
b 689 ( 2004 ) 175 .
d. b. leinweber , these proceedings ; r. d. young _ et al .
_ , phys .
d 66 ( 2002 ) 094507 ; d. b. leinweber _ et al .
_ , arxiv : hep - lat/0501028 . i. a. shushpanov and a. v. smilga , phys .
d 59 ( 1999 ) 054013 .
r. lewis and p. p. ouimet , phys .
d 64 ( 2001 ) 034005 ; b. borasoy and r. lewis , phys .
d 71 ( 2005 ) 014033 .
see e.g. d. b. leinweber _ et al .
_ , arxiv : nucl - th/0406032 . t. burch _ et al . _ [ bgr collab . ] , phys . rev .
d 70 ( 2004 ) 054502 .
d. brommel _ et al .
_ , phys .
d 69 ( 2004 ) 094513 .
t. burch _ et al .
_ , arxiv : nucl - th/0501025 .
c. maynard and d. richards [ ukqcd ] , nucl .
suppl . 119 ( 2003 ) 287 .
m. gckeler _ et al .
_ [ lhp - ukqcd - qcdsf collabs . ] phys .
b 532 ( 2002 ) 63 .
f. x. lee
_ et al . _ , nucl .
suppl . 106 ( 2002 ) 248 .
w. melnitchouk _ et al .
_ , phys .
d 67 ( 2003 ) 114506 ; j. m. zanotti _ et al .
_ [ cssm lattice collab . ] , phys .
d 68 ( 2003 ) 054506 .
d. guadagnoli _ et al .
_ , phys .
b 604 ( 2004 ) 74 and these proceedings .
s. sasaki , t. blum and s. ohta , phys .
d 65 ( 2002 ) 074503 ; s. sasaki , prog .
suppl . 151 ( 2003 ) 143 .
t. w. chiu and t. h. hsieh , these proceedings : arxiv : hep - lat/0501021 ; t. w. chiu and t. h. hsieh , arxiv : hep - lat/0406016 .
r. lewis _ et al .
d 64 ( 2001 ) 094509 and d 66 ( 2002 ) 014502 .
j. flynn , f. mescia and a. tariq [ ukqcd collab .
] , jhep 0307 ( 2003 ) 066 .
m. gckeler
_ et al . _
[ qcdsf collab .
] , phys .
lett . 92 ( 2004 ) 042002 ; arxiv : hep - lat/0409162 ; arxiv : hep - lat/0501029 .
p. hgler _ et al .
_ [ lhp and sesam collabs . ] , phys .
d 68 ( 2003 ) 034505 ; phys .
93 ( 2004 ) 112001 ; arxiv : hep - ph/0410017
. w. schroers , these proceedings : arxiv : hep - ph/0501156 .
m. grtler _ et al .
_ arxiv : hep - lat/0409164 ; r. horsley , arxiv : hep - lat/0412007 . c. alexandrou
_ , phys .
lett . 94 ( 2005 ) 021601 .
|
i comment on progress of lattice qcd techniques and calculations .
recent results on pentaquark masses as well as of the spectrum of excited baryons are summarizned and interpreted .
the present state of calculations of quantities related to the nucleon structure and of electromagnetic transition form factors is surveyed .
|
the evolution of the internet and the resulting large communication , information and social networks has significantly impacted recent advances in graph theory .
the _ clustering coefficient _ of a user in a social network is the proportion of pairs of its friends that are themselves friends .
mean clustering coefficient _ for the network is the average clustering coefficient taken over all users .
this concept was introduced by watts and strogatz @xcite to determine whether a given network is a ` small world ' network .
it was reported in ugander et .
al @xcite that the facebook graph is locally dense but globally sparse .
indeed the clustering coefficient of vertices of small degree is close to @xmath11 and steadily decreases with increasing degree . in this paper
we focus on the global cycle structure of graphs that are locally sufficiently dense . unless specified otherwise , all graphs under consideration are assumed to be connected and of order at least @xmath12 . for graph theory terminology
not introduced here we follow @xcite .
let @xmath1 be a graph and @xmath13 a vertex of @xmath1 .
then the _ neighbourhood _ of @xmath13 , denoted by @xmath14 , is the set of all vertices adjacent with @xmath13 and the closed neighbourhood of @xmath13 is the set @xmath15 = n(v ) \cup\{v\}$ ] .
if @xmath5 is a set of vertices in a graph , then the _ subgraph induced _ by @xmath5 , denoted by @xmath16 , is the subgraph of @xmath1 with vertex set @xmath5 such that two vertices of @xmath5 are adjacent in @xmath16 if and only if they are adjacent in @xmath1 .
let @xmath0 be a graph property .
a graph @xmath1 is said to be _ locally _ @xmath0 ( _ closed locally _
@xmath17 ) , if the subgraph induced by the open neighbourhood ( closed neighbourhood , respectively ) of every vertex in @xmath1 has property @xmath0 .
the _ clustering coefficient _ of a vertex @xmath13 of degree @xmath18 , denoted by @xmath19 , is the ratio of the number of edges in the subgraph induced by @xmath14 to the maximum number of edges in a @xmath20-vertex graph , i.e , @xmath21 .
the _ minimum clustering coefficient _ of a graph @xmath1 with minimum degree @xmath22 , denoted by @xmath23 , is the smallest clustering coefficient among all vertices of @xmath1 . in the sequel
we assume that all graphs under consideration have minimum degree at least @xmath24 .
if the minimum clustering coefficient of a class of graphs , is bounded by a constant @xmath25 , @xmath26 , then these graphs are said to be _ locally dense_. we observe that there are classes of graphs , although being locally dense , may be globally sparse with large diameter .
for example , the class of graphs that are strong products ( see @xcite ) of a path @xmath27 of order @xmath28 and the complete graph @xmath29 of order @xmath24 , denoted by @xmath30 , has minimum clustering coefficient at least @xmath11 , is sparse since @xmath31 as @xmath32 and has diameter @xmath33 . in this paper
we show that connected locally connected graphs with bounded maximum degree , for which the minimum clustering coefficient is at least 1/2 , have a rich cycle structure . in order to make these notions more precise we begin with pertinent definitions and relevant background on cycle properties in graphs that possess certain local properties .
let @xmath1 be a graph of order @xmath34 .
then @xmath1 is _ hamiltonian _ if @xmath1 has a cycle of length @xmath34 .
if , in addition , @xmath1 has a cycle of every length from 3 up to @xmath34 , then @xmath1 is _ pancyclic _ , see @xcite . an even stronger notion than pancyclicity is that of ` full cycle extendability ' , introduced by hendry @xcite . a cycle @xmath7 in a graph @xmath1 is _ extendable _ if there exists a cycle @xmath35 in @xmath1 that contains all the vertices of @xmath7 and one additional vertex .
the graph @xmath1 is _ cycle extendable _ if every nonhamiltonian cycle of @xmath1 is extendable .
if , in addition , every vertex of @xmath1 lies on a 3-cycle , then @xmath1 is _ fully cycle extendable_. a condition weaker than pancyclicity has been the focus of a number of research articles .
let @xmath36 and @xmath37 denote , respectively , the length of a shortest cycle , called the _ girth _ of @xmath1 , and a longest cycle , called the _ circumference _ of @xmath1 .
then @xmath1 is called _ weakly pancyclic _ if @xmath1 has a cycle of every length between @xmath36 and @xmath37 . the problem of determining whether or not a graph has a hamiltonian cycle
is called the _ hamilton cycle
problem_. in 1974 , chartrand and pippert @xcite initiated the study of locally connected graphs .
cycle properties of these graphs with bounded maximum degree have since been studied extensively - see for example @xcite .
another class of graphs that has been widely studied in relation to the hamilton cycle problem , is the class of claw - free , i.e. @xmath38-free , graphs .
these are precisely the graphs @xmath1 for which @xmath39 for all @xmath40 .
thus claw - free graphs can also be described in terms of a local property .
the hamilton cycle problem is np - complete for both locally connected graphs , see @xcite , and for claw - free graphs , see @xcite .
the next result of oberly and sumner demonstrates the strength of combining these two local properties . [ oberly and sumner ] _ @xcite _ if @xmath1 is a connected , locally connected claw - free graph , then @xmath1 is hamiltonian .
clark @xcite showed that connected , locally connected claw - free graph are in fact pancyclic and hendry @xcite observed that clark had actually shown that these graphs are fully cycle extendable .
these results support bondy s well - known ` meta - conjecture ' that almost any condition that guarantees that a graph has a hamilton cycle actually guarantees much more about the cycle structure of the graph . if , in theorem [ oberly and sumner ] , the claw - free condition is dropped , hamiltonicity is no longer guaranteed .
in fact , pareek and skupin @xcite observed that there exist infinitely many connected , locally hamiltonian graphs that are nonhamiltonian .
however , clark s result led ryjek to suspect that every locally connected graph has a rich cycle structure , even if it is not hamiltonian .
he proposed the following conjecture ( see @xcite . )
[ ryjacek ] ( ryjek ) every locally connected graph is weakly pancyclic .
ryjek s conjecture seems to be very difficult to settle , so it is natural to consider weaker conjectures .
this conjecture has been studied , for example , for locally traceable and locally hamiltonian graphs with maximum degree at most 5 and 6 , respectively , see @xcite , neither of which need to be hamiltonian , see @xcite .
one may well ask whether there are local connectedness conditions that guarantee ( global ) hamiltonicity .
one such result was obtained by hasratian and kachatrian @xcite . recall that a graph @xmath1 of order @xmath41 has the ore property if @xmath42 for all pairs @xmath43 of non - adjacent vertices @xmath44 and @xmath13 of @xmath1 and it has the dirac property if @xmath45 for all @xmath40 .
it was shown in @xcite that closed locally ore graphs are hamiltonian and this result was subsequently strengthened in @xcite to full cycle extendability for these graphs .
thus , in particular , all closed locally dirac graphs are fully cycle extendable . if a graph is closed locally dirac , then the clustering coefficient of each vertex is at least @xmath11 .
this prompts the question : can the closed locally dirac condition can be weakened to requiring the graph to have minimum local clustering coefficient at least 1/2 while still guaranteeing hamiltonicity of the graph ?
we show in this paper that locally connected graphs with minimum clustering coefficient at least @xmath11 need not be hamiltonian . nevertheless our results demonstrate that these graphs have a rich cycle structure and they lend support to ryjek s conjecture .
the hamilton cycle problem for graphs with small maximum degree remains difficult , even when additional structural properties are imposed on the graph .
for example , the hamilton cycle problem is np - complete for bipartite planar graphs with @xmath46 ( see @xcite ) , for @xmath47-regular graphs for any fixed @xmath47 ( see @xcite ) and even for planar cubic 3-connected claw - free graphs ( see @xcite ) .
however , some progress has been made for locally connected graphs with small maximum degree .
the first result in this connection was obtained in @xcite where it was shown that every connected , locally connected graph with maximum degree @xmath48 is either hamiltonian or isomorphic to the complete @xmath12-partite graph @xmath49 .
all connected , locally connected graphs with maximum degree at most @xmath50 are described in @xcite .
apart from @xmath49 , all of these graphs are actually fully cycle extendable . since @xmath49 is weakly pancyclic , ryjek s conjecture holds for locally connected graphs with maximum degree at most 4 .
global cycle properties of connected , locally connected graphs with maximum degree 5 were investigated in @xcite .
collectively these results imply that every connected , locally connected graph with @xmath51 and @xmath52 is fully cycle extendable .
a graph @xmath1 is _ locally isometric _ if @xmath53 is an isometric subgraph ( distance preserving subgraph ) of @xmath1 for all @xmath40 .
it was shown in @xcite that the hamilton cycle problem is np - complete even for locally isometric graphs with maximum degree @xmath54 .
nevertheless , it was shown in @xcite that ryjek s conjecture holds for all locally isometric graphs with maximum degree 6 and without true twins ( i.e. , pairs of vertices having the same closed neighbourhood ) ; these graphs are in fact fully cycle extendable .
let @xmath55 be vertices of a graph @xmath1 and @xmath5 a set of vertices of @xmath1 . if @xmath56 is adjacent with @xmath57 ( or every vertex of @xmath5 ) we write @xmath58 ( or @xmath59 , respectively ) .
we use @xmath60 to indicate that @xmath56 is not adjacent with any vertex of @xmath5 .
in this section we obtain a structural characterization of those connected , locally connected graphs with minimum clustering coefficient @xmath11 and maximum degree at most 6 that are fully cycle extendable . to state this characterization
we introduce some useful definitions .
let @xmath3 be a subgraph of a graph @xmath1 and let @xmath4 .
we say that @xmath3 is a _ strongly induced subgraph _ of @xmath1 with _ attachment set _
@xmath5 if @xmath3 is an induced subgraph of @xmath1 and the vertices of @xmath6 are not incident with edges that do not belong to @xmath3 , i.e. , only the vertices of @xmath5 may be incident with edges of @xmath1 that are not in @xmath3 .
let @xmath1 be a connected , locally connected graph with minimum clustering coefficient @xmath62 . if @xmath1 has maximum degree @xmath63 or @xmath12 , then @xmath1 is isomorphic to @xmath64 or either @xmath65 or @xmath66 , respectively ( where @xmath67 is an edge of @xmath65 ) .
so @xmath1 is fully cycle extendable . if @xmath68 , then it is readily seen that @xmath1 is fully cycle - extendable unless @xmath69 . for the remainder of this section we focus on graphs with @xmath70 .
let @xmath71 be a @xmath72-cycle in a graph @xmath1 .
if @xmath73 and @xmath74 , then @xmath75 and @xmath76 denote , respectively , the paths @xmath77 and @xmath78 ( subscripts expressed modulo @xmath72 ) .
let @xmath79 be a non - extendable cycle in a graph @xmath1 .
with reference to a given non - extendable cycle @xmath7 , a vertex of @xmath1 will be called a _ cycle vertex _ if it is on @xmath7 , and an _ off - cycle _ vertex if it is in @xmath80 .
a cycle vertex that is adjacent to an off - cycle vertex will be called an _
attachment vertex_. the following useful result was established in @xcite .
[ l1 ] let @xmath81 be a non - extendable cycle of length @xmath72 in a graph @xmath1 .
suppose @xmath82 and @xmath83 are two distinct attachment vertices of @xmath7 that have a common off - cycle neighbour @xmath56 . then the following hold .
( all subscripts are expressed modulo @xmath72 . )
* @xmath84 and @xmath85 .
* @xmath86 and @xmath87 . * if @xmath88 , then @xmath89 and @xmath90 . * if @xmath91 , then @xmath92 does not have two adjacent neighbours @xmath93 on the path @xmath94 .
with specified attachment sets @xmath95 for @xmath96,width=480 ] the graphs of figure [ forbidden_strong_induced_subgraphs ] show forbidden strong induced subgraphs @xmath97 with specified attachment sets @xmath95 , @xmath96 .
let @xmath98 be the graph @xmath99 and let @xmath100 , @xmath101 and @xmath102 be the graphs shown in figure [ forbidden_graphs ] .
the graphs @xmath97 , @xmath96 , and @xmath103 , @xmath104 , are referred to in lemma [ l2 ] and theorem [ t1 ] .
it is shown in theorem [ t1 ] that a locally connected graph with minimum clustering coefficient @xmath11 and maximum degree 5 or 6 is fully cycle extendable if and only if it is not isomorphic to any of the graphs @xmath103 , @xmath104 , and does not contain any of the @xmath105s as strong induced subgraphs for @xmath106 .
observe that if a connected graph can not be isomorphic to a graph @xmath103 , @xmath104 , this is equivalent to saying that it can not have @xmath103 as strong induced subgraph with empty attachment set .
[ l2 ] let @xmath1 be a connected locally connected graph with maximum degree @xmath107 and minimum clustering coefficient @xmath108 .
let @xmath109 be a non - extendable , nonhamiltonian cycle in @xmath1 and let @xmath110 be an attachment vertex of largest possible degree .
if @xmath111 and @xmath110 has a true twin , then @xmath1 contains @xmath97 as a strong induced subgraph with attachment sets @xmath95 for some @xmath112 , @xmath113 as shown in fig .
[ forbidden_strong_induced_subgraphs ] , or @xmath114 .
first , assume @xmath115 .
since @xmath116 , it follows that @xmath117 .
since @xmath1 is locally connected and no off - cycle neighbour of @xmath110 is adjacent with @xmath118 or @xmath119 , by lemma [ l1](1 ) , @xmath110 has either one or two off - cycle neighbours .
let @xmath120 or @xmath121 , respectively , where @xmath56 and @xmath57 are off - cycle neighbours of @xmath110 and @xmath122 .
by lemma [ l1](1 ) , no off - cycle neighbour of @xmath110 can be a true twin of @xmath110 .
assume @xmath82 is a true twin of @xmath110 . in the case where @xmath110 has one off - cycle neighbour
, it follows that @xmath123 and @xmath124 . by lemmas [ l1](1 ) and [ l1](2 ) , @xmath125 and @xmath126 .
since @xmath127 , it follows that @xmath128 .
since @xmath1 is locally connected and @xmath129 , it follows that @xmath130 .
hence @xmath1 contains @xmath131 as a strong induced subgraph with attachment set @xmath132 . if @xmath110 has two off - cycle neighbours , @xmath133 . by lemmas [ l1](1 ) and [ l1](2 ) , @xmath134 and @xmath135 . since @xmath127
, we have @xmath58 . as @xmath136 and @xmath1 is locally connected , @xmath137 .
hence @xmath1 has @xmath131 as strong induced subgraph with attachment set @xmath138 . next
, assume @xmath139 .
since @xmath140 , it follows , that @xmath141 .
it now follows from the assumption that @xmath110 has a true twin , and by lemmas [ l1](1 ) , and [ l1](2 ) that @xmath110 can not have exactly two off - cycle neighbours ; otherwise @xmath142 .
thus @xmath110 has either exactly one or exactly three off - cycle neighbours .
let @xmath143 or @xmath144 , respectively , where @xmath55 and @xmath145 are off - cycle neighbours and @xmath146 .
if , in the first case , @xmath83 is a true twin of @xmath110 , it follows from lemmas [ l1](1 ) , [ l1](2 ) , and [ l1](3 ) , and the fact that @xmath147 that @xmath148 .
hence , we may assume , in either case , that @xmath82 is a true twin of @xmath110 .
if @xmath110 has exactly three off - cycle neighbours , then it follows since @xmath1 is locally connected and the facts that @xmath147 and @xmath127 and by lemmas [ l1](1 ) and [ l1](2 ) , that @xmath133 .
thus @xmath1 contains @xmath149 as a strong induced subgraph with attachment set @xmath150 .
suppose now that @xmath110 has exactly one off - cycle neighbour .
since @xmath151 and @xmath110 and @xmath82 are true twins , @xmath123 and @xmath124 . by lemmas [ l1](1 ) , [ l1](2 ) ,
@xmath152 and @xmath153 .
we now consider four cases depending on which subset of @xmath154 the vertex @xmath155 is adjacent with . + * case 1 * @xmath156 .
since @xmath127 it follows that in this case , @xmath157 and @xmath128 . since @xmath158 , @xmath110 and @xmath159 do not have neighbours other than those already mentioned .
moreover , since @xmath56 and @xmath82 are adjacent with both @xmath110 and @xmath119 , and @xmath1 is locally connected , neither @xmath56 nor @xmath118 can have any other neighbours .
hence the subgraph induced by @xmath160 is a strong induced subgraph isomorphic to @xmath149 and with attachment set @xmath161 . + * case 2 * @xmath162 and @xmath163 .
then at least two of the three edges @xmath164 @xmath165 belong to @xmath1 .
suppose first that @xmath157 . since @xmath158 , either @xmath166 or @xmath167 .
we may assume the former ; the second case can be argued similarly .
if @xmath128 , then @xmath168 is an extension of @xmath7 .
so @xmath169 . since @xmath151 and @xmath1 is locally connected @xmath170 .
also @xmath171 , for if @xmath118 has a neighbour @xmath172 , then @xmath173 . note , by lemma [ l1](2 ) ,
thus @xmath57 is an off - cycle vertex .
since @xmath175 , @xmath176 and @xmath177 and @xmath178 is an extension of @xmath7 , which is not possible . we now show that @xmath179 .
suppose @xmath119 has a neighbour @xmath57 other than @xmath180 and @xmath155 .
since @xmath1 is locally connected and @xmath158 , @xmath181 . by lemma [ l1](1 ) , @xmath57 is not an off - cycle vertex .
thus @xmath182 ( and by assumption @xmath183 ) .
but now @xmath184 is an extension of @xmath7 .
hence @xmath185 .
so @xmath1 contains @xmath186 as strong induced subgraph with attachment set @xmath187 .
we now assume that @xmath128 and , by the above observation , @xmath155 is adjacent with exactly one of @xmath188 or @xmath119 , say @xmath189 .
the case where @xmath190 and @xmath191 can be argued similarly . since @xmath192 it follows that @xmath193 .
if @xmath194 , then @xmath7 can be extended to the cycle @xmath195 .
so @xmath196 . since @xmath151 and by lemma [ l1](4 ) @xmath197 . by lemma [ l1](1 )
it follows that @xmath155 and @xmath198 do not have a common off - cycle neighbour .
since @xmath199 is connected it now follows that @xmath200 and hence that @xmath201 .
but then @xmath202 is an extension of @xmath7 which is not possible . +
* case 3 * @xmath203 and @xmath204 .
then at least two of the three edges @xmath164 @xmath205 belong to @xmath1 .
suppose first that @xmath206 .
then either @xmath166 or @xmath167 .
we consider the case where @xmath166 . by the symmetry of the structure ,
the case where @xmath167 can be argued similarly . since @xmath166
, it follows from lemma [ l1](4 ) that @xmath169 . if @xmath183 , then it follows since @xmath199 ,
is connected and @xmath158 and by lemmas [ l1](1 ) and [ l1](2 ) , that @xmath207 and thus @xmath208 .
since @xmath1 is locally connected it can now be shown that @xmath209 and @xmath210 .
so the subgraph induced by @xmath211 is a strong induced subgraph isomorphic to @xmath186 with attachment set @xmath212 .
suppose now that @xmath213 .
so @xmath176 .
since @xmath214 does not have a common off - cycle neighbour with either @xmath188 or @xmath215 , @xmath216 .
also @xmath217 since @xmath1 is locally connected and @xmath218 .
hence @xmath1 contains the graph @xmath186 as strong induced subgraph with attachment set @xmath219 .
assume next that @xmath155 is not adjacent with both @xmath188 and @xmath119 .
then @xmath128 and @xmath155 is adjacent with exactly one of @xmath188 and @xmath119 .
we consider the case where @xmath190 ; the case where @xmath189 can be argued similarly . by lemma [ l1](4 )
@xmath193 . by lemma [ l1](1 ) and
the fact that @xmath151 , @xmath220 , and by lemmas [ l1](1 ) and [ l1](2 ) , @xmath221 .
so @xmath222 which is not possible . +
* case 4 * @xmath223 .
suppose first that @xmath166 or @xmath167 .
we consider the case where @xmath166 ; the case where @xmath167 can be argued similarly .
then , by lemma [ l1](4 ) , @xmath224 .
suppose @xmath225 . since @xmath158 and by lemmas [ l1](1 ) and [ l1](2 ) , @xmath226 .
since @xmath199 is connected , this is not possible .
hence @xmath177 . in this case
it can be argued , using lemma [ l1 ] and the fact that @xmath151 , that @xmath227 .
thus @xmath1 is isomorphic to the graph @xmath98 which is not possible .
assume now that @xmath228 , then by lemmas [ l1](1 ) , [ l1](4 ) and the fact that @xmath147 , we see that @xmath229 .
so @xmath222 , a contradiction .
[ t1 ] let @xmath1 be a locally connected graph with maximum degree @xmath230 or @xmath9 and minimum clustering coefficient @xmath108 .
then @xmath1 is fully cycle extendable if and only if @xmath1 is not isomorphic to @xmath103 , @xmath104 , and @xmath1 does not contain the graphs @xmath97 , with attachment sets @xmath95 , for @xmath96 , as strong induced subgraphs .
it is readily seen that if @xmath1 is one of the graphs @xmath103 , @xmath104 , then @xmath1 is not hamiltonian and hence not fully cycle extendable . to see this
observe that each of these graphs contains a vertex cutset @xmath231 such that the number of components of @xmath232 exceeds @xmath233 .
for @xmath98 , @xmath101 , and @xmath102 the set @xmath231 consisting of the vertices of degree @xmath9 is such a cutset and for @xmath100 the set @xmath231 consisting of the vertices of degree 5 is such a vertex cutset .
it follows that @xmath1 is not hamiltonian see @xcite .
if @xmath1 contains @xmath97 as a strong induced subgraph where @xmath234 with the attachment sets as shown in figure [ forbidden_strong_induced_subgraphs ] , then we can again see that @xmath1 contains a vertex cutset @xmath231 such that the number of components of @xmath232 exceeds @xmath233 , and hence @xmath1 is not hamiltonian . for @xmath131 ,
let @xmath231 consist of the vertices having degree 5 in @xmath131 , for @xmath149 let @xmath231 consist of the vertices having degree 6 in @xmath149 , for @xmath186 let @xmath231 consist of the vertices having degree 6 and the vertex having degree 5 in @xmath186 , for @xmath235 let @xmath231 consist of the three vertices having degree @xmath9 in @xmath235 and for @xmath236 let @xmath231 consist of the three vertices of degree 6 in @xmath236 . for the converse , assume that @xmath1 is not isomorphic to one of the graphs @xmath103 and that @xmath1 does not contain the graphs @xmath97 with attachement set @xmath95 as induced subgraphs , @xmath96 .
suppose , to the contrary , that @xmath237 is a non - extendable cycle in @xmath1 for some @xmath238 ( indices taken mod @xmath72 ) .
since @xmath239 and @xmath1 is connected , there must be a vertex of @xmath7 that is an attachment vertex .
we may assume that @xmath110 is an attachment vertex on @xmath7 of maximum degree .
let @xmath56 be an off - cycle neighbour of @xmath110 . since @xmath1 is locally connected and @xmath240 it follows that @xmath241 . since @xmath158 , @xmath242 .
+ * case 1 * suppose @xmath243 .
suppose first that @xmath244 and let @xmath245 .
since @xmath246 , @xmath247 must contain three edges . by lemma [ l1](1 )
@xmath248 , so @xmath249 and @xmath82 is adjacent to at least one of @xmath118 and @xmath119 , say @xmath250 . if @xmath251 , then by lemma [ l1](2 ) , @xmath252 and @xmath253 .
but then @xmath254 , contradicting the fact that @xmath110 is an attachment vertex of maximum degree .
so @xmath255 .
therefore @xmath256 . since @xmath82 has an off - cycle neighbour @xmath56 , @xmath257 by our choice of @xmath258 .
hence @xmath259 and so @xmath260 and @xmath261 .
therefore @xmath133 . by lemmas [ l1](1 ) and
[ l1](2 ) , there are no additional edges joining pairs of vertices of @xmath262 .
suppose @xmath263 .
then there exists a vertex @xmath264 adjacent with a vertex of @xmath7 .
since g is a connected graph , @xmath57 is adjacent to at least one vertex of @xmath262 . by our choice of @xmath110
, it follows that @xmath265 , so @xmath57 is adjacent to at least one of @xmath266 . if @xmath57 is adjacent with @xmath118 or @xmath119 , say @xmath118 , then it follows , since @xmath267 is connected , that some off - cycle neighbour of @xmath118 must also be adjacent with @xmath110 or @xmath159 , contrary to lemma [ l1](1 ) .
so @xmath268 .
thus @xmath269 .
since @xmath1 is locally connected , @xmath1 has no cut - vertices .
so some off - cycle vertex is adjacent with @xmath56 and at least one of @xmath110 and @xmath159 .
this contradicts our choice of @xmath110 .
we next assume that @xmath270 . by our hypothesis and lemma [ l2 ]
, @xmath110 does not have a true twin . since @xmath1 is locally connected , @xmath1 has at most two off - cycle neighbours .
suppose first that @xmath110 has two off - cycle neighbours @xmath56 and @xmath57 .
let @xmath271 .
by lemma [ l1](1 ) @xmath272 .
since @xmath1 is locally connected , @xmath82 must be adjacent with at least one of @xmath56 and @xmath57 and at least one of @xmath118 and @xmath119 .
suppose first that @xmath273 .
if @xmath274 , then @xmath110 and @xmath82 are true twins which is not possible .
we may thus assume that @xmath275 and @xmath276 . by our choice of @xmath110 , @xmath123 .
since @xmath127 , @xmath277 . by lemma [ l1](2 )
the latter is not possible .
thus @xmath82 is adjacent with exactly one of @xmath56 or @xmath57 , say @xmath56 . by lemma [ l1](1 )
@xmath272 and by assumption @xmath278 . since @xmath127 , @xmath279 , @xmath58 and @xmath277 . by lemma [ l1](2 ) , @xmath280 and @xmath281 .
so @xmath282 , contrary to our choice of @xmath110 .
so this case can not occur .
thus all attachment vertices have exactly one off - cycle neighbour .
in particular @xmath110 has exactly one off - cycle neighbour @xmath56 and four cycle neighbours @xmath283 where @xmath284 . by lemma [ l1](1 ) , @xmath285 .
since @xmath1 is locally connected , @xmath56 is adjacent with at least one of the two vertices @xmath82 and @xmath83 , say @xmath82 .
suppose first that @xmath286 .
by lemma [ l1](2 ) , @xmath287 . by lemmas [ l1](1 ) and [ l1](2 ) , @xmath288 .
since @xmath289 , we see that @xmath290 . from our choice of @xmath110 , @xmath291 .
so @xmath82 has a fifth neighbour @xmath292 .
suppose @xmath293 . by our choice of @xmath110 and by lemma [ l1](1 ) , [ l1](2 ) and [ l1](3 ) , @xmath294 .
this contradicts the fact the @xmath295 is connected .
so @xmath296 .
similarly @xmath297 .
if @xmath249 , then we can argue as for @xmath82 that @xmath298 .
however , then there are six non - adjacent pairs in @xmath299 , contrary to the fact that @xmath116 .
so @xmath300 .
since @xmath116 , it follows that @xmath301 . by lemma [ l1](2 ) , @xmath302 .
so @xmath303 .
suppose @xmath304 .
then @xmath305 ; otherwise , @xmath306 is an extension of @xmath7 , a contradiction . since @xmath307 and @xmath308 and @xmath309
, it follows that @xmath88 .
thus @xmath310 is an extension of @xmath7 , which is not possible .
so @xmath311 .
since @xmath10 we see that @xmath312 .
hence @xmath313 .
we also know that @xmath314 .
since @xmath309 , it now follows that @xmath88 and @xmath315 . since @xmath316
, it follows that @xmath91 .
so @xmath317 is an extension of @xmath7 , a contradiction .
we conclude that @xmath318 .
next suppose that @xmath319 .
then , by our choice of @xmath110 , @xmath320 @xmath321 . if @xmath287 , then @xmath322 . by lemma [ l1](1 ) , [ l1](2 ) and [ l1](3 ) , @xmath323 @xmath324 , @xmath325 and @xmath326 , contrary to the fact that @xmath289 .
so @xmath327 .
since @xmath289 , it follows that @xmath328 .
thus @xmath329 .
hence @xmath82 is a universal vertex in @xmath330 . by lemma [ l2 ] , this is not possible .
we conclude that @xmath331 implies @xmath332 , and similarly @xmath333 implies @xmath334 .
suppose @xmath335 . since @xmath127 , @xmath336 and @xmath337 . by our choice of @xmath110 , @xmath123 and @xmath338 .
if @xmath339 , then @xmath7 is extendable . hence @xmath303 . by our choice of @xmath110 and lemmas [ l1](1 ) and [ l1](2 ) , @xmath340 . since @xmath295 is connected @xmath315 . by our choice of @xmath110 , @xmath341 .
so @xmath176 .
since @xmath1 is locally connected and from our choice of @xmath110 , no vertex of @xmath342 is adjacent with a vertex not in this set .
hence @xmath1 is isomorphic to the forbidden graph @xmath100 which is not possible .
hence , if @xmath249 , then @xmath343 and @xmath300 .
moreover , by lemma [ l1](1 ) , @xmath344 . since @xmath127 it follows that @xmath345 and @xmath275 . by lemma [ l1](2 ) , @xmath346 .
so @xmath123 ; otherwise , @xmath347 .
we show next that either @xmath348 or @xmath349 ; otherwise , @xmath350 .
suppose @xmath351 and @xmath352 . since @xmath353 and by lemma [ l1](3 ) , @xmath354 . moreover , by the case we are in , @xmath135 .
also @xmath355 ; otherwise , @xmath356 is an extension of @xmath7 .
thus @xmath350 , which is not possible .
suppose now that @xmath357 .
suppose first that @xmath358 .
we consider @xmath359 .
since @xmath360 and by lemma [ l1](3 ) , @xmath361 also @xmath362 ; otherwise , @xmath363 is an extension of @xmath7 .
since @xmath364 and by lemma [ l1](2 ) @xmath365 . by lemma [ l1](2 ) and the above , @xmath366 .
so @xmath367 .
suppose next that @xmath357 and @xmath368 . as observed , @xmath369 , @xmath135 and @xmath370 .
thus , since @xmath127 we see that @xmath371 . by lemma [ l1](2 ) , @xmath372 and from the case we are in @xmath373 and @xmath374 .
also @xmath375 ; otherwise , @xmath376 is an extension of @xmath7 .
since @xmath214 has five neighbours on @xmath7 and by our choice of @xmath110 , vertex @xmath214 can not have an off - cycle neighbour .
so @xmath377 . since @xmath378 and @xmath1 is locally connected , @xmath379 and @xmath380
so @xmath1 is isomorphic to the graph @xmath100 which is not possible .
suppose @xmath381 and @xmath351 .
since @xmath127 , we can argue as before that @xmath382 . by our choice of @xmath110 ,
vertex @xmath83 has no off - cycle neighbours . by lemma [ l1 ] and
the case we are in we see that in @xmath383 , @xmath384 , and @xmath385 . since @xmath386 , and @xmath387 we have @xmath388 .
now @xmath389 ; otherwise , @xmath390 is an extension of @xmath7 . using this observation , lemma [ l1 ] , and the case we are in
we see that in @xmath391 , @xmath392 , @xmath393 , @xmath370 and @xmath374 .
hence @xmath350 , a contradiction .
hence this case can not occur .
+ * case 2 * assume that @xmath139 . since @xmath127 , it follows that @xmath141 .
hence @xmath299 has at most seven non - adjacencies .
+ * subcase 2.1 * assume @xmath110 has three off - cycle neighbours @xmath55 and @xmath145 . then @xmath110 has exactly one cycle neighbour , @xmath82 say , distinct from @xmath118 and @xmath119 . by lemma [ l1](1 ) , @xmath394 and by lemma [ l2 ] ,
@xmath82 is not a true - twin of @xmath110 . since @xmath127 it follows that @xmath110 is non - adjacent with exactly one neighbour of @xmath110 and @xmath277 . since @xmath158 either @xmath123 or @xmath395 . in either case
it follows from lemma [ l1](2 ) , that @xmath7 is extendable .
so this case can not occur . + * subcase 2.2 * assume @xmath110 has two off - cycle neighbours @xmath56 and @xmath57 .
let @xmath396 @xmath397 where @xmath122 . since @xmath247 is connected , and by lemma [ l1](1 )
, we may assume that @xmath398 .
suppose that @xmath250 .
assume first that @xmath399 and @xmath400 .
since @xmath139 , it follows that @xmath401 . by lemmas [ l1](1 ) , [ l1](2 ) and [ l1](3 ) , @xmath402 and @xmath403 . since @xmath289 and @xmath151 we have @xmath404 .
so @xmath82 has a neighbour not in @xmath405 .
since @xmath289 , this implies that such a neighbour is a true twin of @xmath82 . by lemma [ l2 ] and our hypothesis ,
this is not possible .
hence @xmath260 or @xmath329 .
assume @xmath329 . by lemmas [ l1](1 ) [ l1](2 ) , [ l1](3 ) , @xmath406 , @xmath407 , @xmath408 and @xmath135 .
since @xmath127 , it follows that @xmath82 is a true twin of @xmath110 . by lemma [ l2 ] , this is not possible .
we conclude that @xmath400 and @xmath260 , i.e. @xmath123 . so @xmath409 . next , assume that @xmath410 . by lemmas [ l1](1 ) , [ l1](2 ) , and the above @xmath411 , @xmath412 and @xmath413 .
since @xmath387 and @xmath158 , it follows that @xmath414 .
the neighbour of @xmath159 not in @xmath415 must be adjacent with at least four of the five vertices in this set .
hence such a neighbour must be on @xmath7 .
let @xmath292 be such a neighbour of @xmath159 .
since @xmath159 and @xmath292 are not twins , and @xmath416 , it follows that @xmath417 and @xmath292 has exactly one non - neighbour in @xmath418 .
suppose @xmath419 . thus @xmath420 . consequently , @xmath421 . since @xmath147 , either @xmath422 or @xmath423 .
suppose @xmath422 . by applying lemmas [ l1](1 ) and [ l1](2 ) and using the fact that @xmath158 one can show that , @xmath424 . if @xmath425 , then it can be argued that @xmath426 .
hence , @xmath427 .
suppose @xmath428 .
if @xmath357 , then by lemmas [ l1](1 ) and [ l1](2 ) we see that @xmath429 , @xmath430 and @xmath431 . by lemma [ l2 ] ,
@xmath214 is not a true twin of @xmath159 .
hence either @xmath432 or @xmath433 .
so @xmath426 , a contradiction .
so @xmath351 .
similarly @xmath352 .
this implies that @xmath434 , a contradiction .
we conclude that @xmath435 , and moreover that @xmath436 .
so either @xmath348 or @xmath381 .
suppose @xmath437 . from the case
we are in and by lemmas [ l1](1 ) and [ l1](2 ) we see that @xmath438 , @xmath439 , and @xmath413 . since @xmath387 , @xmath440 .
so , by lemma [ l1](2 ) , @xmath441 . using this observation , lemmas [ l1](1 ) and [ l2 ] and the fact that @xmath158
, we see that @xmath442 , @xmath443 and @xmath135 ; contrary to the fact that @xmath127 .
suppose @xmath304 .
using lemmas [ l1](1 ) and [ l1](2 ) and the above we see that @xmath444 , @xmath445 and @xmath446 . since @xmath387 , it follows that @xmath447 .
so @xmath448 is an extension of @xmath7 , contrary to assumption .
we conclude that @xmath370 .
it follows from a similar argument that we can not have @xmath449 .
as before we see that @xmath450 and @xmath451 . since @xmath127
it follows that @xmath83 is adjacent with at least one of @xmath56 or @xmath57 .
suppose @xmath452 .
we can argue as for @xmath82 , that @xmath338 and that @xmath408 .
so @xmath450 , @xmath453 and @xmath370 .
since @xmath127 we have @xmath454 and @xmath455 . by lemma [ l1](1 ) , @xmath456 .
since @xmath151 and by lemmas [ l1](1 ) and [ l1](2 ) , we have @xmath457 . thus @xmath295 is not connected , a contradiction .
so @xmath458 .
since @xmath459 , @xmath299 is not connected , so this case can not occur .
hence @xmath460 and @xmath458 . by lemma [ l1](1 )
we also know that @xmath406 .
this accounts for six non - edges in @xmath299 . since @xmath127
, there is thus at most one additional non - adjacency in @xmath299 .
so either @xmath461 or @xmath462 , assume the former .
also @xmath83 is adjacent with at least one of @xmath56 or @xmath57 . by lemmas [ l1](1 ) ,
[ l1](2 ) and [ l1](3 ) , @xmath463 , @xmath325 and @xmath326 .
so @xmath464 .
thus this case can not occur .
+ * subcase 2.3 * assume @xmath110 has exactly one off - cycle neighbour @xmath56 .
let @xmath143 where @xmath146 .
by lemma [ l1](1 ) @xmath465 . to prove this case we establishing several useful facts .
* fact 1 * @xmath398 implies @xmath332 .
similarly @xmath466 implies @xmath467 .
+ * proof of fact 1 * we prove the first of these two statements .
the second statement can be proven similarly .
assume , to the contrary , that @xmath468 .
+ * case a * @xmath275 . if @xmath280 , then @xmath469 and @xmath470 @xmath471 . by lemma [ l1](1 ) , @xmath472 and by lemmas [ l1](2 ) and [ l1](3 ) , @xmath473 and @xmath325 . since @xmath474
, it follows that @xmath475 which is not possible if @xmath476 .
hence @xmath123 .
* case a(1 ) * @xmath477 . by lemma [ l1](1 ) , @xmath478 and by lemma [ l1](2 ) , @xmath479 . by assumption @xmath480 .
since @xmath387 it now follows that @xmath481 . if @xmath482 , then necessarily there is a neighbour of @xmath159 that is a true twin of @xmath159 . by lemma [ l2 ] this implies that @xmath1 contains a strong induced subgraph isomorphic to @xmath483 and with attachment set @xmath484 for some @xmath485 , contrary to the hypothesis .
so @xmath440 .
moreover since @xmath387 , the neighbour of @xmath159 that is not one of @xmath486 must be adjacent with at least four of these vertices .
so by lemma [ l1](1 ) this neighbour , is on @xmath7 , call it @xmath292 .
so @xmath292 is adjacent with at least four vertices in @xmath487 . since @xmath147 , @xmath488 or @xmath422 . if @xmath423 , then , by lemma [ l1](2 ) , @xmath489 .
so @xmath292 is adjacent with all vertices of @xmath490 -\{x\}$ ] .
so @xmath491 and @xmath492 .
hence @xmath493 is an extension of @xmath7 , a contradiction .
so @xmath494 .
hence @xmath495 .
since @xmath128 it follows from lemma [ l1](2 ) that @xmath496 and thus @xmath497 .
so @xmath498 .
this now yields the cycle extension @xmath499 of @xmath7 , which is not possible .
* case a(2 ) * @xmath500 .
assume that @xmath501 .
since @xmath416 , it now follows from lemmas [ l1](1 ) , and [ l1](2 ) that @xmath159 has a neighbour @xmath292 adjacent with four vertices in @xmath487 , by an argument similar to case a(1 ) .
if @xmath502 , then @xmath110 and @xmath159 are true twins which , by lemma [ l2 ] , produces a contradiction .
so @xmath503 .
hence , @xmath504 , and it follows that @xmath505 . finally , since @xmath147 and @xmath506 , we must have exactly one of @xmath422 or @xmath507 . in either case
, we can argue as before , using the fact that @xmath147 and lemmas [ l1](1 ) and [ l1](2 ) , that @xmath508 , which is not possible .
so @xmath509 . by lemma [ l2 ] ,
@xmath159 is not a true twin of @xmath110 .
so @xmath510 .
since @xmath140 , it now follows from this observation and lemmas [ l1](1 ) and [ l1](2 ) that @xmath155 is adjacent to three vertices in @xmath511 .
assume that @xmath466 .
it then follows from lemma [ l1](2 ) that @xmath512 and @xmath513 .
we conclude that @xmath514 where @xmath5 is some two - element subset of @xmath515 .
one can argue using the fact that @xmath147 and lemmas [ l1](1 ) , [ l1](2 ) , and [ l1](3 ) , that @xmath516 , for each possible choice of @xmath5 .
( for example suppose @xmath517 .
then , @xmath518 , @xmath519 and @xmath520 .
so @xmath222 . )
we conclude that @xmath521 , and by the above that @xmath522 .
observe that if @xmath523 , or @xmath524 , @xmath7 has the cycle extensions @xmath525 and @xmath526 , respectively .
hence , @xmath513 and @xmath512 . if @xmath527 or @xmath528 , then @xmath7 has the extension @xmath529 or @xmath530 , respectively .
so @xmath531 .
since @xmath532 , it follows from lemma [ l1](4 ) that @xmath533 . by lemma [ l1](2 )
also since @xmath158 and @xmath535 we have @xmath536 .
hence @xmath537 , a contradiction .
* case b * @xmath538 .
suppose first that @xmath303
. then @xmath308 .
moreover , by lemma [ l1](1 ) , @xmath539 , by lemma [ l1](2 ) @xmath325 and by lemma [ l1](3 ) @xmath326 .
this accounts for seven non - edges in @xmath295 .
so @xmath469 .
since @xmath289 there is a vertex in @xmath540 which is adjacent with all vertices in @xmath541 $ ] . by lemma [ l1](1 )
this vertex is on @xmath7 , call it @xmath292 .
thus @xmath82 and @xmath292 are true twins , contrary to lemma [ l2 ] .
so @xmath339 .
as before , we observe that @xmath542 , @xmath326 and @xmath325 .
also since @xmath543 we have @xmath544 .
since @xmath289 and @xmath147 , it follows that @xmath469 . since @xmath289 and , by lemma [ l2 ]
, @xmath82 has no true twin , there is a neighbour @xmath292 of @xmath82 on @xmath7 , where @xmath545 , such that @xmath292 is adjacent with exactly four of the vertices in @xmath546 . since @xmath547 and @xmath1 is locally connected with @xmath147 , @xmath548 . also since @xmath289 , we have @xmath549 .
we show next that @xmath550 and @xmath551 .
if @xmath552 , then @xmath553 is an extension of @xmath7 .
if @xmath422 , then @xmath554 is an extension of @xmath7 . since @xmath151
, we see that @xmath292 does not lie on the path @xmath555 .
so @xmath503 and @xmath292 is on the path @xmath556 .
hence @xmath420 .
so @xmath557 .
since @xmath151 , @xmath558 .
as before we see that @xmath559 , @xmath313 , @xmath560 and @xmath561 .
so @xmath562 , a contradiction .
this completes the proof of fact 1 .
@xmath563 * fact 2 * @xmath564 implies @xmath565 .
+ * proof of fact 2 * assume , to the contrary , that @xmath566 or @xmath337 , say the former . + * case a * @xmath567 and @xmath155 are consecutive on @xmath7 , i.e. , @xmath568 and @xmath569 .
+ by lemma [ l1](1 ) , @xmath570 , by lemma [ l1](2 ) , @xmath571 and @xmath572 and by lemma [ l1](3 ) , @xmath135 . since @xmath127
all remaining pairs of vertices in @xmath330 are adjacent .
so @xmath573 .
this is not possible by lemma [ l1](3 ) .
so @xmath567 and @xmath155 can not be consecutive on @xmath7 .
* case b * no two of the vertices @xmath82 , @xmath83 and @xmath155 are consecutive on @xmath7 , i.e. , @xmath574 and @xmath575 .
+ by lemma [ l1](1 ) , @xmath576 , by lemma [ l1](2 ) , @xmath577 , by lemma [ l1](3 ) , @xmath578 and since @xmath158 , @xmath579 . since @xmath580 it follows that @xmath581 and that there is a vertex in @xmath582 that is necessarily a true twin of @xmath83 . by lemma [ l2
] this is not possible .
* case c * exactly two of the vertices @xmath82 , @xmath83 and @xmath155 are consecutive on @xmath7 .
+ * case c(1 ) * assume that @xmath568 .
if we consider @xmath583 and use lemma [ l1 ] and the fact that @xmath151 , we see that @xmath584 and @xmath585 . since @xmath580 , @xmath83 has a neighbour @xmath586 such that @xmath292 is adjacent with four of the vertices in @xmath587 . by lemmas [ l1](1 ) , and [ l1](3 ) , @xmath296 .
suppose @xmath588 . since @xmath147 , @xmath589 .
note , by fact 1 , we can not have @xmath590 .
hence , @xmath591 .
assume first that @xmath592 .
so @xmath292 is adjacent with exactly one of @xmath110 and @xmath118 .
let us first assume that @xmath492 .
we now determine non - adjacencies in @xmath593 . by lemmas [ l1](1 ) ,
[ l1](2 ) and [ l1](3 ) , @xmath594 , @xmath595 , and @xmath596 .
so @xmath562 , which is not possible . hence @xmath597 and thus @xmath598
so @xmath502 .
since @xmath151 , and from the case we are in and by lemmas [ l1](1 ) , and [ l1](2 ) @xmath599 , @xmath600 . also if @xmath601 , then @xmath602 . since @xmath603 , @xmath604 .
thus @xmath605 is an extension of @xmath7 , contrary to our assumption .
thus @xmath606 . by lemma [ l1 ] and
the fact that @xmath151 , @xmath607 and @xmath608 .
let @xmath609 and @xmath610 .
note that @xmath611 .
moreover , @xmath612 .
since @xmath613 , @xmath60 , @xmath158 and @xmath1 is locally connected , @xmath56 has no neighbours in @xmath1 other than those in @xmath614 .
similarly , @xmath615 and @xmath616 . since @xmath299 has seven non - adjacencies , namely @xmath617 , @xmath618 , and @xmath619 and @xmath127 it follows that @xmath275 .
so @xmath1 is isomorphic to the graph @xmath236 with attachement set @xmath620 .
hence @xmath489 .
thus @xmath621 and once again @xmath502 .
we now consider the non - adjacencies in @xmath299 .
using the previous observations , lemma [ l1 ] , and the fact that @xmath151 we see that @xmath622 , and @xmath623 .
since @xmath127 all other pairs of vertices in @xmath299 are adjacent . hence @xmath624
moreover , @xmath625 and @xmath626 . since @xmath158 , @xmath627 . by lemma [ l1 ] and
the fact that @xmath151 , @xmath628 .
also @xmath629 otherwise , @xmath630 is an extension of @xmath7 .
since @xmath1 is locally connected and @xmath631 , it follows that @xmath632 .
so @xmath1 contains @xmath235 as strong induced subgraph with attachment set @xmath633 , contrary to the hypothesis .
hence @xmath634 .
so @xmath420 .
hence @xmath635 .
since @xmath151 , @xmath636 or @xmath637 . by lemma [ l1 ] , the case we are in , and the fact that @xmath151 it follows in either case that @xmath391 has the following non - adjacencies : @xmath584 , @xmath638 , @xmath639 and @xmath640 .
so @xmath350 , which is not possible .
* case c(2 ) * assume @xmath641 and @xmath569 .
we consider non - adjacencies in @xmath391 . by lemma [ l1 ]
we have @xmath642 , @xmath643 and as @xmath158 , @xmath644 .
so since @xmath645 , @xmath646 and hence there is a neighbour @xmath292 of @xmath83 that has four neighbours in @xmath647 .
assume first that @xmath648 . since @xmath649 , @xmath296 .
as @xmath147 we have @xmath504 , and hence , @xmath650 .
since @xmath147 , it follows that @xmath651 . as before we can argue that @xmath593 has the following non - adjacencies : @xmath652 , @xmath653 , and @xmath654 .
since @xmath603 , @xmath655 .
but now @xmath656 is an extension of @xmath7 .
thus , @xmath657 . since , by lemma [ l2 ]
, @xmath83 and @xmath292 can not be twins , @xmath292 is non - adjacent with exactly one vertex in @xmath647 .
since @xmath292 is also adjacent with @xmath83 and as @xmath151 we see that @xmath637 or @xmath658 . from the case we are considering , by lemmas [ l1](1 ) , [ l1](2 ) and [ l1](3 ) and from the fact that @xmath158 we have the following non - adjacencies in @xmath299 : @xmath659 , @xmath618 , and @xmath458 . also by fact 1
, @xmath82 is not adjacent with both @xmath56 and @xmath119 . since @xmath127
, it follows that all remaining pairs of vertices in @xmath299 are adjacent .
in particular @xmath660 . since @xmath158 , @xmath661 .
assume first that @xmath662
. then @xmath663 .
moreover , since @xmath664 , we also have @xmath665 . since @xmath147
, it necessarily follows that @xmath123 and @xmath666 . since @xmath391 has seven non - adjacencies , namely , @xmath667 , @xmath653 and @xmath644 , all other pairs of vertices in @xmath391 are adjacent .
in particular , @xmath668 .
hence @xmath669 is an extension of @xmath7 .
suppose next that @xmath249 and hence that @xmath571
. then @xmath670 . since @xmath82 and @xmath83 are not true twins , by lemma [ l2 ] , @xmath671 .
so @xmath672 .
since @xmath158 , it follows that @xmath404 and that @xmath123 .
since @xmath673 and @xmath391 is connected , @xmath674 . since @xmath123 , this produces a contradiction to lemma [ l1](2 ) .
so this case can not occur .
this completes the proof of fact 2 .
@xmath563 * fact 3 * if @xmath83 is the only common neighbour of @xmath56 and @xmath110 , then @xmath82 is non - adjacent with some vertex in @xmath675 and @xmath155 is non - adjacent with some vertex of @xmath676 .
+ * proof of fact 3 * we show that @xmath82 is non - adjacent with some vertex of @xmath675 .
it can be argued in a similar manner that @xmath155 is non - adjacent with some vertex of @xmath676 .
assume , to the contrary , that @xmath677 .
since @xmath147 , @xmath260 or @xmath329 .
it follows from lemma [ l1](2 ) that @xmath400 , and hence @xmath260 , i.e. , @xmath123 .
we consider non - adjacencies in @xmath299 . by our assumption @xmath678 and by fact 2
, @xmath679 . if @xmath680 , then by lemma [ l1](2 ) , @xmath572 . since @xmath681
, it follows that @xmath277 . by lemma [ l1](3 ) , this is not possible .
so @xmath682 .
suppose first that @xmath683 .
we now consider the non - adjacencies in @xmath391 . by lemma [ l1](1 ) and
the fact that @xmath151 , @xmath684 . since @xmath580
, it follows that @xmath83 has a neighbour not in @xmath685 that is adjacent with at least four vertices of @xmath686 . by lemma [ l1](1 )
, such a vertex can not be an off - cycle neighbour of @xmath83 and hence lies on @xmath7 .
let @xmath292 be this neighbour .
by lemma [ l2 ] , @xmath292 must be adjacent with exactly four vertices of @xmath686 . since @xmath687 either @xmath688 or @xmath689 . if @xmath502 , then @xmath489 and hence @xmath690 and @xmath691 . since @xmath692 , @xmath693 .
we now consider non - adjacencies in @xmath383 . by an earlier observation , @xmath694 and since @xmath151 , @xmath572 and @xmath695 .
also @xmath135 ; otherwise , @xmath696 is an extension of @xmath7 .
this gives seven non - adjacencies in @xmath383 .
since @xmath387 , it follows that @xmath697 .
thus @xmath698 is an extension of @xmath7 , contrary to our assumption .
so @xmath503 , i.e. , @xmath420 .
thus @xmath699 .
since @xmath700 we see that @xmath701 .
but now there exists eight non - adjacencies in @xmath391 : @xmath702 , @xmath374 , and @xmath703 .
so @xmath350 which is not possible .
this completes the proof of fact 3 .
@xmath563 * fact 4 * @xmath704 . + * proof of fact 4 * assume that @xmath83 is the only common neighbour of @xmath56 and @xmath110 . by this assumption ,
lemma [ l1](1 ) and fact 2 , @xmath705 and @xmath298 .
since @xmath127 , there is only one more non - adjacency in @xmath299 .
by fact 3 , it follows that @xmath706 .
so @xmath707 @xmath708 . by lemma [ l1](2 ) , @xmath709 and @xmath710 .
so @xmath646 . by lemma [ l1](1 ) and
the case we are in @xmath711 . since @xmath151 , @xmath712 . by the above @xmath706 .
hence as @xmath580 we have @xmath713 and @xmath714 .
since @xmath147 , it follows that @xmath715 and @xmath716 , i.e. , @xmath123 , @xmath717 , @xmath718 and @xmath719 .
but now @xmath720 is an extension of @xmath7 which is not possible .
this completes the proof of fact 4 .
@xmath563 .
* fact 5 * @xmath249 implies @xmath721 or @xmath722 and @xmath203 implies @xmath723 or @xmath722 .
+ * proof of fact 5 * we show that @xmath249 implies @xmath721 or @xmath722 .
it can be shown in a similar manner that @xmath203 implies @xmath723 or @xmath722 .
assume that @xmath249 and that @xmath724 . by lemmas [ l1](2 ) and
[ l1](3 ) , @xmath725 and @xmath400 .
moreover , by lemmas [ l1](1 ) , [ l1](2 ) and [ l1](3 ) , @xmath726 , @xmath302 , and @xmath326 . also since @xmath151 , @xmath727 . since @xmath289
it follows that @xmath82 has a neighbour not in @xmath728 and this neighbour must be a true twin of @xmath82 . by lemma [ l2
] this is not possible and completes the proof of fact 5 .
@xmath563 * fact 6 * @xmath729 . + * proof of fact 6 * suppose @xmath730 . by lemma [ l1](1 )
@xmath240 , by fact 1 , @xmath571 and @xmath572 and by fact 2 , @xmath679 . since @xmath249 it follows from fact 5 that either @xmath135 or @xmath460 .
since @xmath203 it also follows from fact 5 that either @xmath135 or @xmath192 . since @xmath127 we conclude that @xmath135 and that @xmath731 , @xmath732 and @xmath190 . since @xmath147 , @xmath123 and @xmath166 .
if @xmath733 , then it follows since @xmath151 , that @xmath734 . by lemmas [ l1](1 ) and
[ l1](2 ) , @xmath735 .
hence @xmath92 is isolated in @xmath295 .
this is not possible since @xmath1 is locally connected .
thus @xmath736 and similarly @xmath737 .
so @xmath719 . by lemmas [ l1](1 ) and
[ l1](2 ) , @xmath738 is an independent set . since @xmath151 and @xmath1 is locally connected , we see that @xmath739 , a contradiction .
this completes the proof of fact 6 .
@xmath563 * fact 7 * @xmath740 and @xmath741 . + * proof of fact 7 * we prove the first of these two statements since the second statement can be proven in a similar manner .
suppose @xmath742 .
by lemma [ l1](1 ) , @xmath303 .
we know from fact 5 that either @xmath135 or @xmath721 .
suppose first that @xmath135 .
by lemma [ l1](1 ) and our assumption , @xmath743 , by fact 1 @xmath571 and by fact 2 @xmath679 . since @xmath127 ,
all other pairs of vertices in @xmath330 are adjacent .
in particular @xmath744 and @xmath745 .
since @xmath162 , it follows from lemma [ l1](2 ) that @xmath710 .
thus , since @xmath151 , @xmath123 and @xmath166 . if @xmath733 , then the fact that @xmath151 and lemmas [ l1](1 ) and [ l1](2 ) imply @xmath746 .
hence @xmath92 is isolated in @xmath295 , contrary to the fact that @xmath1 is locally connected .
so @xmath437 . if @xmath747 , then @xmath151 and lemmas [ l1](1 ) and [ l1](2 ) imply @xmath748 .
this again contradicts the fact that @xmath1 is locally connected .
so @xmath749 . by lemmas [ l1](1 ) and
[ l1](2 ) @xmath750 and @xmath751 are independent sets .
also @xmath752 , otherwise , @xmath753 is an extension of @xmath7 .
since @xmath1 is connected , locally connected and @xmath151 it follows that @xmath754 and @xmath755 .
thus @xmath756 .
assume now that @xmath277 .
by fact 5 , @xmath460 . by lemma [ l1](1 ) , fact 1 , fact 2 , and fact 6 , we have the following seven non - adjacencies in @xmath299 : @xmath757 , @xmath758 and @xmath679 .
so all other pairs of vertices in @xmath330 are adjacent .
in particular , @xmath744 and @xmath759 .
so by lemmas [ l1](1 ) and [ l1](2 ) , @xmath710 and @xmath303 .
since @xmath151 , @xmath166 . by lemma [ l1](1 ) , and the fact that @xmath151 and fact 6 we have the following seven non - adjacencies in @xmath391 : @xmath760 , @xmath90 , @xmath761 and @xmath762 . since @xmath580 all other pairs of vertices in @xmath583 are adjacent .
so @xmath691 , @xmath763 , and @xmath764 .
since @xmath151 , @xmath91 and @xmath693 .
hence @xmath765 is an extension of @xmath7 , a contradiction .
this completes the proof of fact 7 .
@xmath563 * fact 8 * @xmath766 . + * proof of fact 8 * suppose @xmath767 and thus @xmath407 .
by fact 5 either @xmath135 or both @xmath721 and @xmath768 . in the latter case
there are , by fact 1 , lemma [ l1](1 ) and our case , the following non - adjacencies in @xmath299 : @xmath769 , @xmath770 and @xmath771 .
since @xmath127 , all other pairs of vertices in @xmath330 are adjacent .
so @xmath772 , @xmath573 and @xmath277 .
since @xmath151 , @xmath773 or @xmath774 . however , from lemma [ l1](2 ) , @xmath303 and @xmath775 .
hence this case can not occur .
so @xmath135 .
in addition , by lemma [ l1](1 ) , fact 1 and the case we are in , @xmath776 , @xmath571 and @xmath572 .
thus , since @xmath127 , there is at most one additional pair of non - adjacent vertices in @xmath330 .
so either @xmath777 induces a @xmath64 or @xmath778 induces a @xmath64 . without loss of generality
the former case occurs .
the reasoning for the second case is analogous .
since @xmath779 , it follows from lemma [ l1](2 ) , that @xmath303 . if @xmath280 , then by lemmas [ l1](1 ) , [ l1](2 ) , the fact that @xmath151 and from the case we are considering , it follows that @xmath295 has the following seven non - adjacencies : @xmath780 , @xmath781 , @xmath727 . since @xmath289
all other pairs of vertices in @xmath540 are adjacent .
in particular @xmath88 . by lemma [ l1](3 )
, @xmath7 is extendable , a contradiction .
so @xmath123 .
since @xmath779 , it follows from lemma [ l1](2 ) , that @xmath775 .
so if @xmath782 , then , @xmath783 which is not possible .
so @xmath83 is adjacent with at most one of @xmath155 and @xmath119 . since we have already described six non - adjacencies in @xmath299 and @xmath127 , @xmath83 is non - adjacent with exactly one of @xmath155 and @xmath119 .
so @xmath784 .
since @xmath127 , @xmath785 . if @xmath452 , then @xmath786 .
but now @xmath383 has at least seven non - adjacencies , namely : @xmath787 , @xmath788 , and @xmath789 .
since @xmath387 all other pairs of vertices in @xmath487 are adjacent .
so @xmath790 .
since @xmath151 , @xmath736 and @xmath166 .
if @xmath128 , then @xmath791 is an extension of @xmath7 .
but now we can argue as before that @xmath792 has at least seven non - adjacencies , namely : @xmath793 , @xmath794 and @xmath795 .
since @xmath796 it follows that all other pairs of vertices in @xmath797 are adjacent .
hence @xmath798 , contrary to the fact that @xmath799 .
thus @xmath458 and @xmath800 . since @xmath151 , @xmath166 or @xmath680 . by the above @xmath710 .
so @xmath166 . since @xmath151 and @xmath383 has seven non - adjacencies , namely : @xmath801 , @xmath788 , @xmath802 , we have @xmath736 . as @xmath1 is locally connected and @xmath151 , the vertices @xmath803 , and @xmath804 all have degree @xmath9 .
since @xmath151 and by lemma [ l1](4 ) we see that @xmath805 . also @xmath806
; otherwise , @xmath807 is an extension of @xmath7 . since @xmath1 is locally connected , @xmath808 . since @xmath151 , @xmath809 and thus @xmath719 .
since @xmath151 and @xmath1 is locally connected we see that @xmath810 , and @xmath811 all have degree @xmath24 in @xmath1 and the vertices @xmath56 and @xmath118 both have degree @xmath12 in @xmath1 .
so @xmath756 , contrary to the hypothesis .
so this case can not occur .
this completes the proof of fact 8 .
@xmath563 * fact 9 * @xmath812 and @xmath813 . + * proof of fact 9 * we prove the first of these two statements .
the second statement can be proven similarly . by lemma [ l1](1 ) ,
fact 1 , fact 5 , and the case we are in @xmath299 has the following non - adjacencies : @xmath814 , @xmath571 and either @xmath135 or @xmath460 . since @xmath127 it follows that there is at most one more non - adjacency in @xmath299 .
* case a * suppose @xmath572 .
so all other pairs of vertices in @xmath330 are adjacent .
so in particular @xmath815 and @xmath573 . by lemma [ l1](2 ) , and since @xmath779 we have @xmath816 .
thus @xmath817 .
also since @xmath151 , @xmath680 .
we consider two cases depending on whether @xmath721 or @xmath135 . if @xmath721 , then @xmath277 and @xmath280 . from the case we are considering , by lemma [ l1 ] and using the fact that @xmath151 we have the following non - adjacencies in @xmath295 : @xmath818 , @xmath819 , and @xmath820 . since @xmath289 , @xmath88 .
hence @xmath821 is an extension of @xmath7 , which is not possible .
suppose next that @xmath135 and that @xmath626 . since @xmath151 , @xmath123 .
by lemma [ l1](2 ) , @xmath822 and so @xmath823 . as in the previous case
we can argue that in @xmath383 we have the following non - adjacencies : @xmath824 , @xmath385 and by assumption @xmath572 .
since @xmath387 it follows that all other pairs of vertices in @xmath487 are adjacent .
in particular @xmath825 . since @xmath147 , @xmath437 and @xmath826 .
thus @xmath827 is an extension of @xmath7 which is not possible .
* case b * suppose @xmath162 .
as before , we have the following non - adjacencies in @xmath299 : @xmath828 , @xmath829 and either @xmath135 or @xmath460 .
* subcase b(1 ) * suppose @xmath135 . since @xmath140 , there can be at most one additional non - adjacency in @xmath247 .
* subcase b(1.1 ) * @xmath408 .
then @xmath830 and @xmath336 and @xmath452 .
since @xmath147 , it follows that @xmath166 or @xmath680 .
suppose first that @xmath166 .
since @xmath151 we also see that either @xmath773 or @xmath123 .
if @xmath773 , then @xmath831 is an extension of @xmath7 .
so @xmath303 and @xmath123 .
by lemma [ l1](2 ) , @xmath832 .
if @xmath680 , then @xmath833 is an extension of @xmath7 .
so @xmath710 .
hence , using lemma [ l1](1 ) , the fact that @xmath147 and the case we are in there are eight non - adjacencies in @xmath383 , namely , @xmath834 , @xmath835 , and @xmath408 .
this is not possible since @xmath387 and @xmath151 .
so @xmath836 and @xmath680 .
if @xmath303 , it follows since @xmath151 and @xmath275 , that @xmath123 .
again by lemma [ l1](2 ) , @xmath832 .
however , then we see as before that there are eight non - adjacencies in @xmath383 , namely : @xmath837 , @xmath835 and @xmath408 . since @xmath387 and @xmath158
this is not possible .
so @xmath680 and @xmath773 .
thus @xmath838 is an extension of @xmath7 .
* subcase b(1.2 ) * @xmath839 . + * subcase b(1.2.1 ) * @xmath721 .
thus @xmath280 and @xmath840 and @xmath841 .
since @xmath147 , it follows that @xmath680 .
since @xmath779 , it follows from lemma [ l1](2 ) , that @xmath303 . by lemma [ l1](1 ) and
the fact that @xmath151 , @xmath842 . since @xmath295 is connected , @xmath88 . if @xmath733 , then @xmath843 .
but then @xmath295 has eight non - adjacencies , contrary to the fact that @xmath289 .
so @xmath91 .
hence @xmath844 is an extension of @xmath7 , a contradiction .
* subcase b(1.2.2 ) * @xmath626 .
since @xmath779 , it follows from lemma [ l1](2 ) , that @xmath303 .
* subcase b(1.2.2.1 ) * @xmath660 . since @xmath147 it follows that @xmath123 .
so , by lemma [ l1](4 ) , @xmath710 .
also @xmath147 implies that @xmath83 is non - adjacent with at least one of @xmath155 and @xmath119 . if @xmath786 , then there exist seven non - adjacencies in @xmath295 , namely , @xmath845 , @xmath846 and @xmath786 . moreover , there are seven non - adjacencies in @xmath299 .
so @xmath847 .
hence @xmath151 implies @xmath166 and @xmath736 .
thus @xmath848 has the following six non - adjacencies : @xmath849 , @xmath393 and @xmath850 .
if @xmath128 , then @xmath851 is an extension of @xmath7 , which is not possible .
if @xmath852 , then @xmath853 is an extension of @xmath7 , which is not possible .
so @xmath367 , contrary to the hypothesis .
so @xmath800 and @xmath458 .
so @xmath299 has the following seven non - adjacencies : @xmath854 and @xmath855 .
since @xmath127 all other pairs of vertices in @xmath299 are adjacent .
so @xmath190 .
since @xmath710 and @xmath158 , it follows that @xmath166 . by lemma [ l1](1 ) and [ l1](2 ) , the fact that @xmath151 and from the case we are in , @xmath383 has the following non - adjacencies : @xmath856 , @xmath857 . since @xmath387 , @xmath858 .
so @xmath151 implies @xmath717 .
we now consider the non - adjacencies in @xmath859 . if @xmath860 , then @xmath861 . since @xmath1 is locally connected , @xmath862 .
so @xmath863 is an extension of @xmath7 .
so @xmath809 and @xmath719 .
if @xmath864 , then @xmath865 is an extension of @xmath7 .
also , @xmath224 , otherwise , @xmath866 is an extension of @xmath7 . since @xmath147 and by lemma [ l1](2 ) , @xmath867 .
we can argue similarly that @xmath868 . by lemma [ l1](2 ) , @xmath869 .
since @xmath1 is locally connected and @xmath151 , no vertex in @xmath870 is adjacent with a vertex not in this set .
thus @xmath871 , which is not possible .
* subcase b(1.2.2.2 ) * @xmath759 and @xmath706 .
since @xmath127 , it follows that @xmath800 and @xmath872 .
thus @xmath158 implies that @xmath873 .
since @xmath874 , it follows from lemma [ l1](4 ) , that @xmath280 .
since @xmath151 and by lemma [ l1](1 ) we now have seven non - adjacencies in @xmath295 , namely , @xmath875 , @xmath308 and @xmath876 . since @xmath289 , @xmath877 .
since @xmath151 , @xmath348 .
so @xmath878 is an extension of @xmath7 , which is not possible .
* subcase b(1.2.2.3 ) * @xmath573 and @xmath879 .
since @xmath127 , it follows that @xmath880 and @xmath452 . since @xmath151 , @xmath166 or @xmath680 .
if @xmath280 , then by lemmas [ l1](1 ) and [ l1](2 ) and the fact that @xmath151 , we have the following eight non - adjacencies in @xmath295 : @xmath881 , @xmath882 and @xmath781 .
so @xmath464 , contrary to the hypothesis .
so @xmath123 .
hence , by lemma [ l1](4 ) , @xmath710 .
so @xmath166 . from the case
we are in , and the fact that @xmath151 and by lemma [ l1](4 ) we have the following six non - adjcencies in @xmath391 : @xmath883 , @xmath884 and @xmath135 .
if @xmath885 , we also have @xmath886 . since @xmath580 it now follows that @xmath887
hence @xmath888 is an extension of @xmath7 , a contradiction .
so @xmath693 . as before we have the following six non - adjacencies in @xmath391 : @xmath883 , @xmath884 and @xmath135 .
since @xmath580 either @xmath889 or @xmath890 .
if @xmath889 , then @xmath891 is an extension of @xmath7 which is not possible and if @xmath890 , then @xmath892 is an extension of @xmath7 which is not possible . *
subcase b(2 ) * @xmath721 and @xmath277 .
so @xmath280 and , by lemma [ l1](3 ) , @xmath303 . + * subcase b(2.1 ) * @xmath660 .
then @xmath295 has the following six non - adjacencies : @xmath893 , and @xmath308 .
so @xmath894 is adjacent with @xmath83 or @xmath155 .
suppose first that @xmath680 .
if @xmath895 , then @xmath896 is an extension of @xmath7 which is not ossible .
if @xmath313 , then @xmath289 implies that all remaining edges in @xmath295 are present .
so @xmath897 .
so @xmath898 , a contradiction .
so @xmath710 .
suppose now that @xmath895 .
if @xmath779 , then @xmath151 implies @xmath899 .
thus @xmath900 , contrary to the hypothesis .
thus @xmath408 . since @xmath127 it follows that @xmath901
however then @xmath902 .
so @xmath313 and @xmath903 .
since @xmath289 , it follows that @xmath904 , @xmath905 and @xmath906 .
so @xmath907 , contrary to the fact that @xmath151 .
* subcase b(2.2 ) * @xmath82 is non - adjacent with exactly one of @xmath83 or @xmath155 .
suppose @xmath879 and @xmath573 . thus @xmath901 and @xmath190 . since @xmath151 , @xmath680 or @xmath166 .
moreover , there are seven non - adjacencies in @xmath295 , namely , @xmath908 and @xmath909 . since @xmath289
, there is a vertex in @xmath910 that is necessarily a true twin of @xmath82 . by lemma [ l2
] this is not possible .
so @xmath706 and @xmath759 .
since @xmath127 , @xmath911 and @xmath190 . since @xmath303 and @xmath151 , @xmath680 . since @xmath151 , @xmath876 .
so @xmath295 has the following non - adjacencies : @xmath912 , @xmath913 and @xmath409 .
since @xmath289 , there is a vertex @xmath292 in @xmath914 . by lemma
[ l2 ] , @xmath82 and @xmath292 are not true twins .
so @xmath292 is adjacent with all except exactly one vertex of @xmath915 . since @xmath916 and @xmath151 , @xmath917 .
so @xmath918 .
since @xmath158 , @xmath919 . since @xmath779
, we have a contradiction to lemma [ l1](2 ) .
so this case does not occur .
this completes the proof of fact 9 .
@xmath563 .
our result now follows from facts 4 , 6 , 7 , 8 , and 9 .
[ weakly_pancyclic ] if @xmath1 is a connected locally connected graph with @xmath920 and minimum clustering coefficient at least 1/2 , then @xmath1 is weakly pancyclic .
we have already observed that this is the case if @xmath921 .
let @xmath1 have maximum degree @xmath922 or @xmath9 . by theorem [ t1 ] ,
@xmath1 is either fully cycle extendable , or @xmath1 is isomorphic to an @xmath103 for @xmath923 , or @xmath1 contains @xmath97 as a strong induced subgraph with attachment set @xmath95 for @xmath234 .
we proceed by induction on the order of @xmath1 .
let @xmath1 be a graph of order 6 or 7 that is connected , locally connected with @xmath51 or @xmath9 , respectively and minimum clustering coefficient at least 1/2 .
if @xmath1 is fully cycle extendable , then @xmath1 is weakly pancyclic .
we observe that @xmath98 and @xmath100 are weakly pancyclic and that @xmath1 is not isomorphic to @xmath103 for @xmath924 .
so if @xmath1 is not fully cycle extendable , then @xmath1 contains @xmath97 as strong induced subgraph with attachment set @xmath95 for some @xmath925 and hence @xmath1 is isomorphic to @xmath97 for some @xmath925 . in each case
@xmath1 is readily seen to be weakly pancyclic .
suppose now that @xmath1 is a graph of order @xmath926 and that every connected locally connected graph of order @xmath20 , @xmath927 , and maximum degree @xmath928 where @xmath929 and minimum clustering coefficient at least 1/2 , is weakly pancyclic .
let @xmath1 be a connected locally connected graph with @xmath930 and minimum clustering coefficient at least 1/2 .
if @xmath1 is fully cycle extendable , then @xmath1 is weakly pancyclic .
moreover if @xmath1 is isomorphic to @xmath103 for some @xmath931 , @xmath932 , then @xmath1 is readily seen to be weakly pancyclic .
assume thus that @xmath1 contains an @xmath97 with attachment set @xmath95 as strong induced subgraph .
then @xmath1 contains a vertex @xmath13 of degree @xmath24 .
it can be shown in a straightforward manner that @xmath933 is a connected locally connected graph with @xmath934 and minimum clustering coefficient at least 1/2 .
so @xmath933 is weakly pancyclic .
moreover , the circumference of @xmath933 is either @xmath37 or @xmath935 since the neighbours of every vertex of degree 2 are necessarily adjacent . since the girth of both @xmath1 and @xmath933 is @xmath12 , the result now follows . *
remark : * the conclusion of corollary [ weakly_pancyclic ] still holds if @xmath1 is disconnected and each component of @xmath1 has order at least @xmath12 .
in this paper we added more supporting evidence to ryjek s conjecture : which states that every locally connected graph is weakly pancyclic .
we showed that every locally connected graph with minimum clustering coefficient at least 1/2 and maximum degree at most @xmath9 is weakly pancyclic .
indeed we showed that these graphs , in general , have an even richer cycle structure .
we obtained a complete characterization ( in terms of a family of strong induced subgraphs ) of these graphs that are fully cycle extendable .
it remains an open problem to determine whether the problem , of deciding if a locally connected graph with minimum clustering coefficient at least 1/2 is hamiltonian , is np - complete .
hendry @xcite conjectured that all hamiltonian chordal graphs are fully cycle extendable .
this was shown to be true for several subclasses of the chordal graphs , see @xcite .
recently , however , it was shown in @xcite that this conjecture is not true . these results , and the fact that all hamiltonian locally connected graphs with minimum clustering coefficient at least 1/2 and maximum degree at most @xmath9 are fully cycle extendable , prompt the question : which hamiltonian locally connected graphs with minimum clustering coefficient at least 1/2 are fully cycle extendable ?
9999 van aardt , s.a . ,
frick , m. , oellermann , o.r . , and de wet , j. : global cycle properties of locally connected , locally traceable and locally hamiltonian graphs .
discrete appl . math .
|
let @xmath0 be a graph property .
a graph @xmath1 is said to be _ locally _ @xmath0 ( _ closed locally _
@xmath2 ) if the subgraph induced by the open neighbourhood ( closed neighbourhood , respectively ) of every vertex in @xmath1 has property @xmath0 .
the _ clustering coefficient _ of a vertex is the proportion of pairs of its neighbours that are themselves neighbours .
the _ minimum clustering coefficient _ of @xmath1 is the smallest clustering coefficient among all vertices of @xmath1 .
let @xmath3 be a subgraph of a graph @xmath1 and let @xmath4 .
we say that @xmath3 is a _ strongly induced subgraph _ of @xmath1 with _ attachment set _
@xmath5 , if @xmath3 is an induced subgraph of @xmath1 and the vertices of @xmath6 are not incident with edges that are not in @xmath3 .
a graph @xmath1 is _ fully cycle extendable _ if every vertex of @xmath1 lies in a triangle and for every nonhamiltonian cycle @xmath7 of @xmath1 , there is a cycle of length @xmath8 that contains the vertices of @xmath7 .
a complete characterization , of those locally connected graphs with minimum clustering coefficient 1/2 and maximum degree at most @xmath9 that are fully cycle extendable , is given in terms of forbidden strongly induced subgraphs ( with specified attachment sets ) .
moreover , it is shown that all locally connected graphs with @xmath10 and sufficiently large minimum clustering coefficient are weakly pancylic , thereby proving ryjek s conjecture for this class of graphs .
+ _ keywords _ : minimum clustering coefficient ; locally connected ; strongly induced subgraphs ; fully - cycle extendable ; ryjek s conjecture + _ ams subject classification _ : 05c38
|
the problem of time series analysis and prediction has a long and rich history , probably dating back to the pioneering work of yule in 1927 @xcite .
the application scope is vast , as time series modeling is routinely employed across the entire and diverse range of applied statistics , including problems in genetics , medical diagnoses , air pollution forecasting , machine condition monitoring , financial investments , marketing and econometrics .
most of the research activity until the 1970s was concerned with parametric approaches to the problem whereby a simple , usually linear model is fitted to the data ( for a comprehensive account we refer the reader to the monograph of brockwell and davies @xcite ) .
while many appealing mathematical properties of the parametric paradigm have been established , it has become clear over the years that the limitations of the approach may be rather severe , essentially due to overly rigid constraints which are imposed on the processes .
one of the more promising solutions to overcome this problem has been the extension of classic nonparametric methods to the time series framework ( see for example gyrfi , hrdle , sarda and vieu @xcite and bosq @xcite for a review and references ) .
interestingly , related schemes have been proposed in the context of sequential investment strategies for financial markets .
sequential investment strategies are allowed to use information about the market collected from the past and determine at the beginning of a training period a portfolio , that is , a way to distribute the current capital among the available assets . here , the goal of the investor is to maximize their wealth in the long run , without knowing the underlying distribution generating the stock prices . for more information on this subject ,
we refer the reader to algoet @xcite , gyrfi and schfer @xcite , gyrfi , lugosi and udina @xcite , and gyrfi , udina and walk @xcite .
the present paper is devoted to the nonparametric problem of sequential prediction of real valued sequences which we do not require to necessarily satisfy the classical statistical assumptions for bounded , autoregressive or markovian processes .
indeed , our goal is to show powerful consistency results under a strict minimum of conditions . to fix the context
, we suppose that at each time instant @xmath4 , the statistician ( also called the _ predictor _ hereafter ) _ is asked to guess the next outcome @xmath5 _ of a sequence of real numbers @xmath6 with knowledge of the past @xmath7 ( where @xmath8 denotes the empty string ) and the side information vectors @xmath9 , where @xmath10 . in other words , adopting the perspective of on - line learning , the elements @xmath11 and @xmath12 are revealed one at a time , in order , beginning with @xmath13 , and the predictor s estimate of @xmath5 at time @xmath14 is based on the strings @xmath15 and @xmath16 .
formally , the strategy of the predictor is a sequence @xmath17 of forecasting functions @xmath18 and the prediction formed at time @xmath14 is just @xmath19 . throughout the paper
we will suppose that @xmath20 are realizations of random variables @xmath21 such that the process @xmath22 is jointly stationary and ergodic .
after @xmath14 time instants , the ( _ normalized _ ) _ cumulative squared prediction error _ on the strings @xmath23 and @xmath24 is @xmath25 ideally , the goal is to make @xmath26 small
. there is , however , a fundamental limit for the predictability of the sequence , which is determined by a result of algoet @xcite : for any prediction strategy @xmath27 and jointly stationary ergodic process @xmath22 , @xmath28 where @xmath29 is the minimal mean squared error of any prediction for the value of @xmath30 based on the infinite past observation sequences @xmath31 and @xmath32 .
generally , we can not hope to design a strategy whose prediction error exactly achieves the lower bound @xmath33 .
rather , we require that @xmath26 gets arbitrarily close to @xmath33 as @xmath14 grows .
this gives sense to the following definition : a prediction strategy @xmath27 is called universally consistent with respect to a class @xmath34 of stationary and ergodic processes @xmath22 if for each process in the class , @xmath35 thus , universally consistent strategies asymptotically achieve the best possible loss for all processes in the class .
algoet @xcite and morvai , yakowitz and gyrfi @xcite proved that there exist universally consistent strategies with respect to the class @xmath36 of all bounded , stationary and ergodic processes .
however , the prediction algorithms discussed in these papers are either very complex or have an unreasonably slow rate of convergence , even for well - behaved processes . building on the methodology developed in recent years for prediction of individual sequences ( see cesa - bianchi and lugosi @xcite for a survey and references ) , gyrfi and lugosi introduced in @xcite a histogram - based prediction strategy which is `` simple '' and yet universally consistent with respect to the class @xmath36 .
a similar result was also derived independently by nobel @xcite .
roughly speaking , both methods consider several partitioning estimates ( called _ experts _ in this context ) and combine them at time @xmath14 according to their past performance . for this , a probability distribution on the set of experts is generated , where a `` good '' expert has relatively large weight , and the average of all experts predictions is taken with respect to this distribution . the purpose of this paper is to further investigate nonparametric expert - oriented strategies for unbounded time series prediction . with this aim in mind , in section 2.1
we briefly recall the histogram - based prediction strategy initiated in @xcite , which was recently extended to unbounded processes by gyrfi and ottucsk @xcite . in section 2.2 and 2.3
we offer two `` more flexible '' strategies , called respectively _ kernel _ and _ nearest neighbor - based _ prediction strategies , and state their universal consistency with respect to the class of all ( non - necessarily bounded ) stationary and ergodic processes with finite fourth moment . in section 2.4 we consider as an alternative a prediction strategy based on combining generalized linear estimates . in section 2.5
we use the techniques of the previous section to give a simpler prediction strategy for stationary gaussian ergodic processes .
extensive experimental results based on real - life data sets are discussed in section 3 , and proofs of the main results are given in section 4 .
in this section , we briefly describe the histogram - based prediction scheme due to gyrfi and ottucsk @xcite for _ unbounded _ stationary and ergodic sequences .
the strategy is defined at each time instant as a convex combination of _ elementary predictors _ ( the so - called _ experts _ ) , where the weighting coefficients depend on the past performance of each elementary predictor . to be more precise
, we first define an infinite array of experts @xmath37 , @xmath38 as follows .
let @xmath39 be a sequence of finite partitions of @xmath40 , and let @xmath41 be a sequence of finite partitions of @xmath42 . introduce the corresponding quantizers : @xmath43 and @xmath44 to lighten notation a bit , for any @xmath14 and @xmath45 , we write @xmath46 for the sequence @xmath47 and similarly , for @xmath48 we write @xmath49 for the sequence @xmath50 . the sequence of experts @xmath37 , @xmath51 is defined as follows .
let @xmath52 be the locations of the matches of the last seen strings @xmath53 of length @xmath54 and @xmath55 of length @xmath56 in the past according to the quantizer with parameters @xmath56 and @xmath57 : @xmath58 and introduce the truncation function @xmath59 now define the elementary predictor @xmath60 by @xmath61 where @xmath62 is defined to be @xmath63 and @xmath64 here and throughout , for any finite set @xmath65 , the notation @xmath66 stands for the size of @xmath65 .
we note that the expert @xmath60 can be interpreted as a ( truncated ) histogram regression function estimate drawn in @xmath67 ( gyrfi , kohler , krzyak and walk @xcite ) .
the proposed prediction algorithm proceeds with an exponential weighting average method .
formally , let @xmath68 be a probability distribution on the set of all pairs @xmath69 of positive integers
such that for all @xmath56 and @xmath57 , @xmath70 .
fix a learning parameter @xmath71 , and define the weights @xmath72 and their normalized values @xmath73 the prediction strategy @xmath27 at time @xmath14 is defined by @xmath74 it is proved in @xcite that this scheme is universally consistent with respect to the class of all ( non - necessarily bounded ) stationary and ergodic processes with finite fourth moment , as stated in the following theorem . here and throughout the document , @xmath75 denotes the euclidean norm .
[ cons ] assume that * the sequence of partitions @xmath76 is nested , that is , any cell of @xmath77 is a subset of a cell of @xmath78 , @xmath79 ; * the sequence of partitions @xmath80 is nested ; * the sequence of partitions @xmath76 is asymptotically fine , i.e. , if @xmath81 denotes the diameter of a set , then for each sphere @xmath82 centered at the origin @xmath83 * the sequence of partitions @xmath80 is asymptotically fine .
then , if we choose the learning parameter @xmath84 of the algorithm as @xmath85 the histogram - based prediction scheme @xmath27 defined above is universally consistent with respect to the class of all jointly stationary and ergodic processes such that @xmath86 the idea of combining a collection of concurrent estimates was originally developed in a non - stochastic context for on - line sequential prediction from deterministic sequences ( see cesa - bianchi and lugosi @xcite for a comprehensive introduction ) .
following the terminology of the prediction literature , the combination of different procedures is sometimes termed _
aggregation _ in the stochastic context .
the overall goal is always the same : use aggregation to improve prediction . for
a recent review and an updated list of references , see bunea and nobel @xcite and bunea , tsybakov and wegkamp @xcite .
we introduce in this section a class of _ kernel - based _
prediction strategies for ( non - necessarily bounded ) stationary and ergodic sequences .
the main advantage of this approach in contrast to the histogram - based strategy is that it replaces the rigid discretization of the past appearances by more flexible rules .
this also often leads to faster algorithms in practical applications . to simplify the notation , we start with the simple `` moving - window '' scheme , corresponding to a uniform kernel function , and treat the general case briefly later . just like before
, we define an array of experts @xmath37 , where @xmath56 and @xmath57 are positive integers .
we associate to each pair @xmath69 two radii @xmath87 and @xmath88 such that , for any fixed @xmath56 @xmath89 and @xmath90 finally , let the location of the matches be @xmath91 then the elementary expert @xmath60 at time @xmath14 is defined by @xmath92 where @xmath62 is defined to be @xmath63 and @xmath93 the pool of experts is mixed the same way as in the case of the histogram - based strategy .
that is , letting @xmath68 be a probability distribution over the set of all pairs @xmath69 of positive integers such that @xmath94 for all @xmath56 and @xmath57 , for @xmath71 , we define the weights @xmath72 together with their normalized values @xmath95 the general prediction scheme @xmath96 at time @xmath14 is then defined by weighting the experts according to their past performance and the initial distribution @xmath68 : @xmath97 [ th : kernel ] denote by @xmath36 the class of all jointly stationary and ergodic processes @xmath22 such that @xmath98 .
choose the learning parameter @xmath84 of the algorithm as @xmath99 and suppose that ( [ eq : radius1 ] ) and ( [ eq : radius2 ] ) are verified .
then the moving - window - based prediction strategy defined above is universally consistent with respect to the class @xmath36 .
the proof of theorem [ th : kernel ] is in section [ proofs ] .
this theorem may be extended to a more general class of kernel - based strategies , as introduced in the next remark .
define a _ kernel function _ as any map @xmath100 .
the kernel - based strategy parallels the moving - window scheme defined above , with the only difference that in definition ( [ troisun ] ) of the elementary strategy , the regression function estimate is replaced by @xmath101 observe that if @xmath102 is the naive kernel @xmath103 @xmath104where @xmath105 denotes the indicator function and @xmath106 , we recover the moving - window strategy discussed above .
typical nonuniform kernels assign a smaller weight to the observations @xmath107 and @xmath108 whose distance from @xmath109 and @xmath110 is larger .
such kernels promise a better prediction of the local structure of the conditional distribution .
this strategy is yet more robust with respect to the kernel strategy and thus also with respect to the histogram strategy .
this is because it does not suffer from the scaling problems of histogram and kernel - based strategies where the quantizer and the radius have to be carefully chosen to obtain `` good '' performance . to introduce the strategy , we start again by defining an infinite array of experts @xmath37 , where @xmath56 and @xmath57 are positive integers . just like before ,
@xmath56 is the length of the past observation vectors being scanned by the elementary expert and , for each @xmath57 , choose @xmath111 such that @xmath112 and set @xmath113 ( where @xmath114 is the floor function ) . at time @xmath14 ,
for fixed @xmath56 and @xmath57 ( @xmath115 , the expert searches for the @xmath116 nearest neighbors ( nn ) of the last seen observation @xmath109 and @xmath110 in the past and predicts accordingly .
more precisely , let @xmath117 and introduce the elementary predictor @xmath118 if the sum is non void , and @xmath63 otherwise .
next , set @xmath119 finally , the experts are mixed as before : starting from an initial probability distribution @xmath68 , the aggregation scheme is @xmath120 where the probabilities @xmath121 are the same as in ( [ eq : pkln ] ) .
[ th : nn ] denote by @xmath36 the class of all jointly stationary and ergodic processes @xmath22 such that @xmath98 .
choose the parameter @xmath84 of the algorithm as @xmath99 and suppose that ( [ eq : pl ] ) is verified .
suppose also that for each vector @xmath122 the random variable @xmath123 has a continuous distribution function .
then the nearest neighbor prediction strategy defined above is universally consistent with respect to the class @xmath36 .
the proof is a combination of the proof of theorem [ th : kernel ] and the technique used in @xcite .
this section is devoted to an alternative way of defining a universal predictor for stationary and ergodic processes .
it is in effect an extension of the approach presented in gyrfi and lugosi @xcite to non - necessarily bounded processes .
once again , we apply the method described in the previous sections to combine elementary predictors , but now we use elementary predictors which are generalized linear predictors .
more precisely , we define an infinite array of elementary experts @xmath37 , @xmath124 as follows .
let @xmath125 be real - valued functions defined on @xmath126 .
the elementary predictor @xmath60 generates a prediction of form @xmath127 where the coefficients @xmath128 are calculated according to the past observations @xmath129 , @xmath130 , and @xmath119 formally , the coefficients @xmath128 are defined as the real numbers which minimize the criterion @xmath131 if @xmath132 , and the all - zero vector otherwise .
it can be shown using a recursive technique ( see e.g. , tsypkin @xcite , gyrfi @xcite , singer and feder @xcite , and gyrfi and lugosi @xcite ) that the @xmath128 can be calculated with small computational complexity .
the experts are mixed via an exponential weighting , which is defined the same way as earlier .
thus , the aggregated prediction scheme is @xmath120 where the @xmath121 are calculated according to ( [ eq : pkln ] ) .
combining the proof of theorem [ th : kernel ] and the proof of theorem 2 in @xcite leads to the following result : [ th : gl ] suppose that @xmath133 and , for any fixed @xmath56 , suppose that the set @xmath134 is dense in the set of continuous functions of @xmath135 variables .
then the generalized linear prediction strategy defined above is universally consistent with respect to the class of all jointly stationary and ergodic processes such that @xmath86 we give a sketch of the proof of theorem [ th : gl ] in section [ proofs ] .
we consider in this section the classical problem of gaussian time series prediction ( cf .
brockwell and davis @xcite ) . in this context ,
parametric models based on distribution assumptions and structural conditions such as ar(@xmath136 ) , ma(@xmath137 ) , arma(@xmath136,@xmath137 ) and arima(@xmath136,@xmath138,@xmath137 ) are usually fitted to the data ( cf .
gerencsr and rissanen @xcite , gerencsr @xcite , goldenshluger and zeevi @xcite ) .
however , in the spirit of modern nonparametric inference , we try to avoid such restrictions on the process structure .
thus , we only assume that we observe a string realization @xmath15 of a zero mean , stationary and ergodic , gaussian process @xmath139 , and try to predict @xmath5 , the value of the process at time @xmath14 .
note that there is no side information vectors @xmath129 in this purely time series prediction framework .
it is well known for gaussian time series that the best predictor is a linear function of the past : @xmath140 where the @xmath141 minimize the criterion @xmath142 following gyrfi and lugosi @xcite , we extend the principle of generalized linear estimates to the prediction of gaussian time series by considering the special case @xmath143 i.e. , @xmath144 once again , the coefficients @xmath128 are calculated according to the past observations @xmath145 by minimizing the criterion : @xmath146 if @xmath147 , and the all - zero vector otherwise . with respect to the combination of elementary experts @xmath148 , gyrfi and lugosi applied in @xcite the so - called `` doubling - trick '' , which means that the time axis is segmented into exponentially increasing epochs and at the beginning of each epoch the forecaster is reset .
in this section we propose a much simpler procedure which avoids in particular the doubling - trick .
to begin , we set @xmath149 where @xmath150 and combine these experts as before
. precisely , let @xmath151 be an arbitrarily probability distribution over the positive integers such that for all @xmath56 , @xmath152 , and for @xmath71 , define the weights @xmath153 and their normalized values @xmath154 the prediction strategy @xmath27 at time @xmath14 is defined by @xmath155 by combining the proof of theorem [ th : kernel ] and theorem 3 in @xcite , we obtain the following result : [ gauss ] the linear prediction strategy @xmath27 defined above is universally consistent with respect to the class of all jointly stationary and ergodic zero - mean gaussian processes .
the following corollary shows that the strategy @xmath27 provides asymptotically a good estimate of the regression function in the following sense : [ cor1 ] under the conditions of theorem [ gauss],@xmath156 corollary [ cor1 ] is expressed in terms of an almost sure cesro consistency .
it is an open problem to know whether there exists a prediction rule @xmath27 such that @xmath157 for all stationary and ergodic gaussian processes .
schfer @xcite proved that , under some conditions on the time series , the consistency ( [ eq : oq ] ) holds .
we evaluated the performance of the histogram , moving - window kernel , nn and gaussian process strategies on two real world data sets .
furthermore , we compared these performances to those of the standard arma family of methods on the same data sets .
we show in particular that the four methods presented in this paper usually perform better than the best arma results , with respect to three different criteria .
the two real - world time series we investigated were the monthly usa unemployment rate for january 1948 until march 2007 ( 710 points ) and daily usa federal funds interest rate for 12 january 2003 until 21 march 2007 ( 1200 points ) respectively , extracted from the website _ economagic.com_. in order to remove first - order trends , we transformed these time series into time series of _ percentage change _ compared to the previous month or day , respectively . the resulting time series
are shown in figs .
[ unem ] and [ time ] . before testing the four methods of the present paper alongside the arma methods , we tested whether the resulting time series were trend / level stationary using two standard tests , the kpss test @xcite and the pp test @xcite .
for both series using the kpss test , we did not reject the null hypothesis of level stationarity at @xmath158 and @xmath159 respectively , and for both series using the pp test ( which has for null hypothesis the existence of a unit root and for alternative hypothesis , level stationarity ) , the null hypothesis was rejected at @xmath160 and @xmath159 .
we remark that this means the arima@xmath161 family of models , richer than arma@xmath162 is unnecessary , or equivalently , we need only to consider the arima family arima@xmath163 . as well as this , the gaussian process method requires the normality of the data . since the original data in both data sets is discretized ( and not very finely ) , this meant that the data , when transformed into percentage changes only took a small number of fixed values .
this had the consequence that directly applying standard normality tests gave curious results even when histograms of the data appeared to have near - perfect gaussian forms ; however adding small amounts of random noise to the data allowed us to not systematically reject the hypothesis of normality .
given each method and each time series @xmath164 ( here , @xmath165 or @xmath166 ) , for each @xmath167 we used the data @xmath168 to predict the value of @xmath169 .
we used three criteria to measure the quality of the overall set of predictions .
first , as described in the present paper , we calculated the normalized cumulative prediction squared error @xmath170 ( since we start with @xmath171 for practical reasons , this is almost but not exactly what has been called @xmath172 until now ) . secondly , we calculated @xmath173 , the normalized cumulative prediction error over only the last 50 predictions of the time series in order to see how the method was working after having learned nearly the whole time series .
thirdly , since in practical situations we may want to predict only the _ direction _ of change , we compared the direction ( positive or negative ) of the last @xmath174 predicted points with respect to each previous , known point , to the @xmath174 real directions .
this gave us the criteria @xmath175 : the _ percentage of the direction of the last 50 points correctly predicted_. as in @xcite and @xcite , for practical reasons we chose a finite grid of experts : @xmath176 and @xmath177 for the histogram , kernel and nn strategies , fixing @xmath178 and @xmath179 .
for the histogram strategy we partitioned the space into each of @xmath180 equally sized intervals , for the kernel strategy we let the radius @xmath181 take the values @xmath182 and for the nn strategy we set @xmath183 furthermore , we fixed the probability distribution @xmath68 as the uniform distribution over the @xmath184 experts .
for the gaussian process method , we simply let @xmath185 and fixed the probability distribution @xmath151 as the uniform distribution over the @xmath102 experts . used to compare standard methods with the present nonparametric strategies , the arma@xmath162 algorithm was run for all pairs @xmath186 .
the arma family of methods is a combination of an autoregressive part ar@xmath187 and a moving - average part ma@xmath188 .
tables [ unemp ] and [ timetest ] show the histogram , kernel , nn , gaussian process and arma results for the unemployment and interest rate time series respectively .
the three arma results shown in each table are those which had the best @xmath170 , @xmath173 and @xmath175 respectively ( sometimes two or more had the same @xmath175 , in which case we chose one of these randomly ) .
the best results with respect to each of the three criteria are shown in bold .
.results for histogram , kernel , nn , gaussian process and arma prediction methods on the monthly percentage change in usa unemployment rate from january 1948 until march 2007 .
the three arma results are those which performed the best in terms of the @xmath170 , @xmath173 and @xmath175 criteria respectively .
[ unemp ] [ cols="^,^,^,^",options="header " , ] we see via tables [ unemp ] and [ timetest ] that the histogram , kernel and nn strategies presented here outperform all 36 possible arma@xmath162 models ( @xmath189 ) in terms of normalized cumulative prediction error @xmath170 , and that the gaussian process method performs similarly to the best arma method . in terms of the @xmath173 and
@xmath175 criteria , all of the present methods and the best arma method provide broadly similar results . from a practical point of view , we note also that the histogram , kernel and nn methods also run much faster than a single arma@xmath162 trial on a standard desktop computer . for example , the nn method is of the order of 10 to 100 times faster than an arma@xmath162 for a time series with about 1000 points , depending on the values of @xmath136 and @xmath137 .
[ breiman ] let @xmath190 be a stationary and ergodic process . for each positive integer @xmath191 ,
let @xmath192 denote the left shift operator , shifting any sequence @xmath193 by @xmath191 digits to the left .
let @xmath194 be a sequence of real - valued functions such that @xmath195 almost surely for some function @xmath196 .
suppose that @xmath197
. then @xmath198 [ expert ] let @xmath199 be a sequence of prediction strategies ( experts ) .
let @xmath151 be a probability distribution on the set of positive integers .
denote the normalized loss of any expert @xmath200 by @xmath201 where the loss function @xmath202 is convex in its first argument @xmath203 .
define @xmath204 where @xmath205 is monotonically decreasing , and set @xmath206 if the prediction strategy @xmath17 is defined by @xmath207 then , for every @xmath208 , @xmath209 * proof of theorem [ th : kernel ] . * because of ( [ algoet ] ) it is enough to show that @xmath210 with this in mind , we introduce the following notation : @xmath211 for all @xmath132 , where @xmath62 is defined to be @xmath63 , @xmath212 and @xmath213 .
thus , for any @xmath37 , we can write @xmath214 by a double application of the ergodic theorem , as @xmath215 , almost surely , for a fixed @xmath216 and @xmath217 , we may write @xmath218 therefore , for all @xmath219 and @xmath122 , @xmath220 thus , by lemma [ breiman ] , as @xmath221 , almost surely , @xmath222 denote , for borel sets @xmath223 and
@xmath224 , @xmath225 and set @xmath226 next , let @xmath227 denote the closed ball with center @xmath122 and radius @xmath228 .
let @xmath229 then for any @xmath219 and @xmath122 which are in the support of @xmath230 , we have @xmath231 as @xmath232 and for @xmath230-almost all @xmath122 and @xmath219 by the lebesgue density theorem ( see gyrfi , kohler , krzyak and walk @xcite , lemma 24.5 ) .
therefore , @xmath233 observe that @xmath234 ^ 2\\ & \leq \left(\mathbb e\{y_0\,|\,\|x^{0}_{-k}-\bz\| \le r_{k,\ell},\ \|y^{-1}_{-k}-\bs\|\le r'_{k,\ell}\}\right)^2\\ & \qquad \mbox{(since $ |t_{\ell}(z)| \leq |z|$)}\\ & \leq \mathbb e
\{y_0 ^ 2\,|\ , \|x^{0}_{-k}-\bz\| \le r_{k,\ell},\ \|y^{-1}_{-k}-\bs\|\le r'_{k,\ell}\}\\ & \qquad \mbox{(by jensen 's inequality).}\\\end{aligned}\ ] ] consequently , @xmath235 due to the assumptions of the theorem .
therefore , for fixed @xmath56 the sequence of random variables @xmath236 is uniformly integrable and by using the dominated convergence theorem we obtain @xmath237 invoking the martingale convergence theorem ( see , e.g. , stout @xcite ) , we then have @xmath238 and consequently , @xmath239 we next apply lemma [ expert ] with the choice @xmath240 and the squared loss @xmath241 we obtain @xmath242 on one hand , almost surely , @xmath243 on the other hand , @xmath244 therefore , almost surely , @xmath245 summarizing these bounds , we get that , almost surely , @xmath246 and the theorem is proved
. for fixed @xmath56 and @xmath57 , let @xmath247 then , following the proof of theorem 2 in @xcite one can show that for all @xmath248 , @xmath249 where the @xmath128 are defined in ( [ eq : cnj ] ) . using equality ( [ cc ] ) and lemma [ breiman ] ,
for any fixed @xmath56 and @xmath57 we obtain that , almost surely , @xmath250 then , with similar arguments to theorem 2 in @xcite , it can be shown that @xmath251 finally , by using lemma [ expert ] , the assumptions @xmath252 and @xmath253 , and repeating the arguments of the proof of theorem [ th : kernel ] , we obtain @xmath254 as desired .
99 algoet , p. universal schemes for prediction , gambling and portfolio selection , _ ann .
_ , vol . 20 , pp .
901941 , 1992 .
algoet , p. the strong law of large numbers for sequential decisions under uncertainty , _ ieee trans .
inform . theory _
40 , pp . 609633 , 1994 .
bosq , d. _ nonparametric statistics for stochastic processes . estimation and prediction _ , lecture notes in statistics , 110 , springer - verlag , new york , 1996 .
breiman , l. the individual ergodic theorem of information theory , _ ann . math .
_ , vol .
28 , pp . 809811 , 1957 .
correction . _
_ , vol . 31 , pp .
809810 , 1960 .
brockwell , p. and davis , r. a. _ time series : theory and methods _ , second edition , springer - verlag , new york , 1991 .
bunea , f. and nobel , a. sequential procedures for aggregating arbitrary estimators of a conditional mean , ( _ submitted _ ) , florida state university , 2005 . `
http://stat.fsu.edu/ flori / ps / bnapril2005ieee.pdf ` bunea , f. , tsybakov , a. b. and wegkamp , m. h. aggregation for gaussian regression , _ ann .
( accepted ) _ , 2007 .
cesa - bianchi , n. and lugosi , g. _ prediction , learning , and games _ , cambridge university press , new york , 2006 .
devroye , l. , gyrfi , l. and lugosi , g. _ a probabilistic theory of pattern recognition _ , springer - verlag , new york , 1996 .
gerencsr , l. @xmath255 estimation and nonparametric stochastic complexity , _ ieee trans .
inform . theory _
38 , pp . 17681779 , 1992 .
gerencsr , l. on rissanen s predictive stochastic complexity for stationary arma processes , _
j. statist .
plann . inference _ ,
41 , pp .
303325 , 1994 .
gerencsr , l. and rissanen , j. a prediction bound for gaussian arma processes , _ proc . of the 25th conference on decision and control _ , 14871490 , 1986 .
goldenshluger , a. and zeevi , a. nonasymptotic bounds for autoregressive time series modeling , _ ann .
_ , vol . 29 , pp .
417444 , 2001 .
gyrfi , l. adaptive linear procedures under general conditions , _ ieee trans .
inform . theory _ , vol .
30 , pp . 262267 , 1984 .
gyrfi , l. , hrdle , w. , sarda , p. and vieu , p.
_ nonparametric curve estimation from time series _ , lecture notes in statistics , 60 , springer - verlag , berlin , 1989 .
gyrfi , l. , kohler , m. , krzyak , a. and walk , h. _ a distribution - free theory of nonparametric regression _ , springer - verlag , new york , 2002 .
gyrfi , l. and lugosi , g. strategies for sequential prediction of stationary time series , in _ modeling uncertainty _ , internat .
management sci .
, vol . 46 , pp . 225248 , kluweer acad .
boston , 2001 .
gyrfi , l. , lugosi , g. and udina , f. nonparametric kernel - based sequential investment strategies , _ math . finance _ ,
16 , pp . 337357 , 2006 .
gyrfi , l. and ottucsk , g. sequential prediction of unbounded time series , _ ieee trans . inform . theory _
53 , pp . 18661872 , 2007 .
gyrfi , l. and schfer , d. nonparametric prediction , in j. a. k. suykens , g. horvth , s. basu , c. micchelli and j. vandevalle , editors , _ advances in learning theory : methods , models and applications _ , pp . 339354 , ios press , nato science series , 2003 .
` http://www.szit.bme.hu/
oti / portfolio / histog.pdf `
gyrfi , l. , udina , f. and walk , h. nonparametric nearest neighbor based empirical portfolio selection strategies , _ technical report _
oti / portfolio / nn.pdf ` morvai , g. , yakowitz , s. and gyrfi , l. nonparametric inference for ergodic , stationary time series , _ ann .
_ , vol . 24 , pp .
370379 , 1996 .
nobel , a. on optimal sequential prediction for general processes , _ ieee trans .
inform . theory _
49 , pp . 8398 , 2003 .
schfer , d. strongly consistent online forecasting of centered gaussian processes , _ ieee trans .
inform . theory _
48 , pp . 791799 , 2002 .
singer , a. c. and feder , m. universal linear least - squares prediction , _ international symposium of information theory _ , 2000 .
stout , w. f. _ almost sure convergence _
, academic press , new york , 1974 .
tsypkin , ya .
z. _ adaptation and learning in automatic systems _ , academic press , new york , 1971 .
yule , u. on a method of investigating periodicities in disturbed series , with special reference to wlfer s sunspot numbers , _ philos .
_ , vol . a 226 , pp .
267298 , 1927 .
|
time series prediction covers a vast field of every - day statistical applications in medical , environmental and economic domains . in this paper
we develop nonparametric prediction strategies based on the combination of a set of `` experts '' and show the universal consistency of these strategies under a minimum of conditions .
we perform an in - depth analysis of real - world data sets and show that these nonparametric strategies are more flexible , faster and generally outperform arma methods in terms of normalized cumulative prediction error .
+ _ index terms _ time series , sequential prediction , universal consistency , kernel estimation , nearest neighbor estimation , generalized linear estimates .
nonparametric sequential prediction of time series grard biau @xmath0 , kevin bleakley @xmath1 , + lszl gyrfi @xmath2 and gyrgy ottucsk @xmath2 @xmath3 lsta & lpma + universit pierre et marie curie
paris vi + bote 158 , 175 rue du chevaleret + 75013 paris , france + + @xmath1 institut de mathmatiques et de modlisation de montpellier + umr cnrs 5149 , equipe de probabilits et statistique + universit montpellier ii , cc 051 + place eugne bataillon , 34095 montpellier cedex 5 , france + + @xmath2 department of computer science and information theory + budapest university of technology and economics + h-1117 magyar tudsok krt .
2 , budapest , hungary + +
|
a fundamental question of stellar evolution theory is which stars end their lives as supernovae .
current theory for isolated massive stars makes two basic predictions for sne . for one , the zero - age - main - sequence mass ( ) and mass - loss history control whether a sn occurs , and secondly , for single stars there is a clear mapping between and the type of sn @xcite .
in particular , the lower masses explode with their hydrogen envelopes intact ( e.g. ii - p , ii - l , iin ) , and the most massive stars lose much of their envelopes and explode as hydrogen deficient sne ( e.g. iib , ib / c ) .
however , given the complexity of the underlying physics , especially binary evolution , winds , and episodic eruptions , it is unclear whether nature obeys the same well - delineated mass - dependence .
in fact , the relatively high observed rates of h - deficient sne @xcite and low upper limits on progenitor masses of type ibc sne @xcite imply that binary evolution may figure prominently in producing the h - deficient sne .
furthermore , theory predicts that binary evolution can significantly affect the mapping between initial stellar mass as sne type ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
it is clear that progress in understanding both mass loss and sne requires observational constraints linking the progenitor mass with the eventual sn .
unfortunately , of the dozens of sne that have progenitor mass limits , only 17 have masses measured from a directly detected progenitor ( 6 of these 17 overlap this work ) .
@xcite reviewed the mass distribution of 30 core - collapse sne progenitors ( 20 type iip and 10 others ) , but only eight of these had measurements beyond an upper limit .
at that point , there were also 4 other nearby sne progenitors with full mass constraints @xcite , bringing the total to 12 .
since 2009 , 5 additional sne have been measured @xcite , and some measurements improved @xcite making the total 17 measurements .
these mass estimates are based on serendipitous direct imaging of the progenitor .
first , one searches through the hst archive for bright evolved stars at the sn position .
then , if a star is found , the endpoints of stellar evolution models which pass through the color and magnitude of the likely progenitor are used to estimate the star s initial main sequence mass .
if a star is not found , an upper limit on the progenitor luminosity is measured .
even with the limited number of measurements available , there is a hint of a minimum for explosion and that the least massive stars explode as sn ii - p @xcite .
interestingly , these measurements also suggested that the maximum mass for sn ii - p may be lower than expected , with some perhaps having been merged binaries @xcite , although circumstellar dust may also explain the observations @xcite .
counter to expectations , some h - rich sne ( in particular iin ) have been associated with very massive stars @xcite .
while tantalizing , these initial results are poorly constrained , and even the simple @xmath28 m@xmath1 lower mass limit requires more observational constraints .
while direct imaging of progenitors is the standard method for progenitor mass estimation , it suffers from a number of limitations .
first among these is the requirement that the precursor imaging actually exist .
the majority of past sne have neither pre - existing hst imaging , nor sufficiently accurate astrometry .
consequently , of the @xmath3 historic sne within @xmath4 mpc , only a handful have identified progenitors .
some future nearby sne may also lack precursor imaging due to the limited observations in the hst archive .
the second major limitation is that even when precursor imaging is available , interpretation of that imaging depends on modeling of the most uncertain stages of stellar evolution @xcite .
existing studies estimate the mass of a precursor by fitting endpoints of stellar evolution models to its color and magnitude ; however , an evolved star s appearance is not well constrained during the final evolutionary stages .
binarity , mass loss , pulsation , internal mixing , the formation of dust in stellar winds , and convective instabilities in shell - burning layers all contribute to systematic and random uncertainties in such model endpoints . matching individual endpoints of stellar evolution models to a single highly - evolved star on the brink of explosion
therefore places weak constraints on the stellar mass once systematic uncertainties are taken into account . in this paper , we take a complementary approach that obviates both of these limitations .
we build on a technique developed by several investigators over the past two decades , starting with measuring ages for host star clusters @xcite , moving into finding coeval field populations around both sne and supernova remnants ( snrs ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we note that , while we have applied this approach to many snr locations , historic sne are less numerous , but more reliable , targets because they have well - established locations and types , ensuring that every progenitor mass corresponds to a _
bona fide _ core - collapse event .
the technique finds the masses of sne precursors by analyzing the stellar populations of stars surrounding the sne .
using well - established techniques of stellar population modeling , we can age - date the star formation ( sf ) event that led to each sn .
the resulting age places strong constraints on the mass of the precursor , using the well - understood properties of main - sequence stars . as a result , in cases where the specific stellar population can be identified
, our technique provides a more reliable progenitor age than direct imaging , as it does not depend on whether the progenitor was a single star or a binary .
furthermore , the method works even when there is no imaging prior to the sn , or when the sn position is only localized to within a few arcseconds .
hence our technique can be applied to the location of any historic sn that has sufficiently high - resolution and deep imaging to measure resolved stellar photometry of the upper main sequence and/or he - burning sequence . in 2 , we discuss our sample , the data analyzed in our study , detail our analysis technique , and demonstrate its efficacy in test cases . in 3 , we provide our results in the form of age distributions , a table of masses , and a table of probability distributions for each progenitor .
finally , 4 gives a summary of our findings .
we selected all positively - typed historic core - collapse sne within @xmath5 that also have @xmath2arcsecond accuracy in their positions from the asiago supernova catalog @xcite ; higher positional accuracy is not needed , due to the large size of the aperture within which the cmd is constructed .
we have shown in @xcite that our method provides results consistent with direct progenitor detections ( as confirmed by * ? ? ?
* ) in galaxies as distant as 8 mpc . beyond this limit
, our method is not tested , and therefore we have confined our sample to be within this distance .
we cross - referenced the sne catalog with the hst archive and identified sne that have hst imaging in at least 2 broadband filters in acs , wfpc2 , or uvis .
even relatively shallow data can provide constraints given that , for the most massive @xmath250 m@xmath1 progenitors , the surrounding populations are likely to have other very massive stars with m@xmath65 .
we found 22 sne that match these requirements .
another sn , 2011dh , has already been analyzed by our technique in @xcite .
of our sample there are two that are possibly sn impostors .
these are sn1954j @xcite and sn2002 kg @xcite .
we still include these impostors because the classification as sn impostors are not definitive and it is likely that these transients are associated with the last stages of stellar evolution , in which our proposed method to derive progenitor masses is still of interest .
table [ sample ] shows these sne for which we can attempt to derive the sfh and progenitor mass , along with the proposal i d for the dataset used for out photometry .
there were five sne which we attempted to analyze , but were not able to constrain with confidence .
these had relatively shallow data for quite distant events ( sn 1980k , sn1985f , sn2002ap , sn2002bu , and sn2003gd ) with fewer than 5 stars detected within a 50 pc physical radius .
we do not consider these due to the sparsity of data , although it is likely that deeper imaging of these locations would yield photometry that would result in reliable mass estimates .
however it is also possible these progenitors were runaway stars exploding some distance from their co - eval population ( c.f .
our method has been described in several other publications , including its application to sn 2011dh @xcite , 121 sn remnants ( snrs ) in m31 and m33 @xcite , and an unusual transient in ngc 300 @xcite .
we provide a description of the method here as well for convenience . in brief , we fit sfhs to cmds of the population surrounding the site of the sn to determine the age , and thus mass , of the sn progenitor .
the measurements are anchored by the main - sequence stars surrounding the event ; thus , our age estimates do not depend on whether a binary or single - star progenitor is assumed .
furthermore , the measurements are not sensitive to any circumstellar dust present around the progenitor itself . in the following subsections
we first summarize how well the method works and provide a proof of concept .
then we detail how our photometry was performed , how the stellar samples were generated , and how their age distributions were derived .
the method takes advantage of the fact that most stars form in stellar clusters @xcite with a common age ( @xmath7 ) and metallicity .
indeed , over 90% of stars form in rich clusters containing more than 100 members with @xmath8m@xmath1 @xcite .
the stars that formed in a common event remain spatially correlated on physical scales up to @xmath9 during the @xmath10 lifetimes of @xmath11m@xmath1 stars , even if the cluster is not gravitationally bound @xcite ; we have confirmed this expectation empirically in several test cases @xcite .
thus , it is reasonable to assume that most young stars within a 50 parsecs of many sn are coeval .
however , we note that our assumption breaks down for sne from runaway stars @xcite , which would not be coeval with their surrounding population .
the age of a sn s host stellar population can be recovered from its color - magnitude diagram ( cmd ) . in the simplest method
, one can fit a single isochrone to an observed cmd and estimate the turnoff mass of the youngest stars .
however , due to the small numbers of massive stars , one can easily underestimate the mass , since cmds can show an apparent turnoff that is fainter than the true turnoff luminosity simply because of poor poisson sampling of the upper end of the imf .
instead , we adopt more sophisticated methods that take advantage of the entire cmd . these methods fit superpositions of stellar populations to reproduce the observed cmd , using the recovery of artificial stars to generate realistic distributions of stars from theoretical isochrones .
the recovered recent sfh therefore fits not just the turnoff luminosity , but the full luminosity function of the main sequence and the blue and red core helium - burning sequences as well .
including the well - populated lower end of the main sequence adds significant statistical weight when interpreting the sparsely sampled population of massive upper main sequence stars .
the method allows dust extinction to be reliably taken into account .
the main sequence has a well defined color , such that any shift towards redder colors must be produced by foreground reddening , allowing the dust extinction to be inferred from the cmd itself .
differential extinction can be constrained as well , using the observed widening of the main sequence over what is expected from photometric errors .
the resulting reddening constraints are dominated by the young stars in which we are most interested .
an example of the efficacy of the method is in its application to sn 1987a .
we have run our model fits on deep wfpc2 photometry measured from the archival data of proposal i d 7434 ( pi : kirshner ) .
we fit the f555w - f814w cmd in the range 12@xmath0f814w@xmath025 as shown in figure [ 87deep ] , and get a well - constrained median ( 22.0@xmath12 m@xmath1 ) , which is consistent with the mass ( 19@xmath13 m@xmath1 ) derived in direct imaging studies @xcite , and with the combined mass of the binary merger scenario ( 16@xmath143@xmath1519 ; * ? ? ?
this comparison between techniques provides strong verification of our proposed method . while this test is encouraging , the data for 1987a is significantly deeper than that of our more distant objects .
two more tests , however , suggest that our technique works even with much shallower data .
first , we ran our model fits on 1987a only including the photometry for stars brighter than apparent magnitude of 18.5 ( absolute magnitude of 0 ) , comparable to the depth of most of our more distant targets .
the resulting median mass was more poorly constrained ( 22.8@xmath16 m@xmath1 ) , but was still consistent with the known mass .
thus , we may lose precision with shallower data , but we can still obtain useful constraints .
in addition to our tests on sn 1987a , we have verified our technique out to @xmath28 mpc by applying it to sn 2011dh in m51 @xcite , for which we found a progenitor mass of 13@xmath17 m@xmath1 .
@xcite identified a progenitor in archival hst images , which has since vanished @xcite , and fit the bolometric luminosity to stellar - evolution models and derived a progenitor mass of 13@xmath13 m@xmath1 , consistent with our constraints . to generate the cmd ,
we measure resolved stellar photometry of the _ hst _ field containing the location of the historic sn .
this photometry was performed using the packages hstphot ( for wfpc2 data ) or dolphot ( * ? ? ?
* for acs data ; ) .
these packages perform point spread function fitting optimized for the undersampled flat - fielded images that come from _ hst_. all of the photometry we use has been publicly released to the high - level science products in the hst archive through the angst and angrrr programs ( go-10915 and ar-10945 ; pi : dalcanton ) .
the details of the fitting and culling parameters used are provided in @xcite and the angrrr public data archive . as part of these programs ,
hundreds of thousands of artificial star tests were also performed to assess completeness and photometric accuracy .
these tests consist of inserting a single star into the data , rerunning the the data reduction , and assessing whether the fake star was recovered , and if so , how close its measured brightness was to the input brightness .
to isolate the subset of stars from our catalogs that were co - spatial with the historic supernovae , we used the coordinates for the sne from the asiago supernova catalog @xcite , and galaxy distances from @xcite .
we corrected the astrometry in our catalogs by cross - correlating 2mass positions for the bright stars in our catalogs with our positions .
our catalog astrometry is then corrected such that the star positions agree with those of 2mass as precisely as our centroiding for these bright stars will allow ( typically @xmath20.1@xmath18 ) .
this correction to our catalog astrometry made sure that our positions were at least as precise as those in the sne catalog ( within a few tenths of an arcsecond ) . with the location and the distance well - measured
, we were able to pull stars measured within a projected radius of 50 pc of each sne .
in the most distant cases , this radius is only a bit more than 1 arcsecond , making the necessary precision of astrometry only @xmath21@xmath18 .
to provide fake star statistics for the photometric completeness and precision appropriate to our sample , we required a minimum of ten thousand fake stars into our images . to reach this number ,
we included fake stars from a region of up to a factor of 7 larger than the real stars . in fields where the quality of the data varies quickly with position , such as near the center of m82
, we applied additional computing resources to obtain more artificial star tests within the same radius as the stellar sample .
however , for almost all of our fields , changes in stellar density , and therefore photometric quality , were small over the field , making it possible to use a large suite of artificial star tests to improve statistics on our cmd fitting .
our cmd fitting process was very similar to that performed in @xcite .
we used the cmd - fitting package match @xcite to fit each cmd with the stellar evolution models of @xcite with updates in @xcite and @xcite .
the package allows models to be shifted in temperature and luminosity space to mimic systematic uncertainties , and it allows differential extinction to be applied to the models during fitting .
first , we determined the best - fitting amount of differential extinction to apply when fitting each sn .
we fitted the data with a grid of values for the differential extinction ( @xmath19 ) and foreground extinction a@xmath20 .
we chose the @xmath19 value that provided the best fit to the data without requiring an @xmath21 value below the known foreground extinction from @xcite .
we show an example plot summarizing this extinction determination method in figure [ dav ] . with the distance , extinction , and differential extinction values fixed , we fitted the cmd to find the most likely age distribution , allowing metallicities for the young population in the range of -0.6@xmath22[fe / h]@xmath220.1 .
examples for objects with data quality typical of most of our sample are shown in figure [ 93j]-[02hh ] , where we show an image of our extraction region , a plot of the cmd of the 50 pc region , and a final cumulative star formation history from the fitting routine . in figure [ correlations1 ]
we plot our final derived masses against distance and @xmath23 .
the lack of correlations in the derived masses as a function of these parameters suggests they do not introduce any significant bias into our measurements . to assess systematic uncertainties ( due to any model deficiencies ) , we reran the fitting with several changes to the models , following @xcite .
we allowed the effective temperature of the models to vary by @xmath24log(t@xmath25)=0.02 .
we allow the bolometric luminosity of the models to vary by @xmath24log(l@xmath26)=0.17 .
furthermore , we allow the differential extinction applied to the model to vary by @xmath27=0.2 in cases of high @xmath19 ( @xmath280.4 ) .
we run 100 fits to 100 realizations of the model , varying all of these parameters to account for systematic uncertainties resulting from our stellar evolution models and our treatment of the extinction .
then , to measure the random uncertainties due to number of stars and depth of the photometry , we use the hybridmc task within the match package as described in @xcite .
this task determines the star formation rate that would allow an acceptable fit to the data for each age bin , thus providing robust upper - limits for bins where the best - fit star formation rates were 0 . with
both uncertainty determinations complete , we combine the random and systematic uncertainties in quadrature using the match routing _ zcmerge _ to calculate our final uncertainties on the star formation rates in each age bin .
next , we use our total uncertainties on the star formation rates in each age bin to determine the uncertainty on the fraction of stellar mass present in each age bin back to 50 myr .
we perform 1000 monte carlo realizations of the measured sfh from 50 myr to the present .
we then calculate the 16% and 84% ranges in stellar mass fraction in each age bin from these tests .
we adopt these percentiles as the uncertainties on fraction of stellar mass formed in each age bin relative to the total stellar mass produced in the past 50 myr .
once we have the mass fraction ( and associated mass fraction uncertainty ) in each age bin , we calculate our first estimate of the progenitor mass by determining the median age of the best fit .
we then use our uncertainties to determine the age bins consistent with containing the median to assign uncertainties on that median age .
finally , we convert these ages to masses by taking the most massive star remaining in the model isochrone corresponding to the each age ( see * ? ? ?
* for more details ) .
these values provide our the nominal progenitor mass and associated uncertainties for each sn .
these median masses and associated uncertainties ( @xmath29 ) are provided in table [ median ] .
although the assignment of a single progenitor age is of interest , many of our sfhs contain multiple coeval populations , making a more complex distribution of the mass probability desirable for some purposes .
we therefore have also tabulated the uncertainty for each progenitor due to the spread in the recovered age distribution ( @xmath30 ) .
these uncertainties encompass 68% of the total population mass ( about the median value of the best fit ) with ages @xmath050 myr including uncertainties and account for the full distribution ages present at the sn location , similar to the technique adopted in earlier work @xcite . in most cases ,
the stellar mass is relatively well confined to a small age range , but including this second set of uncertainties shows where there are multiple ages present .
for example , the median age of the young population surrounding sn1994i is well - determined , providing a high - precision mass measurement of [email protected] m@xmath1 ; however , there is also a younger population present that represents a significant fraction of the stellar mass . if the presence of this population is taken into account , the uncertainties on the progenitor mass increase substantially to 10.2@xmath32 .
thus , in this case , only under the assumption that progenitor was a member of the dominant young population is the mass of the progenitor well - constrained .
otherwise , it is only a lower limit .
looking at these uncertainties , one can determine which sne would benefit most from improved photometry data .
large spreads in the 68% population mass accompanied by small errors on the median seem to occur for sne with few stars in the cmd .
for example , sn1951h has only 11 stars for fitting , a 10% error on the median , but @xmath30 values that encompass the full mass range of the models .
thus , the well - constrained median suggests that the mass should also be well constrained , but the small number of stars results in large uncertainties for other age bins which would likely be reduced with deeper data and a larger number of detected stars . finally , to provide detailed probability distributions for all sn
, we tabulate the probability that the progenitor was in each age / mass bin , given the sfh and associated uncertainties .
these probability distributions are given in table [ pdfs ] , where each mass bin is assigned a probability which comes from the most likely sfh , and an associated uncertainty on the probability , which comes from applying the uncertainties in the sfh to the mass probability distribution .
thus , in order to account for both sources of uncertainty on the progenitor mass ( the fitting uncertainty and the uncertainty associated with the intrinsic range of ages ) , it is necessary to assign uncertainties to our probabilities .
however , in many cases , the median mass is relatively well defined ( @xmath33% ) , and provides a simple , though less thorough , constraint on the progenitor mass .
six of the events in our sample have previously - measured progenitor masses from direct imaging ( sn1987a , sn1993j , sn2004dj , sn2004et , sn2005cs , sn2008bk , see table [ median ] ) , and while our measurements are less precise in some cases , they are consistent with the previous measurements in all cases , as shown in figure [ comp ] .
indeed , in all cases the previous measurements are consistent with our most optimistic uncertainties the uncertainty on the median age of the young population .
a few sne in our samples stand out as extremely challenging of our ability to measure star formation histories .
for example ,
in figure [ 04am ] we show the image , photometry , and fitting results for our most heavily extincted location , sn2004am . in this case , even though there is a very high amount of differential extinction ( @xmath19=2.5 ) , the relatively large number of stars provides a good constraint on the age distribution .
unfortunately , with this much dust , there is clearly the possibility of a significant number of more massive stars being completely hidden from the sample , which would not be accounted for by our method .
we can not account for stars that are extincted out of our photometry sample .
thus , this amount of dust may make this result less reliable than many of the others .
such examples are unlikely to improve without much deeper data to probe to very high extinctions
. another extreme case is sn2004et , where we only have 6 stars in the cmd due to shallow imaging and a far distance ( meaning a small extraction region on the sky ) .
we show our results for this sn in figure [ 04et ] , where the lack of stars results in very large uncertainties .
although the uncertainties on the progenitor mass are large , the full range of masses are not allowed by our uncertainties , which suggest the mass is @xmath2816 m@xmath1 .
these uncertainties are reliable , as the best - mass is well away from ( but within the large errors of ) the mass measured from direct imaging .
interestingly , even with the large uncertainties , the mass constraint is useful since it rules out masses lower than the best - fit mass from direct imaging .
this example confirms that our uncertainty estimates are reliable , but also demonstrates that attempting this technique with any fewer stars is of little value . finally , our results in this work add further validation to our method .
for the 6 sne with we measure here that have literature measurements , we plot our measurements against those from the literature in figure [ comp ] . in all cases
our measurements are consistent with previous measurements within the uncertainties , and no systematic bias is seen .
we note that our results are consistent with no sn progenitors @xmath2820 m@xmath1 , as are all of the progenitor mass measurements currently available in the literature ( see references in section 1 ) .
while we do have some best estimates that are higher mass , their uncertainties all extend below 20 m@xmath1 .
our most massive central values are for sn2004et and sn1962 m , but these only have 75% and 82% probability of being @xmath2820 m@xmath1 .
furthermore , the direct imaging mass for sn2004et has an upper limit of 20 m@xmath1 , suggesting that the correct mass is indeed at the low end of our uncertainties .
figure [ mass ] plots the masses in ranked order , along with the expected distribution of masses for a @xcite imf with different upper - mass cutoffs .
the large uncertainties on the high progenitor masses severely limit our ability to determine the existence of such a cutoff .
thus , our current sample and data quality does not provide any conclusive evidence that high - mass stars produce core - collapse supernovae .
this lack of conclusive @xmath2820 m@xmath1 progenitors is consistent with findings of several other studies @xcite , hinting that there could be a ceiling to sn production or a mass range that under - produces sne .
however , if we can measure a single progenitor mass @xmath2820 m@xmath1 with even 20% precision , constraints on the progenitor mass distribution would be greatly improved .
we have constrained the progenitor masses of 17 historic sne using cmd fitting of stellar populations measured from hst archival data .
eleven of these are new constraints , making the total number of historic sn progenitor masses 28 .
even with this dramatic increase in mass measurements , there is still not a single high - precision measurement of a progenitor @xmath2820 m@xmath1 , making characterization of the progenitor mass distribution difficult .
this work represents all that is possible with the current state of the hst archive .
the power of the technique is clear , and we hope that future studies will be made possible by more and deeper hst imaging of nearby galaxies containing historic sne .
support for this work was provided by nasa through grants ar-13277 , go-10915 , and hubble fellowship grant 51273.01 from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, for nasa , under contract nas 5 - 26555 . z.g.j .
is supported in part by a national science foundation graduate research fellowship .
lcccccccccc sn1917a & ii & 20:34:46.90 & 60:07:29.00 & ngc 6946 & 11229 & 28.9 & 6.0 & 0.939 & 0.0 & 7 + sn1923a & ii p & 13:37:09.20 & -29:51:04.00 & ngc 5236 ( m83 ) & 11360 & 28.5 & 5.0 & 0.75 & 1.6 & 142 + sn1951h & ii & 14:03:55.30 & 54:21:41.00 & ngc 5457 ( m101 ) & 9490 & 29.3 & 7.2 & 0.024 & 0.0 & 11 + sn1954j & ii pec & 07:36:55.36 & 65:37:52.1 & ngc 2403 & 10182 & 28.0 & 4.0 & 0.45 & 0.2 & 212 + sn1962 m & ii p & 03:18:12.20 & -66:31:38.00 & ngc 1313 & 9796 & 28.0 & 4.0 & 0.4 & 0.0 & 125 + sn1980k & ii l & 20:35:30.07 & 60:06:23.75 & ngc 6946 & 11229 & 28.9 & 6.0 & & & 2 + sn1985f & ib & 12:41:33.01 & 41:09:05.94 & ngc 4618 & 9042 & 29.5 & 7.9 & & & 4 + sn1987a & ii pec & 05:35:28.01 & -69:16:11.61 & lmc & 7434 & 18.5 & 0.05 & 0.206 & 0.7 & 11800 + sn1993j & iib & 09:55:24.95 & 69:01:13.38 & ngc 3031 ( m81 ) & 10584 & 28.0 & 4.0 & 0.45 & 0.0 & 143 + sn1994i & ic & 13:29:54.07 & 47:11:30.50 & ngc 5194 & 10452 & 29.6 & 8.3 & 0.6 & 0.0 & 42 + sn2002ap & ic pec & 01:36:23.85 & 15:45:13.20 & ngc 0628 & 10272 & 29.3 & 7.2 & & & 1 + sn2002bu & iin & 12:17:37.18 & 45:38:47.40 & ngc 4242 & 10272 & 29.5 & 7.9 & & & 0 + sn2002hh & ii p & 20:34:44.29 & 60:07:19.00 & ngc 6946 & 10607 & 28.9 & 6.0 & 1.3 & 0.0 & 66 + sn2002 kg & iin & 07:37:01.83 & 65:34:29.30 & ngc 2403 & 10182 & 28.0 & 4.0 & 0.11 & 0.0 & 267 + sn2003gd & ic pec & 01:36:42.65 & 15:44:19.90 & ngc 0628 & 10272 & 29.3 & 7.2 & & & 4 + sn2004am & ii p & 09:55:46.61 & 69:40:38.10 & ngc 3034 ( m82 ) & 10776 & 28.0 & 4.0 & 0.7 & 2.5 & 37 + sn2004dj & ii p & 07:37:17.02 & 65:35:57.80 & ngc 2403 & 10607 & 28.0 & 4.0 & 0.11 & 0.0 & 127 + sn2004et & ii p & 20:35:25.33 & 60:07:17.70 & ngc 6946 & 11229 & 28.9 & 6.0 & 1.3 & 1.4 & 6 + sn2005af & ii & 13:04:44.06 & -49:33:59.80 & ngc 4945 & 10877 & 27.9 & 3.8 & 0.7 & 0.0 & 6 + sn2005cs & ii p & 13:29:52.85 & 47:10:36.30 & ngc
5194 & 10452 & 29.6 & 8.3 & 0.098 & 0.0 & 33 + sn2008bk & ii p & 23:57:50.42 & -32:33:21.50 & ngc 7793 & 12196 & 28.0 & 4.0 & 0.054 & 0.0 & 248 + sn2008iz & ii & 09:55:51.55 & 69:40:45.80 & ngc 3034 ( m82 ) & 10776 & 28.0 & 4.0 & 0.5 & 0.3 & 56 + [ sample ] lcccccc sn1917a & 7.9 & 6.7 & 0.5 & 25.1 & 0.5 & + sn1923a & 22.0 & 2.3 & 12.8 & 45.6 & 13.8 & + sn1951h & 12.5 & 0.9 & 2.3 & 6.2 & 5.2 & + sn1954j & 12.5 & 3.1 & 0.9 & 16.7 & 4.4 & + sn1962 m & 38.0 & 6.2 & 25.7 & 6.2 & 30.6 & + sn1987a ( shallow ) & 22.8 & 2.5 & 14.4 & 47.4 & 15.5 & 19@xmath313 @xcite + sn1987a ( deep ) & 22.0 & 2.3 & 5.8 & 8.6 & 13.8 & 19@xmath313 @xcite + & & & & & & binary merger 16@xmath143@xmath1519 + & & & & & & @xcite + sn1993j & 11.7 & 5.5 & 0.8 & 21.3 & 2.2 & @xmath217 @xcite + & & & & & & two stars , each 1417 + & & & & & & @xcite + sn1994i & 10.2 & 0.7 & 0.7 & 59.2 & 1.8 & + sn2002hh & 15.9 & 4.8 & 1.3 & 9.9 & 6.4 & + sn2002 kg & 9.7 & 9.1 & 0.6 & 13.5 & 2.3 & + sn2004am & 10.2 & 0.7 & 0.7 & 4.3 & 1.8 & + sn2004dj & 12.9 & 0.9 & 0.9 & 17.7 & 4.7 & @xmath212 @xcite + sn2004et & 56.4 & 12.2 & 39.7 & 12.2 & 48.1 & 15@xmath34 @xcite + sn2005af & 8.7 & 0.5 & 0.5 & 5.2 & 1.3 & + sn2005cs & 12.5 & 10.6 & 2.3 & 54.0 & 4.4 & 10@xmath313 @xcite + & & & & & & 9.5@xmath35 @xcite + sn2008bk & 7.9 & 9.4 & 0.5 & 25.1 & 0.5 & 12.9@xmath36 @xcite + & & & & & & 8.5@xmath311 @xcite + sn2008iz & 11.0 & 0.7 & 0.7 & 4.7 & 2.8 & + [ median ] image of region surrounding sn1993j in m81 .
our 50 pc extraction region is marked with the white circle .
_ center : _ the color - magnitude diagram resulting from our photometry extraction .
_ right _ the cumulative fraction star formation from 50 myr ago to the present as measured by our cmd - fitting analysis described in section 2.,title="fig:",width=192 ] image of region surrounding sn1993j in m81 .
our 50 pc extraction region is marked with the white circle .
_ center : _ the color - magnitude diagram resulting from our photometry extraction .
_ right _ the cumulative fraction star formation from 50 myr ago to the present as measured by our cmd - fitting analysis described in section 2.,title="fig:",width=192 ] image of region surrounding sn1993j in m81 .
our 50 pc extraction region is marked with the white circle .
_ center : _ the color - magnitude diagram resulting from our photometry extraction .
_ right _ the cumulative fraction star formation from 50 myr ago to the present as measured by our cmd - fitting analysis described in section 2.,title="fig:",width=192 ] image of region surrounding sn2002hh in ngc 6946 .
our 50 pc extraction region is marked with the white circle .
_ center : _ the color - magnitude diagram resulting from our photometry extraction .
_ right _ the cumulative fraction star formation from 50 myr ago to the present as measured by our cmd - fitting analysis described in section 2.,title="fig:",width=192 ] image of region surrounding sn2002hh in ngc 6946 .
our 50 pc extraction region is marked with the white circle .
_ center : _ the color - magnitude diagram resulting from our photometry extraction .
_ right _ the cumulative fraction star formation from 50 myr ago to the present as measured by our cmd - fitting analysis described in section 2.,title="fig:",width=192 ] image of region surrounding sn2002hh in ngc 6946 . our 50 pc extraction region is marked with the white circle .
_ center : _ the color - magnitude diagram resulting from our photometry extraction .
_ right _ the cumulative fraction star formation from 50 myr ago to the present as measured by our cmd - fitting analysis described in section 2.,title="fig:",width=192 ] image of region surrounding sn2004am in m82 .
our 50 pc extraction region is marked with the white circle .
_ center : _ the color - magnitude diagram resulting from our photometry extraction .
the lack of any clear feature suggests that our results are becoming less reliable at these high amounts of differential extinction . _
right : _ the cumulative fraction star formation from 50 myr ago to the present as measured by our cmd - fitting analysis described in section 2.,title="fig:",width=192 ] image of region surrounding sn2004am in m82 .
our 50 pc extraction region is marked with the white circle .
_ center : _ the color - magnitude diagram resulting from our photometry extraction .
the lack of any clear feature suggests that our results are becoming less reliable at these high amounts of differential extinction . _
right : _ the cumulative fraction star formation from 50 myr ago to the present as measured by our cmd - fitting analysis described in section 2.,title="fig:",width=192 ] image of region surrounding sn2004am in m82 .
our 50 pc extraction region is marked with the white circle .
_ center : _ the color - magnitude diagram resulting from our photometry extraction .
the lack of any clear feature suggests that our results are becoming less reliable at these high amounts of differential extinction .
right : _ the cumulative fraction star formation from 50 myr ago to the present as measured by our cmd - fitting analysis described in section 2.,title="fig:",width=192 ] image of region surrounding sn2004et in ngc6946 . our 50 pc extraction region is marked with the white circle .
_ center : _ the color - magnitude diagram resulting from our photometry extraction .
_ right : _ the cumulative fraction star formation from 50 myr ago to the present as measured by our cmd - fitting analysis described in section 2 .
the low precision makes it clear that we can not constrain progenitor masses with any fewer stars than this example.,title="fig:",width=192 ] image of region surrounding sn2004et in ngc6946 .
our 50 pc extraction region is marked with the white circle .
_ center : _ the color - magnitude diagram resulting from our photometry extraction .
_ right : _ the cumulative fraction star formation from 50 myr ago to the present as measured by our cmd - fitting analysis described in section 2 .
the low precision makes it clear that we can not constrain progenitor masses with any fewer stars than this example.,title="fig:",width=192 ] image of region surrounding sn2004et in ngc6946 .
our 50 pc extraction region is marked with the white circle .
_ center : _ the color - magnitude diagram resulting from our photometry extraction .
_ right : _ the cumulative fraction star formation from 50 myr ago to the present as measured by our cmd - fitting analysis described in section 2 .
the low precision makes it clear that we can not constrain progenitor masses with any fewer stars than this example.,title="fig:",width=192 ] lccccc sn1917a & 68.6 & 45.3 & 0.0 & 5.7 & 0.0 + sn1917a & 45.3 & 33.0 & 0.0 & 11.0 & 0.0 + sn1917a & 33.0 & 25.9 & 10.6 & 9.9 & 10.6 + sn1917a & 25.9 & 20.7 & 0.0 & 20.7 & 0.0 + sn1917a & 20.7 & 17.3 & 0.0 & 27.2 & 0.0 + sn1917a & 17.3 & 14.6 & 0.0 & 34.7 & 0.0 + sn1917a & 14.6 & 12.5 & 0.0 & 42.4 & 0.0 + sn1917a & 12.5 & 10.9 & 0.0 & 50.4 & 0.0 + sn1917a & 10.9 & 9.6 & 0.0 & 59.2 & 0.0 + sn1917a & 9.6 & 8.4 & 0.0 & 73.1 & 0.0 + ... & ... & ... & ... & ... & ... + [ pdfs ]
|
using resolved stellar photometry measured from archival _ hst _ imaging , we generate color - magnitude diagrams of the stars within 50 pc of the locations of historic core - collapse supernovae that took place in galaxies within 8 mpc .
we fit these color - magnitude distributions with stellar evolution models to determine the best - fit age distribution of the young population .
we then translate these age distributions into probability distributions for the progenitor mass of each sne .
the measurements are anchored by the main - sequence stars surrounding the event , making them less sensitive to assumptions about binarity , post - main - sequence evolution , or circumstellar dust .
we demonstrate that , in cases where the literature contains masses that have been measured from direct imaging , our measurements are consistent with ( but less precise than ) these measurements . using this technique ,
we constrain the progenitor masses of 17 historic sne , 11 of which have no previous estimates from direct imaging .
our measurements still allow the possibility that all sne progenitor masses are @xmath020 m@xmath1 .
however , the large uncertainties for the highest - mass progenitors also allow the possibility of no upper - mass cutoff .
|
fourteen years after its discovery , the double pulsar psr j0737 - 3039 is still studied extensively .
this system is composed of two neutron stars : an old , fast , mildly recycled 22.7 ms pulsar ( hereafter referred to as pulsar a ; @xcite ) and a younger and much slower pulsar with a period of 2.77 s ( hereafter pulsar b ; @xcite ) .
the compact objects orbit each other in a very tight orbit in only 2.4 hr , with mean orbital velocities of about 1 million km / h ( see e.g. * ? ? ?
* for a review ) .
psr j0737 - 3039 represents the most compact relativistic system and the only binary system in which both neutron stars have been detected as radio pulsars .
the keplerian and post - keplerian parameters offer a unique test for theories of strong field gravity @xcite , and constrain the masses of both neutron stars with high accuracy .
the physical riches of the double pulsar have been highlighted from the radio observations @xcite : whereas the pulse profile of pulsar a has been stable , the pulsed emission of pulsar b showed strong orbital flux and profile variations , before disappearing entirely in 2008 @xcite .
moreover , as a direct consequence of the inclination of a system observed nearly edge - on ( i@xmath4 ) , radio eclipses were detected when pulsar a passed behind pulsar b @xcite .
the study of the light curves testified to a new type of interaction between pulsar a s relativistic particle wind and pulsar b s magnetosphere in the previously unexplored close environment between the two neutron stars .
this aroused the interest to carry out high - energy observations in the x - ray band to investigate further the intra - binary environment .
early observations performed by the chandra and xmm - newton satellites @xcite pointed out the difficulty to constrain the origin of the multifold nature of the double pulsar s x - ray emission . the non - thermal pulsed emission from pulsar
a , although very soft ( power - law slope @xmath5 ) , is clearly predominant in the x - ray flux @xcite . in the frame of a first xmm - newton `` large program '' of 235 ks exposure in 2006 , pulsed x - ray emission from pulsar b was also detected in part of the orbit . due to its own low rotational energy loss ,
x - ray emission from pulsar b can be only powered externally through the spin - down energy of pulsar a @xcite .
this emission , consistent with thermal radiation of temperature kt@xmath6 ev and bolometric luminosity of @xmath7 erg s@xmath8 , was ascribed to the heating of pulsar b s surface by pulsar a s wind .
a hotter ( @xmath9 ev ) and fainter ( @xmath10 erg s@xmath8 ) thermal component , possibly originating from backfalling material heating the polar caps of either pulsar a or pulsar b , was also suggested from the spectral analysis by @xcite . in 2011 , a second deep xmm - newton observation of 370 ks was carried out .
comprehensive timing analysis over such a large time span has confirmed x - ray pulsed emission from pulsar b even after its radio disappearance @xcite .
the unusual phenomenology of pulsar b s x - ray emission includes orbital pulsed flux and profile variations as well as a loss of pulsar phase coherence on time scales of years , suggesting orbital - dependent penetration of pulsar a s wind plasma onto pulsar b s closed field lines .
furthermore , timing analysis of the full xmm - newton dataset provided first evidence of orbital flux variability ( @xmath11 ) , possibly involving a bow - shock between pulsar structures @xcite . an additional , possibly hard , spectral component associated with this intra - binary environment
is then expected . in this paper
, we present a comprehensive spectral analysis of the full xmm - newton dataset from both large programs , which allows us for the first time to constrain high - energy components above 4 kev . because of the weakness of the source and the consequent need to scrutinize the low - count statistics , an in - depth revision of the background subtraction procedure is provided .
the standard spectral analysis was coupled with an independent source detection method based on likelihood spatial analysis .
the complementarity of both approaches allows us to speculate for the first time on the possible presence of an iron line in the double pulsar .
clcccccc 2006 & pn & 120 & 115 & & 82 & 76 + & mos1 & 120 & 114 & & 112 & 103 + & mos2 & 120 & 114 & & 112 & 104 & + 2011 & pn & 129 & 106 & 128 & 88 & 68 & 84 + & mos1 & 108 & 103 & 128 & 104 & 84 & 112 + & mos2 & 129 & 103 & 128 & 122 & 84 & 112 + both xmm - newton large programs were carried out with similar instrumental configurations suitable for simultaneous spectral and high - resolution timing analysis of the double pulsar .
for these purposes , the epic - pn @xcite was used in `` small window '' mode whose time resolution of 5.67 ms compares with pulsar a s period of 22.7 ms .
the epic - mos cameras @xcite were also operated in `` small window '' mode with a time resolution of 0.3 s only suitable for pulsar b timing analysis .
medium and thin optical filters were applied to the epic instruments in 2006 and 2011 , respectively .
the thin filter nearly doubles the instrument effective area in the soft band ( @xmath12 kev ) at the expense of higher background contamination : early x - ray observations of the double pulsar showed a very soft source spectrum motivating the use of the thin filter for subsequent observations in order to improve the overall counting statistics .
the xmm - newton data were processed using the science analysis software ( sas ) version 12 .
the calibrated and concatenated epic event lists were obtained by running the meta - tasks epproc and emproc , standard pipeline tasks for epic - mos and epic - pn observations respectively .
we performed a barycentric correction on the event files thanks to the task barycen using the jpl de405 ephemeris to convert the time of arrival of photons from the satellite position to the solar - system barycenter . because of the faintness of the double pulsar in x - rays , careful attention was provided in the assessment and mitigation of background contamination and in the selection of source photons .
strong flares most likely associated with cosmic soft proton events , significantly affected different observations . in order to discard these episodes
, we produced the light curves between 10 and 12 kev to determine the flaring times .
we followed a background rejection procedure according to the general prescriptions by @xcite and further investigating and testing differing filtering options in analogy with the espfilt sas task , and reviewing the procedure adopted by @xcite and @xcite for the same data .
two different photon pattern selections were considered for the distribution of incident photons over the pixels of the epic - pn ( carrying out the analysis for both options ) : single and double events ( pattern @xmath13 ) or single events only ( pattern = 0 ) , slightly reducing the effective area . as for the epic - mos , single , double , triple and quadruple events were selected ( pattern @xmath14 ) .
ccccccccccc pn & 1 & @xmath15 & @xmath16 & @xmath17 & @xmath18 & @xmath19 & & & 239/206 & 1.16 + & 2 & - & & & @xmath20 & @xmath21 & @xmath22 & @xmath23 & 254/206 & 1.23 + & 3 & - & @xmath24 & @xmath25 & @xmath26 & @xmath27 & @xmath28 & @xmath29 & 237/204 & 1.16 + pn+mos & 1 & @xmath30 & @xmath31 & @xmath32 & @xmath33 & @xmath34 & & & 355/318 & 1.12 + & 2 & - & & & @xmath35 & @xmath36 & @xmath37 & @xmath38 & 367/318 & 1.15 + & 3 & @xmath39 1.0 & @xmath40 & @xmath41 & @xmath42 & @xmath43 & @xmath44 & @xmath45 & 346/316 & 1.10 + [ cols="^ " , ]
in order to exclude that the reported hard x - ray features are associated with background emission , the standard spectral analysis was coupled to an _ independent _ source detection method based on likelihood spatial analysis .
the most sensitive procedure to estimate the strength of sources in x - ray data in common with other photon - counting data entails constructing composite models of the spatial distribution of events combining source and background components @xcite .
the background near the double pulsar is made up of cosmic and instrumental contributions and is assumed to be flat .
a point spread function spatial model was applied to epic images between 4 and 8 kev , where the discrepancy between the low - energy spectral models and the hard x - ray component is the most relevant .
the table [ spatial - table ] reports the results separately and combined for each of the 5 long observations comprising the two large programs in 2006 and 2011 .
the exposure time reported , t(s ) , includes the dead - time correction of the small window mode .
the likelihood detection statistic , lnl , shows that the double pulsar is a weak source at high energies that is difficult to detect in a whole spacecraft revolution of nearly two days , but which is consistent in strength with the count rate accumulated in half a million seconds of elapsed time .
the combined log - likelihood detection statistic of 15.1 is a secure @xmath46 detection that amounts to @xmath47 photons above 4 kev from the x - ray source . unlike spectral analysis
, unpredictable background contamination can not significantly affect likelihood analysis results , since background counts are not distributed in the image accordingly to the source point spread function .
moreover , the spatial analysis testifies that there is no detection of source photons above 8 kev .
therefore our analysis proves that hard x - ray photons originate from the double pulsar in the @xmath48 kev energy range .
cccccccc 0405100201 & 54034.037 & 54035.423 & 82710 & @xmath49 & 3.0 + 0405100101 & 54036.037 & 54037.362 & 79067 & @xmath50 & 3.2 + 0670810201 & 55856.863 & 55858.360 & 89353 & @xmath49 & 3.1 + 0670810301 & 55858.828 & 55860.056 & 73249 & @xmath51 & 5.1 + 0670810101 & 55860.854 & 55862.339 & 88675 & @xmath52 & 3.8 + * combined * & & & * 413055 * & * * 0.00038**@xmath53 * 0.0002 * & * 15.1 * +
the x - ray emission from pulsars is typically attributed to a magnetospheric and/or a thermal origin .
because of the small separation between the two neutron stars , a strong interaction is expected between the wind of pulsar a and the magnetosphere of pulsar b , inducing possibly more complex mechanisms of the x - ray emission . in the following , we discuss the origin of the soft x - ray emission ( up to 4 kev ) and hard x - ray emission , including the detection of the spectral feature at @xmath54 kev .
the spectral analysis of the observations performed in 2011 , as well as combining 2006 + 2011 data , provides results comparable with those related to 2006 data alone .
different two - component ( black body plus power law or double black body ) or three - component spectral models ( double black - body plus power law ) can fit the whole dataset .
however , scenarios involving only thermal emission components are disfavoured , since the flux is most likely dominated by non - thermal pulsar a s magnetosphere emission .
a double black - body model implies that the x - ray flux of pulsar b would be about half of pulsar a s flux ( looking at black - body normalizations ) and this is inconsistent with x - ray fluxes derived from timing analysis by @xcite for which x - ray emission from pulsar a is over three times higher with respect to pulsar b. furthermore , pulsar phase - resolved analysis demonstrated that a non - thermal component clearly correlate with pulsar a s light curve peaks @xcite .
no significant spectral variability in soft x - rays is evident over 5 years . in particular , despite a significant relativistic spin precession occurred in the considered time span ( @xmath0 ) , no significant long - term variation of the soft pulsar a s spectral component is observed .
this further supports the hypothesis of a small misalignment between the spin axis and the orbital momentum axis as suggested by timing analysis of the same data @xcite .
the non - thermal x - ray emission of psr j0737 - 3039 is likely attributed to relativistic charged particles accelerated in the magnetosphere of pulsar a , via synchrotron emission in the outer gap model @xcite and/or via inverse compton scattering of the thermal x - ray photons in the polar cap model ( * ? ? ?
* and references therein ) .
no signature of a bow shock component ( power law with @xmath55 as suggested by canonical shock models ; see @xcite and references therein ) between the wind from pulsar a and pulsar b s magnetosphere is present in our soft spectral data , nor evidence of spectral variability as a function of the orbital phase . a weak orbital flux variability ( @xmath567% ) was detected in the timing analysis @xcite and it was possibly attributed to bow - shock emission . in the spectral analysis , the corresponding bow - shock flux could be concealed by the strong pulsar a s luminosity . as for the thermal emission
, it is expected to originate from hot spots around the magnetic poles ( polar caps ) of pulsar a and/or pulsar b , heated up to x - ray temperatures by relativistic particles streaming down onto the surface from the magnetosphere of the pulsar(s ) .
indeed , considering the characteristic ages of pulsar a ( @xmath57 myr ) and pulsar b ( @xmath58 myr ) , and the cooling evolution of such objects ( tsuruta 1998 ) , these pulsars are too old to have an intrinsic thermal radiation from their entire surface or from the atmosphere @xcite . the radio pulsar models @xcite predict polar cap radii @xmath59s@xmath60 km for a neutron star of radius @xmath61 km .
hence , theoretically , the polar cap radii associated with pulsar a and pulsar b are @xmath62 km and @xmath63 m , respectively .
these values are in agreement with black - body emission radii derived from our spectral results .
either assuming the presence of one or two black - body components in our spectral scenario , the expected luminosity of each component is @xmath6410@xmath65 erg / s , a value higher than pulsar b s spin - down luminosity .
this means that the black - body component luminosity would be @xmath66 @xmath67 .
thus , we can confirm that pulsar b s emission can only be powered by an external source ( i.e. pulsar a s spin - down energy ) . for any applicable spectral scenario in the @xmath68 kev range ,
the data above 4 kev are not correctly described by the two or three - component models ( see figure 2 ) .
the complementarity of the independent spatial and spectral approaches allows us to claim for the first time hard x - ray emission from the double pulsar .
this hard emission is well fitted by a gaussian line at @xmath69 kev in the 2006 data .
no other simple model , such as a power law or a black body , is able to fit the high - energy excess .
emission at these energies indicates the likely presence of iron line emission .
the firm detection of hard x - ray emission in the @xmath48 kev range coupled with no source detection @xmath70 kev through spatial analysis , strengthen the reliability of our spectral feature detection .
the presence of such a feature has major importance since it represents the first evidence of an x - ray spectral line around a supposedly `` non - accreting system '' . as for the second long - observation performed in 2011
, we can not claim evidence of a high - energy spectral line , because of a poorer signal - to - noise ratio in hard x - rays mostly due to the use of the thin epic - pn filter . in any case
, the detection of high - energy photons from the double pulsar is also confirmed in these data through spatial analysis , though spectral parameters can not be precisely constrained in the 2011 data alone . as described in section 3 ,
line variability is compatible with our data , though not required .
associated with emission at @xmath3 kev , the fe k@xmath2 line is widely considered as a particular property of accretion - powered sources in which accreting matter forms a disk .
it has been detected in the x - ray spectra of many supermassive black holes and stellar - mass - black - hole and neutron - star x - ray binaries @xcite , and in magnetic cataclysmic variables @xcite .
the emission line is thought to originate in the innermost parts of the accretion disk , where strong relativistic effects broaden and distort its shape .
the production of such a line is fairly simple requiring a source of thermal or non - thermal hard x - rays to illuminate the disk . in this context
our detection of the fe k@xmath2 line in the double pulsar may indicate that at least one of the system components is surrounded by a gaseous disk .
however , what could be the origin of such a disk ? the double pulsar , as the other few confirmed double neutron star binaries , is the descendant of a high - mass binary system that has survived two supernova explosions @xcite .
the two stars under - filled their roche lobes and lose matter at a relatively low rate through winds of relativistic particles .
the conventional scenario of accretion disk formation based on intensive mass - exchange between the system components in this case is not applicable .
however , a relic disk of matter captured from ejecta of the second supernova explosion could be present .
the disk could be formed in two ways .
the first is a so - called fall - back accretion scenario in which the inner parts of the ejecta return towards the newly - formed neutron star forming a disk around its magnetosphere @xcite .
the mass of the remnant disk in this case is about @xmath71 m@xmath72 .
the second scenario suggests the disk to form around the magnetosphere of the old neutron star born during the first supernova explosion as it moves through the supernova ejecta of its exploded companion . as recently pointed out @xcite , the mass of the remnant disk within this scenario can reach @xmath73 m@xmath72 .
thus , current views of the process of supernova explosions do not exclude that one or even both neutron stars could be surrounded by relic disks .
observations give no direct evidence that the neutron stars in the double pulsar are accreting material from a disk ( no radio pulsar timing noise reported by @xcite and @xcite ) .
it is plausible that the relic disk is currently in a dead state @xcite in which mass - transfer towards the star is suppressed by the centrifugal barrier at the boundary of the stellar magnetosphere .
the interaction between the disk and the magnetic field of the radio pulsar in this case results in generation of electrical currents , which connect the inner parts of the disk with polar cap regions at the surface of the neutron star @xcite .
as the currents dissipate in small areas of the polar caps the temperature increases dramatically , enhancing the hot spots visible in x - rays . in contrast , the temperature of the disk increases little since current dissipation occurs over a much larger area .
the disk in this case could contribute to the pulsar emission at lower frequencies by a significant fraction of the pulsar spin - down power .
our spectral analysis of the xmm - newton data demonstrates that the thermal x - ray emission is indeed incompatible with a disk origin , since the disk radius would be too small ( even assuming a disk black - body model _ diskbb _ ) . the surface area associated with the black bodies at 0.1 and 0.3
kev is of the order of 600 and 100 m@xmath74 , respectively , indicating most likely thermal emission from the polar cap(s ) .
the putative disk is most likely colder and thus invisible in x - rays .
it manifests itself only through the iron line that requires external heating
. this could be by illumination by a hard continuum x - rays from pulsar a , although this might be too weak or anisotropic to be observed .
alternatively , unlike accretion - powered pulsars , the disk could be illuminated by high - energy particles from pulsar a s wind . in this case , about @xmath75 of spin - down energy of pulsar a could power the k@xmath2 line .
interestingly enough , this is the same amount of energy powering pulsar b s x - ray emission
. iron line emission could also be in principle associated with an accumulation of matter trapped in the shock layer foreseen between pulsar a s wind and the magnetosphere of pulsar b. a hypothetical spectral line from the surface or atmosphere of the neutron star is not favored , since the gravitational redshift would induce a line at @xmath76 kev accordingly to the lower limit on the neutron stars mass - radius ratio @xcite .
an idea about a relic disk to surrounding a radio pulsar has been invoked by a discovery of two planet - mass companions around the millisecond pulsar psr 1257 + 12 @xcite .
further studies eventually confirmed that the components are likely to originate from a durable remnant disk , which in a previous epoch could be in an accretion or dead state .
the presence of remnant disks around isolated neutron stars has later been suspected in studies of anomalous x - ray pulsars ( axps ) and soft gamma - ray repeaters ( sgrs ) @xcite .
iron line emission at 6.4 kev was detected during a gamma - ray burst in sgr 1900 + 14 @xcite .
the presence of a relic disk was also been invoked to explain the ir / optical radiation discovered from two axps @xcite . only upper limits to the optical flux
were derived for the double pulsar @xcite , which was instead detected in the fuv @xcite . although the thermal or non - thermal nature of this emission is still unclear , it is not related to magnetospheric or surface x - ray pulsar emission since it does not match with extrapolated x - ray spectra .
it is in principle compatible with a relic disk of luminosity l@xmath77 erg / s .
the possible weakening or even disappearance of the spectral feature at @xmath54 kev in the data obtained in 2011 could be related to actual variability of the fe line because of changes in the geometry or precession of the disk , which is not illuminated in the same way over time .
the parameters of interaction between the wind of pulsar a and the magnetosphere of pulsar b are expected to be different between 2006 and 2011 , as manifested by radio disappearance of pulsar b in 2008 .
as the orbit is seen edge - on , changes on the disk could in principle also affect radio pulsar timing ( e.g. observed dispersion measure , pulse shapes and eclipses ) when it crosses the line - of - sight to the pulsars .
a multi - messenger approach is then required to probe the putative relic disk and its time - dependent geometry .
two large programs of observations of the relativistic double pulsar were performed by xmm - newton in 2006 and 2011 .
the total observing time of about 600 ks offers the opportunity to investigate the low luminosity source with a high number of @xmath78 source photon events .
no significant spectral variations in soft x - rays can be claimed in the 5 years between 2006 and 2011 , substantially confirming the previously reported results based on the analysis of the 2006 data alone . two - component ( black body plus power law ) or three - component spectral models ( double black - body plus power law ) can in principle fit the whole dataset , although the latter models seem better suited according to phase - resolved spectral analysis @xcite .
however , for any applicable spectral scenario in the @xmath68 kev range , the data above 4 kev are not correctly described by the models .
we investigated for the first time the high - energy part of the spectrum of the double pulsar .
an intriguing emission feature , possibly variable , is detected at about @xmath54 kev .
spatial analysis confirms the emission of hard energy photons between 4 and 8 kev ascribed to the source .
this feature is most likely attributed to iron line emission , testifying the presence of a relic disk that survived the supernova explosions that terminated the lives of the double pulsar s stellar progenitors .
we acknowledge the referee for his / her constructive suggestions that have helped to improve the content of this paper . this work is based on observations obtained with _ xmm - newton _ , an esa science mission with instruments and contributions directly funded by esa member states and the usa ( nasa ) .
the authors thank the members of the _ xmm - newton _ user support group .
e. egron and m. marongiu acknowledge financial support from the autonomous region of sardinia through a research grant under the program crp-25399 po sardegna fse 2007 - 2013 , l.r .
7/2007 , promoting scientific research and innovation technology in sardinia .
ikhsanov acknowledges support of the russian scientific foundation under the grant no .
14 - 50 - 00043 .
arnaud , k. a. 1996 , in asp conf .
astronomical data analysis software and systems v , ed .
g. h. jacoby & j. barnes , 101 , 17 arons , j. 1981 , , 248 , 1099 beskin , v. s. , gurevich , a. v. , & istomin , y. n. 1993 , physics of pulsar magnetosphere ( cambridge : cambridge univ.press ) bisnovatyi - kogan , g. s. & ikhsanov , n. r. 2015 , astronomy reports , 59 , 503 breton , r. p. , kaspi , v. m. , kramer , m. et al .
2008 , science , 321 , 5885 , 104 burgay , m. , damico , n. , possenti , a. , et al .
2003 , , 426 , 531 cackett , e. m. , miller , j. m. , ballantyne , d. r. 2010 , , 720 , 205 campana , s. , possenti , a. , & burgay , m. 2004 , , 613 , l53 cash , w. 1979 , , 228 , 939 chatterjee , s. , gaensler , b. m. , melatos , a. , et al .
2007 , , 670 , 1301 chatterjee , p. , hernquist , l. , narayan , r. 2000 , , 534 , 373 cheng , a. f. & ruderman , m. a. 1980 , , 235 , 576 cheng , k. s. & zhang , l. 1999 , , 515 , 337 chevalier , r. a. 1989 , , 346 , 847 colgate , s. a. 1971 , , 163 , 221 de luca , a. & molendi , s. 2004 , , 419 , 837 durant , m. , kargaltsev , o. & pavlov , g. g. , 2014 , , 783 , 22 ezuka , m. 1999 , , 120 , 277 ferraro , f. r. , mignani , r. p. , pallanca , c. et al .
2012 , , 749 , 84 granot , j. & mszros , p. 2004
, , 609 , l17 iacolina m. n. , pellizzoni , a. , egron , e. et al .
2016 , , 824 , 87 kaplan , d. l. , chakrabarty , d. , wang , z. & wachter , s. 2009 , , 700 , 149 kaspi , v. m. , ransom , s. m. , backer , d. c. , et al .
2004 , , 613 , l137 kramer , m. , stairs , i. h. , manchester , r. n. , et al .
2006 , science , 314 , 97 kramer , m. & stairs , i. h. 2008 , , 46 , 541 kramer , m. & wex , n. 2009 , classical and quantum gravity , 26 , 7 lyne , a. g. , burgay , m. , kramer , et al .
2004 , science , 303 , 1153 lyutikov , m. 2004 , , 353 , 1095 mclaughlin , m. a. , lyne , a. g. , lorimer , d. r. , et al .
2004 , , 616 , l131 michel f. c. , & dessler , a. j. 1981 , , 251 , 654 michel f. c. , & dessler , a. j.1983 , , 303 , 48 michel f. c. , & dessler , a. j. 1985 , sience , 228 , 1015 michel f. c. 1985 , , 290 , 721 michel f. c. 1988 , , 333 , 644 miller , j. m. 2007 , , 45 , 441 pellizzoni , a. , de luca , a. , mereghetti , s. , et al .
2004 , , 612 , l49 pellizzoni , a. , tiengo , a. , de luca , a. , et al .
2008 , , 679 , 664 perera , b. b. p. , mclaughlin , m. a. , kramer , et al .
2010 , , 721 , 1193 pollock , a. m. t. 1987 , , 320 , 283 possenti , a. , rea , n. , mclaughlin , m. a. , et al .
2008 , , 680 , 654 protassov , r. , van dyk , d. a. , connors , a. et al .
2002 , , 571 , 545 stairs , i. h. 2004 , science , 304 , 547 strohmayer , t. e. & ibrahim a. i. 2000 , , 537,111 strder , l. , briel , u. , dennerl , k. , et al .
2001 , , 365 , l18 sturner , s. j. , skibo , j. g. , dermer , c. d. , mattox , j. r. 1997 , , 490 , 619 sunyaev , r. a. & shakura , n. i. 1977 , soviet astronomy letters 3 , 138 verbiest , j. p. w. , weisberg , j. m. , chael , a. a. , et al .
2012 , , 755 , 39 tsuruta , s. 1998 , , 292 , 1 turner , m. j. l. , abbey , a. , arnaud , m. , et al .
2001 , , 365 , l27 wang , z. , chakrabarty , d. & kaplan , d. l. , 2006 , , 440 , 772 wilms , j. , allen , a. , & mccray , r. 2000 , , 542 , 914 wolszczan , a. & frail , d. a. 1992 , , 355 , 145 zavlin , v. e. 2009 , theory of radiative transfer in neutron star atmospheres and its applications , astrophysics and space science library , ed .
becker w. , 357 , 181 zeldovich , y. b. , ivanova , l. n. & nadezhin , d. k. 1972 , soviet astronomy 16 , 209 zhang , b. & harding , a. k. 2000 , , 532 , 1150 zhao , h. h. & song , l. m. 2012 , proceedings iau symposium n.290 , 8 , pp 369 - 37
|
we present a long - term spectral monitoring of the unique double pulsar binary psr j0737 - 3039 corresponding to two `` large programs '' performed by xmm - newton in 2006 and 2011 .
spectral variability of pulsar emission in soft x - rays is not evident over 5 years , despite the significant relativistic spin precession in the considered time span ( @xmath0 ) .
we provide , for the first time , evidence of hard x - ray emission from the system in the @xmath1 kev energy band .
the standard spectral analysis was coupled to the energy dependent spatial analysis to confirm this excess , most likely ascribed to iron line emission .
the fe k@xmath2 emission line at @xmath3 kev was previously unheard - of in non - accreting binary systems and could testify to the presence of a relic disk that survived the supernova explosions that terminated the lives of the double pulsar s stellar progenitors .
the existence of a relic disk in this system reinforces speculation about the presence of similar structures around other peculiar classes of isolated neutron stars .
|
the x - ray luminosity function of galaxy clusters in the nearby universe provides a powerful cosmological probe .
the observed luminosity function , @xmath7 , can be combined with a relation linking the observed x - ray luminosity and mass , and the mass function , @xmath8 , predicted by simulations , to obtain tight constraints on the combination of cosmological parameters @xmath1 and @xmath0 , where @xmath1 is the mean matter density of the local universe and @xmath0 is the root - mean - square ( rms ) variation of the density field smoothed by a top hat window function of size 8@xmath9mpc .
observationally , the keys to such studies are precise determinations of the local x - ray luminosity function of clusters and the relation linking the observed x - ray luminosities and total masses .
x - ray selection currently offers the best way to identify massive galaxy clusters , and the local x - ray luminosity function has now been precisely determined by the bcs ( ebeling 1997 ; ebeling 2000 ) and reflex ( bhringer 2002 ) studies .
the flux - limited bcs and reflex samples , which are based on data from the rosat all - sky survey ( rass ; trmper 1993 ) , together include @xmath10 clusters and cover approximately two thirds of the sky .
recently , significant effort has also been invested into measuring the local temperature function of clusters ( markevitch 1998 ; pierpaoli 2001 ; ikebe 2002 ) , which offers a complementary method for determining cosmological parameters . at present , however , the temperature function samples are significantly smaller than the combined bcs - plus - reflex luminosity function data set , and the analysis of the temperature function is complicated by the fact that these samples are selected according to x - ray flux as well as temperature , requiring the use of both mass - temperature and temperature - luminosity relations in the analysis .
recent years have also seen significant efforts directed towards a precise calibration of the ` virial ' relations linking the observed luminosities , temperatures and masses of galaxy clusters ( horner , mushotzky & scharf 1999 ; nevalainen , markevitch & forman 2000 ; finoguenov , reiprich & bhringer 2001 ; allen , schmidt & fabian 2001b ; sanderson 2003 ) .
in particular , the launch of the chandra x - ray observatory has permitted the first precise measurements of the temperature and mass profiles of relaxed clusters from x - ray data . using a combination of chandra and gravitational lensing data , allen ( 2001b ) confirmed that luminous , relaxed galaxy clusters follow the simple scaling relations predicted by theory , but that the normalization of the observed mass - temperature relation measured within @xmath11 ( where the mean enclosed mass density is 2500 times the critical density of the universe at the redshifts of the clusters ) is approximately 40 per cent lower than predicted by standard adiabatic simulations .
this highlights the likely importance of additional physics such as cooling and pre - heating in the intracluster gas ( see also pearce 2000 ; thomas 2002 ; voit 2002 ; muanwong 2002 ) .
theoretically , the primary requirement for cosmological studies using the observed luminosity and/or temperature functions of clusters is a precise prediction of the mass function .
this has now been achieved for flat @xmath6cdm ( and @xmath12cdm ) cosmologies using the hubble volume simulations of jenkins ( 2001 ) and evrard ( 2002 ) .
in this paper we present precise constraints on @xmath0 and @xmath1 based on the observed local luminosity function of the most x - ray luminous clusters in the extended bcs ( ebeling 2000 ) and reflex samples , and a new calibration , using pointed chandra and rosat x - ray observations and weak gravitational lensing results , of the mass - luminosity relation linking the masses of clusters measured within @xmath13 to their total @xmath14 kev rosat luminosities .
having determined our combined constraint on @xmath0 and @xmath1 , we show that the degeneracy between these parameters can be broken using chandra results on the x - ray gas mass fractions in the most dynamically relaxed clusters . including gaussian priors on the mean baryon density of the universe ( @xmath15 ; omeara 2001 ) , the hubble constant ( @xmath16 ; freedman 2001 ) , and a theoretical bias factor ( @xmath17 ; bialek , evrard & mohr 2001 ) relating the asymptotic baryon fraction in the most x - ray luminous clusters to the mean value for the universe as a whole , we obtain @xmath4 and @xmath5 ( marginalized 68 per cent confidence limits for an assumed flat @xmath6cdm cosmology ) .
we compare our results to other measurements based on the local number density of clusters , evolution of the x - ray luminosity and temperature functions , the 2df galaxy redshift survey , cosmic microwave background anisotropies , and measurements of cosmic shear . throughout this paper , a flat @xmath6cdm cosmology with @xmath18
is assumed . in order to facilitate a direct comparison with previous x - ray studies ,
results on the masses , x - ray luminosities and x - ray gas mass fractions of individual clusters are quoted for a hubble parameter @xmath19@xmath20 or , equivalently , @xmath21@xmath22 .
jenkins ( 2001 ) show that the predicted mass function of galaxy clusters of mass @xmath23 at redshift @xmath24 can be written as a function of @xmath25 , where @xmath26 is the variance of the linearly evolved density field smoothed by a spherical top - hat filter of comoving radius @xmath27 , enclosing a mass @xmath28 . here
@xmath29 is the mean comoving matter density of the universe , @xmath30(0 ) is the mean , present matter density in units of the critical density , and @xmath31 is the critical density at redshift zero . using a spherical overdensity algorithm to measure the masses of clusters within radii @xmath13 , where the mean enclosed mass density is 200 times
the critical density of the universe at the redshift of interest , evrard ( 2002 ) show that for a flat @xmath6cdm cosmology , the mass fraction @xmath32 can be written as @xmath33 \,,\ ] ] where , for @xmath34 and @xmath35 , @xmath36 , @xmath37 and @xmath38 .
evrard ( 2002 ) also provide simple interpolations for @xmath39 and @xmath40 for other values of @xmath1 and @xmath24 .
the differential number density of clusters with mass @xmath23 at redshift @xmath24 is @xmath41 where @xmath42 is the mean mass density of the universe at redshift @xmath24 .
following viana & liddle ( 1999 ) , we write @xmath43 where @xmath44\,\ ] ] and @xmath45 is the shape parameter of the cold dark matter transfer function . following sugiyama ( 1995 )
, we set @xmath46 where @xmath47k is the temperature of the cosmic microwave background ( mather 1994 ) and @xmath48 is the mean , present - day baryon density in units of the critical density ( omeara 2001 ) .
for this calculation , we set @xmath49 and fix the primordial spectral index @xmath50 .
the redshift - dependent quantity @xmath51 is related to its present value , @xmath52 , by @xmath53 where , for a flat @xmath6cdm universe @xmath54^{-1 } \,\ ] ] and @xmath55 viana & liddle ( 1996 ) . combining equations 2 and 3 ,
we obtain @xmath56 in section [ section : ml ] we show that for clusters with x - ray luminosities , @xmath57 , exceeding @xmath58 in the @xmath14 kev rosat band , the mass @xmath23 measured within @xmath13 can be related to the x - ray luminosity by a power - law model of the form @xmath59 } ^\alpha,\ ] ] where the best fit values , uncertainties on @xmath60 and @xmath61 , and the intrinsic scatter in the relation , are determined from chandra and rosat x - ray data and weak gravitational lensing results . the evolution parameter @xmath62 accounts for the expected cosmological scaling of the relationship ( bryan & norman 1998 ; mathiesen & evrard 2001 ) . applying the chain rule to equations 9 and 10 ,
we obtain @xmath63^{\alpha+1 } f(\sigma^{-1}).\ ] ] the predicted differential luminosity function ( the comoving number density of clusters at redshift @xmath24 with luminosities in an interval @xmath64 around @xmath57 ) given by equation 12 can be compared with the observed x - ray luminosity function from the bcs and reflex studies to constrain the combination of cosmological parameters @xmath65 and @xmath52 , hereafter referred to as @xmath1 and @xmath0 , respectively .
for this study , we concentrate on the local ( @xmath66 ) x - ray luminosity function and restrict ourselves to the most luminous clusters , with @xmath67 in the @xmath14 kev rosat band ( for a flat @xmath6cdm cosmology with @xmath34 and @xmath68 ) .
this selection is facilitated by the large sample size and well - determined selection functions of the bcs and reflex data sets .
the restriction to high luminosities reduces systematic uncertainties by matching the luminosity range of the luminosity function data to the range over which the mass - luminosity relation has been calibrated ( section [ section : ml ] ) . at lower luminosities ,
the effects of pre - heating and cooling in the intracluster gas are expected to become important and cause the mass - luminosity and mass - temperature relations to deviate from simple power - law forms ( cavaliere , menci & tozzi 1997 ) .
since the most massive clusters provide the most powerful constraints on cosmological parameters , relatively little information is lost by restricting ourselves to the largest systems . the binned x - ray luminosity functions for clusters with @xmath69 from the northern extended bcs ( ebeling 2000 ) and southern reflex ( bhringer 2002 ) studies are summarized in table [ table : lumfunc ] .
the mean redshift of the bcs clusters in this luminosity range is @xmath70 .
, the mean redshift of the bcs clusters with @xmath71 . shifting this redshift by @xmath72 does not significantly change the results . ]
-0.1truein .the observed , binned x - ray luminosity function of the most x - ray luminous ( @xmath71 ) galaxy clusters from the extended bcs and reflex studies .
column 2 gives the mean @xmath14 kev x - ray luminosity for each bin , @xmath73 , in @xmath74 .
error bars indicate the bin boundaries .
column 3 gives the number of clusters in each bin and column 4 the comoving space density in @xmath75mpc@xmath76(@xmath77)@xmath78 .
a @xmath6cdm cosmology with @xmath34 and @xmath68 is assumed . [ cols="^,^,^,^,^ " , ] in constructing the mass - luminosity relation , we have used pointed rosat observations to determine the total , intrinsic @xmath14 kev luminosities of the clusters .
this minimizes systematic uncertainties by matching the observing band and ( as far as possible ) detector technology to that used for the rass observations .
the agreement between the bcs and reflex fluxes , which are based on rass data , and the fluxes determined from deep , pointed rosat observations is good ( ebeling 1998 ; bhringer 2002 ) .
the details of the pointed rosat observations are summarized in table [ table : rosat ] .
the data were analysed using the xselect package ( version 2.0 ) and xspec ( version 11.0 ) .
the emission - weighted temperatures and metallicities of the clusters were set to the values measured with chandra . where chandra data were not available ,
the metallicity was set to 0.3 solar and the temperature was determined iteratively from the luminosity - temperature relation of allen & fabian ( 1998 ) .
note , however , that the precise settings of the temperatures and metallicities have little effect on the measured @xmath14 kev luminosities for such hot , massive clusters .
the absorbing column densities were set to the galactic values determined by dickey & lockman ( 1990 ) .
the intrinsic @xmath14 kev luminosities for a @xmath6cdm cosmology with @xmath79 and @xmath80 ( the same cosmology assumed in the bcs and reflex luminosity functions in table [ table : lumfunc ] ) are listed in table [ table : rosat ] . the mass - luminosity relation for
the 17 x - ray luminous clusters in our sample , measured within radii @xmath13 corresponding to a density contrast @xmath81 with respect to the critical density of the universe at the redshifts of the clusters , is shown in fig .
[ fig : ml ] .
those clusters with mass measurements from chandra x - ray data are indicated by filled circles .
clusters with weak lensing mass measurements are marked with open squares .
the four clusters with chandra mass measurements and consistent weak lensing results are indicated by filled circles surrounded by open squares .
a flat @xmath6cdm cosmology with @xmath34 and @xmath82 is assumed .
we have fitted the data using a model of the form @xmath83~ = ~\alpha\ , { { \rm log_{10 } } \left [ \frac{l}{e(z ) \,10^{44}\,h_{50}^{-2}{\rm \ergps } } \right ] } \nonumber\\ + ~\,{\rm log_{10}}\left[\frac{m_0}{h_{50}^{-1}\msun}\right ] .
\hspace{2.0cm}\end{aligned}\ ] ] using the bces(@xmath84 ) estimator of akritas & bershady ( 1996 ) , which accounts for errors in both axes and the presence of possible intrinsic scatter , we obtain best - fitting values and 68 per cent confidence limits from @xmath85 bootstrap simulations of log@xmath86 and @xmath87 .
the distribution of @xmath60 and @xmath61 values are shown in fig .
[ fig : m0_alpha ] .
the observed slope is in good agreement with the expected slope of the mass_bolometric _ luminosity relation of @xmath88 , from models of simple gravitational collapse . the rms scatter of the observed log[@xmath89 values about the best fitting curve is @xmath90 for the full sample of 17 clusters , 0.15 for the 10 dynamically relaxed clusters studied with chandra , and @xmath91 for the 8 clusters with weak lensing measurements from dahle ( 2002 ) .
the scatter in the full sample is similar to that measured by reiprich & bhringer ( 2001 ) using asca and rosat x - ray data .
note , however , that our best - fitting curve implies a slightly higher x - ray luminosity for a given mass .
using the theoretical prescription for the mass function of galaxy clusters described in section [ section : theory ] , the observed bcs and reflex x - ray luminosity functions summarized in table 1 , and the mass - luminosity data discussed in section [ section : ml ] , we can determine @xmath0 as a function of @xmath1 .
in determining our results on cosmological parameters , we have used a monte carlo approach . for each iteration of the code
, we construct a random bootstrap sample of the @xmath92 and @xmath93 values listed in tables @xmath94 and examine a grid of ( @xmath95 ) @xmath1 and @xmath0 values , covering the plane @xmath96 and @xmath97 . for each value of @xmath1 , we scale the mass and luminosity values appropriately and determine the best - fitting mass - luminosity relation . for each @xmath1 , @xmath0 parameter pair , we construct a model x - ray luminosity function , including the effects of random scatter in the mass - luminosity relation , which we characterize using a log - normal distribution . the model x - ray luminosity function is then compared with the observed bcs and reflex data , also scaled to the appropriate cosmology , and the @xmath98 difference between the two is calculated . in this way , the best fitting @xmath1 , @xmath0 pair for the grid is determined , which provides us with a single sample result .
the whole process is repeated for @xmath85 iterations to produce the final results , discussed below .
[ fig : contour1 ] shows the 68.3 and 95.4 per cent confidence contours in the @xmath99 plane from one million iterations of the monte carlo code . the results exhibit the well known degeneracy between @xmath0 and @xmath1 , which can be approximated ( for @xmath100 ) by the simple fitting formula @xmath101 .
our analysis favours low values for @xmath1 : marginalizing over @xmath0 , we find @xmath102 at 68 per cent confidence .
note that when neglecting the effects of scatter and uncertainties in the normalization and slope of the mass - luminosity relation , we obtain the best fit ( @xmath103 for 4 degrees of freedom ) for @xmath104 and @xmath105 .
the usual approach , discussed in the literature , to break the degeneracy between @xmath0 and @xmath1 ( fig .
[ fig : contour1 ] ) is to use the redshift evolution of the luminosity and/or temperature function of clusters , which depends strongly on the mean mass density of the universe ( see references in section [ section : evolution ] ) .
however , this approach is both observationally challenging in terms of identifying complete , high - redshift cluster samples , and prone to systematic uncertainties due to potentially increased levels of dynamical activity and contaminating agn emission at high redshifts , which introduce additional scatter into the mass - luminosity and mass - temperature relations .
fortunately , the chandra data offer a powerful alternative method to break the degeneracy between @xmath0 and @xmath1 using the observed x - ray gas mass fractions , @xmath108 , in the clusters and their apparent redshift dependence ( white & frenk 1991 ; white 1993 ; sasaki 1996 ; pen 1997 ; ettori & fabian 1999 ; allen 2002a ; erdogdu , ettori & lahav 2002 ; ettori , tozzi & rosati 2003 ) .
the matter content of rich clusters of galaxies is thought to provide an almost fair sample of the matter content of the universe ( white 1993 ) . the observed ratio of baryonic to total mass in clusters is therefore expected to closely match the ratio of the cosmological parameters @xmath109 , where @xmath110 is the mean baryon density of the universe .
the apparent redshift dependence of the @xmath108 measurements arises from the fact that the measured @xmath108 values depend upon the assumed angular diameter distances to the sources as @xmath111 .
thus , although we expect the measured @xmath108 values to be invariant with redshift , this will only appear to be the case when the assumed cosmology matches the true , underlying cosmology .
the observed @xmath108 profiles for the ten relaxed clusters studied with chandra , for an assumed @xmath80 standard cold dark matter ( scdm ) cosmology with @xmath112 and @xmath113 , are shown in fig .
[ fig : fgas_r ] .
( the six clusters previously studied by allen 2002a are shown in a lighter shading . ) with the possible exception of abell 963 , we see that the @xmath108 profiles appear to have converged , or be close to converging , within @xmath11 .
we note that the data for the two nearest clusters , abell 478 and pks0745 - 191 , do not extend to @xmath11 .
however , their @xmath108 profiles appear to be close to converging within @xmath114 . fig .
[ fig : fgas_z ] shows the @xmath108 measurements at @xmath11 ( or at the outermost radii studied for abell 478 and pks0745 - 191 ) as a function of redshift , for the nine clusters with convergent @xmath108 profiles . following allen ( 2002a )
, we have fitted these data with the model @xmath115^{1.5}\end{aligned}\ ] ] and determined the best - fitting value of @xmath1 , for an assumed flat @xmath6cdm cosmology . the parameter @xmath116 is a bias factor that is motivated by gasdynamical simulations , which suggest that the baryon fraction in clusters is slightly depressed with respect to the universe as a whole ( cen & ostriker 1994 ; eke , navarro & frenk 1998a ; frenk 1999 ; bialek 2001 ) .
we include a gaussian prior on the bias factor , @xmath17 , a value appropriate for hot ( @xmath117 kev ) , massive clusters in the redshift range @xmath118 from the simulations of bialek ( 2001 ) .
we also include gaussian priors on the hubble constant , @xmath16 , the final result from the hubble key project reported by freedman ( 2001 ) , and @xmath119 ( omeara 2001 ) , from cosmic nucleosynthesis calculations constrained by the observed abundances of light elements at high redshifts . the constraints on @xmath1 determined from this analysis are shown as the dark , solid curve in fig .
[ fig : fgas_chi ] .
we obtain @xmath120 at 68 per cent confidence .
also shown ( dashed curve ) are the results obtained when fixing the bias parameter @xmath121 , for which @xmath122
. we can now combine ( by multiplying the relevant probability densities ) the constraints on @xmath1 from fig .
[ fig : fgas_chi ] with the joint constraints on @xmath0 and @xmath1 , shown in fig .
[ fig : contour1 ] , to obtain our final results on @xmath0 and @xmath1 .
these are shown in fig .
[ fig : contour2 ] . using the gaussian prior on the bias parameter ,
we obtain @xmath4 and @xmath5 ( marginalized 68 per cent confidence limits ) . with
@xmath121 fixed , we obtain @xmath123 and @xmath124 . an important aspect of the present work is the reduced level of systematic uncertainty with respect to most previous studies based on the local abundance of galaxy clusters . in the first case , the independent bcs and reflex studies have well determined selection functions ( ebeling 1998 , 2000 ; bhringer 2002 ) and provide precise , consistent results on the local x - ray luminosity function . with the large size of the combined bcs - plus - reflex data set ( which covers two thirds of the sky ) we have been able to limit our analysis to the most luminous clusters , with @xmath125 ( for an @xmath34 , @xmath68 , cosmology ) , for which the mass - luminosity relation has been calibrated using chandra and rosat x - ray data and weak lensing observations ( section [ section : ml ] ) .
the complicating effects of cooling and pre - heating are minimized for such massive clusters and a simple power - law model provides a reasonable description of the mass - luminosity relation ( fig . [ fig : ml ] ) , albeit with some residual intrinsic scatter that is in part related to the different dynamical histories of the clusters and residual systematic effects in the mass measurements ( sections [ section : dahle ] , [ section : ml ] ; see also below ) . repeating the analysis using only the bcs luminosity function data , or only the reflex data , or using a higher luminosity cut in the luminosity function ( @xmath126 ) , leads to consistent results ( although with larger statistical uncertainties ) .
consistent results on cosmological parameters are also obtained if we replace the evrard ( 2002 ) critical spherical overdensity mass function ( section [ section : theory ] ) with the jenkins ( 2001 ) mean spherical overdensity ( so324 ) prescription , scaling the cluster masses accordingly .
the dominant statistical and systematic uncertainties in the analysis are associated with the mass - luminosity relation .
however , care has been taken to minimize the systematic uncertainties .
in particular , we have limited our study of the mass - luminosity relation to clusters with precise mass measurements from chandra or wide - field gravitational lensing studies .
the clusters studied with chandra have been selected from the rass as the most luminous , dynamically relaxed ( in terms of their x - ray and optical morphologies ) clusters known .
the relaxed natures of these systems means that they are the clusters for which x - ray mass measurements are most reliable . for abell 963 , 1835 , 2390 and rxj1347 - 1145
, independent confirmation of the chandra mass measurements at @xmath13 is available from weak gravitational lensing studies ( dahle 2002 ; squires 1996 ; fischer & tyson 1997 ; allen 2001a , 2002b ) .
a programme of weak lensing measurements for the other clusters in our chandra sample is underway ( gray , in preparation ) .
in addition , consistent strong lensing mass measurements are available for abell 1835 , 2390 , rxj1347.5 - 1145 , ms2137.3 - 2353 and pks0745 - 191 ( schmidt 2001 , allen 2001a , 2002b , schmidt , in preparation ) . the presence of significant non - thermal pressure support on large spatial scales in these clusters can therefore be excluded .
as well as increasing the size of our calibration sample , the inclusion of the dahle ( 2002 ) weak lensing data is important in that it allows us to examine the distribution of less dynamically relaxed systems in the mass - luminosity plane .
whereas x - ray data can only be used to obtain precise mass measurements for dynamically relaxed clusters ( the x - ray mass measurements are based on the assumption of hydrostatic equilibrium in the x - ray gas ) , lensing mass measurements are essentially independent of the dynamical state of the gravitating matter and so can be extended to systems undergoing merger events .
the dahle ( 2002 ) sample includes clusters with a range of dynamical states .
inspection of rosat and chandra images of these systems shows that many exhibit significant dynamical activity , with abell 141 , 209 , 520 , 1351 and 1576 undergoing major merger events .
( similar conclusions are drawn by dahle 2002 from their optical data ) .
the bcs and reflex samples , from which the x - ray luminosity functions have been constructed ( section [ section : bcs ] ) , include all clusters above the respective x - ray flux limits , independent of their dynamical states .
it is therefore important that the mass - luminosity relation provides a fair translation between luminosity and mass for all clusters included in the samples .
although the 17 clusters in our mass - luminosity relation do not represent a complete subsample of the bcs and reflex data sets , inspection of fig .
[ fig : ml ] suggests that the dahle ( 2002 ) clusters exhibit a similar mean mass per unit x - ray luminosity to the relaxed systems studied with chandra , albeit with two significant outliers from the best - fit curve ( see below ) and , therefore , that the application of the mass - luminosity relation determined here to the bcs and reflex samples is reasonable .
we have carried out a more rigorous test of the validity of the mass - luminosity relation by repeating the determination of cosmological parameters using other mass - luminosity relations constructed from subsamples of the calibration data . in the first case , a mass - luminosity relation based on the chandra mass measurements ( 10 clusters ) was examined . in the second case ,
only the weak lensing measurements were used ( 8 clusters ) . )
were fitted , as the dahle ( 2002 ) clusters do not populate the highest luminosity bin . ]
these results were then compared with those obtained from the full sample ( 17 clusters , excluding the lensing result for abell 963 ) .
the normalized , marginalized probability distributions for @xmath0 are shown in fig .
[ fig : marginal_s8 ] . for the chandra subsample
, we obtain @xmath127 .
for the lensing subsample , we find @xmath128 ( 68 per cent confidence limits ) .
both results are consistent with those obtained for the whole sample , @xmath129 , although for the lensing subsample the increased scatter in the mass - luminosity relation leads to larger error bars .
it appears , then , that the inclusion of chandra mass measurements for dynamically relaxed clusters does not bias the best - fit result on @xmath0 significantly , but does reduce the formal statistical uncertainties . assessing whether this leads us to underestimate the true uncertainties in @xmath0 will require more data and the ability to disentangle whether the scatter in the lensing results arises primarily from the effects of correlated substructure ( or similar systematic problems in the lensing analysis ) , or the effects of dynamical activity on the x - ray gas ( see below ) .
ricker & sarazin ( 2001 ) , ritchie & thomas ( 2002 ) and randall , sarazin & ricker ( 2002 ) present simulations of the effects of mergers on the x - ray properties of clusters for a variety of mass ratios , impact parameters and central gas densities .
these authors suggest that major mergers can lead to significant short - term increases in the luminosities of clusters during the periods of closest approach , which are then followed by dips in luminosity as the merging dark matter cores move apart , before the cluster returns to equilibrium . from fig .
[ fig : ml ] we see that abell 209 appears to have an unusually high x - ray luminosity / mass ratio , which may have been boosted by the ongoing merger activity in this cluster .
( we note , however , that ms2137.3 - 2353 , which is a highly relaxed cluster with no obvious merger activity and a sharp central density peak , also appears to have a high luminosity / mass ratio and lies below the best - fitting curve . in this case ,
the offset may indicate a relatively early formation epoch for the cluster . ) in contrast abell 1351 , which is also undergoing a major merger event , has an unusually low x - ray luminosity / mass ratio and lies above the best - fit curve in fig .
[ fig : ml ] . detailed simulations and further observations are required to improve our understanding of the effects of mergers on the x - ray properties of clusters .
however , the indications from the present study are that the effects of mergers on the mean mass - luminosity relation for the most luminous clusters are relatively small . chandra measurements of the x - ray gas mass fraction , @xmath108 , in dynamically relaxed clusters provide one of the most simple and robust methods by which to measure @xmath1 ( see allen 2002a and references therein ) . the largest systematic uncertainties in the analysis lie in how well the measured baryonic mass fractions in the clusters approximate the universal mean . lowering the bias factor , @xmath116 , by @xmath130 per cent would cause the best - fitting value of @xmath1 to rise by a similar amount .
similarly , although the @xmath108 profiles in fig . [ fig : fgas_r ] appear to have converged , or be close to converging , at the outer measurement radii , any rise in the @xmath131 values beyond these points would cause a corresponding reduction in @xmath1 .
any change in the best - fitting value of @xmath1 would then lead to a change in @xmath0 , as shown in fig .
[ fig : contour1 ] .
finally , we note that our monte carlo analysis ( section 4.1 ) takes account of the intrinsic scatter and uncertainties in the normalization and slope of the mass - luminosity relation .
( the scatter is modelled as a log - normal distribution about the best - fitting curve , which provides a good description of the measured offsets . )
neglecting such effects would cause to the best - fitting value of @xmath0 to rise slightly , and significantly reduce the formal statistical uncertainties on the measured parameters .
[ fig : seljak ] shows a comparison of the results on @xmath0 as a function of @xmath1 from the present study ( thick , solid curves : as in fig .
[ fig : contour1 ] ) , with the findings from five other recent studies based on the observed local number density of galaxy clusters .
the dot - dashed curve in fig . [ fig : seljak ] shows the result of seljak ( 2002 ) , @xmath132 , using the observed local x - ray temperature function of rich clusters ( pierpaoli , scott & white 2002 ) and a mass - temperature relation determined from asca and rosat observations ( finoguenov 2001 ) .
the result of seljak is consistent with ours for @xmath133 , although the present study leads to tighter constraints .
the result of pierpaoli ( 2001 ) , @xmath134 ( dotted curve ) , based on the same local temperature function data but normalized by a theoretical mass - temperature relation , lies significantly above ours .
the result of reiprich & bhringer ( 2001 ) , @xmath135 ( short dashes ) , based on asca and rosat observations of rass selected clusters , is in good agreement with the present study for @xmath133 .
the result of viana , nichol & liddle ( 2002 ) , @xmath136 ( long dashes ) , using the local reflex x - ray luminosity function and a mass - luminosity relation determined from rosat x - ray observations and a stacked weak lensing analysis of relatively low - mass clusters identified in sloan digitized sky survey ( sdss ) commissioning data , is consistent with the present study for @xmath133 .
the result of bahcall ( 2002 ) , @xmath137 , obtained by combining the number density of optically - selected , relatively low - mass clusters observed in sdss commissioning data with a mass - optical richness correlation , is in good agreement with this work for @xmath133 .
finally , our results are in good agreement with the recent findings of schuecker ( 2003 ; not plotted ) , @xmath138 , @xmath139 , from a combined analysis of the x - ray luminosity function and large - scale clustering in the reflex sample .
borgani ( 2001 ) present constraints on @xmath0 and @xmath1 determined from an analysis of evolution in the x - ray luminosity function of clusters in the rosat deep cluster survey , which spans the redshift range @xmath140 .
their results of @xmath141 and @xmath142 for a flat @xmath6cdm cosmology are consistent with those reported here .
donahue & voit ( 1999 ) report results from an analysis of evolution in the temperature function of clusters within @xmath143 , using the low - redshift cluster sample of markevitch ( 1998 ) and a high redshift sample identified from the einstein observatory extended medium sensitivity survey ( emss ) .
their results , for an assumed flat @xmath6cdm cosmology , of @xmath144 and @xmath145 are in good agreement with ours .
eke ( 1998b ) obtain @xmath146 and @xmath147 ( flat @xmath6cdm ) from an analysis of the temperature function within @xmath148 , which is consistent with the present work .
our results are marginally consistent with the values of @xmath149 and @xmath150 reported by henry ( 2000 ) from an analysis combining the local temperature function of clusters with the properties of emss clusters within @xmath151 .
[ fig : shear ] shows a comparison of the results on @xmath0 as a function of @xmath1 determined from the present study ( dark , solid curves : as in fig .
[ fig : contour1 ] ) together with the findings from seven recent studies based on measurements of weak gravitational lensing due to large scale structure ( cosmic shear ) . the upper dotted curve in fig .
[ fig : shear ] shows the result of van waerbeke ( 2002 ) , @xmath152 .
the upper short - dashed curve shows the result of refregier , rhodes & groth ( 2002 ) , @xmath153 .
the long - dashed curve shows the result of bacon ( 2002 ) , @xmath154 . the dot - dashed curve shows the result of hoekstra , yee & gladders ( 2002 ) , @xmath155 .
the results on @xmath0 from these four studies lie @xmath156 per cent above the present work for @xmath133 .
( note that in order to obtain agreement with these results , the mean mass per unit x - ray luminosity for the clusters included in the present study would need to be raised by a factor @xmath157 , well beyond the systematic uncertainties , which are @xmath158 per cent . ) the lower three curves in fig .
[ fig : shear ] show the results from the three most recent cosmic shear studies , which have appeared on preprint servers after this paper was submitted .
the lower dotted curve shows the result of hamana ( 2002 ) , @xmath159 .
the lower dashed curve shows the result of brown ( 2002 ) , @xmath160 .
finally , the solid curve shows the result of jarvis ( 2002 ) , @xmath161 .
importantly , the brown ( 2002 ) and jarvis ( 2002 ) studies include improved estimates of the redshifts of the lensed , background sources .
( a flat @xmath6cdm cosmology is assumed in all cases . )
the results from the three most recent cosmic shear studies are in good agreement with the present work .
[ fig : sarah ] shows a comparison of the 68.3 and 95.4 per cent confidence constraints on @xmath0 and @xmath1 obtained from the present study ( inner , shaded contours ) , including the chandra @xmath107 data and gaussian priors on the bias parameter ( @xmath17 ) , hubble constant ( @xmath16 ) and @xmath110 ( @xmath162 ) , with the results from analyses of cosmic microwave background ( cmb ) anisotropies and 2df galaxy redshift survey data . for this comparison ,
we have used the published markov chain monte carlo samples of lewis & bridle ( 2002 ) .
the cmb data consist of a combination of cobe ( bennett 1996 ) , boomerang ( netterfield 2002 ) , maxima ( hanany 2000 ) , dasi ( halverson 2002 ) , cosmic background imager ( pearson 2002 ) and very small array ( scott 2002 ) data .
the 2df galaxy redshift survey constraints are from percival ( 2002 ) .
the results obtained from the cmb data alone , using the 6 parameter model of lewis & bridle ( 2002 ) and fixing the optical depth to reionization @xmath163 and marginalizing over @xmath164 , @xmath165 and @xmath166 , are shown as the outer , shaded contours .
the results obtained from the combined cmb+2df data set , marginalizing over the same parameters , are shown as solid lines .
a flat @xmath6cdm cosmology is assumed throughout .
we see that the results from all three data sets are consistent at the 68 per cent confidence level .
our results are also in good agreement with the independent analyses of cmb+2df data reported by lahav ( 2002 ) , percival ( 2002 ) and melchiorri & silk ( 2002 ) , the results from cmb data and type ia supernovae presented by jaffe ( 2001 ) , and the analysis of the iras point source catalogue redshift survey presented by plionis & basilakos ( 2001 ) .
the constraints on cosmological parameters reported here ( @xmath4 and @xmath5 for an assumed flat @xmath6cdm cosmology ) are consistent with , though ( in most cases ) tighter than , those obtained from a number of other , recent studies based on the observed x - ray temperature and luminosity functions of galaxy clusters . ( schuecker 2003 report results with similar precision to those presented here )
. our results are also consistent with current findings from studies of anisotropies in the cmb , the distribution of galaxies in the 2df galaxy redshift survey , the properties of type ia supernovae , early results on large scale structure from the sdss , and the most recent results from studies of cosmic shear .
our results have a number of implications for other cosmological work .
firstly , the agreement between the cluster , cmb and 2df results in fig .
[ fig : sarah ] suggests that the optical depth to reionization is not large ( @xmath167 ; this issue will be examined in more detail in future work ) .
secondly , our result on @xmath0 implies that the possible excess power detected at high multipoles ( @xmath168 ) in the cmb anisotropy power spectrum with the cosmic background imager ( mason 2002 ; bond 2002 ) is unlikely to be due to the sunyaev - zeldovich ( sz ) effect .
our results also have important implications for future x - ray and sz clusters surveys , since the number of clusters detected at high redshifts with high x - ray luminosities , temperatures and sz fluxes will be much lower in a @xmath169 universe than predicted by simulations with @xmath170 .
finally , we suggest that given the precision of the constraints on @xmath1 available from current chandra @xmath107 data for relaxed clusters , and the relatively small systematic uncertainties involved , the @xmath107 data should be considered as a powerful probe in future cosmological work , complementary to measurements of redshift evolution in the x - ray temperature and luminosity functions of galaxy clusters .
we thank george efstathiou , sarah bridle and jerry ostriker for a number of helpful discussions and suggestions .
we also thank sarah bridle for providing the cmb and 2df results shown in fig .
[ fig : sarah ] .
we thank alastair edge and carolin crawford for their efforts in compiling the extended bcs sample , and ofer lahav , john peacock and andrew liddle for comments on the manuscript .
swa and acf acknowledge the support of the royal society .
he acknowledges financial support from nasa grant nag 5 - 8253 and cxo grant go1 - 2132x .
akritas m.g . ,
bershady m.a . , 1996 ,
apj , 470 , 706 allen s.w .
, fabian a.c . , 1998 , mnras , 297 , l57 allen s.w . , ettori s. , fabian a.c . ,
2001a , mnras , 324 , 877 allen s.w . , schmidt r.w . , fabian a.c . , 2001b , mnras , 328 , l37 allen s.w . , schmidt r.w . , fabian a.c . , 2002a , mnras , 334 , l11 allen s.w . , schmidt r.w . , fabian a.c . , 2002b , mnras , 335 , 256 arnaud , k.a . , 1996 , in astronomical data analysis software and systems v , eds . jacoby g. and barnes j. , asp conf . series volume 101 , p17 bacon d. , massey r. , refregier a. , ellis r. , 2002 , mnras , submitted ( astro - ph/0203134 ) bahcall n.a . , 2002 ,
apj , in press ( astro - ph/0205490 ) balucinska - church m. , mccammon d. , 1992 , apj , 400 , 699 bennett c.l . , 1996 ,
apj , 464 , l1 bialek j.j . ,
evrard a.e .
, mohr j.j . , 2001 , apj , 555 , 597 bhringer h. 2001 , a&a , 368 , 86 bhringer h. 2002 , apj , 566 , 93 bond j.r . , 2002 , apj , submitted ( astro - ph/0205386 ) borgani s. , 2001 , apj , 561 , 13 brown m.l . , taylor a.n . ,
bacon d.j . ,
gray m.e . , dye s. , meisenheimer k. , wolf c. , 2002 , mnras , submitted ( astro - ph/0210213 ) bryan g.l . , norman m.l . , 1998 , apj , 495 , 80 cavaliere a. , menci n. , tozzi p. , 1997
, apj , 484 , l21 cen r. , 1997 , apj , 485 , 39 cen r. , ostriker j.p . , 1994 , apj , 429 , 4 dahle h. , kaiser n. , irgens r.j . , lilje p.b . , maddox s. , 2002 , apjs , 139 , 313 dickey j.m . , lockman f.j . , 1990 ,
ara&a , 28 , 215 donahue m. , voit g.m . , 1999 , apj , 523 , l137 ebeling h. , edge a.c . , henry j.p . , 2001 , apj , 553 , 668 ebeling h. , edge a.c . , fabian a.c . ,
allen s.w .
, crawford c.s . , bhringer h. , 1997 , apj , l101 .
ebeling h. , edge a.c .
, bhringer h. , allen s.w .
, crawford c.s . , fabian a.c . , voges w. , huchra j.p . , 1998 , mnras , 301 , 881 ebeling h. , edge a.c . , allen s.w .
, crawford c.s . , fabian a.c . , huchra j.p . ,
2000 , mnras , 318 , 333 eke v.r . , navarro j.f .
, frenk c.s . , 1998a , apj , 503 , 569 eke v.r . ,
cole s. , frenk c.s .
, henry j.p . , 1998b , mnras , 298 , 1145 erdogdu p. , ettori s. , lahav o. , 2002 , mnras , in press ( astro - ph/0202357 ) ettori s. , fabian a.c . , 1999 , mnras , 305 , 834 ettori s. , tozzi p. , rosati p. , 2003
, a&a , 398 , 879 evrard a.e . , 2002 , apj , 573 , 7 finoguenov a. , reiprich t.h . , bhringer h. , 2001 , a&a , 368 , 749 fischer p. , tyson j.a . , 1997 , aj , 114 , 14 freedman w. , 2001 , apj , 553 , 47 frenk c.s . , 1999 ,
apj , 525 , 554 fukugita m. , hogan c.j . ,
peebles p.j.e . , 1998 ,
apj , 503 , 518 hanany s. , 2000 , apj , 545 , l5 halverson n.w . , 2002 , apj , 568 , 38 hamana t. 2002 , apj , submitted ( astro - ph/0210450 ) henry j.p . , 2000 , apj , 534 , 565 hoekstra h. , 2001 , a&a , 370 , 743 hoekstra h. , yee h.k.c . , gladders m.d . , 2002 ,
apj , 577 , 595 horner d.j . ,
mushotzky r.f .
, scharf c.a . , 1999 ,
apj , 520 , 78 ikebe y. , reiprich t.h . , bhringer h. , tanaka y. , kitayama t. , 2002 , a&a , 383 , 773 jaffe a.h . , 2001 , phys .
86 , 3475 jarvis m. , bernstein g.m . , fischer p. , smith d. , jain b. , tyson j.a .
, wittman d. , 2002 , aj , in press ( astro - ph/0210604 ) jenkins a. , frenk c.s .
, white s.d.m . , colberg j.m . ,
cole s. , evrard a.e .
, couchman h.m.p . , yoshida n. , 2001 , mnras , 321 , 372 kaastra j.s . , mewe r. , 1993 , legacy , 3 , heasarc , nasa kayser r. , helbig p. , schramm t. , 1997 , a&a , 318 , 680 lahav o. , 2002 , mnras , 333 , 961 lewis a. , bridle s. , 2002 , phys .
d , in press ( astro - ph/0205436 ) liedhal d.a . , osterheld a.l . , goldstein w.h . , 1995 , apj , 438 , l115 markevitch m. , 1998 , apj , 504 , 27 mason b.s . , 2002 ,
apj , submitted ( astro - ph/0205384 ) mather j.c . , 1994 , apj , 420 , 439 mathiesen b.f . , evrard a.e .
, 2001 , apj , 546 , 100 melchiorri a. , silk , j. , 2002 , phys .
d , 66 , 041301r ( astro - ph/0203200 ) metzler c.a .
, white m. , loken c. , 2001 , apj , 547 , 560 muanwong o. , thomas p.a .
, kay s. , pearce f.r . , 2002 , mnras , 336 , 527 navarro j.f . ,
frenk c.s . ,
white s.d.m .
, 1997,apj , 490 , 493 netterfield c.b .
, 2002 , apj , 571 , 604 nevalainen j. , markevitch m. , forman w. , 2000 , apj , 536 , 73 omeara j.m . , tytler d. , kirkman d. , suzuki n. , prochaska j.x . , lubin d. , wolfe a.m. , 2001 , apj , 552 , 718 pearce f.r . ,
thomas , p.a .
, couchman h.m.p . ,
edge a.c . , 2000 ,
mnras , 317 , 1029 pearson t.j .
2002 , apj , submitted ( astro - ph/0205388 ) pen u. , 1997 , newa , 2 , 309 percival w.j . , 2002 , mnras , 337 , 1068 pierpaoli e. , scott d. , white m. , 2001 , mnras , 325 , 77 piffaretti r. , jetzer o. , schindler s. , 2003 , a&a , 398 , 41 plionis m. , basilakos s. , 2001 , mnras , 327 , l32 randall s.w . , sarazin c.l . , ricker p.m. , 2002 , apj , 577 , 579 reblinsky k. , bartelmann m. , 1999 , a&a , 345 , 1 refregier a. , rhodes j. , groth e.j . , 2002 , apjl , 572 , l131 reiprich t.h . ,
bhringer h. , 2002 , apj , 567 , 716 ricker p.m. , sarazin c.l . , 2001 , apj , 561 , 621 ritchie b.w . ,
thomas p.a . , 2002 ,
mnras , 329 , 675 sanderson a.j.r . ,
ponman t.j .
, finoguenov a. , lloyd - davies e.j . , markevitch m. , 2003 , mnras , in press ( astro - ph/0301049 ) sasaki s. , 1996 , pasj , 48 , l119 schmidt r.w . ,
allen s.w .
, fabian a.c . , 2001 ,
mnras , 327 , 1057 schuecker p. , bhringer h. , collins c.a . , guzzo l. , 2003 , a&a , 398 , 867 scott p.f . , 2002 , mnras , submitted ( astro - ph/0205380 ) seljak u. , 2002 , mnras , 337 , 769 squires g. , kaiser n. , fahlman g. , babul a.,woods d. , 1996 , apj , 469 , 73 sugiyama n , 1995 , apjs , 100 , 281 thomas p.a .
, muanwong o. , kay s. , liddle a.r . , 2002 , mnras , 330 , l48 trmper j. , 1993 , science , 260 , 1769 van waerbeke , l , mellier y. , pello r. , pen u .-
, mccracken h.j . , jain b. , 2002 , a&a , 393 , 369 viana p.t.p . ,
liddle a.r .
, 1996 , mnras , 281 , 323 viana p.t.p .
, liddle a.r . , 1999 , mnras , 303 , 535 viana p.t.p . , nichol r.c . ,
liddle a.r . , 2002 ,
apj , 569 , l75 voit g.m . ,
bryan g.l . , balogh m.l . ,
bower r.g . , 2002 ,
apj , 576 , 601 white s.d.m . ,
frenk c.s . , 1991 , apj , 379 , 52 white s.d.m . , navarro j.f .
, evrard a.e . ,
frenk c.s . , 1993 ,
nature , 366 , 429
|
we present precise constraints on the normalization of the power spectrum of mass fluctuations in the nearby universe , @xmath0 , as a function of the mean local matter density , @xmath1 . using the observed local x - ray luminosity function of galaxy clusters from the extended bcs and reflex studies , a mass - luminosity relation determined from chandra and rosat x - ray data and weak gravitational lensing observations , and the mass function predicted by the hubble volume simulations of evrard
, we obtain @xmath2 , with @xmath3 at 68 per cent confidence .
the degeneracy between @xmath0 and @xmath1 can be broken using chandra measurements of the x - ray gas mass fractions in dynamically relaxed clusters . using this information and including gaussian priors on the mean baryon density of the universe and the hubble constant , we obtain @xmath4 and @xmath5 , for an assumed flat @xmath6cdm cosmology ( marginalized 68 per cent confidence limits ) .
our results are in good agreement with some recent studies based on the local x - ray temperature function of clusters , the redshift evolution of the x - ray luminosity and temperature functions of clusters , early results from the sloan digitized sky survey , the most recent results from studies of cosmic shear , and combined analyses of the 2df galaxy redshift survey and cosmic microwave background anisotropies . # 1to 0pt#1 = " 2218 # 1 # 1 to size # 1 to size = cmr10 scaled2 = cmbx10 scaled2 = cmti10 scaled2 ps . cosmological parameters x - rays : galaxies : clusters gravitational lensing large - scale structure of the universe x - rays : galaxies : clusters
|
in this paper , we prove the global well - posedness of the energy - transport equations @xmath0 where @xmath1 , @xmath2 , and @xmath3 with @xmath4 is a bounded domain .
this system describes the evolution of a fluid of particles with density @xmath5 and temperature @xmath6 .
the parameter @xmath7 is the relaxation time , which is the typical time of the system to relax to the thermal equilibrium state of constant temperature .
the system arises in the modeling of semiconductor devices in which the elastic electron - phonon scattering is dominant .
the above model is a simplification for vanishing electric fields .
the full model was derived from the semiconductor boltzmann equation in the diffusion limit using a chapman - enskog expansion around the equilibrium distribution @xcite .
the parameter @xmath8 appears in the elastic scattering rate ( * ? ? ?
* section 6.2 ) .
certain values were used in the physical literarure , for instance @xmath9 @xcite , @xmath10 @xcite , and @xmath11 ( * ? ? ?
* chapter 9 ) .
the choice @xmath9 leads in our situation to two uncoupled heat equations for @xmath12 and @xmath13 and does not need to be considered .
we impose physically motivated mixed dirichlet - neumann boundary and initial conditions @xmath14 where @xmath15 models the contacts , @xmath16 the union of insulating boundary segments , and @xmath17 is the exterior unit normal to @xmath18 which is assumed to exist a.e .
the mathematical analysis of - is challenging since the equations are not in the usual divergence form , they are strongly coupled , and they degenerate at @xmath19 .
the strong coupling makes impossible to apply maximum principle arguments in order to conclude the nonnegativity of the temperature @xmath20 . on the other hand
, this system possesses an interesting mathematical structure .
first , it can be written in `` symmetric '' form by introducing the so - called entropy variables @xmath21 and @xmath22 . then , setting @xmath23 and @xmath24 , is formally equivalent to @xmath25 where the diffusion matrix @xmath26 is symmetric and positive semi - definite .
second , system possesses the entropy ( or free energy ) @xmath27 = \int_\omega n\log\frac{n}{\theta^{3/2}}dx,\ ] ] which is nonincreasing along smooth solutions to .
even more entropy functionals exist ; see @xcite and below .
however , they do not provide a lower bound for @xmath20 when @xmath12 vanishes .
we notice that both properties , the symmetrization via entropy variables and the existence of an entropy , are strongly related @xcite .
equations resemble the diffusion equation @xmath28 , which was analyzed by pierre and schmitt @xcite . by pierre
s duality estimate , an @xmath29 bound for @xmath30 in terms of the @xmath29 norm of @xmath31 has been derived . in our situation , we obtain even @xmath32 estimates for @xmath33 and @xmath34 . in spite of the above structure , there are only a few analytical results for - . in earlier works , drift - diffusion equations with temperature - dependent mobilities but without temperature gradients @xcite
( also see @xcite ) or nonisothermal systems containing simplified thermodynamic forces @xcite have been studied .
xu included temperature gradients in the model but he truncated the joule heating to allow for a maximum principle argument @xcite .
later , existence results for the complete energy - transport equations ( including electric fields ) have been achieved , see @xcite for stationary solutions near thermal equilibrium , @xcite for transient solutions close to equilibrium , and @xcite for systems with uniformly positive definite diffusion matrices .
this assumption on the diffusion matrix avoids the degeneracy at @xmath19 .
a degenerate energy - transport system was analyzed in @xcite , but only a simplified ( stationary ) temperature equation was studied .
all these results give partial answers to the well - posedness problem only . in this paper , we prove for the first time a global - in - time existence result for any data and with physical transport coefficients . surprisingly , the above logarithmic entropy structure does not help .
our key idea is to use the new variables @xmath35 and @xmath36 and nonlogarithmic entropy functionals .
then system becomes @xmath37 where @xmath38 , @xmath39 , and @xmath40 .
discretizing this system by the implicit euler method and employing the stampacchia truncation method and a particular cut - off test function , we are able to prove the nonnegativity of @xmath41 , @xmath42 , and @xmath20 . in the following
, we detail our main results and explain the ideas of the proofs .
let @xmath43 , @xmath44 , and @xmath45 is relatively open in @xmath18 .
furthermore , let @xmath46 we define the space @xmath47 as the closure of @xmath48 in the @xmath32 norm ( * ? ? ?
* section 1.7.2 ) .
this space can be characterized by all functions in @xmath49 which vanish on @xmath15 in the weak sense .
this space is the test function space for the weak formulation of .
our first main result reads as follows .
[ thm.ex ] let @xmath50 , @xmath4 , @xmath1 , @xmath7 and let - hold .
then there exists a weak solution @xmath51 to - such that @xmath52 , @xmath53 in @xmath54 , @xmath55 , satisfying @xmath56 the idea of the proof is to employ the implicit euler method with time step @xmath57 and the new variables @xmath58 and @xmath59 , which approximate @xmath35 and @xmath36 at time @xmath60 , respectively .
we wish to solve @xmath61 to simplify the presentation , we ignore the boundary conditions and a necessary truncation of the temperature ( see section [ sec.ex ] for a full proof ) .
a nice feature of this formulation is that we can apply a stampacchia truncation procedure to prove the strict positivity of @xmath62 and @xmath63 ( see step 2 in the proof of theorem [ thm.ex ] ) .
the main difficulty is to show the positivity of @xmath64 .
we define a nondecreasing smooth cut - off function @xmath65 such that @xmath66 if @xmath67 and @xmath68 if @xmath69 for some @xmath70 .
we use the test functions @xmath71 and @xmath72 in the weak formulation of , respectively , and we subtract both equations to find after a straightforward computation ( see step 3 in the proof of theorem [ thm.ex ] ) that @xmath73 since @xmath74 , there exists @xmath57 sufficiently small such that the first summand becomes nonnegative .
the third and last summands are nonnegative , too .
( recall that we need to truncate @xmath75 with positive truncation . ) hence , the integral over the second term is nonpositive .
then , choosing @xmath76 , @xmath77 because @xmath78 for @xmath79 , this is only possible if @xmath80 or @xmath81 .
clearly , the bound @xmath82 depends on @xmath83 , and in the de - regularization limit @xmath84 , the limit of @xmath75 becomes nonnegative only .
a priori estimates which are uniform in the approximation parameter @xmath57 are obtained by proving a discrete version of the entropy inequality @xcite @xmath85 for some @xmath86 and @xmath87 , @xmath88 . choosing a variant of the sum of two entropies @xmath89 , we are able to derive gradient estimates for @xmath90 , @xmath91 , and @xmath92 ( see step 4 of the proof of theorem [ thm.ex ] ) .
together with aubin s lemma and weak compactness arguments , the limit @xmath84 can be performed .
theorem [ thm.ex ] can be generalized in different ways .
first , the boundary data may depend on time .
we do not consider this case here to avoid too many technicalities .
we refer to @xcite for the treatment of time - dependent boundary functions .
second , we may allow for temperature - dependent relaxation times , @xmath93 where @xmath94 and @xmath95 .
this expression can be derived by using an energy - dependent scattering rate ( * ? ? ?
* example 6.8 ) . for this relaxation time
, the conclusion of theorem [ thm.ex ] holds .
[ coro.ex ] let the assumptions of theorem [ thm.ex ] hold except that the relaxation time is given by
. then there exists a weak solution to - with the properties stated in theorem [ thm.ex ] .
however , we have not been able to include electric fields in the model .
for instance , in this situation , the first equation in becomes @xmath96 where @xmath97 is the electric potential which is a given function or the solution of the poisson equation @xcite .
the problem is the treatment of the drift term @xmath98 for which the techniques developed for the standard drift - diffusion model ( see , e.g. , @xcite ) do not apply .
our second main result concerns the long - time behavior of the solutions .
[ thm.long ] let @xmath4 , @xmath99 , @xmath7 , and @xmath100 , @xmath101 .
let @xmath51 be the weak solution constructed in theorem [ thm.ex ] .
then there exist constants @xmath87 , @xmath88 , which depend only on @xmath8 , @xmath102 , @xmath103 , and @xmath104 , such that for all @xmath55 , @xmath105 the proof of this theorem is based on discrete entropy inequality estimates . the main difficulty is to bound the entropy dissipation . usually , this is done by employing a convex sobolev inequality ( e.g. the logarithmic sobolev or beckner inequality ) .
however , these tools are not available for the cross - diffusion system at hand , and we need to employ another technique . our idea is to estimate the entropy dissipation by using another entropy ( choosing different values for @xmath106 in the discrete version of ) . denoting the discrete ( nonlogarithmic ) entropy at time @xmath107 by @xmath108 $ ] , we arrive at the inequality @xmath109 - s[n_{j-1},n_{j-1}\theta_{j-1 } ] \le ch s[n_j , n_j\theta_j]^2,\ ] ] where @xmath110 is independent of the time step size @xmath111 .
a discrete nonlinear gronwall lemma then shows that @xmath108 $ ] behaves like @xmath112 , and in the limit @xmath84 , we obtain the result . the paper is organized as follows .
we prove theorem [ thm.ex ] and corollary [ coro.ex ] in section [ sec.ex ] .
sec.long ] is devoted to the proof of theorem [ thm.long ] .
the numerical results in one space dimension presented in section [ sec.num ] indicate that the existence of solutions still holds for @xmath113 and @xmath114 and that the solutions converge exponentially fast to the steady state .
we prove theorem [ thm.ex ] and corollary [ coro.ex ] .
_ step 1 : reformulation .
_ let @xmath50 , @xmath115 , and set @xmath116 .
we consider the semi - discrete equations @xmath117 with the boundary conditions .
the idea is to reformulate the elliptic equations in terms of the new variables @xmath118 observing that @xmath119 and @xmath64 , equations - are formally equivalent to @xmath120 the boundary conditions become @xmath121 in order to show the existence of weak solutions to this discretized system , we need to truncate . for this , let @xmath122 and let @xmath123 , @xmath124 be given such that @xmath125 , @xmath126 , @xmath127 , and @xmath128 . we define @xmath129 and @xmath130 .
the truncated problem reads as @xmath131 where @xmath132 .
note that if @xmath133 and @xmath134 in @xmath54 then - are equivalent to - .
_ step 2 : solution of the truncated semi - discrete problem .
_ we define the operator @xmath135\to l^2(\omega)$ ] by @xmath136 , where @xmath137 is the unique solution to the linear system @xmath138 where @xmath139 , with the boundary conditions @xmath140 we have to prove that the operator @xmath141 is well defined .
first , observe that does not depend on @xmath42 and that the right - hand side is an element of @xmath142 .
therefore , by standard theory of elliptic equations , we infer the existence of a unique solution @xmath143 to with the corresponding boundary conditions in .
with given @xmath41 , there exists a unique solution @xmath144 to with the corresponding boundary conditions .
it remains to show that @xmath41 and @xmath42 are strictly positive in @xmath54 such that the quotient @xmath145 is defined and an element of @xmath142 . to this end
, we employ the stampacchia truncation method .
let @xmath146 note that @xmath147 because of our boundedness assumptions on @xmath148 and @xmath149 .
then @xmath150 is an admissible test function in the weak formulation of yielding @xmath151 taking into account @xmath152 ( observe that @xmath153 ) and the definition of @xmath154 .
this implies that @xmath155 and consequently @xmath156 in @xmath54 . defining @xmath157 and employing the test function @xmath158 in the weak formulation of , a similar computation as above and @xmath159 yield @xmath160 since
the integrals on the left - hand side are nonnegative , we conclude that @xmath161 in @xmath54 .
this shows that @xmath41 and @xmath42 are strictly positive with a lower bound which depends on @xmath162 and @xmath83 . because of @xmath163 and @xmath164 , @xmath165 for @xmath4 .
hence , the operator @xmath141 is well defined and its image is contained in @xmath166 .
standard arguments and the compact embedding @xmath167 ensure that @xmath141 is continuous and compact .
when @xmath168 , it follows that @xmath169 and @xmath170 and thus , @xmath171 . let @xmath172 be a fixed point of @xmath173 . then @xmath174 . by standard elliptic estimates , we obtain @xmath32 bounds for @xmath41 and @xmath42 independently of @xmath175 .
since @xmath41 is strictly positive , we infer an @xmath29 bound for @xmath20 independently of @xmath175 .
thus , we may apply the leray - schauder fixed - point theorem to conclude the existence of a fixed point of @xmath176 , i.e. of a solution @xmath177 to - with boundary conditions - . in order to close the recursion
, we need to show that @xmath178 and @xmath179 .
we employ the following result which is due to stampacchia @xcite : let @xmath180 be the unique solution to @xmath181 with mixed dirichlet - neumann boundary conditions and let @xmath182 be nonnegative and @xmath183 with @xmath184
. then @xmath185 with a bound which depends only on @xmath186 , @xmath54 , and the boundary data .
since the right - hand side of is an element of @xmath142 and @xmath4 , we find from the above result that the solution @xmath41 to is bounded .
furthermore , @xmath42 solves ( see ) @xmath187 taking advantage of the @xmath188 bound for @xmath41 . by stampacchia
s result , @xmath189 .
this shows the desired bounds .
_ step 3 : removing the truncation .
_ we introduce the function @xmath190 where we recall the definition of @xmath82 .
in particular , @xmath191 satisfies @xmath192 in @xmath193 . since @xmath194
, we have @xmath195 on @xmath15 .
because @xmath196 vanishes outside of the interval @xmath197 $ ] , it holds that @xmath198 , @xmath199 .
consequently , @xmath200 and @xmath201 are admissible test functions in @xmath47 for and , respectively , which gives the two equations @xmath202 we take the difference of these equations : @xmath203 since @xmath153 , we have @xmath204 .
therefore , we can choose @xmath205 which implies that @xmath206 , and the first integral is nonnegative .
the same conclusion holds for the second integral in since @xmath207 also the third integral in is nonnegative .
hence , the fourth integral is nonpositive , which can be equivalently written as @xmath208 taking into account definition of @xmath82 , we infer that the integral on the right - hand side is nonpositive , which shows that @xmath209 where @xmath210 for @xmath211 , employing @xmath212 for @xmath213 .
now , @xmath214 for @xmath215 , and we conclude that @xmath216 and @xmath213 in @xmath54 .
since @xmath130 , this means that @xmath217 and @xmath218 .
consequently , we have proven the existence of a weak solution @xmath219 to the discretized problem - with the boundary conditions - , which also yields a weak solution @xmath220 to - with the boundary conditions
. _ step 4 : entropy estimates .
_ let @xmath86 and define the functional @xmath221 = \int_\omega\left(f_b(n , n\theta ) - f_{b , d } - \frac{\pa f_{b , d}}{\pa n}(n - n_d )
- \frac{\pa f_{b , d}}{\pa ( n\theta ) } ( n\theta - n_d\theta_d)\right)dx,\ ] ] where @xmath222 and we have employed the abbreviations @xmath223 the function @xmath224 is convex if @xmath225 or @xmath226 since @xmath227 and @xmath228 .
we wish to derive a priori estimates from the so - called entropy functionals @xmath229 = \frac{1}{|b_1|}\phi_{b_1}[n , n\theta ] + \frac{1}{|b_2|}\phi_{b_2}[n , n\theta].\ ] ] the parameters @xmath230 are chosen from the following set : @xmath231 where @xmath232 consists of all @xmath86 such that @xmath233 and @xmath234 the set of all @xmath235 such that @xmath236 is illustrated in figure [ fig.nstarbeta ] . in particular , we have @xmath225 or @xmath226 for all @xmath236 with @xmath237 .
it is not difficult to check that @xmath238 for all @xmath237 .
such that @xmath236.,width=264 ] [ lem.s ] let @xmath239 .
then @xmath240 + c_1 h\int_\omega\big(\theta_j^{b_1 + 1/2-\beta } + \theta_j^{b_2 + 1/2-\beta}\big)|\na n_j|^2 dx \nonumber
\\ & \phantom{xx}{}+ c_1 h\int_\omega n_j^2\big(\theta_j^{b_1 - 3/2-\beta } + \theta_j^{b_2 - 3/2-\beta}\big)|\na\theta_j|^2 dx \nonumber \\ & \le c_2 h + s_{b_1,b_2}[n_{j-1},n_{j-1}\theta_{j-1 } ] , \label{ex.ei}\end{aligned}\ ] ] where @xmath241 depends on @xmath106 and @xmath8 and @xmath88 depends on @xmath242 , @xmath102 , and @xmath243 .
the constant @xmath244 vanishes if @xmath245 . and @xmath101 .
we abbreviate @xmath246 let @xmath247 or @xmath248 .
we already observed that @xmath225 or @xmath226 .
hence , @xmath249 is convex , and using - , we compute @xmath250 - \phi_b[n_{j-1},n_{j-1}\theta_{j-1}]\big ) \nonumber \\ & = \frac{1}{h}\int_\omega\left((f_{b , j}-f_{b , j-1 } ) - \frac{\pa f_{b , d}}{\pa n}(n_j - n_{j-1 } ) - \frac{\pa f_{b , d}}{\pa(n\theta ) } ( n_j\theta_j - n_{j-1}\theta_{j-1})\right)dx \nonumber \\ & \le \frac{1}{h}\int_\omega\bigg(\left(\frac{\pa f_{b , j}}{\pa n } -\frac{\pa f_{j , d}}{\pa n}\right)(n_j - n_{j-1 } ) + \left(\frac{\pa f_{j , b}}{\pa(n\theta)}-\frac{\pa f_{b , d}}{\pa(n\theta)}\right ) ( n_j\theta_j - n_{j-1}\theta_{j-1})\bigg)dx \nonumber \\ & = -\int_\omega\na\left(\frac{\pa f_{b , j}}{\pa n } -\frac{\pa f_{j , d}}{\pa n}\right)\cdot\na(n_j\theta_j^{1/2-\beta})dx \nonumber \\ & \phantom{xx}{}- \kappa\int_\omega\na \left(\frac{\pa f_{j , b}}{\pa(n\theta ) } -\frac{\pa f_{b , d}}{\pa(n\theta)}\right)\cdot\na(n_j\theta_j^{3/2-\beta})dx \nonumber \\ & \phantom{xx}{}+ \frac{1}{\tau}\int_\omega \left(\frac{\pa f_{j , b}}{\pa(n\theta ) } -\frac{\pa f_{b , d}}{\pa(n\theta)}\right)n_j(1-\theta_j)dx . \label{ex.phi}\end{aligned}\ ] ] we estimate these integrals term by term .
first , we compute @xmath251 where , taking into account that @xmath2 , @xmath252 the above integrand defines a quadratic form in @xmath253 and @xmath254 which is positive definite if and only if @xmath255 and @xmath256 .
these two conditions are equivalent to @xmath257 and these inequalities define the set @xmath232 . we infer that there exists a constant @xmath241 such that @xmath258 the first two terms on the right - hand side of involving the boundary contributions only are estimated by using the young inequality with @xmath259 : @xmath260 where @xmath110 depends only on @xmath8 . it remains to investigate the last integral in involving the relaxation term .
since @xmath153 , we have @xmath261 and @xmath262 . then @xmath263 since @xmath264 , the first expression on the right - hand side is nonpositive .
the second integral is written as @xmath265 where the functions @xmath266 only depend on the boundary data . then the young and poincar inequalities ( with constant @xmath110 )
give @xmath267 putting together the above estimations and using @xmath268 , it follows that @xmath269 - s_{b_1,b_2}[n_{j-1},n_{j-1}\theta_{j-1}]\big ) \nonumber \\ & \phantom{xx } { } + c_1\int_\omega\big((\theta_j^{b_1 + 1/2-\beta}+\theta_j^{b_2 + 1/2-\beta } ) |\na n_j|^2 + n_j^2(\theta_j^{b_1 - 3/2-\beta}+\theta_j^{b_2 - 3/2-\beta})|\na\theta_j|^2\big)dx \nonumber \\ & \le c\delta\int_\omega\big((1+\theta_j^{1 - 2\beta}+\theta_j^{3 - 2\beta } ) |\na n_j|^2 + n_j^2(1+\theta_j^{-1 - 2\beta}+\theta_j^{1 - 2\beta})|\na\theta_j|^2\big)dx + c_2 , \label{ex.aux2}\end{aligned}\ ] ] where the constant @xmath270 vanishes if @xmath245 . and @xmath101 . the conditions @xmath271 and @xmath272 are equivalent to @xmath273 and @xmath274 as well as to @xmath275 and @xmath276 .
thus , there exists a positive constant @xmath110 , which depends on @xmath277 , @xmath278 , and @xmath8 , such that for all @xmath279 , @xmath280 therefore , choosing @xmath259 sufficiently small , the integral on the right - hand side of can be absorbed by the corresponding integral on the left - hand side .
this finishes the proof of the lemma .
_ step 5 : the limit @xmath84 .
_ we define the piecewise constant functions @xmath281 and @xmath282 for @xmath283 and @xmath284 $ ] , where @xmath285 .
the discrete time derivative of an arbitrary function @xmath286 is defined by @xmath287 for @xmath283 , @xmath288 .
then - can be written as @xmath289 the entropy inequality for @xmath290 becomes , after summation over @xmath83 , @xmath291 & + c_1\int_0^t\int_\omega\big((1+\theta_h^{11/2-\beta})|\na n_h|^2 + n_h^2(\theta_h^{-2}+\theta_h^{7/2-\beta})|\na\theta_h|^2\big)dx\,ds \nonumber \\ & \le c_2 t + s_{b_1,b_2}[n_0,n_0\theta_0 ] .
\label{ex.ei2}\end{aligned}\ ] ] we will exploit this inequality to derive @xmath111-independent estimates for @xmath292 and @xmath293 . [ lem.est ] there exists a constant @xmath110 such that for all @xmath57 , @xmath294 first , we observe that there exists a constant @xmath110 , which depends only on @xmath295 , such that @xmath296 we claim that for @xmath297 , @xmath298 \ge -c + c\int_\omega n_h^2(\theta_h^{\beta-1/2}+\theta_h^5)dx,\ ] ] where @xmath110 is a ( generic ) constant independent of @xmath111 .
indeed , it holds @xmath299 and @xmath300 , and the terms involving the boundary data can be estimated according to @xmath301 where we employed the young inequality with @xmath259 .
we infer from that @xmath302 and these terms can be absorbed for sufficiently small @xmath259 by the corresponding terms coming from @xmath303 and @xmath304 .
this proves .
now , we multiply by @xmath305 , integrate over @xmath54 , and employ : @xmath306\big).\end{aligned}\ ] ] taking into account the entropy inequality , estimates - follow .
next , we compute , using - , @xmath307 hence , young s inequality gives @xmath308 where @xmath309 depends on the @xmath29 norms of @xmath310 , @xmath311 , @xmath312 , and @xmath313 .
note that @xmath314 if @xmath102 and @xmath243 are constant in @xmath54 .
we integrate over @xmath315 and employ the poincar inequality to find that @xmath316 because of the entropy inequality .
this shows - .
finally , estimate follows from @xmath317 using and . by weak compactness and the aubin lemma in the version of (
* theorem 1 ) , the uniform bounds of lemma [ lem.est ] imply the existence of subsequences of @xmath292 and @xmath293 , which are not relabeled , such that as @xmath84 , @xmath318 we wish to identify the limit functions @xmath319 , @xmath320 , and @xmath321 . to this end
, we observe that ( subsequences of ) @xmath322 and @xmath323 converge pointwise a.e . in @xmath54 for a.e .
hence , by fatou s lemma and the cauchy - schwarz inequality , @xmath325 since and show that the integral of @xmath326 is bounded uniformly in @xmath111 and @xmath327 .
we infer that @xmath328 a.e . in @xmath329 .
we define @xmath330 for @xmath331 and @xmath19 for @xmath332 .
then @xmath333 .
we pass to the pointwise a.e .
limit @xmath84 in @xmath334 to find that , thanks to the pointwise a.e .
convergence of the sequences @xmath292 , @xmath293 , @xmath335 , and @xmath336 , @xmath337 now , we are in the position to perform the limit @xmath84 in which finishes the proof of theorem [ thm.ex ] .
_ step 6 : temperature - dependent relaxation times . _
it remains to prove corollary [ coro.ex ] .
the proof is exactly as in steps 1 - 5 except at two points .
first , we need to ensure in step 2 that @xmath338 is bounded from below to obtain @xmath339 for all @xmath340 , which is needed to estimate .
second , we need to pass to the limit @xmath84 in the relaxation time term in step 5 .
this is more involved since we can not perform the limit in @xmath341 .
the idea is to expand the fraction @xmath342 and to consider @xmath343 the pointwise convergences of @xmath344 and @xmath345 imply that @xmath346 converges pointwise to @xmath347 as @xmath84 .
consequently , we have the pointwise convergence @xmath348 furthermore , by , @xmath349 is uniformly bounded such that , together with the above pointwise convergence and up to a subsequence , it follows that @xmath350 this ends the proof .
we prove theorem [ thm.long ] . the proof is divided into several steps . _ step 1 : _ let @xmath351 be a solution to with boundary conditions .
we recall that both @xmath90 and @xmath75 are strictly positive . observing that the constant boundary data gives @xmath352 in , we obtain for @xmath353 , @xmath354 & + c_1 h\int_\omega\big((\theta_j^{b_1 + 1/2-\beta } + \theta_j^{b_2 + 1/2-\beta})|\na n_j|^2 \\ & + n_j^2(\theta_j^{b_1 - 3/2-\beta}+\theta_j^{b_2 - 3/2-\beta})|\na\theta_j|^2 \big)dx \le s_{b_1,b_2}[n_{j-1},n_{j-1}\theta_{j-1}].\end{aligned}\ ] ] in particular , for @xmath355 , @xmath354 & + c_1 h\int_\omega\big(1+\theta_j^{3 - 2\beta})|\na n_j|^2 + n_j^2(\theta_j^{-2}+\theta_j^{1 - 2\beta})|\na\theta_j|^2\big)dx \nonumber \\ & \le s_{b_1,b_2}[n_{j-1},n_{j-1}\theta_{j-1 } ] , \label{lo.beta-12}\end{aligned}\ ] ] and for @xmath356 ( here , we need @xmath357 ) , @xmath354 & + c_1 h\int_\omega\big((\theta_j^{-5/2-\beta}+\theta_j^{11/2-\beta } ) |\na n_j|^2 \nonumber
\\ & + n_j^2(\theta_j^{-9/2-\beta}+\theta_j^{7/2-\beta})|\na\theta_j|^2 \big)dx \le s_{b_1,b_2}[n_{j-1},n_{j-1}\theta_{j-1 } ] . \label{lo.-3}\end{aligned}\ ] ] _ step 2 : _ we show that the integral involving the gradient terms in can be bounded from below by , up to a factor , the entropy @xmath358 . to this end , we observe that , by the convexity of @xmath359 for @xmath226 or @xmath225 , @xmath360 this implies that @xmath361 & \le \int_\omega\bigg(\left(\frac{\pa f_{b , j}}{\pa n}-\frac{\pa
f_{j , d}}{\pa n } \right)(n_j - n_d ) \\ & \phantom{xx}{}+ \left(\frac{\pa f_{b , j}}{\pa(n\theta ) } -\frac{\pa f_{j , d}}{\pa(n\theta)}\right)(n_j\theta_j - n_d)\bigg)dx \\ & = \int_\omega\big((2-b)(n_j\theta_j^b - n_d)(n_j - n_d ) + b(n_j\theta_j^{b-1}-n_d)(n_j\theta_j - n_d)\big)dx \\ & \le c\big(1+\|n_j\theta_j^b\|_{l^2(\omega)}\big)\|n_j - n_d\|_{l^2(\omega ) } \\ & \phantom{xx}{}+ c\big(1+\|n_j\theta_j^{b-1}\|_{l^2(\omega)}\big ) \|n_j\theta_j - n_d\|_{l^2(\omega)}.\end{aligned}\ ] ] hence , we obtain for @xmath362 , @xmath363 \le c\phi_{b_1}[n_j , n_j\theta_j ] + c\phi_{b_2}[n_j , n_j\theta_j ] \nonumber \\ & \le c\big(1+\|n_j\theta_j^{\beta-1/2}\|_{l^2(\omega ) } + \|n_j\theta_j^{5/2-\beta}\|_{l^2(\omega)}\big)\|n_j - n_d\|_{l^2(\omega ) } \nonumber \\ & \phantom{xx}{}+ c\big(1+\|n_j\theta_j^{\beta-3/2}\|_{l^2(\omega ) } + \|n_j\theta_j^{3/2-\beta}\|_{l^2(\omega)}\big ) \|n_j\theta_j - n_d\|_{l^2(\omega)}. \label{lo.aux3}\end{aligned}\ ] ] noting that ( again using @xmath357 ) @xmath364 for some generic constant @xmath110 not depending on @xmath111 , we infer from , after summation over @xmath83 , that @xmath365 and @xmath110 does not depend on @xmath83 or @xmath111 . thus , becomes , with @xmath362 , @xmath366 \le c\|n_j - n_d\|_{l^2(\omega ) } + c\|n_j\theta_j - n_d\|_{l^2(\omega)}.\ ] ] taking the square and employing the poincar inequality yields @xmath363 ^ 2 \le c\int_\omega\big(|n_j - n_d|^2 + |n_j\theta_j - n_d|^2\big)dx \\ & \le c\int_\omega\big(|\na n_j|^2 + |\na(n_j\theta_j)|^2\big)dx \le c\int_\omega\big((1+\theta_j^2)|\na n_j|^2 + n_j^2|\na\theta_j|^2 \big)dx \\ & \le c\int_\omega\big((1+\theta_h^{3 - 2\beta})|\na n_h|^2 + n_h^2(\theta_h^{-2}+\theta_h^{1 - 2\beta})|\na\theta_h|^2\big)dx.\end{aligned}\ ] ] the last inequality follows from elementary estimations using the fact that @xmath153 .
this is the desired estimate .
thus , it follows from that @xmath366 + ch s_{b_1,b_2}[n_j , n_j\theta_j]^2 \le s_{b_1,b_2}[n_{j-1},n_{j-1}\theta_{j-1}],\ ] ] where still @xmath362 .
we employ the following lemma which is a consequence of lemma 17 in @xcite .
let @xmath367 be a sequence of nonnegative numbers such that @xmath368 for @xmath369 .
then @xmath370 hence , with @xmath371 $ ] , @xmath366 \le \frac{s_0}{1 + chj s_0/(1 + 2ch s_0)},\ ] ] which can be written as @xmath372 \le \frac{s_0}{1 + ct s_0/(1 + 2ch s_0 ) } , \quad t>0.\ ] ] _ step 3 : _ it remains to prove a lower bound for @xmath373 .
we employ again the convexity of @xmath224 : @xmath374 \ge c\int_\omega\lambda\big(|n_h - n_d|^2 + |n_h\theta_h - n_d|^2\big)dx,\ ] ] where @xmath375 is the minimal eigenvalue of the hessian @xmath376 and @xmath377 , @xmath378 for some @xmath379 .
we recall the following results from linear algebra . if @xmath380 and @xmath381 are two symmetric matrices in @xmath382 with minimal eigenvalues @xmath383 and @xmath384 , respectively , then @xmath385 ( since the minimal eigenvalue is the minimum of the rayleigh quotient ) .
furthermore , a simple computation shows that @xmath386 .
consequently , since @xmath387 with @xmath388 , we conclude that @xmath389 since @xmath153 , the function @xmath390 has a positive lower bound .
therefore , @xmath375 is strictly positive independent of @xmath111 .
going back to , we infer the lower bound @xmath391
\ge c\int_\omega\big(|n_h - n_d|^2 + |n_h\theta_h - n_d|^2\big)dx.\ ] ] together with , this shows that @xmath392 in view of lemma [ lem.est ] , the sequences @xmath292 and @xmath293 are bounded in @xmath393 .
therefore , by fatou s lemma , we obtain @xmath394 which concludes the proof .
in this section we present some numerical results related to . according to theorem [ thm.long ]
, the solution @xmath395 converges to @xmath396 in @xmath142 as @xmath397 if @xmath102 , @xmath243 are constants and @xmath101 .
we want to check this behavior in the numerical simulations if the particle density and temperature are close to zero in some point initially .
we consider system in one space dimension with @xmath398 , and we impose dirichlet boundary conditions at @xmath399 and initial conditions .
we choose the boundary data @xmath400 and the initial functions @xmath401 for @xmath402 , @xmath403 for @xmath404 , and @xmath405 .
both initial functions are very small at @xmath406 ; it holds @xmath407 .
the equations are discretized in time by the implicit euler method with time step @xmath408 and in space by central finite differences with space step @xmath409 .
the discretized nonlinear system is solved by the newton method .
the time step is chosen in an adaptive way : it is multiplied by the factor 1.25 when the initial guess in the newton iterations satisfies already the tolerance imposed on the residual , and it is multiplied by the factor 0.75 when the solution of the newton system is not feasible ( namely , not positive ) . the space step is chosen as @xmath410 ( 501 grid points ) and the maximal time step is @xmath411 . figures [ fig.betaminus025 ] and [ fig.betaplus025 ] illustrate the temporal behavior of the partical density @xmath12 and the temperature @xmath20 for @xmath412 and @xmath413 , respectively , at various small times . for larger times , the functions approach the constant steady state .
the diffusion causes the singularity at @xmath406 to smooth out quickly , and the solution converges to the steady state . in figure [ fig.longtime ] , the decay of the relative @xmath414 difference to the steady state is illustrated in a semi - logarithmic plot .
even for @xmath113 or @xmath114 , the decay to equilibrium seems to be exponentially fast , at least after an initial phase .
this may indicate that the decay rate of theorem [ thm.long ] is not optimal .
moreover , the results indicate that there may exist solutions to - even for @xmath113 and @xmath114 . and the temperature @xmath20 ( semi - logarithmic plot ) at various times for @xmath412.,title="fig:",width=302,height=226 ] and the temperature @xmath20 ( semi - logarithmic plot ) at various times for @xmath412.,title="fig:",width=302,height=226 ] and the temperature @xmath20 ( semi - logarithmic plot ) at various times for @xmath413.,title="fig:",width=302,height=226 ] and the temperature @xmath20 ( semi - logarithmic plot ) at various times for @xmath413.,title="fig:",width=302,height=226 ] distance to the equilibrium functions @xmath415 and @xmath416 for various values of @xmath8 ( semi - logarithmic plot).,title="fig:",width=302,height=226 ] distance to the equilibrium functions @xmath415 and @xmath416 for various values of @xmath8 ( semi - logarithmic plot).,title="fig:",width=302,height=226 ] 11 w. allegretto and h. xie .
nonisothermal semiconductor systems . in : _
comparison methods and stability theory _
( waterloo , on , 1993 ) , lect .
notes pure appl .
162 , pp .
dekker , new york , 1994 .
d. chen , e. kan , u. ravaioli , c. shu , and r. dutton .
an improved energy transport model including nonparabolicity and non - maxwellian distribution effects .
_ ieee electr .
device letters _ 13 ( 1992 ) , 26 - 28 .
|
a class of energy - transport equations without electric field under mixed dirichlet - neumann boundary conditions is analyzed .
the system of degenerate and strongly coupled parabolic equations for the particle density and temperature arises in semiconductor device theory .
the global - in - time existence of weak nonnegative solutions is shown .
the proof consists of a variable transformation and a semi - discretization in time such that the discretized system becomes elliptic and semilinear .
positive approximate solutions are obtained by stampacchia truncation arguments and a new cut - off test function .
nonlogarithmic entropy inequalities yield gradient estimates which allow for the limit of vanishing time step sizes . exploiting the entropy inequality ,
the long - time convergence of the weak solutions to the constant steady state is proved . because of the lack of appropriate convex sobolev inequalities to estimate the entropy dissipation , only an algebraic decay rate is obtained .
numerical experiments indicate that the decay rate is typically exponential .
|
the black hole solutions of three dimensional gravity have been investigated during recent years .
the first black hole solutions , known as the btz black holes , were found in @xcite .
these solutions were found in the presence of a negative cosmological constant . adding the higher derivative terms to the einstein - hilbert action in the presence of a cosmological constant changes the solutions and their asymptotic behaviors and their physical properties .
the topological massive gravity ( tmg ) describes propagation of the massive gravitons around the flat , de sitter or anti de sitter background metrics .
this theory is constructed by adding a parity - violating chern - simons term to the einstein - hilbert action @xcite .
the cosmological tmg solutions contain either the btz black holes @xcite or the warped @xmath1 black holes @xcite .
the charged black hole solutions for topologically massive ( tmge ) presented in @xcite .
the new massive gravity ( nmg ) was found in @xcite .
this theory was constructed by adding a parity - preserving higher derivative term to the tree level action . in the nmg theory
there are the btz and warped @xmath1 solution too , @xcite .
the charged black hole solutions for new massive gravity ( nmge ) is given in @xcite .
several attempts have been done to extend the three dimensional gravity theories to the higher curvature corrections .
one of the most recent extensions is the born - infeld extension of the new massive gravity @xcite [ binmg ] s&=&d^3 x(-(1 + ) ) & & & + & d^3 x^a__a _ , where we have added a gauge field strength and a chern - simons term in order to study the charged solutions of this theory .
our computations shows a similar behavior with a complicated form of solutions depending on @xmath2 and @xmath3 parameters . ] . in the above action @xmath4
is the metric and @xmath5 is the einstein tensor for the curvature tensor @xmath6 .
@xmath7 is the field strength for a @xmath8 gauge field .
we consider @xmath9 as a mass parameter and @xmath10 and @xmath11 as two constant parameters , here @xmath12 . to have a positive coefficient for the scalar curvature we choose @xmath13 . by inserting @xmath14 , and to second order of expansion for the small curvature parameter of the above action
, one finds the new massive gravity action in @xcite .
the next order terms add other extensions to the new massive gravity , which are consistent with deformations of nmg obtained from ads / cft correspondence @xcite .
the other extension of the nmg using ads / cft method in 3d is given by @xcite .
the uncharged @xmath15 black hole solutions for this theory has been found in @xcite ( see also @xcite ) . in this paper and in section two
, we review a method for finding the black hole solutions for this theory .
this method has been used in @xcite to find the warped solutions in new massive gravity . in sections
three and four we expand the action ( [ binmg ] ) to four and six derivative terms . by solving the equations of motion we find the charged and uncharged black holes .
we discuss the domain of validity of each solution and find the physical parameters ( mass , angular momentum , temperature and entropy ) of our black holes .
we show that at each order of expansion the behavior of the solutions ( the physical parameters ) depend on different range of the parameters of the theory . in section five
we do the same steps of sections three and four but for the bi action ( [ binmg ] ) ( unexpanded action ) .
we show that the behavior of the solutions is totally different when one considers the bi action ( so we need to study the theory at each order of expansion to find the behavior of the black holes at that order ) . in section six
we find the central charges for the cft duals . in section seven
we summarize our results .
to find the stationary circularly symmetric solutions for the lagrangian ( [ binmg ] ) at different orders we use the dimensional reduction procedure , presented in @xcite , @xcite .
we use the following ansatz for the metric and the @xmath8 gauge field [ metr ] ds^2=_ab()dx^a dx^b + ^-2()r^-2 ( ) d^2 , = a_a ( ) dx^a , where ( @xmath16 ) and ( @xmath17 , @xmath18 ) .
the parameter @xmath19 can be expressed as a @xmath20 matrix = ( cc t+x & y + y & t - x ) , and @xmath21 is the minkowski pseudo - norm of the `` vector '' @xmath22 .
we do the following steps to find the solutions : \1 .
we insert the ansatz ( [ metr ] ) into the lagrangian and reduce the action to @xmath23 . by variation with respect to @xmath24 and @xmath25
we find the equations of motion .
the black hole solutions for equations of motion in step 1 can be found by choosing the following ansatz for the vector field @xmath26 and the gauge field @xmath27 ( this behavior is general , for example see the gauge field solutions in @xcite and @xcite ) = ^2++,=c(2z dt-(+2z ) d ) .
[ x ] inserting this ansatz into the equations of motion for @xmath28 or @xmath25 or putting into the equation of motion for @xmath29 , we always find two conditions @xmath30 .
we will find the values of @xmath3 and @xmath31 for the above ansatz by using the equations of motion .
\3 . without lose of generality
we choose @xmath32 in all equations of motion .
\4 . following to @xcite
we can choose a rotating frame and a length - time scale such that = ( 12,-12,0),=(,-,-1),=(z+u , z - u,-2z ) , where @xmath33 .
by these parameters one finds @xmath34 . by the above parameters , we are able to write the metric in the warped-@xmath1 adm form @xcite ds^2 = - _ 0 ^ 2dt^2 + r^2 ^ 2 + 1 _ 0 ^ 2 ^ 2 , where @xmath35 ( describes the location of the horizon ) together with @xmath36 are two parameters of the theory and @xmath37 . \5 . as indicated in @xcite , in order to avoid the closed time - like solutions , we must have @xmath38 or @xmath39 .
we impose this condition in order to find the domain of validity of our solutions . for charged solutions
we consider the reality condition for the gauge field strength .
\6 . for each black hole solution
we can find the entropy according to the wald s formula s & = & 4a_h((g^00g^22)^-1)_h , where @xmath40 $ ] is the area of the horizon .
the computation of the mass and angular momentum is possible if we could linearize the field equations and use the adt approach @xcite and @xcite .
equivalently , we can follow the clment s approach in @xcite . the main idea of this approach is the fact that the lagrangian has @xmath41 symmetry .
this symmetry allows us to write the metric as ( [ metr ] ) .
by this symmetry we can find a conserved current and its conserved charge . under infinitesimal symmetry transformation we can find the transformations of the gravitational and electromagnetic fields . by using these transformations
we can find the conserved current , called the super angular momentum vector , @xmath42 .
we have discussed the details of computing the super angular momentum in the appendix a. the conserved charge is the angular momentum and it can be read from the super angular momentum by using the relation @xmath43 . here
@xmath44 is the difference between the values of the super - angular momentum for the black hole and for the background solution .
the background solution is given by the values @xmath45 .
using these values we can find from ( [ admmetr ] ) a horizon - less background metric ds^2=- _ 0 ^ 2dt^2+^2 ( d - dt)^2 + , with @xmath46 . to compute the mass ,
one could use the first law of the thermodynamics for black holes in the modified smarr - like formula @xcite , which is appropriate for the warped @xmath1 black holes , i.e , [ fl ] m = t_hs+2_hj . using the adm form of the metric
we can read the hawking temperature and the horizon angular velocity ( see @xcite ) [ th ] t_h = , _ h = . inserting ( [ th ] ) into ( [ fl ] ) and by using the value of the angular momentum we can find the mass of the black holes . for every solution
, we must check that , the physical parameters satisfy the differential form of the first law of the thermodynamics for black holes , i.e , dm = t_h ds+_h dj . in our solutions ,
the physical parameters of mass , entropy and angular momentum are functions of two free parameters , @xmath35 and @xmath36 , so the differentiations are with respect to these two free parameters . for more details of computations
see the appendix b. * note : * for every lagrangian we will use , we find three types of black holes .
the first one is the uncharged black hole when @xmath47 .
the second one is the maxwell charged ( m - charged ) black hole , where we consider @xmath48 but @xmath49 .
the third one is the maxwell - chern - simons charged black hole ( mcs - charged ) . in the later case we consider @xmath48 and @xmath50 , but to write the lagrangian in the canonical form we choose @xmath51 and @xmath52 .
note that we are able to perform all the calculations for general values of @xmath10 and @xmath11 .
expanding the general lagrangian ( [ binmg ] ) up to four derivative terms , gives us the following lagrangian ( @xmath53 ) l_4 & = & + ^a__a_. the equations of motion coming form the variation with respect to the gauge field are = ( a^4c^2m^2z+(z+m^2 + 18)a^2 + 12)=0,==0 . variation with respect to @xmath29 gives = \{-a^2 c^2z(2 + 3a^2c^2z)m^4+(14 + -74a^2c^2z-4a^2c^2z^2)m^2 + -32z - z^2}=0 , and finally variations with respect to @xmath54 and @xmath25 give the following equations & = & = ( 17 + 8z-8m^2 + 2a^2c^2m^2(8m^2 + 9 + 12z)+16a^4c^4m^4z)=0,&&&=&0 . in this case
we find the known nmg solution @xcite z = -+m^2,=. with the above values , the domain of validity of the solution for @xmath55 will be @xmath56 and for the cosmological constant it is @xmath57 the super angular momentum is given by _
bh&=&-(1+^2,1-^2,2)+(1,-1,0 ) .
+ using the method reviewed in section 2 , we can find the entropy , the angular momentum and the mass as follows [ unch ] s&=&,j=(-^2 ) , & & m&=&- . in this case we consider @xmath48 and @xmath49 .
the equations of motion give the following solution z = -1-,=,c^2a^2=1- .
the domain of validity of this solution for @xmath55 will be @xmath58 , which looks totally different from the uncharged case .
the cosmological constant is limited between @xmath59 , so it has both plus and minus signs .
it changes its sign at @xmath60 .
we also find from the above solution that @xmath61 .
using the method in section 2 the super angular momentum can be found as _
bh&=&-(1+^2,1-^2,2)+(1,-1,0 ) , and the entropy , angular momentum and mass are given by s&=&,j=(-^2 ) , & & m&=&- . in this case we consider @xmath51 and @xmath52 .
we find the following parameters from the equations of motion z = -(1+),=-(16+),c^2=. the domain of validity of this solution will be @xmath62 . inserting this into the cosmological constant relation , we find that it is always negative @xmath63 in this domain .
finally we find from the solution that @xmath64 .
the super angular momentum for mcs - charged black hole is given by _
bh&=&-(1+^2,1-^2,2)+(1,-1,0 ) , and the entropy , angular momentum and mass are as follows s&=&,j=(-^2),&&m&=&- .
as we see , each of the above solutions has its own domain of validity for @xmath55 and the cosmological constant . in all the above cases the entropy is proportional to the area of the horizon .
the angular momentum and mass have the same functionality in terms of @xmath36 and @xmath35 but with different coefficients .
by expanding the lagrangian ( [ binmg ] ) up to six derivative terms we find the the following lagrangian l_6&=&l_4 + . the gauge field equations of motion are given by & = & = 0,==0 .
variation with respect to @xmath29 gives & = & = 0 , and variations of the lagrangian with respect to @xmath54 and @xmath25 give the following relations & = & = ( ( z^2+z-)-12(32a^4c^4z^2+a^2c^2z+1)a^2c^2m^6 + & -&(a^4c^4(z+)z+34a^2c^2(z+34)-)m^4 + & + & ( ( z^2-z-)a^2c^2 - -z)m^2 ) = 0 , + & = & 0 .
we now try to solve the above equations of motion and find different ( un)charged black hole solutions . as the previous section we will find the super angular momentum , and by computing the value of the entropy from the wald s formula and by using the first law of the thermodynamics for black holes we will be able to find the angular momentum and the mass of our solutions .
our calculations for all the solutions show that , there is a general behavior .
in fact we find that the super angular momentum can be written as [ genjbh ] j_bh = u(1+^2,1-^2,2)+v_0 ^ 2(1,-1,0 ) , and the entropy , angular momentum and mass are given by [ gensjm ] s= , j=(u^2+v_0 ^ 2 ) , m= , where @xmath65 and @xmath66 depend on the solution . for uncharged black holes ,
when @xmath67 one finds the following values for @xmath31 and @xmath68 [ z ] m^2&=&+ z(11695 - 208z-320z^2)^12 + [ lambdasol ] & = & ( + 54m^2 - 73m^4+m^6(m^4-)(81 + 144m^2 - 80m^4)^12 ) .
+ to have a causally regular warped black hole one needs to consider @xmath39 , so we find the following domains by looking at equation ( [ z ] ) : * the upper sign of ( [ z ] ) : * in this case @xmath69 .
the lower bound happens at @xmath70 and the upper bound is located at the point of @xmath71 .
inserting this domain into the equation ( [ lambdasol ] ) , for the upper sign of ( [ lambdasol ] ) we find @xmath72 and for the lower sign of ( [ lambdasol ] ) one obtains @xmath73 . * the lower sign of ( [ z ] ) : * in this case @xmath74 . here the lower bound happens at @xmath70 and the upper bound is located at @xmath75 . inserting again the above domain into the equation ( [ lambdasol ] ) , for the upper sign of ( [ lambdasol ] ) one finds @xmath76 and for the lower sign we have @xmath77 . the super angular momentum ( [ genjbh ] ) and the entropy , angular momentum and mass in ( [ gensjm ] ) all are given by the following values for @xmath78 and @xmath79 [ uv1 ] u&=&z(m^4-(18 + 5z)m^2 - -58z+z^2 ) , + v&=&(z-12)(m^4-(18+z)m^2 - -18 z+32 z^2 ) . to find the maxwell charged black holes we consider @xmath48
we find the following equation for the value of @xmath31 in terms of @xmath9 [ poly1 ] & & 6144(z+1)^2m^8-(8192z^3 + 15360z^2 + 5376z-1792)m^6+(8192z^4 + 14336z^3 + & + & 5376z^2 - 640z+272)m^4+(1024z^3 + 768z^2 - 16z+24)m^2 - 15z^2 - 12z=0 .
+ by finding the roots of the above polynomial one is able to find the values for @xmath68 and @xmath3 . for simplicity
we define @xmath80 then [ lambda3 ] & = & , + [ c3 ] c^2a^2&=&(-8m^2 - 8z-54 ) , where the plus sign in ( [ lambda3 ] ) corresponds to minus sign in ( [ c3 ] ) and vice versa . to find the domain of validity for this black hole one needs to find the roots of ( [ poly1 ] ) .
a numerical analysis shows that there are two real solutions for @xmath55 .
the figure 1 ( left ) shows the results of this numerical analysis . for the negative roots we find that @xmath81 . for the positive roots we have @xmath82 , where in this case the extremum point is near to the point @xmath83 .
the numerical calculations show the following values of the cosmological constant ( see the figure 2 ) .
* the upper sign : * for the equation ( [ lambda3 ] ) with the upper sign the cosmological constant is always negative ( left diagram in figure 2 ) .
for @xmath84 we have @xmath85 and for @xmath86 we find @xmath87 , where the extremum is located at @xmath88 . * the lower sign : * the numerical analysis shows that for the equation ( [ lambda3 ] ) with the lower sign the cosmological constant is always positive ( the right diagram of the figure 2 ) . for @xmath84 we have @xmath89 and for @xmath86 we have @xmath90 , where the extremum is located at @xmath91 .
it remains to find the behavior of @xmath92 in equation ( [ c3 ] ) .
for the upper or lower signs of ( [ c3 ] ) one could find either positive or negative values ( see the figure 3 ) . since the value of @xmath3 must be real ( @xmath93 )
so depending on @xmath94 sign one may choose either the left or the right diagram .
the superangular momentum ( [ genjbh ] ) and the values for the entropy , angular momentum and mass in equation ( [ gensjm ] ) are given by the following values [ uv2 ] u&=&13 z((2z - m^2)+14m^4 + 2(1 - 4z)m^2 + 8z^2-z+ ) , + v&=&19(z-12)((2z - m^2 - 14)+14m^4+(1 - 8z)m^2 + 8z^2 - 2z+ ) .
the maxwell - chern - simons black holes with @xmath51 and @xmath52 are given by the following relation [ poly2 ] & & z(5z+4)-(z^3 + 34z^2-z+)m^2 - 4096(z^4 + 56z^3-z^2 + & -&z+)m^4 + 4096(z^3+^2-z-)m^6 - 1792(z^2+z + & + & ) m^8 - 3072(z+)m^10=0 , by finding the roots of the above polynomial again , one could find the values for @xmath68 and @xmath3 .
we define @xmath95 , then we find [ lambda4 ] & = & , & & & & and [ c4 ] c^2&=&(m^2+z+5812 ) , where the plus sign in ( [ lambda4 ] ) corresponds to minus sign in ( [ c4 ] ) and vice versa .
similar to previous sections we can find the domain of validity of our solution .
looking at the equation ( [ poly2 ] ) one finds that this equation has two negative and one positive roots , figure 1 ( right ) . inserting these roots into the relation ( [ lambda4 ] )
one finds : * the positive root : * for @xmath86 we have @xmath96 for upper sign of ( [ lambda4 ] ) and for the lower sign we find @xmath97 ( see the figure 4 ) .
* the negative roots : * for @xmath84 we have both signs ( depending on @xmath31 ) . for upper sign of ( [ lambda4 ] )
we can draw the figure 5 ( left ) and for lower sign we find the figure 5 ( right ) .
the @xmath78 and @xmath79 values in this case are given by [ uv3 ] u&=&13z((2m^4+m^2 - 2z)+4m^6 + 4(z+)m^4 + 2(1 - 4z)m^2 + 8z^2-z+ ) , + v&=&19(z-12)(-(2z - m^2 - 14)+11m^4+(1 - 8z)m^2 + 8z^2 - 2z+ ) .
as we saw in previous sections , at each level of expansion we have different properties for our solutions .
it is interesting to consider the bi action ( [ binmg ] ) without expansion and find its physical properties too .
if we use our ansatz then we will find the following sets of equations of motion : the gauge field equations of motion are given by = ( + 4a^2m^3()^12)=0,==0 .
variation with respect to @xmath29 gives & = & ( 1+)&&&-&=0 , and variations of the lagrangian with respect to @xmath54 and @xmath25 give the following relations & = & = = 0 , + & = & 0 .
inserting @xmath14 in the above equations of motion we find ( see also @xcite ) [ all1 ] & & = 0 , + & & = 0 . the solution for these equations is m=32,=. unlike the previous cases here the value of @xmath9 is fixed . to have a regular black hole we need either @xmath98 when @xmath99 or @xmath100 when @xmath101 .
the entropy of such a black hole can be found by using the wald formula for @xmath102 [ unchbi ] s = . as we see in order to have a positive value entropy we must choose @xmath103 .
we can find the super angular momentum as before = ( ( 1+^2,1-^2,2 ) -_0 ^ 2(1,-1,0 ) ) . using the above results the angular momentum and the mass of the solution
can be found as j = ( _ 0 ^ 2 - 4z^2 ^ 2),m=. solving the equations of motion when @xmath48 the only consistent solution to the equations of motion will be @xmath104 . inserting this value to other equations of motion gives @xmath75 and @xmath105 with @xmath106 .
so in this case the gauge field vanishes and we have not charged black hole .
solving equations of motion gives the following solution z=-,=-13,c^2=. as we see in this case the value of cosmological constant is fixed .
the values of @xmath55 are limited between @xmath107 , also we find @xmath108 .
the super angular momentum for this black hole is equal to = -(1+^2,1-^2,2 ) + _ 0 ^ 2(1,-1,0 ) .
the entropy , angular momentum and mass of this black hole are given by s= , j=(- ) , m=- .
in previous sections we found the warped @xmath1 solutions . it is natural to ask what are the properties of the 2 dimensional cft duals to these solutions . in the pure einstein gravity , the global @xmath109 symmetry
is enhanced to two copies of an infinite dimensional virasoro algebra with the @xmath110 symmetry which has the central charge @xcite , @xmath111 where @xmath112 is the length of @xmath15 space . when one considers the higher derivative terms or add the gauge fields , the value of the central charge changes .
some properties of the cft dual theories for warped @xmath1 and btz black holes have been found in @xcite,@xcite,@xcite,@xcite .
the isometery group of the dual cft in the asymptotic limit of the warped ads black hole is @xmath113,@xcite . to find the central charge we use the cardy s formula .
[ card1 ] s_bh=(c_lt_l+c_rt_r ) , where @xmath114 are the left and right temperatures which we define them for this kind of solutions as [ lrtemps ] t_l_0,t_r2z .
one can also find the same results for central charge by using another form of the cardy s formula @xmath115 where @xmath116 and @xmath117 are the left and right energies defined in terms of physical mass and angular momentum of the black hole , i.e. @xmath118 , @xmath119 @xcite . for each type of solution we can find separate central charges .
for the uncharged solution by substituting ( [ lrtemps ] ) into ( [ card1 ] ) and using the entropies found in ( [ unch ] ) and ( [ unchbi ] ) we obtain c^(4)=,c^(bi)=,similarly for the m - charged solution we have c_m^(4)=,c_m^(bi)=,and finally for the mcs - charged solution they become c_mcs^(4)=,c_mcs^(bi)=.for the sixth derivative lagrangian in terms of parameter @xmath79 , we find the following central charge c^(6)=-where @xmath79 are given by ( [ uv1]),([uv2 ] ) and ( [ uv3 ] ) .
note that in this paper we have considered the case where @xmath52 , but in general one can keep it in equations and find the central charges in terms of @xmath11 , @xcite .
in this paper we have found different black hole solutions for the born - infeld extension of new massive gravity .
we have extended the nmg in two directions , gravity and electromagnetism .
the electromagnetic part contains maxwell term and chern - simons term .
we have found three types of warped @xmath0 solutions .
these are uncharged , maxwell charged and maxwell - chern - simons charged black holes . for each of them we have found the domain of validity ( ctc - free and reality of the gauge field strength ) . we have found these black holes for expanded ( up to four and six derivative ) and unexpanded bi action .
the physical properties of the solutions in each case are totally different and one can not find the four or six derivative properties form the bi properties .
in this theory we have two types of parameters .
the first type includes parameters in the lagrangian .
the @xmath55 which is the mass parameter for our massive gravity theory and the cosmological constant @xmath68 .
the second type includes the parameters which are coming from the solutions . from ( [ admmetr ] )
we see three parameters .
the parameter @xmath120 or equivalently @xmath31 which is limited by ctc free condition , and two free parameters @xmath35 and @xmath36 .
there is another parameter @xmath3 , in ( [ x ] ) , which corresponds to the electric or magnetic charge of the solutions . among these parameters in all the solutions ,
the values of @xmath55 , @xmath68 and @xmath3 are controlled by the value of the @xmath31 parameter .
our computation shows that each of these parameters may have positive or negative values .
so we may have de sitte or anti de sitte solutions in our theory which depends on the level of expansion and existence of the m - charges or mcs - charges in the theory .
our results is summarized as follows : in four derivative case we have only one set of roots for @xmath55 which is either positive or negative depending on the solution .
but in six derivative case for uncharged and m - charged cases we have two sets of roots one positive and one negative . for mcs - charged solutions
we have three sets of roots , one positive and two negative . in bi case
the situation is totally different . for uncharged case
we find only a single point for @xmath55 .
for the bi case there is not m - charged solution and for mcs - charged black holes there is only one set of roots for @xmath55 . in four derivative action for uncharged solutions @xmath86 and @xmath96 . for m - charged solutions @xmath84 but @xmath68 changes its sign in a specific point @xmath31 . for mcs - charged black holes @xmath84 and @xmath97 . in six
derivative action for uncharged solutions @xmath55 and @xmath68 can have both signs , this is true for m - charged solutions . for mcs - charged black holes we have the same behavior but there are situations where @xmath68 changes its sign in a specific value of @xmath31 . in unexpanded bi action , for uncharged solutions
@xmath55 has fixed value and the cosmological constant has both plus and minus signs . for m - charged solutions there is no solution .
for mcs - charged black holes @xmath84 and the cosmological constant has a fixed value @xmath121 .
for all the above solutions we have found the entropy , angular momentum and mass .
our results satisfy the differential form of the first law of the thermodynamics for black holes .
we have used the free parameters of the theory , @xmath35 and @xmath36 to show this . in all of the solutions ,
the entropy is proportional to the area of the horizon . using the cardy s formula we have found the central charges of the cft duals . for our solutions in despite of presence of maxwell - chern - simons term the left and right central charges are equal . in general as shown in @xcite if there is a gravitational chern - simons term ( tmg model ) then this equality will be broken and we obtain different left and right central charges .
to compute the angular momentum it is essential to note that the lagrangian has @xmath41 symmetry and the super angular momentum is a @xmath41 conserved current . under the infinitesimal transformation , we find the following transformation for the fields & & t= ^1 y-^2 x , x= ^0 y-^2 t , y=-^0 x+^1 t , + & & a_0= 12(^0+^1 ) a_1 - 12 ^ 2 a_0,a_1= 12(-^0+^1 ) a_0 + 12 ^ 2 a_1,a_2=0 .
+ using the above relations one can find the conserved currents ( super angular momentum ) . for the gravity part
gr= , + where prims denote the derivatives with respect to @xmath122 variable .
one may write this result as @xmath123 ( see @xcite ) .
also for the electromagnetic part one finds _ em=12 .
the total super angular momentum then will be @xmath124 .
as we mentioned in section 2 the black hole solutions in this paper have two free parameters @xmath35 and @xmath36 . in order to check that the entropy , angular momentum and mass satisfy the differential form of the first law of the thermodynamics for black holes we need to know every parameter in terms of these two free parameters .
the area of the horizon is equal to @xmath125 .
the temperature of the black holes is given by @xmath126 , and the angular velocity can be read from the metric and it is equal to @xmath127 . for all solutions in this paper we must check the following relations [ dform ] - t_h - _ h = 0,- t_h - _ h = 0 .
as an example ( we have checked the above relations for every solution , the computations are very lengthy in most cases ) for uncharged solutions in four derivative case we find & & - t_h - _ h = -12=0 , + & & - t_h - _ h = -14 = 0 , where we have used the equation of motion @xmath128 in this case . as another example for uncharged solution in the six derivative case
we find & & - t_h - _ h = + & & - = 0 , + & & - t_h - _ h = + & & = 0 , where we have used the equation of motion again . for charged solutions , for example consider the m - charged solution at four derivative case , we find & & - t_h - _ h = -=0 , + & & - t_h - _ h = = 0 . in all cases in this paper , by using the equations of motion one can show that the equations ( [ dform ] ) are correct .
we would like to thank mohammad moghaddasi for discussion .
we would like to thank shahin sheikh - jabbari for his comments .
this work was supported by ferdowsi university of mashhad under the grant 2/15734 ( 30/08/1389 ) .
99 m. banados , c. teitelboim and j. zanelli , `` the black hole in three - dimensional space - time , '' phys .
lett . * 69 * , 1849 ( 1992 ) [ arxiv : hep - th/9204099 ] . m. banados , m. henneaux , c. teitelboim and j. zanelli , `` geometry of the ( 2 + 1 ) black hole , '' phys .
d * 48 * , 1506 ( 1993 ) [ arxiv : gr - qc/9302012 ] .
s. deser , r. jackiw and s. templeton , `` three - dimensional massive gauge theories , '' phys .
lett . * 48 * , 975 ( 1982 ) .
k. a. moussa , g. clement and c. leygnac , `` the black holes of topologically massive gravity , '' class .
* 20 * , l277 ( 2003 ) [ arxiv : gr - qc/0303042 ] .
a. bouchareb and g. clement , black hole mass and angular momentum in topologically massive class .
* 24 * , 5581 ( 2007 ) [ arxiv:0706.0263 [ gr - qc ] ] .
d. anninos , w. li , m. padi , w. song and a. strominger , `` warped @xmath1 black holes , '' jhep * 0903 * , 130 ( 2009 ) [ arxiv:0807.3040 [ hep - th ] ]
. k. a. moussa , g. clement , h. guennoune and c. leygnac , `` three - dimensional chern - simons black holes , '' phys .
d * 78 * , 064065 ( 2008 ) [ arxiv:0807.4241 [ gr - qc ] ] .
e. a. bergshoeff , o. hohm and p. k. townsend , `` massive gravity in three dimensions , '' phys .
lett . *
102 * , 201301 ( 2009 ) [ arxiv:0901.1766 [ hep - th ] ] .
g. clement , `` warped @xmath1 black holes in new massive gravity , '' class .
* 26 * , 105015 ( 2009 ) [ arxiv:0902.4634 [ hep - th ] ] .
i. gullu , t. c. sisman and b. tekin , `` born - infeld extension of new massive gravity , '' class .
* 27 * , 162001 ( 2010 ) [ arxiv:1003.3935 [ hep - th ] ] .
i. gullu , t. c. sisman and b. tekin , `` c - functions in the born - infeld extended new massive gravity , '' phys .
d * 82 * , 024032 ( 2010 ) [ arxiv:1005.3214 [ hep - th ] ] .
a. ghodsi and m. moghadassi , `` charged black holes in new massive gravity , '' phys .
b * 695 * , 359 ( 2011 ) [ arxiv:1007.4323 [ hep - th ] ] .
a. sinha , `` on the new massive gravity and ads / cft , '' jhep * 1006 * , 061 ( 2010 ) [ arxiv:1003.0683 [ hep - th ] ] .
s. nam , j. d. park and s. h. yi , `` ads black hole solutions in the extended new massive gravity , '' jhep * 1007 * , 058 ( 2010 ) [ arxiv:1005.1619 [ hep - th ] ] . m. alishahiha , a. naseh and h. soltanpanahi , `` on born - infeld gravity in three dimensions , '' phys .
d * 82 * , 024042 ( 2010 ) [ arxiv:1006.1757 [ hep - th ] ] .
l. f. abbott and s. deser , `` stability of gravity with a cosmological constant , '' nucl .
b * 195 * , 76 ( 1982 ) .
s. deser and b. tekin , `` gravitational energy in quadratic curvature gravities , '' phys .
lett . * 89 * , 101101 ( 2002 ) [ arxiv : hep - th/0205318 ] .
m. f. paulos , `` new massive gravity , extended , '' arxiv:1005.1646 [ hep - th ] .
d. anninos , w. li , m. padi , w. song and a. strominger , `` warped @xmath1 black holes , '' jhep * 0903 * , 130 ( 2009 ) [ arxiv:0807.3040 [ hep - th ] ] .
a. garbarz , g. giribet and y. vasquez , `` asymptotically @xmath1 solutions to topologically massive gravity at special values of the coupling constants , '' phys .
d * 79 * , 044036 ( 2009 ) [ arxiv:0811.4464 [ hep - th ] ] .
j. m. maldacena and a. strominger , `` ads(3 ) black holes and a stringy exclusion principle , '' jhep * 9812 * , 005 ( 1998 ) [ arxiv : hep - th/9804085 ] .
r. fareghbal , `` hidden conformal symmetry of warped @xmath1 black holes , '' phys .
b * 694 * , 138 ( 2010 ) [ arxiv:1006.4034 [ hep - th ] ] .
j. d. brown and m. henneaux , `` central charges in the canonical realization of asymptotic symmetries : an example from three - dimensional gravity , '' commun .
* 104 * , 207 ( 1986 ) .
|
in this paper we find different types of black holes for the born - infeld extended new massive gravity .
our solutions include ( un)charged warped @xmath0 black holes for four and six derivative expanded action .
we also look at the black holes in unexpanded bi action . in each case
we calculate the entropy , angular momentum and mass of the black holes .
we also find the central charges for the cft duals .
ahmad ghodsi and davood mahdavian yekta + +
|
the theoretical determination of ion distributions in aqueous solutions was initiated almost a century ago by guoy @xcite and chapman @xcite , who applied the poisson - boltzmann ( pb ) theory to calculate the spatial distribution of monovalent ions near a uniformly charged interface .
ever since their seminal work , the topic remains central in statistical mechanics , physical chemistry , and biophysics@xcite .
the original pb theory is a mean field theory with some simplified assumptions such as , point like particles and uniform surface charge density . to account for the finite ionic radius , stern introduced phenomenologically a layer of ions at the charged interface with a different dielectric constant , the stern layer @xcite .
the effect of excess salt concentration and the resulting screening was extended by debye - hckel @xcite .
grahame generalized the gouy - chapman theory to multivalent ions @xcite .
subsequently , more refined theories and numerical simulations were developed to incorporate short - range interactions , image charges , finite size ionic radius , and ion - ion correlations @xcite .
more recently , modifications of pb theory have been developed to incorporate hydration forces @xcite .
some first - principles calculations of surface - tension for amphiphilic monolayers assume a pb theory with one or more layers of varying dielectric constant @xcite .
experimental support for the validity of pb theory was provided by electrokinetic , visco - electric effects , and other techniques ( see ref .
@xcite for a review ) .
mclaughlin and collaborators @xcite have shown good agreement between @xmath4-potentials computed from pb theories and electrophoretic measurements in lipid vesicles .
other techniques , such as radiotracer experiments @xcite , x - ray reflectivity @xcite , or infrared spectroscopy @xcite allow the determination of the total amount of ions in the immediate vicinity of a charged interface .
it is noteworthy that all the experimental data for monovalent ions ( at moderate salt concentrations @xmath5 0.1 m ) outlined above are adequately described by the guoy - chapman theory ( with the generalization of excess salt ) with no need for further corrections @xcite .
a close inspection , however , shows the agreement between theory and experiment is either based on fitting variables such as surface charge or interfacial dielectric values , that are not known in advance or / and based on integrated quantities . as an example , it has been shown recently that the degree of proton dissociation of arachidic acid spread on a sodium salt solution is adequately described by pb theory , but this agreement only involves the integral ( over the entire space ) of the sodium distribution @xcite .
thus , local deviations that preserve the integral of the distribution ( i.e. , total number of ions ) are not discriminated by these experiments . on the other hand ,
force measurements between two charged membranes separated by salt solutions , although well described by pb theory at large distances , show strong deviations at short distances ( 1 - 2 nm)@xcite .
the origin of these hydration forces is still controversial . in some cases
, it has been suggested to extend pb theory to incorporate the restructuring of water , resulting in ion distributions that deviate from pb at short distances from the interface @xcite .
it is therefore imperative to determine the ion distribution itself to establish the degree of accuracy of pb theory . in this manuscript
, we experimentally determine the distribution of cs@xmath3 ions next to charged interfaces by anomalous x - ray reflectivity @xcite for several decades in bulk cesium concentrations .
we compare our results with the predictions of pb theory and present a discussion on the sensitivity of the experimental results to quantify the magnitude of possible deviations from the pb distribution .
this kind of investigation has become feasible only with the advent of the second generation x - ray synchrotron sources with novel insertion devices ( i.e. undulator ) and improved optics , which readily produce variable - energy x - ray beams with brilliances capable of detecting a single atomic - layer even if not closely packed .
the manuscript is organized as follows : in section [ theory ] , we review several theoretical results relevant to this study . in section [ experimntal ] , the experimental details are provided . in section [ results ] , the x - ray reflectivity ( xr ) and grazing incidence x - ray diffraction ( gixd ) measurements are presented with detailed structural analysis .
the experimental results are compared with the theoretical pb theory in section [ analysis ] .
the implications of the present study are discussed in section [ conclusions ] .
we briefly review the pb theory of an aqueous solution containing monovalent ions ( 1:1 positive and negative electrolytes ) of average bulk concentration @xmath6 in the presence of an ideally flat charged interface of known surface density @xmath7 , at @xmath8 .
as the surface charge is uniformly distributed , the electric potential @xmath9 depends only on the distance from the interface @xmath10 , and the boltzmann distribution is @xmath11=-2en_{b}\sinh{\phi(z)},\ ] ] where @xmath12 is the boltzmann constant and @xmath13 .
the ion distribution is calculated self - consistently from the poisson equation @xmath14 where @xmath15 is the debye screening length .
equation ( [ pb1 ] ) can be solved analytically , yielding @xmath16 , \label{psi_z}\ ] ] @xmath17 where @xmath18$]=@xmath19 , and @xmath20 is the gouy - chapman length .
figure [ pb](a ) shows calculated counter - ion distributions using eq.([n_z ] ) for a fixed surface charge density , typical of fully deprotonated closed packed phospholipids @xmath21 @xmath22 for several salt concentrations .
it is worth noting that for this surface density ( @xmath23 ) and bulk salt concentrations ( @xmath24 and 9.6 at 10@xmath25 and 0.1 m , respectively ) @xmath26 , and @xmath27 , the concentration of ions next to the interface is independent of bulk concentration @xmath28 ( a ) calculated monovalent ion distributions @xmath29 near a negatively charged surface as obtained from pb theory , eq .
( [ n_z ] ) for different bulk concentration .
surface charge density @xmath30 is one electron charge per 40 @xmath31 , @xmath32 , and @xmath33 k. note that for this surface charge density the value of the distribution at the @xmath34 is practically a constant , @xmath35 ( b ) the convolutions of the distributions @xmath29 using eq.([conv_gauss ] ) assuming two @xmath36 values as indicated . for this strongly charged surface the convoluted distributions for different bulk ionic concentrations
are practically indistinguishable.,width=3 ] we have recently argued @xcite that theoretical pb distributions should be convoluted with the experimental resolution function @xmath37 for comparison with experiments , as follows , @xmath38 in the particular case of a gaussian resolution function the convolution is @xmath39 the width in the gaussian function is given as a sum , @xmath40 , where each @xmath41 s account for surface - roughness due to capillary waves and imperfections in the monolayer and other contributions . since surface - roughness is the dominant contribution we use @xmath42 , where @xmath43 is the average roughness determined from the reflectivities . from the statistical mechanical perspective , the above convolution maybe justified if the fluctuations of the interface are gaussian and the wavelength of the capillary waves @xmath44 is large ( @xmath45 , where @xmath46 is typical ion size ) so that the ions `` see '' an effective flat charged interface .
alternative methods for computing distributions next to flexible interfaces have recently been proposed @xcite .
figure [ pb](b ) shows convolution of the distributions shown in fig .
[ pb](a ) for two different values of @xmath36 as indicated .
it is interesting to note that the convoluted calculations are practically indistinguishable as a function of bulk salt concentration . in section
[ analysis ] we discuss quantitatively to what extent the experimental resolution limits our ability to compare with theoretical distributions . the system we describe in this paper differs from simple pb in that one of the ions species , the hydronium h@xmath2o@xmath1 , can bind to the interface , and as a result , out of the @xmath47 phosphate groups forming the interface , only @xmath48 are electrically charged .
the density of charge in the system ( including the interfacial charge @xmath34 ) is therefore @xmath49 the index @xmath50 runs over the different species of ions present in solution and @xmath51 is their valence ( i.e. the solution contains monovalent ions only ) . in the free energy , the ions are treated as an ideal gas within the solvent , and subjected to electrostatic contributions ( the standard pb ) @xcite .
we also incorporate two additional terms that represent the favorable binding of hydroniums to the interface and the mixing entropy of the charged and neutral phosphate groups at the interface , @xmath52 where @xmath53 is the gain in free energy for a hydronium binding to the phosphate group .
the quantity @xmath54 has dimensions of volume and defines the standard state .
the condition that the free energy is a minimum for variations of the different number densities @xmath55 under the constraint of a fixed number of particles ( canonical ensemble ) leads to the boltzmann distribution @xmath56 with chemical potential @xmath57 , and also to the poisson equation eq .
( [ pb1 ] ) .
further minimization of the free energy with respect to the parameter @xmath58 yields the additional equation @xmath59 where @xmath60 is the value of the electric potential at the interface ( the contact value ) .
this equation may be rewritten as @xmath61 and expresses the equilibrium of the process h@xmath2o@xmath62po@xmath63 @xmath64 po@xmath65h@xmath66h@xmath67o @xcite .
if the standard state is defined as @xmath68 molar , the binding free energy @xmath53 is related to the @xmath69 of the molecules forming the interface by @xmath70 and the hydronium concentration is related to the @xmath71 of the solution by n@xmath72 .
substitution of these values into eq .
( [ chemical_binding ] ) yields @xmath73 the fraction of charged phosphate groups is then @xmath74 as a result , the pb equation eq .
( [ pb1 ] ) needs to be solved with the self - consistent boundary condition eq .
( [ alpha_g ] ) @xmath75 the ion distribution is given by eq .
( [ n_z ] ) with a renormalized guoy - chapman length @xmath76 ( rpb ) . at high surface charge ,
the contact point potential is significantly larger than @xmath77 , and the consequences of eq .
( [ self_consistent ] ) are dramatic .
the ion distribution becomes strongly dependent on ion concentration for points very close to the interface , as shown in fig .
this should be contrasted with fig.[pb ] , where the ion distribution is independent of bulk concentration .
to manipulate ion bulk - concentrations , we used csi ( 99.999% , sigma corp .
, catalog no .
203033 ) solutions in ultra - pure water , taking advantage of the @xmath78 resonance of cs ions at 5.012 kev in anomalous reflectivity measurements . to control surface charge density , monolayers of dihexadecyl - hydrogen - phosphate ( dhdp ,
see fig .
[ iso1 ] ) ( c@xmath79h@xmath80o@xmath81p ; mw = 546.86 , sigma corp . , catalog no .
d2631 ) were spread from 3:1 chloroform / methanol solutions at the air - water interface in a thermostated langmuir trough @xcite .
dhdp was chosen for this study , since it forms a simple in - plane structure at high enough surface pressures @xcite and its hydrogen - phosphate head - group ( _ r_-po@xmath65h ) has a pk@xmath82 = 2.1 , _ presumably _ guaranteeing almost complete dissociation [ po@xmath83/[_r_-po@xmath65h ] @xmath84 0.99999 , with one electron - charge per molecule ( @xmath85 0.4 c / m@xmath22 ) .
monolayers of dhdp were prepared in a thermostatic , solid teflon langmuir trough , and kept under water - saturated helium environment .
ultrapure water ( nanopure , barnstead apparatus ; resistivity , 18.1 m@xmath86 cm ) was used for all solution preparations .
monolayer compression , at a rate of @xmath87 1 @xmath88(molecule@xmath89min . ) , was started 10 - 15 minutes after spreading to allow solvent evaporation .
the monolayer was then compressed to the desired surface pressure . during the compression , the surface pressure was recorded by a microbalance using a filter - paper wilhelmy plate .
the langmuir trough is mounted on a motorized stage that can translate the surface laterally with respect to the incident beam to allow the examination of different regions of the sample to reproduce results and monitor radiation damage of the monolayer .
x - ray studies of monolayers at gas / water interfaces were conducted on the ames laboratory liquid surface diffractometer at the advanced photon source ( aps ) , beam - line 6id - b ( described elsewhere @xcite ) .
the highly monochromatic beam ( 16.2 and 5.012 kev ; @xmath90 and @xmath91 ) , selected by a downstream si double crystal monochromator , is deflected onto the liquid surface to a desired angle of incidence with respect to the liquid surface by a second monochromator [ ge(220 ) and ge(111 ) crystals at 16.2 and 5.012 kev , respectively ] located on the diffractometer @xcite .
prior to the measurements , the absolute scale of the x - ray energy was calibrated with six different absorption edges to better than @xmath923 ev .
x - ray reflectivity and gixd are commonly used to determine the monolayer structure @xcite .
specular xr experiments yield the electron density profile across the interface , and can be related to molecular arrangements in the film .
herein , the electron density profile across the interface is extracted by a two - stage refinement of a parameterized model that best fits the measured reflectivity by non - linear least - squares method .
a generalized density profile @xmath93 of the electron - density ( ed ) and the absorption - density ( ad ) ( real and imaginary parts , respectively ) is constructed by a sum of error functions as follows : @xmath94 where n+1 is the number of interfaces , @xmath95 ; @xmath96 and @xmath97 are the ed and ad of @xmath98th slab , @xmath99 and @xmath100 are the position and roughness of @xmath98th interface , respectively , @xmath101 is the electron density of the solution ( @xmath102 e / @xmath103 ) , and @xmath104 is the electron density of the gaseous environment .
the use of a different roughness @xmath100 for each interface preserves the integral of the profile along @xmath105 or the electron density per unit area , thus conserving the chemical content per unit area a. although small variations are expected in @xmath100 for interfaces that separate rigid portion of a molecule ( hydrocarbon - chains / gas interface and hydrocarbon - chain / headgroup interface , for instance ) , somewhat larger variations can occur at different interfaces ( such as , gas / hydrocarbon chains interface versus headgroup / subphase interface ) .
the ad profile is particularly important at the cs resonance ( 5.012 kev ) as demonstrated below .
the reflectivity is calculated by recursive dynamical methods @xcite of the discretized ed and ad in eq .
( [ rho_z ] ) . in the first stage of the refinement
, the variable parameters used to construct the electron density across the interface @xmath106 are the thickness values of the various slabs @xmath107 , and interfacial roughness @xmath108 . by non - linear square fit ( nlsf )
we determine the minimum number of slabs or ` boxes ' required for obtaining the best fit to the measured reflectivity .
the minimum number of slabs is the one for which the addition of another slab does not improve the quality of the fit , i.e. , does not improve @xmath109 . in the second stage , we apply space filling and volume constraints @xcite to calculate @xmath110 by assigning to each slab a different portion of the molecule , the ions and water molecules , to a profile that has the same number of slabs as obtained in stage one of the analysis . as described in detail in section [ results ] , the model eliminates the @xmath110 s as parameters , and is self consistent with the molecular structure , providing detailed information on the constituents of each slab .
the gixd measurements are conducted to determine the lateral organization in the film .
the angle of the incident beam with respect to the surface , @xmath58 , is fixed below the incident and reflected critical angle ( @xmath111 cm , where @xmath112 is the electron density of subphase ) for total reflection , while the diffracted beam is detected at a finite azimuthal in - plane angle , @xmath113 , and out - of plane , @xmath114 ( the angle of the reflected beam with respect to the surface ) .
rod scans along the surface normal at the 2d bragg reflections are used to determine the average ordered chain length and tilt with respect to the surface normal .
the intensity along the rod of the 2d bragg reflection is analyzed in the framework of the distorted wave born approximation ( dwba ) @xmath115 where @xmath116 and @xmath117 ( @xmath118 ) are the fresnel transmission functions , which give rise to an enhancement at the critical angle . in modeling the rod scans ,
the length and tilt of the tails are varied , examining two tilt directions : one toward nearest neighbors ( nn ) and the second toward next nn ( nnn)@xcite .
the form factor for the tails is given by @xmath119 where @xmath120 is defined along the long axis of the tail , and @xmath121 is the length of the tail .
surface pressure versus molecular area for dhdp spread on csi solutions at various concentrations as indicated .
reflectivity and gixd were performed at constant surface - pressures 30 mn / m and 40 mn / m.,width=3 ] surface pressure versus molecular - area ( @xmath122 ) isotherms of dhdp at various csi salt concentrations ( @xmath6 ) are shown in fig .
[ iso1 ] . for @xmath123 ,
the isotherm exhibits two distinct slopes , associated with crystalline tilted and untilted acyl - chains with respect to the surface normal .
the transition from tilted to untilted chains at @xmath124 , occurs at a constant @xmath125 @xmath22 , whereas @xmath126 increases with salt concentration @xmath6 .
csi and other electrolytes ( nacl and cscl ) in solution significantly influence the isotherm causing an increase of the monolayer - coalescence area @xmath127 , ( i.e. , @xmath123 ) with the increase in @xmath6 . for @xmath128 39 @xmath22 , approximately the cross - section of the two acyl - chains of dhdp , ( constant @xmath129 mn / m )
the monolayer is in the yet poorly characterized state of _
collapse_. in the present study we focus on the untilted crystalline phase ( @xmath130 mn / m ) , where the molecular area variation at a fixed @xmath131 is less than 1.5% ( in the data analysis we allow @xmath92 5% variation in molecular area ) .
gixd scans versus the modulus of the inplane momentum transfer @xmath132 , at surface pressure @xmath133 mn / m ( curves are shifted by a decade for clarity ) .
the bragg peaks are independent of bulk salt concentration indicating no significant change in in - plane molecular packing .
gixd scan for @xmath134 m csi ( without dhdp ) shows a broad peak at @xmath135 @xmath136 , due to the surface structure of water.,width=3 ] gixd experiments provided additional insight into the molecular packing of the acyl chains within the langmuir monolayers , namely , the average inplane density of the headgroups and the surface charge - density . figure [ gid1 ] shows gixd scans from dhdp on pure water and on csi solution ( 10@xmath137 m , at @xmath138 mn / m ) and from bare surface of csi solution ( 10@xmath137 m ) before spreading the monolayer as a function of in - plane momentum transfer @xmath139 @xcite .
the broad peak centered at @xmath140 @xmath136 is due to the structure - factor of the aqueous - solution interface
. the spreading and compression of the monolayer slightly modifies the water structure factor peak at @xmath141 @xmath136 , and brings about two prominent bragg reflections due the ordering of acyl - chains superimposed on a modified surface liquid structure - factor @xcite .
the main features of the diffraction pattern are independent of ionic concentration , consistent with the isotherms at the 30 - 40 mn / m region , that show very small variations in the molecular packing .
figure [ gid2 ] shows the 2d diffraction pattern consists of a strong bragg reflection at @xmath142 @xmath136 and a weaker peak at @xmath143 @xmath136 , corresponding to 4.145 , and 2.392 @xmath144-spacings , respectively ( table [ gid1_t ] ) .
the shape , spacing , and location of the two intense peaks correspond closely with literature values for ( 1,0 ) and ( 1,@xmath145 ) planes in a hexagonal unit cell of ordered alkyl chains , also confirmed by the ratio @xmath146 .
this unit cell ( molecular area 19.83 @xmath31 ) agrees with the cross - sectional area of alkyl chain @xcite .
each headgroup has two alkyl chains , giving a molecular area of 39.66 @xmath31 , in agreement with values obtained from the @xmath147 isotherm ( @xmath148 @xmath31 , @xmath149 mn / m ) .
the small variations in molecular areas are attributed to the existence of domain boundaries , defects and minute impurities .
the peaks in fig .
[ gid2 ] were fitted to lorentzians the parameters of which are given in table.[gid1_t ] .
the peak linewidth @xmath150 is significantly larger for the higher order peak .
this is expected in simple 2d crystals @xcite , and is even more pronounced for 2d crystals fluctuating in 3d space ( fluctuating tethered membranes , see ref .
the inset in fig .
[ gid2 ] shows the rod - scan at @xmath132 ( 1,0 ) bragg reflection . the the rod - scan analysis ( using eqs .
( [ nume ] ) and ( [ form ] ) ) yields an average chain - length @xmath151 , and a tilted angle with respect to the surface normal @xmath152 , consistent with the reflectivity and previous reports @xcite .
.[gid1_t]best - fit parameters to high - resolution diffraction scan of dhdp on @xmath134 m csi solution ( @xmath153 mn / m ) [ cols="^,^,^,^ " , ]
we first compute the integrated number of cs@xmath3 per dhdp over the first 15 next to the charged interface .
this number can be obtained by integrating the experimental distribution along the @xmath10-axis and can be checked self - consistently from the model used in the analysis of the combined data set ( table [ para2 ] ) .
the number of ions per dhdp versus csi bulk concentration are plotted in fig.[cs1](b ) with square symbols . in fig .
[ cs1](b ) the integrated values obtained by pb theory with the surface charge corresponding to the fully deprotonated phosphate groups is also plotted ( triangles connected with a dashed line ) for comparison . as shown , the experimental integrated number of cs@xmath3 at the interface varies roughly as a power - law of bulk concentration , which is well described with rpb ( solid line ) without fitting parameters , as described in section [ theory ] .
the ion distribution predicted from rpb theory using eq .
( [ n_z ] ) with the renormalized gouy - chapman length ( dashed line ) is compared with the experimental distribution in figs .
[ ref4](c ) and [ cs1](a ) ( solid line ) .
as discussed in section [ theory ] the theoretical distribution needs to be convoluted with the effective experimental resolution function .
the distribution resulting from the convoluted rpb , with no fitting parameters , with @xmath36 ( @xmath154 ) obtained from the analysis of the reflectivity , is shown to reproduce the experimental data remarkably well , as shown in fig .
[ ref4](c ) .
the value for @xmath155 , used in the calculation , is consistent within the range of our measured values for the @xmath156 .
the distributions corresponding to the three bulk cs@xmath3 concentrations @xmath157 , and @xmath158 m are shown in fig .
[ cs1](a ) with a solid line .
the agreement with the rpb ( dashed lines ) convoluted as described is remarkable except for points far from the interface .
we attribute this error to the difficulty to include slowly decaying tails of the pb theory to the ed profile modeled by the faster decaying error functions .
this section deals with the examination of the sensitivity of the anomalous reflectivity in determining the ion distribution , in addition to the integrated number of ions at the interface .
the results presented in fig . [ cs1](a ) show the unique capability of the anomalous reflectivity technique in providing ion distributions . whereas the integrated number of ions , such as the ones shown in fig .
[ cs1](b ) , can be obtained from standard reflectivity measurements and other experimental techniques , as discussed in section[introduction ] , the determination of ion distributions requires the use of anomalous reflectivity .
in fact , data such as shown in fig . [
cs1](b ) has been used to assess the validity of pb theory in the past . as discussed in the introduction , the excellent agreement maybe somewhat deceptive in that it may hide significant short - distance deviations from rpb theory because only the total integrated ion density is involved .
( a ) solid line corresponds to the rpb profile at bulk concentration @xmath134 m. step - like functions preserve the integral ( up the first 15 ) of the continuous rpb distribution .
( b ) corresponding convoluted distributions with a gaussian of width @xmath159 .,width=3 ] our experimental resolution constrain possible deviations from rpb distributions to within 3.8 .
if such deviations are non - monotonic , showing bumps or oscillations , then the constraints are even more stringent . as an example , in fig .
[ comp5](a ) we construct two step - wise distributions ( @xmath160 , and @xmath161 steps of 2.5 and 1.9 width , respectively ) whose total integrated area is the same as for rpb ( continuous line ) .
although the histograms have the same area as the rpb , they incorporate hypothetical non - monotonic decay of the distribution in the form of bumps . as it is shown in fig .
[ comp5](b ) , despite the fact that the step - size are smaller than the 3.8 resolution , such distributions can be ruled out by the experiments .
this demonstrates that if actual ion distributions are non - monotonic , their maxima or minima must be short - ranged in nature ( shorter than @xmath162 ) to be consistent with our data .
the goal of this study was to explore the accuracy of pb theory for monovalent ions .
for that , we selected a system with a relatively high surface charge density ( one electron per 40 @xmath22 , lattice constant @xmath163 ) and a low pk@xmath164 system , which we expected would provide the most favorable scenario to observe deviations from pb theory . from the results obtained in our experiments
we conclude that pb with the renormalization of surface charge density due to proton - transfer and release processes ( rpb ) is strikingly accurate .
certainly , the accuracy of our results is limited by the effective experimental resolution , which is dominated by the natural surface - roughness of the gas - amphiphile - water interface ( @xmath165 ) , due mainly to thermally activated capillary - waves .
we should point out , however , that this resolution is comparable to the diameter of one cesium ion ( @xmath166 ) or a single water molecule ( @xmath167 ) , and therefore our experimental distributions constraint deviations from rpb theory to very short - range variations involving one , at most two , water molecules or cesium ions .
if the actual ionic distribution is non - monotonic , departures from rpb theory are even more constrained as it follows from the discussion in fig .
[ comp5 ] .
our results show that theoretical effects that are usually suggested to modify pb theory , such as finite ion - size , in - plane modulations of surface charge density , hydration forces , short - range interactions , the roughness of the surface or image charges are not necessary to describe the experimental data .
this is not to be understood as implying that such effects are not present , but rather that their significance is entirely limited to a characteristic spatial distance of the order of @xmath84 4 or less . as for claims based on the modification of pb by hydration forces
( see for example , ref.@xcite ) , our experimental results conclusively rule out the possibility of modifications of pb within the 10 - 20 range from the interface and confine such corrections , if present , to within the first 3 - 4 from the interface as discussed .
although the experimental counter - ion distributions are well described within the rpb theory , we point out that the reflectivity and gixd hint at water restructuring at the interface . future theoretical or numerical work may clarify this issue .
a recent report analyzing the accuracy of pb near charged liquid - liquid interfaces by the use of x - ray surface sensitive techniques shows that ion distributions are well described by pb at concentrations of 10 mm , but marked deviations are found at higher concentrations ( of the order 100 mm ) , where ion - ion correlations have to be included in the theoretical analysis.@xcite we did not perform anomalous reflectivity for concentrations above 10 mm , so we are not able to provide ionic distributions for these concentrations .
we point out , however , the good agreement found for the integrated quantities at these concentrations ( fig .
[ cs1](b ) ) , which provides an example of integrated quantities possibly hiding deviations from actual distributions as pointed out in the introduction .
the results presented in this study enhance our understanding of the electrostatics in aqueous media , and also show the strength of surface sensitive x - ray synchrotron techniques in obtaining high resolution data .
we thank d. s. robinson and d. wermeille for technical support at the 6-id beamline .
the midwest universities collaborative access team ( mucat ) sector at the aps is supported by the u.s .
department of energy , basic energy sciences through the ames laboratory under contract no .
use of the advanced photon source is supported by the u.s .
department of energy , basic energy services , office of science , under contract no .
w-31 - 109-eng-38 .
the work of at is partially supported by nsf grant dmr-0426597 .
guoy , a. _ j. phys . _
( paris ) * 1910 * , _ 9 _ , 457 .
chapman , d. l. _ phil . mag . _
* 1913 * , _ 25 _ , 475 .
israelachvili , j. _ intermolecular and surface forces _ , academic press , london , 2000 .
stern , o. _ z.elektrochem . _
* 1924 * , _ 30 _ , 508 .
debye , p. p. ; hckel , f. _ phys . z. _ * 1923 * , _ 24 _ , 185 .
grahame , d. c. _ chem .
_ * 1947 * , _ 1 _ , 103 .
outhwaite , c. _
j. c. s. faraday _ * 1978 * , _ 74 _ , 1214 .
outhwaite , c. ; bhuiyan , l. b. ; levine , s. _
j. c. s. faraday ii _ * 1980 * , _ 76 _ , 1388 .
torrie , g. m. ; valleau , j. p. ; patey , g. n. _ j. chem .
phys . _ * 1982 * , _ 76 _ , 4615 .
kjellander , r. ; marcelja , s. _ j. chem .
phys . _ * 1984 * , _ 82 _ , 2122 .
leikin , s. ; parsegian , v. ; rau , d. c. _ annu .
chem . _ * 1993 * , _ 44 _ , 369 .
manciu , m. ; ruckenstein e. ; _ adv . in coll . and int .
* 2004 * , _ 112 _ , 109 .
faraudo , j. ; bresme f. ; _ phys .
_ * 2004 * , _ 92 _ , 236102 .
proesser , a. j. ; franses , e. i. _ colloids and surfaces a _ * 2001 * , _ 178 _ , 1 . hunter , r. j. _ colloid science _ , academic press , london , 1981 .
mclaughlin , s. _ annu .
chem . _ * 1989 * , _ 18 _ , 113 .
tajima , k. _ bull .
_ * 1971 * , _ 44 _ , 1767 .
tajima , k. ; muramatsu , k. ; sasaki , m. _ bull .
jpn . _ * 1970 * , _ 43 _ , 1991 .
kobayashi , k. ; takaoka , k. _ bull .
jpn . _ * 1986 * , _ 59 _ , 93 .
kjaer , k. ; als - nielsen , j. ; helm , c. ; tippman - krayer , p. ; mhwald , h. _ j. phys .
* 1989 * , _ 93 _ , 3200 .
bloch , j.m . ;
yun w. _ phys .
* 1990 * , _ 41 _ 844 .
bedzyk , m. j. ; bommarito , g. m. ; caffrey , m. ; penner , t. l. _ science _ * 1990 * , _ 248 _ , 52 .
lecalvez , e. ; blaudez , d. ; buffeteau , t. ; desbat , b. _ langmuir _ * 2001 * , _ 17 _ , 670 .
travesset , a. ; vaknin d. _ europhys .
_ * 2006 * , _ 74 _ , 181 .
gur , y. ; ravina , i. ; babchin , a. j. _ j. colloid interface sci . _ * 1978 * , _ 64 _ 333 .
paunov , v. n. ; dimova , r. i. ; kralchevsky , p. a. ; broze , g. ; mehreteab , a. _ j. colloid interface sci . _ * 1996 * , _ 182 _ 239 .
vaknin , d. ; krger , p. ; lsche , m. _ phys .
lett . _ * 2003 * , _ 90 _ , 178102 .
bu , w. ; vaknin , d. ; travesset a. _ phys.rev .
e _ * 2005 * , _
72r _ , 060501 .
fleck , c. c. ; netz , r. r. ; _ phys .
lett . _ * 2005 * _ 95 _ 128101 .
safran , s. _ statistical thermodynamics of surfaces , interfaces , and membranes _ , frontiers in physics , perseus publishing , 1994 .
healy , t. w. ; white , l. r. _ adv . in coll . and int .
* 1978 * , _ 9 _ , 303 .
vaknin , d. _ j. am .
soc . _ * 2003 * , _ 125 _ , 1313 .
gregory , b. w. ; vaknin , d. ; gray , j. d. ; ocko , b. m. ; stroeve , p. ; cotton , t. m. ; struve , w. s. _ j. phys .
b _ * 1997 * , _ 101 _ , 2006 .
gregory , b. w. ; vaknin , d. ; gray , j. d. ; ocko , b. m. ; cotton , t. m. ; struve , w. s. _ j. phys .
b _ * 1999 * , _ 103 _ , 502 .
vaknin , d. in _ methods in materials research _ , edited by kaufmann , e. n. _ et al .
_ ; wiley , new york , 2001 ; p 10d.2.1 . als - nielsen , j. ; pershan , p. s. _ nucl .
methods phys .
rev . _ * 1983 * , _ 208 _ , 545 .
als - nielsen , j. ; kjaer , k. in _ phase transitions in soft condensed matter _ , edited by riste , t. ; sherrington , d. ; plenum , new york , 1989 .
jacquemain , d. ; wolf , s. g. ; leveiller , f. ; deutsch , m. ; kjaer , k. ; als - nielsen , j. ; lahav , m. ; leiserowitz , l. _ angew .
chem . _ * 1992 * , _ 31 _ , 130 .
kjaer , k. _ physica b _ * 1994 * , _ 198 _ , 100 .
parratt , l. g. _ phys .
rev . _ * 1954 * , _ 59 _ , 359 .
born , m. ; wolf , e. _ principles of optics _
, macmillan , new york , 1959 .
lavoie , h. ; blaudez , d. ; vaknin , d. ; desbat , b. ; ocko , b. m. ; salesse , c. _ biophys . j. _ * 2002 * , _ 83 _ , 3558 .
the 3 density parameters @xmath110 are now replaced by 5 parameters in eqs . ( 18 ) and ( 20 ) ( @xmath168 , in the headgroup slab and the slab below , and the molecular area @xmath169 which is restricted to @xmath170 @xmath22 ) .
the two volume constraints in eq .
( 22 ) compensate for the use of two extra parameters .
vaknin , d. ; kjaer , k. ; als - nielsen , j. ; lsche , m. _ biophys . j. _ * 1991 * , _ 59 _ , 1325 .
vaknin , d. ; kjaer , k. ; als - nielsen , j. ; lsche , m. _ makromol .
, macromol .
symp . _ * 1991 * , _ 46 _ , 383 .
kaganer , m. v. ; mhwald , h. ; dutta , p. _ rev .
phys . _ * 1999 * , _ 71 _ , 779 . the coordinate system used in this manuscript is such that @xmath171 is normal to the liquid surface , @xmath172 is parallel to the horizontal ( untilted ) incident x - ray beam , and @xmath173 is orthogonal to both @xmath172 and @xmath171 .
the hydrocarbon chains form two dimensional poly - crystals giving rise to a diffraction pattern that depends on the modulus of the in - plane momentum transfer @xmath174 and is practically independent of sample rotation over the z - axis .
jancovici , b. _ phys .
* 1967 * , _ 19 _ , 20 .
bowick , m. j. ; travesset , a. _ phys .
rep . _ * 2001 * , _ 344 _ , 255 .
luo , g. m. ; malkova , s. ; yoon , j. ; schultz , d. g. ; lin , b. h ; meron , m. ; benjamin , i. ; vanysek , p. ; schlossman , m. l. ; science , * 2006 * , _ 311 _ , 216 ( this report became available to us after the submission of this manuscript ) .
|
surface sensitive synchrotron x - ray scattering studies were performed to obtain the distribution of monovalent ions next to a highly charged interface .
a lipid phosphate ( dihexadecyl hydrogen - phosphate ) was spread as a monolayer at the air - water interface to control surface charge density . using anomalous reflectivity off and at the @xmath0 cs@xmath1 resonance ,
we provide spatial counterion ( cs@xmath1 ) distributions next to the negatively charged interfaces .
five decades in bulk concentrations are investigated , demonstrating that the interfacial distribution is strongly dependent on bulk concentration .
we show that this is due to the strong binding constant of hydronium h@xmath2o@xmath1 to the phosphate group , leading to proton transfer back to the phosphate group and to a reduced surface charge .
the increase of cs@xmath3 concentration modifies the contact value potential , thereby causing proton release .
this process effectively modifies surface charge density and enables exploration of ion distributions as a function of effective surface charge - density .
the experimentally obtained ion distributions are compared to distributions calculated by poisson - boltzmann theory accounting for the variation of surface charge density due to proton release and binding .
we also discuss the accuracy of our experimental results in discriminating among possible deviations from poisson - boltzmann theory .
|
the control of quantum superpositions in coupled electronic systems has been suggested as a possible way to realize quantum logic gates in semiconductor structures .
an example are coupled quantum dots @xcite , where coulomb interactions between electrons can be exploited @xcite to define very small effective hilbert spaces such as that of two tunnel - splitted ground states , separated by a large energy gap from the remaining many - particle states@xcite . if parameters of the hamiltonian ( like the tunnel - coupling between dots ) are slowly changed as a function of time , adiabatic control of the state vector @xcite , swap operations @xcite , and the controlled transfer from an initial to a final state @xcite become possible .
in addition , by coupling such a system to external electronic reservoirs , one can pump electrons through the system in a controlled , adiabatic manner which in principle can serve as a ` read out ' of the state vector in form of an electric current . in the simplest case
, this can be achieved through a simultaneous variation of two parameters as a function of time .
both adiabatic control of rotations or swaps , and coherent pumping of electrons evidentally are very sensitive to decoherence .
the coupling to dissipative degrees of freedom such as phonons disturbs the coherent time - evolution and therefore leads to a loss of control over the desired superposition .
decoherence and dissipation in quantum xor gate operations have been discussed recently by thorwart and hnggi @xcite in a powerful numerical scheme .
they found that properties like gate fidelities are very sensitive to the dissipative bath coupling constant , but only weakly depend on temperature . in this article , we quantitatively investigate the role of dissipation for one- and two - qubit adiabatic ` rotations ' of an electron between two coupled quantum dots .
for the electron charge one - qubit , we discuss a ` quantum electron pump ' for dots which are coupled to external leads ( electron reservoirs ) . in our scheme , decoherence is mainly due to absorption of bosons in excitations of the instantaneous ground state which leads to an exponential temperature dependence . for a sinusoidal pulse ,
we use an exact solution and perturbation theory in the phonon coupling to predict how quantum mechanical oscillations between the dots can be made visible in a ` read out ' electronic current , similar to the experiment by nakamura _ et al . _ in a superconducting cooper pair box @xcite and the recent electron spin resonance scheme for dots by engel and loss @xcite .
first , we compare results for both weak and strong coupling to the bosons and derive an analytic expression for the influence of the bath on the fidelity of the operation .
second , we investigate two - qubit swap operations in spin - based two - electron quantum dots .
these _ spin _ qubits sensitively depend on _ charge _ decoherence due to intermediate states where charge has tunneled between the dots .
piezoelectric phonons coupled to the electron charge incoherently mix states in the singlet sector and lead to a loss of fidelity of the swap operation .
adiabatic transfer and pumping of charges through small metallic islands or semiconductor quantum dots @xcite has already been demonstrated experimentally .
furthermore , in the strong coulomb blockade regime of coupled quantum dots , the non - adiabatic coupling to ac fields in photo - assisted transport @xcite has been established in experiments @xcite . in the opposite regime of weak coulomb correlations , experiments in open dots
@xcite have demonstrated the feasability of an ` adiabatic quantum electron pump ' .
these systems can be described as non - interacting mesoscopic scatterers @xcite .
a _ combination _ of strong coulomb blockade and the adiabatic control of the wave function in a triple quantum dot has been suggested recently @xcite .
furthermore , in superconducting josephson junction qubits @xcite , adiabatic quantum computation with cooper pairs @xcite and adiabatic controlled - not gates for quantum computation have been proposed by averin @xcite . the decoherence properties of adiabatic one- and two - qubit operations are closely related to the dissipative landau - zener problem ( see , e.g. , the review @xcite for further references ) . in time - dependent operations , decoherence rates in general become time - dependent themselves and therewith usually involve the whole ( or a large part ) of the spectrum of frequencies in the effective spectral density @xmath0 of the bosonic bath .
as we show below , adiabatic pulses can be chosen such that qubit rotations run at a _ constant _ energy gap @xmath1 to the excited state whereby dissipation is due to @xmath2 , i.e. , one boson frequency only .
this can be used to extract @xmath0 from the pump current by using a series of pulses with different @xmath1 . a suppression of @xmath0 at certain frequencies
@xmath3 has been predicted in free - standing ` phonon cavities ' due to symmetry and geometrical confinement @xcite . here
, we demonstrate that a ` rabi ' rotation pulse tuned to constant energy difference @xmath4 effectively ` switches off ' the decoherence in such systems , at least within second order in the coupling constant .
consequently , this defines a one - dimensional ` decoherence - free manifold ' ( curve ) on the adiabatic groundstate energy surface of the system , which might be of interest for adiabatic quantum computation schemes suggested recently @xcite .
the paper is organized as follows : in section [ section_one_qubit ] , we discuss one - qubit rotations , introduce the time - dependent spin - boson model , and describe the adiabatic transfer of an electron between quantum dots with and without dissipation . in section [ sectionfidelity ] , we derive a perturbative , analytical expression for the one - qubit fidelity , compare with the strong - coupling case , and suggest a scheme to extract quantum oscillations as a ` read out ' electron pump current . in section [ section_two_qubit ] , we turn to spin - qubit swaps and charge decoherence for the gate discussed recently by schliemann , loss , and macdonald . finally , section [ conclusion ] is a short conclusion .
adiabatic transfer in a two - level system consists in the rotation of a chosen initial state @xmath5 into a final state @xmath6 by an adiabatic variation of system parameters .
the simplest example is a spin @xmath7 in a slowly rotating magnetic field .
using the analogy with adiabatic steering in atomic three - level systems , the possibility of adiabatic pumping of electrons through _ triple _ quantum dots has been suggested recently @xcite , using a suitable ` design ' of instantaneous energies through two time - dependent tunnel couplings .
here , we consider the transfer of an electron through a _ double _ quantum dot via quasi - stationary adiabatic eigenstates .
this ` charge qubit ' @xcite , fig .
( [ 2surface ] ) left , is defined by two electron states @xmath8 and @xmath9 with a time - dependent energy difference @xmath10 and coupled by a tunnel matrix element @xmath11 , as described by the time - dependent hamiltonian @xmath12 with @xmath13 and @xmath14 . experimental control of ( constant ) @xmath15 and
@xmath16 has been demonstrated in double quantum dots @xcite .
the initial state is an additional electron in the left dot with an energy @xmath17 well below the chemical potential of the left lead , the final state is an additional electron in the right dot which then ( keeping @xmath18 ) is lifted above the chemical potential of the right lead .
such a transfer cycle of the open system ( coupled to electron reservoirs ) in the coulomb blockade regime requires a hilbert space @xmath19 spanned by three states @xmath20 , @xmath8 , and @xmath9 @xcite . here
, the two basis states @xmath21 and @xmath22 describe one additional electron in the left ( right ) dot above a ground state @xmath23 ( ` empty state ' ) . for the remainder of this section
, we only consider the first part of the transfer cycle , i.e. , the dynamics of the double dot isolated from the leads , and turn to the full transfer cycle including tunneling to and from the leads in section [ tunnelsection ] .
left : double dot with time - dependent energy level difference @xmath24 and tunnel matrix element @xmath11 , connected to electron reservoirs .
right : surface of the lower energy eigenvalue @xmath25 of the two - level hamiltonian @xmath26 , eq .
( [ h0define ] ) . to adiabatically transfer an electron from the left to the right dot , @xmath15 and @xmath16
are varied as a function of time as in eq .
( [ etpulse ] ) , corresponding to the curve on the @xmath25 surface .
, title="fig:",scaledwidth=20.0% ] left : double dot with time - dependent energy level difference @xmath24 and tunnel matrix element @xmath11 , connected to electron reservoirs .
right : surface of the lower energy eigenvalue @xmath25 of the two - level hamiltonian @xmath26 , eq .
( [ h0define ] ) . to adiabatically transfer an electron from the left to the right dot , @xmath15 and @xmath16
are varied as a function of time as in eq .
( [ etpulse ] ) , corresponding to the curve on the @xmath25 surface .
, title="fig:",scaledwidth=20.0% ] an adiabatic transfer from left to right requires the simultaneous change of at least two parameters such as @xmath27.\end{aligned}\ ] ] this corresponds to a change of @xmath28 with a simultaneous switching of the tunnel coupling @xmath11 between the dots .
the precise form of the pulse , eq.([etpulse ] ) , is not important and has been chosen for convenience here .
the instantaneous , hybridized eigenstates of the isolated coupled quantum dot are @xmath29\\ n_{\pm}&:= & \sqrt{4|t_c|^2+(\delta \mp \varepsilon)^2},\quad \delta:=\sqrt{\varepsilon^2 + 4|t_c|^2}\nonumber.\end{aligned}\ ] ] the eigenvalues @xmath30 of the coupled system represent two energy surfaces over the @xmath16-@xmath15 plane , the lower of which ( ground state ) is shown in fig .
1a . the pulse eq .
( [ etpulse ] ) corresponds to a curve on the @xmath25 surface . the corresponding change of the ` inversion ' @xmath31 is obtained from a numerical integration of the ( coherent ) equation of motion for the density matrix of the system .
the result is shown in fig .
2 , together with the form of the pulses eq .
( [ etpulse ] ) . in accordance with the adiabatic theorem ,
the initial groundstate @xmath8 of the system is rotated into the instantaneous superposition @xmath32 of @xmath8 and @xmath9 , eq .
( [ eq : twobytworesult ] ) , if the rotation is ` slow ' , i.e. , @xmath33 . in this case
, the time enters as a parameter into the state @xmath32 which is used to calculate the approximate expectation value @xmath34 which excellently reproduces the overall form of the numerically obtained @xmath31
. the form eq .
( [ adiabaticsz ] ) corresponds to crisp s solution for the adiabatic following of an atom in a near resonance light pulse @xcite .
the exact solution exhibits the expected rabi oscillations with frequency @xmath35 around the adiabatic value , which are strongest when the tunnel coupling is fully switched on .
inversion @xmath36 for transfer from left ( @xmath37 ) to right ( @xmath38 ) in a two - level system , eq . ( [ h0define ] ) .
inset : time - dependent tunnel matrix element @xmath11 and energy splitting @xmath28 , eq .
( [ etpulse ] ) .
energies ( times ) are in units of the amplitude @xmath16 ( @xmath39 ) in eq .
( [ etpulse ] ) ; the other parameters are @xmath40 , @xmath41 , @xmath42 , @xmath43 , @xmath44 . the adiabatic approximation eq .
( [ adiabaticsz ] ) is shown as dotted line.,scaledwidth=50.0% ] due to landau - zener tunneling from the adiabatic ground state @xmath32 to the excited state @xmath45 , there is always a finite albeit small probability @xmath46 for the electron to remain in the left dot , i.e. , the excited state after the rotation .
@xmath46 can be made exponentially small for large enough level splitting @xmath1 and slow pulses , but depends on the exact pulse shape @xmath47 .
the adiabatic rotation discussed above is an idealization and valid only for a two - level system completely isolated from its environment . in semiconductor quantum dots
, the coupling to low - energy bosonic excitations of the surrounding electron system , to photons , and to phonons leads to deviations from the coherent time - evolution .
these excitations are characterized by an effective spectral density @xmath48 of modes @xmath49 with frequency @xmath50 and constants @xmath51 coupling to the electron charge in the dot . in this paper , we only consider the coupling of phonons to the charge density which has been found to be dominant coupling mechanism in double quantum dots @xcite .
other coupling mechanisms like inelastic spin - flip scattering @xcite lead to additional dephasing channels , so that our results can only be considered as a lower bound on dephasing . on the other hand ,
the time - dependent spin - boson hamiltonian ( @xmath52 throughout ) @xmath53 employed here ( and derived in appendix a ) is of sufficient general form for our results being transferable to similar dissipative one - qubit rotations in other two - level systems .
the function @xmath54 can be calculated explicitely if one assumes sharply peaked electron densities of negligeable width around the dot centers @xmath55 , having a distance @xmath56 and zero extension into the @xmath57 ( growth ) direction . for bulk piezo - electric phonons ,
one obtains @xcite @xmath58 \theta(\omega),\end{aligned}\ ] ] where @xmath59 is a dimensionless coupling constant and @xmath60 , with @xmath61 the velocity of the phonon mode . in general
, a finite extension @xmath62 of the electron densities in lateral or growth direction leads to a form factor that cuts off phonons with frequencies @xmath63 ( ` phonon bottleneck ' ) .
a microscopic determination of @xmath0 would require ( apart from details of the microscopic electron - phonon interaction potential ) exact knowledge of the many - body electron density in the dots .
we argue that the assumption of relatively sharply localized positions between which the additional electron tunnels is justified by the strong intra - dot electron - electron repulsion @xcite . the origin of the oscillatory form of eq .
( [ rhoomegadefinition ] ) @xcite lies in the double - slit like scattering of the phonons ( waves with wave vector @xmath49 ) at the electron charge that is delocalized between the two dots . in the spin - boson problem ,
one often assumes a parametrized form of the spectral density @xcite , @xmath64 where @xmath65 is a high - frequency cut - off .
notice that the ohmic case @xmath66 describes the microscopic form @xmath67 , eq . ( [ rhoomegadefinition ] ) , in the limit @xmath68 .
however , phonons interacting with the dot through other coupling mechanisms like the deformation potential coupling , surface acoustic waves @xcite , or rayleigh - lamb waves in confined geometries @xcite , lead to other forms of the spectral density @xmath0 . throughout the discussion of the one - qubit dynamics
, we will assume the simple ohmic form , eq.([rhosdefinition ] ) with @xmath66 , and return to the form eq .
( [ rhoomegadefinition ] ) in the discussion of the two - qubit in section [ section_two_qubit ] .
the coupling to the bosonic bath introduces decoherence and in general ( exceptions are discussed below ) leads to a loss of fidelity of the adiabatic rotation , i.e. even for very slow and adiabatic pulses @xmath47 , the transfer from the left to the right state remains imperfect with the final expectation value @xmath36 considerably deviating from @xmath69 . during the rotation ,
the pure adiabatic state decays into a mixture .
we solve for the reduced density matrix @xmath70 of the two - level system coupled to external electron reservoirs @xcite .
different techniques can be applied for weak boson coupling ( perturbation theory ) and strong boson coupling ( polaron transformations or path integral in ` niba ' approximation @xcite ) . in general , to obtain the solution for ime - dependent spin - boson problems even numerically is a non - trivial task @xcite . from the liouville - von neumann equation ,
the diagonal elements of @xmath70 are easily obtained as @xmath71 \\ & + & \gamma_l\left[1-\rho_{ll}(t)-\rho_{rr}(t)\right ] -\bar{\gamma}_l\rho_{ll}(t)\nonumber\\ \frac{\partial}{\partial t}\rho_{rr}(t)&=&-it_c(t)\left [ \rho_{rl}(t)-\rho_{lr}(t ) \right ] \\ & + & \gamma_r\left[1-\rho_{ll}(t ) - \rho_{rr}(t)\right ] -\bar{\gamma}_r\rho_{rr}(t)\nonumber,\end{aligned}\ ] ] where we have included the coupling to the electron reservoirs ( see fig .
1 ) with the tunnel rates ( @xmath72 ) @xmath73,\end{aligned}\ ] ] where @xmath74 is the fermi distribution in lead @xmath75 and @xmath76 the rate for tunneling between dot @xmath75 and lead @xmath75 .
the equations for the off - diagonal elements @xmath77 can be obtained only approximatively by performing a perturbation expansion in either @xmath59 or @xmath16 : in the first approach , one starts from the basis of the hybridized states @xmath78 and considers the term @xmath79 in the hamiltonian eq.([modelhamiltonian ] ) as a weak perturbation .
if @xmath15 and @xmath16 were constant , one could easily define an interaction picture with respect to @xmath80 and proceed in the standard way , i.e. second order perturbation theory and tracing out of the bosonic degrees of freedoms .
however , when @xmath28 and @xmath11 become a function of time , the time - evolution of the unperturbed system in general can not be written down analytically which makes the evaluation of the electron - boson terms very tedious . here , we use an adiabatic approximation by regarding the time @xmath81 in @xmath28 and @xmath11 as a parameter for the derivation of the incoherent ( electron - boson ) part of the master equation for @xmath82 .
we neglect memory effects of the bosonic system and derive @xmath82 up to second order in @xmath83 ( born - markov approximation ) .
the bosonic environment enters solely via the correlation function of the operator @xmath84 , eq .
( [ modelhamiltonian ] ) , in the interaction picture , @xmath85 which yields @xmath86 , \ ] ] where @xmath87^{-1}$ ] is the bose distribution at temperature @xmath88 ( @xmath89 ) .
the result is @xmath90 here , the coefficients @xmath91 and @xmath92 ( suppressing the parameter @xmath81 in the following ) are defined as @xmath93 the coupling to the bosonic bath introduces a dephasing rate @xmath91 of the off - diagonal element @xmath94 , @xmath95 and additional terms ( @xmath92 ) in the coupling to the diagonals . in the ohmic case @xmath66 of the spectral density @xmath96 an additional contribution to @xmath91 linear in the temperature @xmath88
appears which , however , is unphysical in the sense that for the microscopic @xmath97 this term does not appear , cf .
eq.([rhoomegadefinition ] ) and appendix b. note that in order to be consistent within the born - markov approximation , the scattering rates in eq .
( [ gammarates ] ) have to fulfill @xcite @xmath98 .
one should also keep in mind that within our adiabatic approximation , these rates are time - dependent through the time - dependence of @xmath28 , @xmath11 , and @xmath99 .
furthermore , the imaginary parts of @xmath92 can be understood as a renormalization of the tunneling rate @xmath16 , as can be seen from eq .
( [ rholr ] ) . at finite temperatures
, they have to be computed numerically .
one can derive an equation for the off - diagonal @xmath82 in a second , alternative approach by performing a unitary polaron transformation of the original hamiltonian .
this method basically is a perturbation theory in the tunnel - coupling @xmath16 and suitable for strong coupling to the bosonic bath .
the unitary transformation @xmath100 is applied to the hamiltonian , leading to @xmath101 + h_b,\end{aligned}\ ] ] where @xmath102 , @xmath103 , and a constant c - number energy term in @xmath104 has been dropped .
the transformation eq .
( [ polarontrafo ] ) has removed the linear term @xmath105 and instead introduced the product @xmath106 of unitary displacement operators for the bosonic modes @xmath49 into the tunneling term . the master equation for @xmath94
is now derived in this transformed picture in lowest order perturbation theory in @xmath16 .
the result is @xmath107\ ] ] where @xmath108 is an equilibrium correlation function with respect to the bosonic bath ( inverse temperature @xmath109 , spectral density @xmath0 , eq .
( [ rhodefinition ] ) ) , @xmath110.\nonumber\end{aligned}\ ] ] in order to discuss the effect of dissipation on the adiabatic transfer , we now numerically evaluate and compare the equations of motion for the density matrix @xmath70 in both the perturbative and the polaron transformation approach . here
, we consider a closed two - level system without coupling to external electron reservoirs ( @xmath111 ) and comment on the ` open ' system case ( ` quantum pump ' ) below .
the boson spectral density @xmath112 , eq .
( [ rhosdefinition ] ) , always has ohmic form .
first , the strong coupling equations eq .
( [ rhollrr ] ) and eq .
( [ eom3new ] ) can be condensed into a closed equation for @xmath31 as @xmath113f_{\pm}(t , t')\\ f_{\pm}(t , t')&:= & \pm 2t_c(t)t_c(t ' ) { \rm re } \left\{e^{\pm i\int_{t'}^t ds \varepsilon(s ) } c(t - t')\right\}.\nonumber\end{aligned}\ ] ] this single integro - differential equation @xcite can be solved by standard numerical techniques .
results from both the perturbative approach , eq.([rholr ] ) and eq.([rhollrr ] ) , and the polaron transformation , eq.([polaronsigmaz ] ) are shown in fig.([integro_3.eps ] ) for the weak coupling regime ( @xmath114 ) , where the same form of pulses eq.([etpulse ] ) as in the coherent case , fig.([integro_2.eps ] ) , has been chosen .
we first checked in a separate calculation that the influence of the imaginary parts @xmath115 , eq .
( [ raten - ohmsch ] ) , on the solution of the master equation eq .
( [ rholr ] ) is negligible .
the perturbative inclusion of dissipation at low temperatures leads to the expected , small change of the time - evolution of @xmath36 , with @xmath36 saturating slightly above the value @xmath38 of the coherent case .
this means that in the dissipative case , the one - qubit rotation , i.e. the transfer of the electron from left to right , is never complete even if the rotation is performed adiabatically .
we will derive an analytic expression for this loss of fidelty in presence of dissipation in section [ sectionfidelity ] . in comparison to the perturbative approach ,
the polaron transformation approach breaks down at weak couplings @xmath59 and low temperatures where it predicts inversions below @xmath69 towards the end of the rotation .
this breakdown if well - known from the static case ( @xmath16 , @xmath15 fixed ) . on the other hand ,
the results from the polaron approach agree fairly well with the perturbative results at higher temperatures , as can be clearly infered from fig .
( [ integro_3.eps ] ) which again is consistent with the spin - boson dynamics for fixed @xmath16 and @xmath15 . two - level system as in fig . ( [ integro_2.eps ] ) , but with ( weak ) dissipation at finite temperatures @xmath88 .
comparison between the strong coupling polaron transformation ( pt ) solution and the perturbative solution is made.,scaledwidth=50.0% ]
the previous discussion has shown that the adiabatic transfer remains incomplete in the presence of dissipation , i.e. @xmath36 can considerably deviate from its non - dissipative value @xmath69 in the perfectly adiabatic case where no landau - zener transitions occur .
our aim in this section is to analyze how the deviation @xmath116 depends on temperature , boson spectral density @xmath0 , and the duration of the swap operation .
intuitively , it is clear that for too long swap duration , inelastic transitions to the excited level will have sufficient time to destroy the coherent transfer . on the other hand ,
if the swap operation proceeds too fast , landau - zener transitions become stronger and again lead to deviations @xmath117 .
this means that for any given inelastic rate @xmath118 there should be an optimal swap duration such that these two competing processes balance each other .
one problem in quantifying this argument is that @xmath118 naturally is time - dependent itself .
silvestrini and stodolsky nevertheless have suggested @xcite to extract decoherence rates in experiments by performing an adiabatic sweep and determining the maximum sweep duration for which the inversion is still successfull .
however , their model of a bloch equation ( with @xmath118 as a constant parameter ) is too simple in order to make detailed quantitative predictions here . in the following
, we calculate the deviation @xmath119 for both weak and strong electron - boson coupling and derive analytic expressions that allow to quantify the arguments above . the main physical idea in
this discussion is the introduction of curves ( ellipses in the @xmath15-@xmath16 plane ) where the excitation energies @xmath120 remain constant as a function of time and consequently , dissipation is determined by a constant boson spectral density @xmath96 . in the weak coupling case ,
we follow grifoni and hnggi @xcite and perform a unitary transformation of the original hamiltonian ( [ modelhamiltonian ] ) to a hamiltonian @xmath121 , @xmath122 where the matrix @xmath123 contains the columns of the hybridized eigenstates @xmath32 and @xmath45 , eq . ( [ eq : twobytworesult ] ) , @xmath124 , @xmath125 , and the time - dependence of @xmath11 and @xmath28 is again considered as parametric .
we introduce the amplitude @xmath126 for an inelastic transition from the adiabatic ground state @xmath32 at time @xmath127 to the excited state @xmath45 at time @xmath128 , i. e. a transition induced by the coupling to the bosons .
the corresponding probability @xmath129 describes the deviation @xmath130 of the inversion due to the coupling to the bosons .
the amplitude @xmath126 is given by the non - vanishing matrix element of the time evolution operator , expanded to lowest order , @xmath131 where the interaction picture is with respect to @xmath132 .
with @xmath133 = @xmath134 , we find the corresponding probability for a transition from @xmath135 to @xmath136 due to the interaction with the bosons as @xcite @xmath137 f(-\omega , t)\big\ } \nonumber\\ f(\omega , t ) & : = & \left|\int_0^t dt ' \frac{t_c(t')}{\delta(t ' ) } e^{-i\int_{0}^{t'}ds \left[\delta(s)-\omega\right]}\right|^2.\end{aligned}\ ] ] ` elliptic ' pulses @xmath138 , eq . ( [ ellipse ] ) ,
are defined by curves on the @xmath139-surface ( fig .
[ 2surface ] ) with constant energy difference @xmath1 to the excited state @xmath140 . for the particular sinusoidal form @xmath141 of @xmath11 and @xmath28 , the time - dependent hamiltonian @xmath142 $ ] , eq.([h0define ] ) , becomes exactly integrable .
it corresponds to a spin @xmath7 in a magnetic field that rotates within the @xmath143-@xmath57-plane around the @xmath144-axis with frequency @xmath145 .
the solution for the inversion @xmath31 is easily obtained by transforming the schrdinger equation into a rotating frame ( rabi solution ) .
one obtains @xmath146\cos\omega t\\ & + & \frac{\omega}{\omega_r}\sin\omega_r t\sin\omega t,\quad \omega_r:=\sqrt{\omega^2+\delta^2}\nonumber.\end{aligned}\ ] ] here , the rabi frequency @xmath147 corresponds to maximal detuning @xmath148 since there is no static ` magnetic field ' @xmath149 in @xmath144-direction . the use of the harmonic pulse eq .
( [ harmonicpulses ] ) has the further advantage that the quantity @xmath150 , eq .
( [ p+- ] ) , can be evaluated analytically .
the swap operation requires a pulse acting half a period from the initial time @xmath127 to the final time @xmath151 . using @xmath152 in @xmath150
, we obtain @xmath153 ^ 2,\quad x : = ( \delta-\omega)/\omega.\end{aligned}\ ] ] in the adiabatic limit @xmath154 , we find an approximation to the integrals eq .
( [ p+- ] ) . in that limit
, one has @xmath155 whence the dissipation induced change of the inversion becomes @xmath156 .
the sum of the coherent contribution @xmath130 from the rabi solution at @xmath152 , eq .
( [ sigmazrabi ] ) , and the perturbative dissipative contribution @xmath157 leads to @xmath158\\ & + & 2\frac{c}{\omega}\frac{\rho(\delta)}{\exp({{\delta}/{k_bt}})-1},\quad{\omega}\ll{\delta } , \quad c= 2.4674\nonumber.\end{aligned}\ ] ] inversion change @xmath159 after time @xmath151 for sinusoidal pulses eq.([harmonicpulses ] ) as obtained from the master equation , eq.([rholr ] ) .
dotted curves correspond to the analytical prediction , eq .
( [ sigmazanalytic]).,scaledwidth=50.0% ] results for the inversion change @xmath160 from the ideal value @xmath161 as a function of the pulse frequency @xmath145 for a half - period sweep ( duration @xmath151 ) are shown in fig .
( [ ddotzener_omega.eps ] ) .
we compare the results from the ( weak - coupling ) master equation , eq.([rholr ] ) , and the analytical prediction eq.([sigmazanalytic ] ) .
the agreement between the numerical and the analytical prediction ( dotted lines ) , is extremely good for small coupling constants ( @xmath162 ) , but deviations become strong at small @xmath145 where the simple second order perturbation theory becomes worse .
the @xmath163 dependence of the dissipative contribution to @xmath160 is clearly visible at small @xmath145 , indicating that for too long pulse duration the electron swap remains incomplete due to incoherent dissipation .
on the other hand , if the pulse duration is too short ( larger @xmath145 ) , the oscillatory coherent contribution from @xmath164 dominates .
one should bear in mind , however , that eq .
( [ sigmazanalytic ] ) is an approximation that only holds in the limit of an infinitely slow adiabatic change , i.e. @xmath165 .
in fact , for any finite pulse duration @xmath166 , even in the limit of zero temperature @xmath167 , eq .
( [ p+- ] ) yields @xmath168 which shows that there is a small , but finite probability for inelastic processes during transitions from @xmath32 to @xmath45 even at zero temperature , in agreement with @xcite .
these processes are due to the spontaneous emission of bosons which occur _ during _ landau - zener transitions from @xmath32 to @xmath45 with a finite probability as long as @xmath169 is finite . for strong electron - boson coupling constants
@xmath59 , we used eq .
( [ polaronsigmaz ] ) to extract the final inversion @xmath170 after the time @xmath152 for the pulse eq.([etpulse ] ) .
an interesting observation can be made in the temperature behavior of @xmath170 : for couplings @xmath171 , a temperature increase leads to an increase of @xmath170 , which is as in the weak coupling case .
however , above @xmath172 , the temperature dependence changes in that larger temperatures @xmath88 lead to smaller values of @xmath170 .
in fact , for large coupling constants @xmath59 , the system tends to remain localized in the left dot state @xmath8 and no tunneling to the right state @xmath9 occurs . in this regime , higher temperatures destroy the localization and lead to smaller @xmath170 .
this behavior again is consistent with the borderline @xmath173 ( @xmath174 ) @xcite in the dissipative two - level dynamics @xcite for static parameters @xmath15 and @xmath16 .
inversion @xmath175 for strong electron - boson coupling after application of the pulse @xmath11 and @xmath28 , eq .
( [ etpulse ] ) , as in figs .
( [ integro_2.eps ] ) and ( [ integro_3.eps ] ) .
clearly visible is the crossover at @xmath176 where the temperature dependence changes.,scaledwidth=50.0% ] in the remainder of this section , we discuss how results for the closed double dot system , and in particular the coherent and incoherent inversion swap deviation @xmath160 like the one for the rabi rotation , eq .
( [ sigmazanalytic ] ) , relate to electron transport in a double dot system coupled to external leads . here
, we only consider the weak dissipation case .
the main idea is to apply time - dependent pulses such that the quantum mechanical time evolution of the two - level system is well separated from a merely ` classical ' decharging and charging process .
one complete cycle of such an operation is sketched in fig .
[ rabi.eps ] .
the cycle starts with an additional electron in the left dot and an adiabatic rotation of the parameters @xmath177 such as , e.g. , in eq .
( [ harmonicpulses ] ) or in eq .
( [ etpulse ] ) , cf .
[ rabi.eps ] a - b .
this completely quantum - mechanical part of the cycle is performed in the ` save haven ' of the coulomb- and the pauli - blockade @xcite , i.e. , with the left and right energy levels of the two dots well below the chemical potentials @xmath178 of the leads ( which are assumed identical here for simplicity ) .
the cycle continues with closed tunnel barrier @xmath18 and increasing @xmath179 ( fig .
[ rabi.eps ] c ) ; the two dots then are still in a superposition of the left and the right state . the subsequent lifting of the right level above the chemical potential of the right lead ( fig . [ rabi.eps ]
d ) constitutes a measurement of that superposition ( collapse of the wave - function ) : the electron is either in the right dot ( with a high probability @xmath180 ) and tunnels out , or the electron is in the left dot ( and nothing happens because the left level is still below @xmath178 and the system is coulomb blocked ) . if the tunnel rates @xmath181 to the right and left leads are sufficiently larger than the inverse of the cyle duration @xmath182 , @xmath183 the decharging of the right dot and the re - charging of the left dot from the left lead is fast enough to bring the system back into its initial state with one additional electron on the left dot . in this case
, the precise value of @xmath181 , and the precise shape of the @xmath28-pulse for @xmath184 has no effect on the total charge transfered within one cycle .
then , since the probability to transfer one electron from the left to the right in one cycle is given by @xmath180 , on the average an electron current @xmath185 flows from left to right .
note that the leads essentially act as classical measurement devices of the quantum - mechanical time - evolution between the two dots .
measuring the current @xmath186 as a function of the pulse length @xmath151 then offers a scheme to make quantum mechanical oscillations such as those predicted in eq .
( [ sigmazanalytic ] ) visible in the electronic current , similar to the recent experiment by nakamura _
_ in a superconducting cooper pair box @xcite .
adiabatic scheme for cycling electrons from left to right through a double quantum dot with energy level difference @xmath24 .
left and right energy levels are assumed to change symmetrically with respect to @xmath187.,scaledwidth=50.0% ] the ` rabi - pulse ' , eq .
( [ harmonicpulses ] ) , keeps the energy difference to the excited state @xmath45 constant throughout the adiabatic rotation . in this case
, it follows that the dissipative contribution to @xmath160 , eq.([sigmazanalytic ] ) , is due to @xmath2 , i.e. , one fixed boson frequency only .
furthermore , the analytic result , eq.([sigmazanalytic ] ) , in principle allows to extract the value of the phonon spectral density , @xmath96 , for ` rabi'-pulses at fixed energy difference @xmath1 .
another important observation is the fact that the rotation remains dissipation - free if @xmath1 is chosen to coincide with a zero of @xmath0 .
this defines a ` decoherence - free manifold ' in the parameter space of the system .
zeroes in @xmath0 at certain frequencies @xmath3 have been predicted @xcite in free - standing , two - dimensional _ phonon cavities _ ( slabs ) .
in fact , the control of vibrational properties of quantum dot qubits has been suggested @xcite , and considerable progress has been made in the fabrication of nano - structures that are only partly suspended or even free - standing @xcite . in a thin - plate cavity
model@xcite , the zeroes in @xmath0 are due to symmetry and geometrical confinement both for deformation potential and piezoacoustic phonon scattering in second order of the coupling constant @xmath59 .
this means that within a phonon cavity , the dissipative contribution to @xmath160 , eq . ( [ sigmazanalytic ] ) , can be ` switched off ' if @xmath1 is tuned to the energy @xmath188 of a decoupled phonon mode . for gaas slabs of width @xmath189 m
, this energy has been predicted to be of the order @xmath190ev .
an energy @xmath191ev corresponds to @xmath192mk such that the temperatures shown in fig .
[ ddotzener_omega.eps ] are attainable in current experiments .
a typical pulse frequency @xmath193 then would correspond to @xmath194 mhz , the corresponding current being of the order @xmath195pa .
furthermore , we mention that the weak coupling regime with @xmath196 seems to be justified to describe recent experiments on phonon coupling in double quantum dots @xcite .
in this section , we apply our formalism to spin - qubit swaps and charge decoherence in two - qubit operations .
these requires larger hilbert spaces and can be realized by _ two _ electrons on a double dot .
loss and divincenzo have introduced a detailed scheme for ` quantum computation ' with spin - states of coupled single - electron quantum dots @xcite .
dephasing of spin degrees of freedom due to spin - orbit coupling or the coupling to nuclear spins is expected to be much weaker than dephasing of charge degrees of freedom .
nevertheless , spin and charge become coupled during switching operations whereby charge dephasing also effects spin - based qubits @xcite . here
, we consider a specific two - qubit swap operation as discussed recently by schliemann , loss and macdonald : two electrons with spin are localized on two coupled quantum dots @xmath197 and @xmath198 , giving rise to a basis of six states . during the operation _ charge _
decoherence occurs for intermediate states that get involved in the swap operation when charge is tunneling between the dots .
piezoelectric phonons then couple to the electron charge and incoherently mix states in the singlet sector which leads to a loss of fidelity of the swap operation . as in the previous section ,
we only consider the coupling of phonons to the charge density which yields a lower bound for dephasing .
the four basis vectors with the two electrons on different dots are the spin singlet and triplets @xmath199 the remaining two states with two electrons on dot @xmath197 ( ` left ' ) or dot @xmath198 ( ` right ' ) are @xmath200\nonumber\\ |r\rangle & : = & c^{\dagger}_{b\uparrow}c^{\dagger}_{b\downarrow}|0\rangle = 2^{-1/2 } \left [ |s_2\rangle - |s_3\rangle \right]\end{aligned}\ ] ] which are superpositions of two spin singlets @xmath201 , @xmath202 that differ in their orbital wave function . as in @xcite , we specify the two - qubit swap as an adiabatic rotation from an initial state @xmath203 to a final state @xmath204 , @xmath205\end{aligned}\ ] ] that can be achieved @xcite by an adiabatically opening and then closing of the tunnel barrier between the two dots as a function of time .
this operation leads out of the subspace span@xmath206 since it involves intermediate doubly occupied states in span@xmath207 (= span@xmath208 ) which altogether define a four - dimensional hilbert space , @xmath209 , in which the two - qubit swap takes place . to be more specific @xcite
, we use a basis of @xmath209 as given by the three singlets @xmath210 and the triplet @xmath211 , @xmath212 to define the time - dependent hamiltonian @xmath213,\end{aligned}\ ] ] where @xmath214 denotes the energies of the spin singlet states , @xmath215 , @xmath216 , @xmath217 with the spin triplet energy @xmath218 , the on - site coulomb repulsion @xmath219 , the exchange term @xmath220 , and the time - dependent tunnel coupling element between the dots @xmath11 .
the total hamiltonian in presence of bosons coupling to the charge degree of freedom is derived in appendix b , @xmath221 it has exactly the same form as in the one - qubit case , eq.([modelhamiltonian ] ) , but with the free hamiltonian @xmath26 replaced by @xmath222 , the coupling constants @xmath51 replaced by @xmath223 , and @xmath224 now referring to the two - particle states eq .
( [ lrdefine ] ) . in analogy with the one - qubit case
, we introduce the matrix elements of the reduced density operator in the interaction picture with respect to @xmath225 by @xmath226 the coherent part of the time evolution of the @xmath227 is obtained trivially from the liouville equation , using the hamiltonian @xmath222 , eq . ( [ h02define ] ) .
the non - trivial part are the additional terms due to the electron - phonon coupling .
a systematic perturbation theory in the latter starts from the four adiabatic eigenstates of the unperturbed hamiltonian @xmath222 , eq .
( [ h02define ] ) . here , we restrict ourselves to the case of small tunnel coupling , @xmath228 where inelastic transitions are determined by the dynamics in the subspace spanned by the states @xmath229 and @xmath230 and admixtures from @xmath231 through the hybridization between @xmath232 and @xmath229 can be neglected .
this case is particularily interesting since the ( adiabatic ) energy difference @xmath233 between @xmath229 and @xmath230 remains constant throughout the operation . as we show below , if @xmath233 coincides with a zero @xmath234 of the boson spectral density , the operation is again dissipation - less in second order of the electron - boson coupling . within the born and markov approximation one finds @xmath235 here ,
the rates @xmath236 and @xmath237 are defined in analogy with eq .
( [ gammarates ] ) , @xmath238 with the energy difference @xmath1 given by @xmath239 notice that due to the limit eq .
( [ tclimit ] ) , @xmath1 does not depend on time .
expectation value @xmath240 for a two - electron double dot qubit swap from initial state @xmath203 to final state @xmath204 , eq .
( [ swapdefinition ] ) , as a function of time ( in units of @xmath241 , @xmath242 : onsite coulomb repulsion ) .
the corresponding dephasing times ( @xmath243 ) are ( from top to bottom ) @xmath244 , @xmath245 and @xmath246 .
inset : @xmath247 and @xmath248 for vanishing phonon coupling @xmath249.,scaledwidth=50.0% ] in fig .
( [ swap3.eps ] ) , we show results of a numerical evaluation of the two - electron master equation .
the spectral density @xmath250 , eq .
( [ rhoomegadefinition ] ) , was used to obtain the rates , eq .
( [ ratesdefinition2 ] ) , for different coupling strengths @xmath59 and energies @xmath251 ( @xmath60 , with @xmath61 the speed of sound and @xmath56 the dot center distance ) .
using @xmath252 as energy scale , our results coincide with those of schliemann et al . for @xmath249 with a pulse of the form @xcite @xmath253 as shown in the inset . here , we chose @xmath254 , @xmath255 , @xmath256 , @xmath257 , and the temperature @xmath258 .
a value of @xmath259mev corresponds to a time of @xmath260 s. the diagonal element @xmath261 is a measure of the fidelity of the swap operation , transfering @xmath203 into @xmath204 , cf .
( [ swapdefinition ] ) . even in absence of dissipation ,
the non - adiabacity of the operation results in a finite value of @xmath262 after the swap @xcite .
the electron - phonon interaction acts when charge between the dots is moved during the opening of the tunneling barrier .
consequently , the two states @xmath229 and @xmath230 become mixed incoherently , leading to a finite , irreversible occupation probablity of the energetic lower state @xmath230 even after the pulse operation .
we mention that the data shown here correspond to a worst case scenario where the inelastic rate @xmath236 , eq .
( [ ratesdefinition2 ] ) , is essentially determined by a large energy difference @xmath233 between @xmath229 and @xmath230 which enters into @xmath263 .
then , spontaneous emission of phonons occuring during the slow swap ( cf .
( [ tclimit ] ) ) leads to a dephasing rate @xmath264 . in this case ,
even relatively small values of @xmath59 ( @xmath265 in experiments on lateral dots @xcite ) can lead to a considerable fidelity loss of the operation . on the other hand
, our results also demonstrate that a complete suppression ( at least to lowest order in @xmath59 ) , similar to the one - qubit case discussed above , is theoretically possible for the two - qubit swap in phonon cavities .
if @xmath1 is tuned to the energy @xmath188 of a decoupled phonon mode , the inelastic rates , eq . ( [ ratesdefinition2 ] ) , can be ` switched off ' and the operation again is ` dissipationless ' .
in conclusion , we have studied adiabatic quantum state rotations in coupled quantum dots in presence of dissipation . for the one - qubit case
, we suggested a scheme to extract quantum oscillations in the inversion change @xmath160 from the average pump current through the dots , eq .
( [ pumpcurrent ] ) . the analytic result , eq.([sigmazanalytic ] ) , furthermore allows to extract the value of the phonon spectral density , @xmath96 , for ` rabi'-pulses at fixed energy difference @xmath1 .
for the two - qubit case , we have quantified the effect of dissipation on a swap operation , where states in the singlet sector are incoherently mixed .
although we have only considered the case of small tunnel coupling between the dots , as in the one - qubit case we have found that two - qubit rotations can be performed such that dissipation is due to @xmath2 , i.e. , one fixed boson frequency only .
if @xmath1 is chosen to coincide with a zero of @xmath0 as in free - standing phonon cavities , this defines a ` decoherence - free manifold ' in the parameter space of the system .
this work was supported by the eu via tmr and rtn projects fmrx - ct98 - 0180 and hprn - ct2000 - 0144 , the german dfg project br 1528/4 - 1 , the uk project epsrc gr44690/01 and the uk quantum circuits network .
discussions with r. h. blick , s. de franceschi , e. m. hhberger , l. p. kouwenhoven , and f. renzoni are gratefully acknowledged .
phonons coupling to the electron _ charge _ in a double quantum dot are described by a spin - conserving electron - phonon hamiltonian @xmath266,\end{aligned}\ ] ] where @xmath267 is the creation operator for a phonon mode @xmath49 , @xmath268 the fourier transform of the operator @xmath269 of the electron density , and @xmath270 is the microscopic electron - phonon interaction matrix element .
the electron - phonon interaction potential in real space in first quantization is given by @xmath271 .
we consider the simplest case of an electron - phonon interaction with only diagonal terms @xmath272\left[a_{-\bf q } + a^{\dagger}_{\bf q}\right],\end{aligned}\ ] ] where @xmath273 ( @xmath274 ) , @xmath275 , and @xmath276 here , @xmath277 , @xmath274 , is the density of electrons with spin @xmath278 in the left ( right ) dot as obtained from the ( many - body ) ground states @xmath8 ( @xmath9 ) . assuming two identical dots with @xmath279 to be smooth functions centered around the left ( right ) dot centers @xmath55
, we obtain @xmath280 where @xmath281 is the ` form factor ' of the electron density in the left and in the right dot . in the electron - phonon interaction , eq . ( [ hepab ] ) , we write @xmath282\end{aligned}\ ] ] and recognize that the term proportional to the unity operator @xmath283 in the two - dimensional hilbert space @xmath284 leads to a mere renormalization of the total energy by the phonons in eq .
( [ hepab ] ) : the energies @xmath285 in both dots are shifted by the same amount . in fact , if the coupling constants @xmath286 and @xmath287 were the same , eq.([hepab ] ) would describe a trivial coupling of the phonons to @xmath288 which would not affect the dynamics of the system , while here ( due to eq .
( [ pq ] ) ) we find the phase relation @xmath289 where @xmath290 is the vector pointing from the center of the left to the center of the right dot .
defining @xmath291 , the term @xmath292 in eq .
( [ aplusb ] ) yields the interaction term @xmath105 in the hamiltonian @xmath293 , eq .
( [ modelhamiltonian ] ) .
the ohmic form eq.([rhosdefinition ] ) with @xmath66 for the spectral function @xmath294 is used in the discussion of the one - qubit dynamics in section [ section_one_qubit ] . in this case , the corresponding rates @xmath91 and @xmath92 , eq .
( [ gammarates ] ) , are given by @xmath295 note that the ohmic case @xmath66 is peculiar in the sense that a contribution @xmath296 linear in the temperature @xmath88 appears in @xmath91 and @xmath297 due to the limit @xmath298 under the integral in eq .
( [ gammarates ] ) , which gives a finite contribution for @xmath66 .
however , the microscopic calculation @xcite of @xmath250 , eq.([rhoomegadefinition ] ) , shows that the ohmic spectral density as an ` envelope ' of @xmath250 fails to be correct for small @xmath299 where @xmath97 for piezoelectric phonons in double quantum dots . the contribution to @xmath91 and
@xmath297 linear in @xmath88 therefore is unphysical but of minor relevance in the numerical calculations anyway , as we have checked .
as in the one - qubit case , we start from an electron - phonon interaction hamiltonian with only diagonal terms @xmath300,\end{aligned}\ ] ] where the matrix element @xmath301 is calculated with the single particle states @xmath302 .
we assume that the single electron densities in eq.([alphadef2 ] ) are spin - independent , @xmath303 whence @xmath304\left[a_{-\bf q } + a^{\dagger}_{\bf q}\right]\end{aligned}\ ] ] with the number operators @xmath305 .
the non - vanishing matrix elements of the number operators are @xmath306,@xmath307 where @xmath308 , @xmath309 , and @xmath310 . here
, we used the states @xmath311 $ ] with two electrons on the left ( right ) dot , eq .
( [ lrdefine ] ) . using the completeness relation in @xmath209 yields @xmath312 where again we defined @xmath224 .
the total hamiltonian now can be written as @xmath313\nonumber\\ & + & x\sigma_x + \frac{1}{2}\hat{a}\sigma_z + h_b,\end{aligned}\ ] ] where @xmath314 , @xmath315 and @xmath316 . comparing the coupling constant @xmath317 in the one - qubit case and @xmath318 in the two - qubit case , and keeping in mind the definitions for @xmath319 and @xmath320 ,
( [ alphadef1 ] ) and eq .
( [ alphadef2 ] ) , we realize that for spin - independent electron densities fulfilling @xmath321 we obtain identical coupling constants @xmath322 : the spin sum in eq .
( [ alphadef1 ] ) yields a factor @xmath323 .
an exception is the one - qubit with only one electron having a fixed spin ( so that there is no spin sum ) . in that case and with eq .
( [ denscondition ] ) holding , the one - qubit coupling @xmath51 is half the two - qubit coupling @xmath223 .
|
we investigate the influence of dissipation on one and two qubit rotations in coupled semiconductor quantum dots , using a ( pseudo ) spin boson model with adiabatically varying parameters . for weak dissipation ,
we solve a master equation , compare with direct perturbation theory , and derive an expression for the ` fidelity loss ' during a simple operation that adiabatically moves an electron between two coupled dots .
we discuss the possibility of visualizing coherent quantum oscillations in electron ` pump ' currents , combining quantum adiabaticity and coulomb blockade . in two qubit spin swap operations where the role of intermediate charge states has been discussed recently , we apply our formalism to calculate the fidelity loss due to charge tunneling between two dots .
|
globular clusters contain many x - ray sources at lower luminosities , @xmath3ergs@xmath4 .
these sources were first discovered with the einstein satellite ( hertz & grindlay 1983 ) , and many more were found with rosat ( for a compilation , see johnston & verbunt 1996 ) .
the nature of these low - luminosity sources is the subject of debate , because various types of objects can emit x - rays at such luminosities , such as soft x - ray transients in quiescence , cataclysmic variables , rs cvn binaries , and recycled neutron stars ( see e.g. fig .
8 in verbunt et al .
the most compelling identification of a dim x - ray source with an object observed at other wavelengths is the recycled radio pulsar in m28 : the x - ray flux varies on the pulse period ( danner et al .
plausible identifications with cataclysmic variables have been suggested for dim x - ray sources in ngc6397 , ngc6752 , ngc5904 and 47tuc ( cool et al . 1995b , grindlay 1993 , hakala et al .
1997 , verbunt & hasinger 1998 ) .
these identifications are based on the proximity of the x - ray position to that of a cataclysmic variable , and thus their probability depends on the accuracy of the x - ray position .
the accuracy of the rosat position of a detected x - ray source is determined by two factors : the statistical accuracy of the position of the source on the detector , and the accuracy with which the position of the detector as a whole is projected on the sky . for a sufficient number of photons
the statistical error is less than an arcsecond , but the projection in general has a typical error of @xmath5 .
secure identification of a source in the detector field reduces the error in the projection to the statistical error of the identified source , provided that the optical ( or radio ) position has better accuracy .
only one identification is necessary , because the roll angle of the detector ( i.e. the north - south direction ) is accurately known ; nonetheless , identification of more than one source is preferable to allow checks on internal consistency . in a globular cluster
the surface density of possible counterparts is so high that chance coincidence usually can not be excluded ; a secure identification can usually be made only for x - ray sources detected well outside the cluster .
this method has been used by verbunt & hasinger ( 1998 ) to improve the positional accuracy of the sources in the core of 47tuc from @xmath6 to @xmath7 , whereby the area in which the source is expected to lie is reduced sufficiently to exclude several proposed identifications , and increase the probability of others , including two possible cataclysmic variables . in this paper
we investigate three clusters known to contain multiple dim x - ray sources in their core , which have been observed in long exposures with the rosat hri , and one cluster known to harbour a transient .
we analyse hitherto unpublished observations and detect both previously published and new x - ray sources .
all source positions are checked in the simbad data base versus positions of other objects , and we find objects in the hipparcos or tycho catalogues ( esa 1997 , perryman et al .
1997 , hg et al .
1997 ) with each cluster , i.e. counterparts with very accurate positions . in sect.2
we describe the observations and our data reduction procedures ; sections 3 to 5 describe the results for , , and , respectively . in sect.6
we discuss an observation of .
a discussion of our results is given in sect.7 .
l@ll@rl@l & & @xmath8(s ) & @xmath9 & @xmath10 + + 1992 & aug & 2448835.935837.027 & 3469 & & + 1993 & jan & 2449008.688008.702 & 1243 + 1994 & jul & 2449547.617550.695 & 5997 + 1995 & jan & 2449735.780744.786 & 17653 + 1995 & jul & 2449912.812941.103 & 75481 & @xmath11 & @xmath12 + 1996 & feb & 2450120.247130.418 & 12990 + 1997 & jan & 2450461.343461.358 & 1317 + + 1991 & mar & 2448338.820340.054 & 7499 + 1992 & mar & 2448696.631697.646 & 12958 + 1995 & mar & 2449793.872810.440 & 77204 & @xmath13 & @xmath14 + + 1992 & mar & 2448697.719703.626 & 31300 & @xmath15 & @xmath16 + 1995 & mar & 2449800.941815.431 & 17411 & + 1996 & apr & 2450182.957186.041 & 23278 & + + 1996 & sep & 2450332.541337.810 & 16412 + the x - ray observations were obtained with the rosat x - ray telescope ( trmper et al .
1991 ) in combination with the high - resolution imager ( hri , david et al .
the list of the observations is given in table [ ta ] .
the standard data reduction was done with the extended scientific analysis system ( zimmermann et al .
1996 ) , as follows . to take into account the re - calibration of the pixel size ( hasinger et al . 1998 ) , we multiply the @xmath17 pixel coordinates of each photon with respect to the hri center with 0.9972 . a search for sources
is made by comparing counts in a box with the counts in a ring surrounding it , and by moving this detection box across the image .
the sources thus detected are excised from the image and a background map is made for the remaining photons . a search for sources is then made by comparing the number of photons in a moving box with respect to the number expected on the basis of the background map .
finally , at each position in which a source was found , a maximum - likelihood technique is used to compare the observed photon distribution with the point spread function of the hri ( cruddace et al .
this produces a maximum - likelihood value ml such that the probability that the source is due to chance at one trial position is given by @xmath18 .
we retain sources for further discussion if ml@xmath19 .
( to make sure that all such sources are found , we enter in the maximum likelihood technique all sources that have ml@xmath2010 according to the sliding box searches . ) upper limits at the position of known sources were determined by counting the number @xmath21 of actually detected photons at the position of the source ( and in an area surrounding it corresponding to the uncertainty in the position ) ; we then assign as upper limit the lowest expected number @xmath22 for which the probability of measuring a number @xmath21 or smaller is less than 5% according to the poisson distribution .
the maximum - likelihood technique also provides an indication whether the source is extended .
if such indication is present , we apply further analysis to test whether the source is a multiple point source .
the further analysis is also based on maximum likelihood techniques , but the analysis is limited to a small area of the detector , near its center .
this allows the simplifications that the background in the analysed area is a constant ( as opposed to a polynomial function of the @xmath17 pixel coordinates ) , and that the point spread function is that for the center of the image ( david et al .
suppose that a model to be tested predicts @xmath23 photons at detector pixel @xmath24 .
the probability that @xmath25 photons are observed is then given by the poisson probability : @xmath26 the probability that the model describes the observations is given by the product of the probabilities for all @xmath24 in the region considered : @xmath27 . for computational ease
we maximize the logarithm of this quantity : @xmath28 the last term in this equation does nt depend on the assumed model , and in terms of selecting the best model may be considered as a constant .
thus maximizing @xmath29 is equivalent to minimizing @xmath30 , where @xmath31 if one compares two models a and b , with number of fitted parameters @xmath32 and @xmath33 and with likelihoods of @xmath34 and @xmath35 , respectively , the difference @xmath36 is a @xmath37 distribution with @xmath38 degrees of freedom , for a sufficient number of photons ( cash 1979 , mattox et al . 1996 ) .
our analysis of possibly multiple sources thus proceeds as follows .
first we compute @xmath39 for a model with constant background and for the best model with background plus one source , and compare it with the @xmath37-distribution with 3 degrees of freedom . if @xmath40 , the presence of one source has a significance more than three sigma .
next we compare the best model with two sources with the best model with one source , to prove the significance of a second source ; the best models with three and two sources to prove the significance of a third source , etc . until no more significant sources are found .
the addition of one source adds three fitted parameters , one for its number of counts and two for its position . in the case of ngc6397 optical counterparts
have been suggested for three x - ray sources .
for these we also make a fit in which the distances in right ascension and declination between these three sources is fixed to the optically determined values .
the three sources in that case only add five fitted parameters , two for the position of one of them , and three for the fluxes . to determine the error in a parameter , we start from the best fit value @xmath41 .
we then fix the parameter at @xmath42 and make a new fit , allowing all other parameters to vary . the value of @xmath43 for which @xmath44 increases by 1 is quoted as the 1-sigma error .
@xmath1cen is a very massive globular cluster , with a relatively low central density .
hertz & grindlay ( 1983 ) reported five sources .
sources a , d and e all are well outside the cluster core , and it appears that only sourcec is clearly related to the globular cluster ( e.g. verbunt et al .
sourcee has tentatively been identified with a foreground k star ( margon & bolte 1987 ) .
the sourcesa and d have been identified with foreground m stars , on the basis of better positions for the x - ray sources obtained with einstein and rosat hri observations ( cool et al .
a rosat pspc pointing indicates that sourcec , near the cluster center , is composed of two sources of comparable luminosity ( johnston et al .
1994 ) .
rr@ r@ rr@ r@ rccr@ l + x & & & @xmath45 & @xmath46 & cts / ksec + 4 & 13 & 27 & 27.72 & @xmath4747 & 19 & 8.0 & 0.4 & 4.6 & [email protected] & d + 3 & 13 & 25 & 52.05 & @xmath4747 & 19 & 8.6 & 0.5 & 5.1 & [email protected]&a + 18 & 13 & 26 & 46.33 & @xmath4747 & 19 & 45.9 & 1.0 & 3.4 & [email protected] + 19 & 13 & 27 & 21.21 & @xmath4747 & 23 & 22.0 & 1.8 & 3.1 & [email protected] + 6 & 13 & 27 & 29.27 & @xmath4747 & 25 & 54.5 & 1.3 & 3.0 & [email protected] + 20 & 13 & 26 & 48.70 & @xmath4747 & 27 & 46.5 & 1.0 & 0.4 & [email protected] + 9a & 13 & 26 & 53.39 & @xmath4747 & 29 & 1.5 & 0.9 & 0.5 & [email protected]&c + 7 & 13 & 26 & 19.75 & @xmath4747 & 29 & 11.9 & 0.8 & 1.7 & [email protected]&b + 9b & 13 & 26 & 52.04 & @xmath4747 & 29 & 37.2 & 0.9 & 0.6 & [email protected]&c + 21 & 13 & 26 & 48.99 & @xmath4747 & 31 & 28.1 & 2.1 & 1.1 & [email protected] + 10 & 13 & 26 & 25.20 & @xmath4747 & 32 & 28.9 & 1.4 & 2.0 & [email protected] + 13 & 13 & 26 & 11.12 & @xmath4747 & 37 & 11.1 & 3.7 & 4.0 & [email protected] + + gc & 13 & 26 & 45.9 & @xmath4747 & 28 & 37 & + 4 & 13 & 27 & 27.37 & @xmath4747 & 19 & 6.2 & + 4 & 13 & 27 & 27.68 & @xmath4747 & 19 & 6.3 & + 3 & 13 & 25 & 51.79 & @xmath4747 & 19 & 7.0 & + 3 & 13 & 25 & 52.23 & @xmath4747 & 19 & 7.5 & + 18 & 13 & 26 & 46.33 & @xmath4747 & 19 & 45.9 & + e & 13 & 29 & 18.54 & @xmath4747 & 22 & 50.5 & +
we have analysed all observations listed in table [ ta ] .
the 1992 and 1993 data have been reported on before by cool et al .
( 1995a ) .
as expected , we detect the largest number of sources in the longest observation , that of july 1995 .
we use this observation as the basis for comparison with the other observations . in the july 1995 observation
we detect twelve sources , listed in table[tabomc ] .
eight of these sources have been detected before with einstein or with the rosat pspc observation reported by johnston et al .
( 1994 ) ; four sources are new . the sources x@xmath493 and x@xmath494 are the einstein sources a and d , respectively , identified with foreground m dwarfs by cool et al .
( 1995a , see table[tabomc ] ) .
both stars can be found in the usno - a2 catalogue ( monet et al .
1998 , see table[tabomc ] ) . the new source x@xmath4918 can be identified with ; this is star tyc825246271 in the tycho reference catalogue , and thus its position and proper motion are well known ( hg et al .
1998 ) . for this reason we use this star to determine the bore sight correction , i.e. the offset between the x - ray coordinates and the optical coordinates .
the result is listed in table[ta ] .
this bore sight correction has been applied to the x - ray positions , and the resulting positions are given in table[tabomc ] . with a statistical accuracy for the x - ray position of x@xmath4918 of @xmath50@xmath51 , and taking into account small additional systematic errors ( see hasinger et al .
1998 ) , we estimate the systematic error in the positions given in table[tabomc ] to be less than @xmath7 ; for each individual source its statistical uncertainty should be added in quadrature to this systematic error . the positional accuracy can be improved if accurate astrometry of x@xmath493/a and x@xmath494/d is obtained , which will allow computation of their positions at epoch 1995.5 . in the rosat deep survey ( hasinger et al .
1998 ) , an area with 12@xmath525 radius contains 25 sources brighter than our approximate detection limit of @xmath53 ; we thus expect @xmath541 serendipitous source within the core radius , i.e. the faintest source within the core radius , x@xmath4920 , may well be a serendipitous background source .
the brighter sources within the core radius probably are associated with @xmath1cen . outside the core radius
a source is more likely to be a fore- or background source than a cluster member .
the x - ray image of the inner area of @xmath1cen is shown in fig.[figomc ] .
the half - mass radius of the cluster contains five sources .
source x@xmath499 from johnston et al .
( 1994 ) is clearly separated into two sources , which we denote x@xmath499a and x@xmath499b for the northern and southern source , respectively .
we have determined the countrates or upper limits for the six central sources in all rosat hri observations of @xmath1cen listed in table[ta ] . for the very short observations , no useful upper limits are obtained ; the long - term lightcurves as determined from the other observations are shown for four of the central sources in fig.[figlt ] .
we also show the pspc observation , dividing the pspc counts for pspc - x@xmath499 equally between x@xmath499a and x@xmath499b . for the absorption towards @xmath1cen and a black body spectrum of 0.6kev
the pspc countrate is about 2.8 times the hri countrate ; for smaller reddening the pspc - to - hri count ratio varies rapidly , and therefore we do not show pspc points of the foreground m dwarfs , whose absorption is unknown .
there is marginal evidence for variability in sources x@xmath499b and x@xmath4910 ; and no evidence for variability of x@xmath497 and x@xmath499a .
x@xmath4920 is detected only in july 1995 and in 1996 , x@xmath4921 only in july 1995 , and all upper limits in the other observations are compatible with the faint fluxes of these sources .
the july 1995 observation was obtained in two parts , separated by about 10 days .
we have analysed the two parts separately , and find marginal variation of x@xmath497 ( @xmath55 and @xmath56 ctsksec@xmath4 in the first and second half , respectively ) and of x@xmath499b ( @xmath57 and @xmath58 ctsksec@xmath4 , respectively ) .
in addition , virtually all counts of x@xmath4921 are from the first part of the observation .
the number of counts of x@xmath4921 is too small for further subdivision . to convert the observed countrates into x - ray luminosities
, we assume a column and distance to @xmath1cen as given in table[tabomc ] . for an 0.6kev blackbody ( see the analysis of the pspc spectrum of x@xmath499 in johnston et al .
1994 ) 1 ctsksec@xmath4 in the rosat hri corresponds to @xmath59ergs@xmath4 in the 0.52.5kev band .
the two sources x@xmath499a and x@xmath499b thus have about this luminosity , source x@xmath497 is 40% brighter , and sources x@xmath4920 and x@xmath4921 are 40% fainter .
the foreground dwarfs ( einstein x - ray sources a and d ) are highly variable , as has been pointed out before ( koch - miramond & aurire 1987 , cool et al .
these sources also vary between the first and second half of the july 1995 observation .
the extremely high flux of a in 1996 is due to a flare , which lasts almost a day ( see fig.[flare ] ) .
hd116789 is an a0v star . from its magnitude @xmath60 and
colour @xmath61 we estimate @xmath62 and a distance of about 310pc . for an assumed bremsstrahlung spectrum of 1.4kev
the observed countrate corresponds to a luminosity in the 0.52.5kev band of @xmath63ergs@xmath4 .
it is not expected for an a0v star to emit such a flux ; perhaps this star has a white dwarf companion which is responsible for the x - ray emission , as various other a0v stars detected with rosat ; on the other hand , various apparently single a0v stars in the bright star catalogue have been detected at similar and higher luminosities as hd116789 ( e.g. hd17864 , hnsch et al .
hd116789 was detected with exosat by verbunt et al .
( 1986 ) , who interpreted the detection as due to the ultraviolet leak of the cma detector .
the exosat cma countrate of this source , @xmath64ctss@xmath4 , converts to an x - ray luminosity at the distance of hd116789 of about @xmath65ergs@xmath4 , much higher than the luminosity derived for this star from the rosat observations ; we therefore still think that the exosat countrate is due to the ultraviolet flux .
the ultraviolet leak in the rosat hri is far too small ( berghfer et al . 1999 ) to explain the rosat detection .
the positions of the sources within the half - mass radius of @xmath1cen as given in table[tabomc ] are more accurate than previously published positions , and may be used to search for optical counterparts .
we have done this among the variables ( contact binaries , detached binaries , and suspected rs cvn stars ) found in @xmath1cen by kaluzny et al .
( 1996 , 1997 ) : no counterpart is among these stars .
( only one of these variables is in the area shown in fig.[figomc ] , viz . the contact binary oglegc13 . )
our non - detection of these binaries is not surprising , considering that our detection limit is above @xmath66ergs@xmath4 : all of the contact binaries hitherto detected in x - rays ( mcgale et al .
1996 ) , and many rs cvn systems ( dempsey et al .
1993 ) are less luminous than this .
cool et al .
( 1995a ) argue that x@xmath497/b is an extended source .
this source is detected in the rosat pspc observation ( johnston et al .
1994 ) and in the rosat hri observations of 1994 , 1995 january and july , and 1996 .
all of these observations are more sensitive than the 1992 and 1993 observations used by cool et al .
( 1995a ) ; in all of them x@xmath497/b is compatible with being a point source . with the identification of sources x@xmath493 and x@xmath494 with foreground stars , we can reinvestigate the suggested identification of x@xmath495/e with a foreground k star , as suggested by margon & bolte ( 1987 ) .
we use the optical positions of a and d to determine the offset between the x - ray positions of the pspc observation as listed in johnston et al .
( 1994 ) and the optical positions .
we then apply this offset to the position of x@xmath495 , and find that the resulting position is at 6@xmath485 arcseconds from the optical star .
we take the position of the optical star ( table[tabomc ] ) from usnoa20375 - 18334783 ( monet et al .
identification of x@xmath495 with the foreground star is therefore a distinct possibility .
ngc6397 is a nearby cluster , with a collapsed core , in or close to which cool et al .
( 1993 ) detected with a rosat hri observation .
photometry with the hubble space telescope enabled cool et al .
( 1995b ) to find eight candidate counterparts for these sources , on the basis of high ultraviolet flux or of h@xmath67 emission .
the h@xmath67 emission of three stars has been confirmed spectroscopically by grindlay et al .
( 1995 ) who argue that these stars are cataclysmic variables , and responsible for the x - ray emission close to the core .
we analyse first the longest observation , obtained in 1995 , and use this as a reference for our discussion of the earlier , shorter observations .
the standard analysis provides 14 sources , listed in table[taba ] .
identifications with earlier x - ray sources or optical objects are indicated ; 7 sources are new .
x@xmath496 has been identified by cool et al .
( 1993 ) as sao244944 .
this star is identical to , and is in the hipparcos catalogue as hip86569 .
its position and proper motion are thus very accurately known , and we use it to determine the bore sight correction .
this bore sight correction is given in table[ta ] , and is applied to the x - ray positions ; the resulting positions are given in table[taba ] .
the statistical uncertainty in the x - ray position of x@xmath496 is about 0.5@xmath51 ; we therefore estimate that systematic error of the x - ray positions listed in table[taba ] is better than 1@xmath51 ; this error should be added in quadrature to the statistical error for each individual source position .
the quasar identified by cool et al .
( 1993 ) with x@xmath495 coincides within the error with our position for x@xmath495 . however , the active galaxy identified by cool et al .
( 1993 ) with x@xmath492 is 10@xmath51 from our x - ray position , mainly in right ascension ; and we conclude that it is not the x - ray source .
the explanation probably lies in the new scale for the size of the hri pixel that we use ( see sect.2 ) , which modifies positions of sources at large distance from the center of the hri image .
rr@ r@ rr@ r@ rrrr@ l x & & & @xmath45 & @xmath46 & cts / ksec + + 2 & 17 & 41 & 34.79 & @xmath4753 & 32 & 2.0 & 0.8 & 230 & [email protected] & g + 11 & 17 & 40 & 51.16 & @xmath4753 & 33 & 49.0 & 2.0 & 135 & [email protected] + 12 & 17 & 40 & 48.85 & @xmath4753 & 39 & 46.0 & 1.2 & 25 & [email protected] + 13@xmath68 & 17 & 40 & 41.56 & @xmath4753 & 40 & 6.0 & 0.5 & 6.3 & [email protected] & b + 4 & 17 & 40 & 42.24 & @xmath4753 & 40 & 23.8 & 0.4 & 2.2 & [email protected] & c + 14 & 17 & 40 & 1.43 & @xmath4753 & 42 & 25.6 & 1.6 & 125 & [email protected] + 15 & 17 & 39 & 16.54 & @xmath4753 & 43 & 11.3 & 3.7 & 258 & [email protected] + 5 & 17 & 40 & 33.18 & @xmath4753 & 43 & 46.6 & 0.4 & 72 & [email protected] & d + 16 & 17 & 41 & 23.73 & @xmath4753 & 46 & 17.2 & 1.0 & 172 & [email protected] & e + 17 & 17 & 39 & 32.92 & @xmath4753 & 46 & 48.7 & 3.5 & 240 & [email protected] + 18 & 17 & 40 & 30.81 & @xmath4753 & 47 & 51.4 & 0.8 & 152 & [email protected] + 6 & 17 & 41 & 27.66 & @xmath4753 & 48 & 10.6 & 0.5 & 207 & [email protected] & f + 19 & 17 & 40 & 14.53 & @xmath4753 & 50 & 31.7 & 2.8 & 217 & [email protected] + 8 & 17 & 40 & 10.46 & @xmath4753 & 50 & 54.7 & 1.7 & 229 & [email protected] + + + 13 & 17 & 40 & 41.44 & @xmath4753 & 40 & 3.3 & 0.3 & & [email protected] & b + 4a & 17 & 40 & 42.47 & @xmath4753 & 40 & 28.6 & 0.5 & & [email protected] + 4b & 17 & 40 & 42.55 & @xmath4753 & 40 & 19.1 & 0.5 & & [email protected] & id3 + 4c & 17 & 40 & 41.68 & @xmath4753 & 40 & 19.0 & 0.9 & & [email protected] & id1 + 4d & 17 & 40 & 41.64 & @xmath4753 & 40 & 27.7 & 0.7 & & [email protected] + + 13 & 17 & 40 & 41.44 & @xmath4753 & 40 & 3.3 & 0.3 & & [email protected] & b + 4a & 17 & 40 & 42.24 & @xmath4753 & 40 & 28.6 & 0.4 & & [email protected] & id2 + 4b & 17 & 40 & 42.57 & @xmath4753 & 40 & 19.3 & & & [email protected] & id3 + 4c & 17 & 40 & 41.52 & @xmath4753 & 40 & 19.4 & & & [email protected] & id1 + 4d & 17 & 40 & 41.56 & @xmath4753 & 40 & 27.7 & 0.8 & & [email protected] & + 4e & 17 & 40 & 42.65 & @xmath4753 & 40 & 27.6 & & & [email protected] & id6 + + + gc & 17 & 40 & 41.3 & @xmath4753 & 40 & 25 & + 5 & 17 & 40 & 33.4 & @xmath4753 & 43 & 45 . &
+ 2@xmath69 & 17 & 41 & 35.9 & @xmath4753 & 32 & 4 . & + 6 & 17 & 41 & 28.0 & @xmath4753 & 48 & 13 . & + 6 & 17 & 41 & 27.66 & @xmath4753 & 48 & 10.6 & + + the flux detection limit is about @xmath70 outside the blended central region , similar to that obtained for @xmath1cen .
analogous to our argument for @xmath1cen , we find that all objects detected within @xmath71 are probably cluster members , whereas we expect 1.4 background sources within @xmath72 from the center of of ngc6397 ; the sources at @xmath73 therefore may be background sources .
we thus can not decide whether x@xmath4912 is a cluster member .
outside the half - mass radius , the sources are more likely to be background or foreground sources .
x@xmath495 , just outside the half - mass radius , is a quasar ( cool et al .
1993 ) . in fig.[figngca ]
we show the x - ray contours of the center of ngc6397 together with the ultraviolet and/or h@xmath67-emission stars discovered by cool et al .
( 1995b ) .
the first models we investigated as fits to the central area of @xmath74 , containing sources x@xmath4913/b and x@xmath494/c , are those with successively one , two , three , four and five sources ; all with free positions . using the @xmath39 criterion for significance ( see sect .
2 ) we find that five sources are required .
we refer to the fit with five sources as modeli .
the parameters of the five sources of this model are given in table[taba ] .
we do not detect source a of cool et al .
( 1993 ) in the 1995 observation .
cool et al .
( 1995b ) resolved source x@xmath494/c into three components c1 - 3 . from a list of h@xmath67 emission and/or ultraviolet excess objects ( their table 2 ) , they suggest identifications of id1 with c2 , id2 for c3 and id3 for c1 . comparing the positions of the sources in modeli we find that the positions of x@xmath494b and x@xmath494c are compatible with those of id3 and id1 , respectively ; we thus identify x@xmath494b with c1 and x@xmath494c with c2 .
x@xmath494d is a new source .
( the offset required to match these positions from table [ taba ] with those given by cool et al .
( 1995b ) is slightly larger than our claimed accuracy of @xmath75 ; the remaining difference may be explained by an offset between the guide star catalogue coordinate system and the more accurate hipparcos coordinate system . )
the position of x@xmath494a is not compatible with that of id2 .
the reason for this may be seen in fig.[figngca ] : the two brightest components of source x@xmath494/c have a smaller difference in right ascension than id2 and id3 .
if id2 is the correct identification for c3 , we conclude that x@xmath494a is not identical to c3 . to further investigate this
, we note that if the identifications are correct , the distances between the x - ray sources must match the distances between the proposed optical counterparts , which are accurately known from the hst observations . in modelii
we fit five sources to the x - ray data of the center of ngc6397 , of which three are forced to be at fixed relative positions , corresponding to the distances between id1 , id2 and id3 .
modelii thus has four fitted parameters less than modeli .
the @xmath44 of modelii is 26 higher than that of modeli , i.e. it is a significantly ( 4-sigma ) worse fit .
this confirms that x@xmath494a is not id2 . in modeliii
we assume that id6 of table 2 in cool et al .
( 1995b ) rather than id2 is the counterpart of c3 , and fix the distances between the sources accordingly .
this fit has the same @xmath44 as modelii , and thus also is significantly worse than the fit of modeli .
again , the reason for the bad fit is the mismatch in the difference in right ascensions of the two brightest x - ray sources with that between the proposed counterparts : x@xmath494a is not id6 .
we note that the best position of x@xmath494a is between id2 and id6 , and in modeliv we fit six sources , of which four are forced to be at the relative distances of id1 - 3 and id6 .
modeliv thus has three fitted parameters less than modeli .
its @xmath44 is 6 higher than that of modeli , i.e. it is marginally worse at less than 2-sigma .
the parameters of the six sources of this model are also given in table[taba ] .
it is seen that the positions of x@xmath494b , x@xmath494c and x@xmath494d are the same ( within the error ) in modeliv as in modeli .
thus , we have two acceptable models . in both models
we confirm the possible identifications of id3 with c1 ( @xmath76x@xmath494b ) and of id1 with c2 ( @xmath76x@xmath494c ) , and we find one new source ( x@xmath494d ) . in modeli the remaining flux is ascribed to one source ( x@xmath494a ) which is not identical to id2 . in modeliv the remaining flux
is ascribed to two sources , one of which is id2/c3 and one is a second new source , x@xmath494e / id6 .
the two acceptable models are illustrated in fig.[figngcb ] .
.results of fitting four models to the three data sets of ngc6397 .
the table lists the number @xmath21 of fitted parameters , and the difference @xmath39 with respect to the best model for a given data set . in all fits ,
the fluxes of all sources are fitted parameters .
for the 1995 observation modeli has four sources with free positions , modeliv has six sources of which three have free positions and three have fixed positions relative to id3 , corresponding to the offsets of id1 , id2 , and id6 with respect to id3 .
modelsii and iii are as modeliv , after removing id6 and id2 , respectively . for each model ,
the same positions as in the best fit for 1995 are used for the 1991 and 1992 data .
[ tabfit ] [ cols= " < , > , > , > , > , > , > " , ] with these points in mind , we note from fig.[fxlum ] that in all clusters except possibly @xmath1cen the most luminous sources appear to be in the cluster core .
the main difference between @xmath1cen and the other clusters is that the collision frequency in @xmath1cen is so low that one expects no low - mass x - ray binaries in it , and that most cataclysmic variables in it will be evolved from primordial binaries ( verbunt & meylan 1988 , davies 1997 ) .
in addition , the mass segregation in this cluster is very low .
thus in @xmath1cen there is no marked difference between the core and the regions outside the core . in each cluster
we detect sources down to the detection limit ; this suggests that more sensitive observations will detect more sources . in the cores of ngc6397 and
ngc6752 the detection of more source will also require better imaging , so that the faint sources can be detected against the brighter ones .
we do not detect a difference between the luminosities of sources in the collapsed globular cluster ngc6397 and the much less concentrated globular cluster ngc6752 .
on the other hand , the highly concentrated cluster 47 tuc contains three sources which are an order of magnitude brighter than the brightest sources in ngc6397 and ngc6752 .
viable optical counterparts have been suggested for only five among the 26 sources shown in fig.[fxlum ] , all of them probable cataclysmic variables .
we compare the ratio of x - ray flux to optical flux of these sources with the ratios measured for cataclysmic variables and for rs cvn systems in the galactic disk in fig.[fcompa ] .
it is seen that the suggested optical counterparts for the sources in ngc6397 and ngc6752 lead to ratios which are compatible with those of cataclysmic variables , whereas those in 47 tuc are too bright in x - rays , in agreement with fig.[fxlum ] .
if these sources are indeed cataclysmic variables , their excessive x - ray luminosity needs to be explained ; alternatively , the suggested identifications may be chance coincidences ( as discussed by verbunt & hasinger 1998 ) .
all suggested counterparts lead to higher x - ray to optical flux ratios than those of rs cvn binaries .
the accurate positions that we determine for individual sources are valid for separately detected sources in particular . in the case of overlapping sources , we do not have unique solutions .
thus , in the core of ngc6397 fits with 5 and 6 sources are both acceptable , at similar quality ; and we can not exclude that more sources contribute to the observed flux , which would invalidate our derived positions .
binaries may reside away from the core either because the cluster has undergone little mass segregation , or because a three - body interaction ( i.e. a close encounter of a binary with a single star ) in the core has expelled the binary from the core ( e.g. hut et al .
1992 ) . in the latter case
the binary is expected to be eccentric immediately after being expelled ; tidal forces may in time circularize the orbit again .
such binaries are only a minority of the overall binary population of a cluster ; however , x - ray observations may preferably select such binaries if tidal forces act in them . since sources away from the core can be fore- or background sources , optical identification of them is required to settle whether they belong to the cluster or not .
our accurate positions should help in finding such counterparts .
we have made use of the rosat data archive of the max planck institut fr extraterrestrische physik at garching ; of the simbad database operated at centre de donnes astronomiques in strasbourg ; and of the digitized sky surveys . for hipparcos and tycho data
the cd - rom celestia , provided by esa , was very helpful .
we further thank lucien kuiper for discussions about the maximum likelihood algorithms , marten van kerkwijk for help with use of the digitized sky surveys , and dr .
asai for communication of the re - determination of the position of the asca source in liller1 .
zimmermann , h. , becker , w. , belloni , t. , dbereiner , s. , izzo , c. , kahabka , p. , schwentker , o. 1996 , ser s guide : extended scientific analysis system to evaluate data from the astronomical x - ray satellite rosat , edition 5 , technical report , mpe
|
we detect stars from the hipparcos and tycho catalogues in the field of view of observations with the rosat hri of three globular clusters .
we use the positions of these stars to reduce the systematic error in the positions of x - ray sources in the clusters to @xmath0 for @xmath1cen and ngc6752 , and @xmath2 for ngc6397 .
we detect three x - ray sources in the core of @xmath1cen , and four in the core of ngc6752 ; the data for the center of ngc6397 may be fitted with five or six sources . outside the cores , but within the half - mass radius of the clusters , we detect two sources in @xmath1cen , one in ngc6397 and two in ngc6752 ; these may or may not be cluster members .
a rosat hri observation of liller1 does not detect a low - luminosity source , at a limit below a detection with asca .
we discuss the nature of the low - luminosity x - ray sources in globular clusters in the light of these new results .
|
_ rectifiability _ is a fundamental concept in geometric measure theory .
rectifiable sets extend the class of surfaces considered in classical differential geometry ; while admitting a few edges and sharp corners , they are still smooth enough to support a rich theory of local analysis . however , for certain question of more global nature the boundedness of singular integrals being the main example
the notion of rectifiability is too qualitative . in a series of influential papers around the year 1990 , @xcite
, g. david and s. semmes developed an extensive theory of _ quantitative rectifiability _ in euclidean spaces .
one of their main objectives was to find geometric criteria to characterize the @xmath5-dimensional subsets of @xmath6 , @xmath7 , on which `` nice '' singular integral operators ( sio ) are @xmath8-bounded . here ,
`` nice '' refers to sios with smooth , odd caldern - zygmund kernels , the archetype of which is the riesz kernel @xmath9 , @xmath10 .
notice that for @xmath11 and @xmath12 , the riesz kernel essentially coincides with the cauchy kernel @xmath13 in the complex plane .
a motivation for the efforts of david and semmes was the significance of sios for the problem of finding a geometric characterization of _ removable sets _ for bounded analytic functions and lipschitz harmonic functions .
due to seminal papers by david @xcite , david and p. mattila @xcite , and f. nazarov , x. tolsa and a. volberg @xcite , @xcite , it is now known that these removable sets coincide with the purely @xmath14-unrectifiable sets in @xmath6 , i.e. the sets which intersect every @xmath15 hypersurface in a set of vanishing @xmath14-dimensional hausdorff measure
. the geometric characterization of removability , and its connections to geometric measure theory and harmonic analysis , has a very interesting history ; we refer to the excellent survey by volberg and v. eiderman @xcite , and to the recent book of tolsa @xcite .
the problem of characterizing removable sets for lipschitz harmonic functions has a natural analogue outside the euclidean setting in certain non - commutative lie groups , of which the heisenberg group is the simplest example . in this group ,
the role of the euclidean laplace operator is played by the _ sub - laplacian _ and harmonic functions are , by definition , the solutions to the sub - laplacian equation .
the question of characterizing removability for lipschitz harmonic functions was considered in @xcite for heisenberg groups @xmath16 endowed with a sub - riemannian metric . it was shown that sets with hausdorff dimension lower than @xmath17 are removable , while those with dimension higher than @xmath17 are not .
moreover , there exist both removable and non - removable sets with hausdorff dimension equal to @xmath17 .
hence , as in the euclidean case , the dimension threshold for removable sets is @xmath18 , where @xmath19 denotes the hausdorff dimension of @xmath16 .
the results from @xcite were extended in @xcite from heisenberg groups to a larger class of lie groups , the _ carnot groups_. there exists a well developed theory of sub - laplacians in this setting , see for instance the book @xcite by a. bonfiglioli , e. lanconelli and f. uguzzoni . in order to characterize removable sets for lipschitz harmonic functions in @xmath4 ,
one has to characterize the sets on which the sio associated with the riesz kernel @xmath20 is bounded in @xmath21 . in @xmath22 , one would need to face a sio with kernel @xmath23 for @xmath24 . at this point ,
the knowledge about the action of this sio on @xmath2-dimensional subsets of @xmath22 ( i.e. subsets of co - dimension @xmath25 ) is extremely limited . in the present paper
, we will not address this question further , but it motivates the study of quantitative rectifiability in @xmath22 .
the main purpose of the present paper is to initiate this study , and to introduce some new , relevant concepts in @xmath26 .
our aims are twofold .
first , we demonstrate that some parts , at least , of the euclidean theory of quantitative rectifiability carry over to @xmath0 . to us , this gives hope that some day in the distant future questions on the boundedness of sios on subsets of @xmath0 may be understood as well as they currently are in @xmath4 .
our second aim is somewhat more philosophical : we want to demonstrate that building a theory of quantitative rectifiability in @xmath0 is worth the effort . in particular , the proofs are not , merely , technically challenging replicas of their euclidean counterparts .
new phenomena appear .
in particular , proving our main result , theorem [ main ] , involved studying non - smooth solutions of the ( planar ) non - linear pde @xmath27 known as the ( or rather `` a '' ) _ burgers equation_. in proving the euclidean counterpart of theorem [ main ] ,
such considerations are not required . at least to us , any connection between the innocent - looking statement of theorem [ main ] , and the pde , was quite a surprise at first sight . in the terminology of david and semmes , theorem [ main ]
is the _ weak geometric lemma _ for certain subsets of @xmath0 , called _ intrinsic lipschitz graphs_. for now , we just briefly explain the meaning of these concepts ; precise definitions are postponed to sections [ s : prelim ] and [ s : bpilg ] .
we consider two kinds of subgroups of @xmath0 : _ horizontal _ and _ vertical_. writing @xmath28 , the horizontal subgroups are lines through the origin inside @xmath29 , while the vertical subgroups are planes spanned by a horizontal subgroup and the `` @xmath30-axis '' @xmath31 . in the present paper , we are mainly concerned with _ intrinsic graphs _ over a fixed ( but arbitrary ) vertical subgroup @xmath32 , which we often take to be the `` @xmath33-plane '' @xmath34 .
let @xmath35 be the horizontal subgroup @xmath36 , and consider a function @xmath37 .
the intrinsic graph of @xmath38 ( over @xmath39 ) is the set @xmath40 where `` @xmath41 '' refers to the heisenberg product ( see section [ s : prelim ] ) .
note that @xmath42 is , in general different , from the `` euclidean graph '' @xmath43 , and in fact @xmath42 need not be representable as the euclidean graph of any function ( see example 2.5 in @xcite ) .
recall that a function @xmath44 is ( euclidean ) lipschitz , if and only if there exists a cone , which , when centered at any point @xmath45 on the graph of @xmath46 , only intersects the graph at @xmath45 .
the notion of _ intrinsic lipschitz function _ in @xmath0 is * defined * with this characterization in mind , with `` graph '' replaced by `` intrinsic graph '' , and `` cone '' replaced by a natural @xmath0-analogue , see .
intrinsic lipschitz functions were introduced by b. franchi , r. serapioni and f.serra cassano in @xcite , and they turned out to be very influential in the evolution of geometric analysis in heisenberg groups , see for instance @xcite .
curiously , the definition does not guarantee that an intrinsic lipschitz function is ( metrically ) lipschitz between the spaces @xmath39 and @xmath35 .
if the reader is familiar with the theory of rectifiability in metric spaces , but not with that in @xmath0 , she may wonder why such `` intrinsic '' notions are necessary in the first place ; why can not one study ( metrically ) lipschitz images @xmath47 ?
the reason is simple : a lipschitz image @xmath48 has vanishing @xmath49-dimensional measure for @xmath50 .
this is a result of l. ambrosio and b. kirchheim @xcite .
the work of mattila , serapioni and serra cassano @xcite and franchi , serapioni and serra cassano @xcite suggests that intrinsic lipschitz graphs , instead , are the correct class of sets to consider in connection with @xmath2-rectifiability in the heisenberg group .
we believe that this is true also in the quantitative setting .
lipschitz graphs in @xmath4 are , arguably , the most fundamental examples of quantitatively rectifiable sets in the sense of david and semmes . in the present paper
, we propose that intrinsic lipschitz graphs play the same role in @xmath0 . in @xmath4 ,
the term `` quantitatively rectifiable '' has many meanings ; the fundamental results of david and semmes show that a certain class of sets enjoys and can be characterized by a wide variety of properties , both geometric and analytic , each of which could be taken as the definition of `` quantitatively rectifiable '' . in @xmath0 , no such results are available , yet , so we have to specify our viewpoint to `` quantitative rectifiability '' .
it will be that of `` quantitative affine approximation '' .
theorem [ main ] , informally stated , says that intrinsic lipschitz graphs admit good affine approximations `` at most places and scales '' .
the traditional way to quantify such a statement is via the notion of @xmath3-numbers , introduced by p. jones in @xcite in order to control the cauchy singular integral on @xmath25-dimensional lipschitz graphs .
they were later used by jones @xcite and david and semmes @xcite in order to characterize quantitative rectifiability .
the same approach works in @xmath0 , if the @xmath3-numbers are defined correctly . in definition [ verticalbetas ] below , we introduce the _
vertical @xmath3-numbers_. these nearly coincide with the usual ( euclidean ) @xmath3-numbers , defined with respect to the metric in @xmath0 of course : the single , crucial , difference is that approximating affine planes are restricted to sets of the form @xmath51 , where @xmath52 and @xmath53 is a vertical subgroup . viewing @xmath0 as @xmath54 for a moment ,
these sets are simply ( euclidean ) translates of the sets @xmath53 .
so , they are quite literally vertical planes
. here is , finally , the main result : [ main ] an intrinsic lipschitz graph in @xmath0 satisfies the weak geometric lemma for the vertical @xmath3-numbers .
for a more precise restatement see theorem [ wglintlg ] . in brief , the weak geometric lemma states that , for any fixed @xmath55 , the vertical @xmath3-numbers of the graph have size at most @xmath55 in all balls , except perhaps a family satisfying a carleson packing condition ( with constants depending on @xmath55 ) .
this manner of quantifying the `` smallness '' of an exceptional family of balls is ubiquitous in the theory of david and semmes .
theorem [ main ] does not explain our need to define the vertical @xmath3-numbers ; since the vertical @xmath3-numbers are , evidently , at least as large as the `` usual '' ones ( with no restrictions on the approximating affine planes ) , the statement of theorem [ main ] merely becomes a little weaker , if the word `` vertical '' is removed .
however , it turns out that the weak geometric lemma for the vertical @xmath3-numbers , combined with a condition on _ vertical projections _ , essentially * characterizes * intrinsic lipschitz graphs .
this is the content of our second main result , a counterpart of a theorem of david and semmes @xcite from 1990 : [ main2 ] assume that a @xmath2-regular set @xmath56 satisfies the weak geometric lemma for the vertical @xmath3-numbers , and has big vertical projections .
then @xmath57 has big pieces of intrinsic lipschitz graphs . as before
, we postpone explaining the notions of _ big vertical projections _ ( definition [ bvp ] ) and _ big pieces of intrinsic lipschitz graphs _ ( definition [ bgilg ] ) .
the latter condition does not guarantee that @xmath57 is an intrinsic lipschitz graph ( such a statement would be false , rather obviously ) .
instead , @xmath58 contains a measure - theoretically big piece of an intrinsic lipschitz for every ball @xmath59 centered on @xmath57 .
finally , we mention that theorem [ main2 ] admits a converse , which follows from theorem [ main ] by standard considerations , outlined at the end of the paper : [ maincor ] assume that a @xmath2-regular set @xmath56 has big pieces of intrinsic lipschitz graphs .
then @xmath56 satisfies the weak geometric lemma for the vertical @xmath3-numbers , and has big vertical projections .
so , the property of having big pieces of intrinsic lipschitz graphs is characterized by the combination of the weak geometric lemma for the vertical @xmath3-numbers , and the big vertical projections condition .
we remark that other versions of sub - riemannian @xmath3-numbers have been defined before .
f. ferrari , b. franchi , and h. pajot @xcite , n. juillet @xcite and recently s. li and r. schul @xcite , @xcite , considered horizontal @xmath3-numbers in order to study curves in the heisenberg group .
we stress that these quantities differ significantly from the vertical @xmath3-numbers . in the case of the horizontal @xmath3-numbers
, approximation is restricted to _ horizontal lines _ , which are @xmath25-dimensional objects . as we mentioned earlier , for the vertical @xmath3-numbers ,
the approximating sets are translated vertical planes , which are @xmath2-dimensional objects in the heisenberg metric .
part of our motivation stems from facilitating the study of removability questions in @xmath0 ; and since the critical dimension for this problem is @xmath2 , we believe that the vertical @xmath3-numbers will be the relevant tool .
the paper is organized as follows . in section [
s : prelim ] we lay down the necessary background in the heisenberg group and we also discuss intrinsic lipschitz graphs and some of their main properties . in section [ s :
bpilg ] , we give sufficient conditions for a @xmath2-dimensional set in the first heisenberg group to have big pieces of intrinsic lipschitz graphs .
section [ s : wgl ] is devoted to the proof of theorem [ main ] and is definitely the most technical part of the paper . to facilitate the reader s navigation through the somewhat lengthy section [ s : wgl ]
a second introductory part appears in section [ outline ] .
* acknowledgements . *
part of the research for this paper was completed while various subsets of the authors visited the universities of bern , helsinki and connecticut .
the hospitality of these institutions is acknowledged .
the _ heisenberg group _ @xmath0 is @xmath60 endowed with the group law @xmath61 for @xmath62 .
we will sometimes identify @xmath60 with @xmath63 and denote points in the heisenberg group by @xmath64 for @xmath65 and @xmath66 .
we use the following metric on @xmath0 : @xmath67 where @xmath68 the closed balls in @xmath69 will be denoted by @xmath70 .
we will also denote by @xmath71 the @xmath72-dimensional hausdorff measure in @xmath69 .
the reader who is not familiar with the notion of hausdorff measures should have a look at @xcite . for more information on the heisenberg group ,
see for instance the book @xcite by capogna , danielli , pauls and tyson .
here we just mention that @xmath73 , and the usual lebesgue measure on @xmath1 coincides ( up to a constant ) with @xmath74 on @xmath0 .
the distance @xmath75 is invariant with respect to left translations @xmath76 and homogeneous with respect to dilations @xmath77 recall that a @xmath78-measurable set @xmath79 is _ @xmath2-(ahlfors - david)-regular _ , if there exists a constant @xmath80 , the _ regularity constant _ of @xmath57 , such that @xmath81 for all @xmath82 and @xmath83 .
we stress once more that metric concepts in @xmath0 , such as `` ball '' , `` hausdorff measure '' or `` ahlfors - david regular '' are always defined with respect to @xmath84 , unless explicitly stated otherwise .
identify @xmath0 with @xmath85 for a moment .
if @xmath86 is a line through the origin , then @xmath87 is called a _ horizontal subgroup of @xmath0 . _
a _ vertical subgroup of @xmath0 _ is a set of the form @xmath88 , where @xmath86 is a line through the origin .
note that both @xmath32 and @xmath89 are subgroups of @xmath0 , and closed under the action of @xmath90 . under the identification of @xmath32 with @xmath91
, the subgroup @xmath32 can be endowed with the @xmath92-dimensional lebesgue measure @xmath93 .
this turns out to be a haar measure on @xmath94 , and it agrees ( up to a multiplicative constant ) with @xmath78 on @xmath95 , see ( * ? ? ?
* proposition 2.20 ) . given a vertical subgroup @xmath96 of @xmath0
, we define the _ complementary horizontal subgroup _
@xmath97 where @xmath98 is the orthogonal complement of @xmath99 in @xmath100 .
then every point @xmath101 can be written * uniquely * as @xmath102 with @xmath103 and @xmath104 .
one could also consider other splittings of the heisenberg group , but in this paper we will always assume that the groups @xmath32 and @xmath105 are _ orthogonal _ ; by this we mean that @xmath99 and @xmath106 are orthogonal as above .
let @xmath88 be a vertical subgroup with complementary horizontal subgroup @xmath107 .
as we observed , every point @xmath101 can be written uniquely as @xmath108 with @xmath103 and @xmath104 .
this gives rise to the _ vertical projection _ @xmath109 and _ horizontal projection _
@xmath110 , defined by @xmath111 the mappings @xmath109 and @xmath110 have the following explicit formulae : @xmath112 and @xmath113 for @xmath114 . here
@xmath115 and @xmath116 are the usual orthogonal projections onto the lines @xmath99 and @xmath106 in @xmath100 .
we also used the abbreviating notation @xmath117 for @xmath118 and @xmath119 .
the horizontal projections @xmath110 are both ( metrically ) lipschitz functions , and group homomorphisms .
the vertical projections @xmath109 are neither . however , as we will see many times in this paper , the projections @xmath109 interact well with intrinsic lipschitz graphs , defined below .
let @xmath32 be a vertical subgroup with complementary horizontal subgroup @xmath89 .
@xmath0-cone perpendicular to @xmath32 and with aperture @xmath120 _ is the following set @xmath121 : @xmath122 [ d : intrlip ] a subset @xmath123 is an _
intrinsic @xmath124-lipschitz graph over a vertical subgroup @xmath32 _ , if @xmath125 if @xmath126 is any set , and @xmath89 is the complementary horizontal subgroup of @xmath32 , we say that a function @xmath127 is _ an intrinsic @xmath124-lipschitz function _ , if the _ intrinsic graph of @xmath38 _ , namely @xmath128 is an intrinsic @xmath124-lipschitz graph .
the _ intrinsic lipschitz constant _ of @xmath38 ( or @xmath129 ) is defined as the infimum over all constants @xmath124 for which @xmath38 ( or @xmath129 ) is intrinsic @xmath124-lipschitz . for a nice picture of intrinsic lipschitz graphs and the @xmath0-cones ,
see section 3 in @xcite .
[ remk : para ] an intrinsic lipschitz graph can be uniquely parametrised by an intrinsic lipschitz function .
more precisely , given an intrinsic @xmath124-lipschitz graph @xmath129 over a vertical subgroup @xmath32 , there exists a unique intrinsic @xmath124-lipschitz function @xmath130 such that @xmath131 . to see this , one first checks that the cone condition implies the injectivity of @xmath132 .
indeed , assume that @xmath133 for some @xmath134 . writing ( uniquely ) @xmath135 and @xmath136 , this gives @xmath137 since @xmath138 , and @xmath109 annihilates @xmath89 .
hence , @xmath139 for any @xmath140 , and so @xmath141 , implying @xmath142 .
consequently , the following mapping @xmath130 is well - defined : @xmath143 the mapping @xmath144 clearly satisfies @xmath131 , and thus @xmath144 is intrinsic @xmath124-lipschitz by definition .
the uniqueness of @xmath144 follows from the uniqueness of the representation @xmath135 , @xmath145 .
we refer to @xmath144 as the _ parametrisation of @xmath129_. a key property of intrinsic lipschitz graphs is that they are invariant under left translations and dilations in @xmath0 ; if @xmath129 is an intrinsic @xmath124-lipschitz graph , then @xmath146 is also an intrinsic @xmath124-lipschitz graph for any @xmath101 and @xmath147 .
this is why these sets are called `` intrinsic '' !
[ rotationalinvariance ] we will often `` without loss of generality '' assume that the intrinsic lipschitz functions and graphs under consideration are defined over the specific vertical plane @xmath148 .
this is legitimate , because the notions are invariant under the rotations @xmath149 around the @xmath30-axis , given by @xmath150 .
the rotations are both group homomorphisms and isometries with respect to the metric @xmath84 .
the homomorphism property implies that @xmath151 , and then the isometry property gives the following : if @xmath129 is an intrinsic @xmath124-lipschitz graph over @xmath32 , then @xmath152 is an intrinsic @xmath124-lipschitz graph over @xmath153 .
the class of intrinsic lipschitz functions has ( in greater generality ) been introduced and studied by franchi , serapioni and serra cassano .
the following facts are special cases of the results in @xcite and @xcite : * for @xmath126 , an intrinsic @xmath124-lipschitz function @xmath154 can be extended to an intrinsic @xmath155-lipschitz function @xmath156 , where @xmath155 depends only on @xmath124 . *
an intrinsic lipschitz function @xmath157 is intrinsically differentiable @xmath93 almost everywhere on @xmath32 . *
an intrinsic @xmath124-lipschitz graph is @xmath2-regular with regularity constant depending only on @xmath124 .
we will write more about intrinsic differentiability and the related notion of _ intrinsic gradient _ in section [ ss : intr_grad ] .
if @xmath158 , we will use the notation @xmath159 to signify that there exists a constant @xmath160 depending only on @xmath161 such that @xmath162 .
if the constant @xmath163 is absolute , we write @xmath164 .
the two - sided inequality @xmath165 is abbreviated to @xmath166 .
in this section , we start proving our two main results . to warm up ,
we begin with the significantly easier theorem [ main2 ] : if a @xmath2-regular set @xmath56 satisfies the weak geometric lemma for vertical @xmath3-numbers , and has big vertical projections ( see definitions below ) , then @xmath57 has bpilg .
the argument is very similar to the euclidean counterpart , due to david and semmes @xcite .
in fact , the greatest surprise here is probably the similarity of the arguments itself : considering that the vertical projections @xmath167 are not lipschitz , one might expect a rockier ride ahead .
we start with a few central definitions and auxiliary results .
[ bvp ] a @xmath2-regular set @xmath168 is said to have _ big vertical projections _
( bvp in short ) if there exists a constant @xmath169 with the following property : for all @xmath82 and for all @xmath170 there exists a vertical subgroup @xmath32 such that @xmath171 [ bgilg ] a @xmath2-regular set @xmath168 has _
big pieces of intrinsic lipschitz graphs _
( bpilg in short ) if there exist constants @xmath172 and @xmath173 with the following property
. for all @xmath82 and @xmath170 , there exists an intrinsic @xmath124-lipschitz graph @xmath129 over some vertical subgroup such that @xmath174 [ verticalbetas ] let @xmath175 be a set , let @xmath176 be a ball with radius @xmath177 centered on @xmath57 , let @xmath32 be a vertical subgroup , and let @xmath52 .
we write @xmath178 and then we define the _ vertical @xmath3-number _ as @xmath179 the infimum is taken over all vertical subgroups @xmath32 , and all points @xmath52 .
[ changeofmidpoint ] the following observation is useful , and not quite as trivial as its euclidean counterpart .
assume @xmath176 is a ball centered on @xmath57 , @xmath32 is any vertical subgroup , and @xmath52 . then @xmath180 to prove this , observe that @xmath181 for any point @xmath101 .
in particular , if @xmath182 , we have @xmath183 hence , if further @xmath184 and @xmath185 , we have @xmath186 since this holds for all @xmath185 , we have @xmath187 as claimed .
[ wgldef ] we say that a @xmath2-regular set @xmath168 satisfies the _ weak geometric lemma for vertical @xmath3-numbers _ ( wgl in short ) , if @xmath188 for all @xmath189 , @xmath82 and @xmath190 .
the following lemma shows that even if the vertical projections @xmath109 are not lipschitz , they still can not increase @xmath78-measure ( too much ) .
this is rather surprising , as it is easy to find less than three - dimensional sets @xmath56 such that @xmath191 , see example 4.1 in @xcite .
[ lipschitzlemma ] let @xmath32 be a vertical subgroup in @xmath0 .
then there exists a constant @xmath192 such that for all @xmath193 , one has @xmath194 the lemma follows from ( * ? ? ?
* lemma 2.20 ) , which , when specialized to the heisenberg group , states that there is a constant @xmath195 such that @xmath196 for all @xmath197 and @xmath198 .
see also ( * ? ? ?
* lemma 3.14 ) . given a set @xmath199 of positive @xmath200-measure ,
choose a covering of @xmath201 by closed balls @xmath202 , @xmath203 , such that @xmath204 then , we find that @xmath205 this completes the proof .
we recall the construction of _ david cubes _
, first introduced by david in @xcite , which can be defined on any regular set in a geometrically doubling metric space .
let @xmath56 be a @xmath2-regular set .
then , there exists a family of partitions @xmath206 of @xmath57 , @xmath207 , with the following properties : * if @xmath208 , @xmath209 and @xmath210 , then either @xmath211 , or @xmath212 . * if @xmath209 , then @xmath213 . * every set @xmath209 contains a set of the form @xmath214 for some @xmath215 , and some constant @xmath216 .
the sets in @xmath217 are called david cubes , or just cubes , of @xmath57 . for @xmath209 , we define @xmath218 .
thus , by ( ii ) , we have @xmath219 for @xmath220 . given a fixed cube @xmath221 , we write @xmath222 it follows from ( ii ) , ( iii ) , and the @xmath2-regularity of @xmath57 that @xmath223 for @xmath209 .
it is an easy fact , needed a bit later , that the following holds : if @xmath224 are distinct points , there exists an index @xmath207 , and disjoint cubes @xmath225 , containing @xmath45 and @xmath226 , respectively , with the properties that @xmath227 , @xmath228 indeed , let @xmath207 be the largest integer such that @xmath229 , and let @xmath225 be the unique cubes containing @xmath45 and @xmath226
. then @xmath230 , and since @xmath231 , we have @xmath232 .
the same holds with the roles of @xmath45 and @xmath226 reversed . in the sequel
, we write @xmath233 the weak geometric lemma ( definition [ wgldef ] ) implies the following reformulation in terms of david cubes .
write @xmath234 then @xmath235 for any @xmath189 and @xmath221 .
deriving this property from definition [ wgldef ] is an easy exercise using the properties of david cubes , and we omit the details .
here is the statement of theorem [ main2 ] once more : [ t : main1_bpilg ] a @xmath2-regular set in @xmath0 with bvp and satisfying the wgl has bpilg .
let @xmath236 be a system of david cubes on @xmath57 , let @xmath237 be constants , and let @xmath32 be a vertical subgroup . throughout this section
, a cube @xmath220 will be called _ good _ ( more precisely @xmath238-good ) , if @xmath239 and @xmath240 we outline the proof of theorem [ t : main1_bpilg ] .
the proof is divided into two parts , a geometric one , and an abstract one .
the geometric part shows that good cubes @xmath220 are already `` almost '' intrinsic lipschitz graphs in the following sense : if @xmath241 and @xmath242 satisfy @xmath243 , then @xmath244 for some small @xmath140 .
the abstract part uses the wgl and bvp assumptions to infer ( cutting a few corners here ) that only a small fraction of @xmath245 , @xmath246 , meets @xmath247 for some non - good cube @xmath248 .
hence , a large set @xmath249 meets @xmath247 only for good cubes @xmath248 .
unfortunately , this is not literally true , and additional ( technical ) considerations are needed . ignoring these for now
, we can complete the proof as follows .
fixing @xmath250 , we can use the discussion in section [ davidcubes ] to find a cube @xmath220 with @xmath241 , @xmath251 and @xmath252 .
since @xmath250 , we know that @xmath248 is a good cube , and it follows from the geometric part that @xmath253 .
consequently , @xmath254 is an intrinsic lipschitz graph .
the following lemma is our counterpart of lemma 2.19 in david and semmes proof in @xcite .
[ l : ds_lem ] assume that @xmath248 is a @xmath238-good cube , @xmath241 and @xmath242 with @xmath243 .
then @xmath244 , if @xmath55 is sufficiently small with respect to @xmath255 , and @xmath140 is small enough ( depending on the constants @xmath256 ) .
we start with a reduction to `` unit scale '' .
assume that the statement of the lemma fails for certain parameters @xmath257 , and a certain @xmath238-good cube @xmath220 . by this
, we mean that and hold for @xmath248 , yet @xmath258 for some @xmath241 and @xmath242 with @xmath259 .
consider the set @xmath260 , and observe that it , also , satisfies , since @xmath261 and @xmath262 .
the first equation is not altogether trivial , but it follows from the equations @xmath263 and the fact that the mapping @xmath264 , @xmath265 , has unit jacobian for any fixed @xmath101 , see the proof of lemma 2.20 in @xcite , or below .
further , if @xmath266 , then @xmath267 . here , @xmath3 denotes the vertical @xmath3-number associated with @xmath268 .
note that @xmath269 is @xmath2-regular with the same constants as @xmath57 .
finally , note that @xmath270 , and @xmath271 with @xmath272 . to sum up ,
if the lemma fails for @xmath248 , then we can construct another @xmath2-regular set @xmath269 , and another good david cube @xmath273 ( for @xmath269 ) with @xmath270 and @xmath274 , such that the lemma fails for @xmath273 .
thus , it suffices to prove the lemma for a david - cube @xmath248 with the additional properties @xmath275 and @xmath276 . to this end , assume to the contrary that @xmath277 with @xmath278 .
we will use this to show that the entire projection @xmath279 is contained in a small neighbourhood of the @xmath30-axis @xmath280 .
this will violate . somewhat abusing notation
, we write @xmath281 let @xmath101 and @xmath53 be such that @xmath282 the first task is to show that the angle @xmath283 between @xmath284 and @xmath285 satisfies @xmath286 write @xmath287 .
the plan is to use the smallness of @xmath288 in order to find a point @xmath289 on @xmath53 , but close to @xmath226 , such that latexmath:[\[\label{eq : angle } the angle between @xmath106 and @xmath291 is @xmath292 and thus , as claimed . in order to show , we first observe that the assumption of @xmath277 implies @xmath293 recalling that @xmath278 , this is only possible if @xmath294 indeed , we even have @xmath295 ; otherwise the left hand side of can be bounded from below by @xmath296 , while the upper bound is then @xmath297 .
for small enough @xmath298 , this is impossible .
thus we may suppose .
further , by remark [ changeofmidpoint ] we find for all @xmath299 that @xmath300 by our choice of @xmath161 and @xmath53 .
in particular , for @xmath301 , there exists a vector @xmath302 with @xmath303 by , this gives @xmath304 , and finally , using @xmath277 , @xmath305 this proves and .
so , we know that * @xmath248 is close to @xmath53 ( by ) , * @xmath53 is close to @xmath285 ( by ) . as we will next demonstrate , @xmath279 is close to @xmath306 . indeed ,
since we do not care about the best constants here , we can finish the proof very quickly : let @xmath307 be a number such that if @xmath308 , then @xmath309 for some @xmath310
. recalling , we can pick @xmath311 arbitrarily small by choosing @xmath312 small enough .
now , if @xmath313 , then by and the triangle inequality , we have @xmath314 for some @xmath315 . since @xmath109 is locally @xmath316-hlder continuous , it follows that @xmath317 the same holds with @xmath280 replaced by @xmath318 .
finally , the @xmath319-measure of the @xmath320-neighbourhood of @xmath321 is bounded by a constant depending only on @xmath312 , and this constant tends to zero as @xmath322 . for sufficiently small values of @xmath312 , this violates , and the proof is complete . in this section , we apply lemma [ l : ds_lem ] to good cubes inside a set @xmath175 satisfying the weak geometric lemma for vertical @xmath3-numbers .
this is a counterpart for theorem 2.11 in @xcite , which in turn is modelled on a result of p. jones @xcite .
the proof below is very similar to that in @xcite ; given lemmas [ lipschitzlemma ] and [ l : ds_lem ] , the argument does not really see the difference between @xmath0 and @xmath4 .
we still record the full details .
[ t : main2_inv ] assume that @xmath168 is a @xmath2-regular set satisfying the wgl and let @xmath323 .
then there exist numbers @xmath140 and @xmath324 , depending only on @xmath325 and the @xmath2-regularity and wgl constants of @xmath57 , such that the following holds : for every david cube @xmath326 in @xmath57 and for all vertical projections @xmath109 , there exist intrinsic @xmath327-lipschitz graphs @xmath328 , @xmath329 , @xmath330 let @xmath57 and @xmath325 be as in the assumptions of theorem [ t : main2_inv ] .
let further @xmath331 be a small number to be chosen later ( based on lemma [ l : ds_lem ] ) .
fix an arbitrary cube @xmath221 , and an arbitrary vertical subgroup @xmath32 .
first , we will group the cubes in @xmath332 into `` good '' and `` bad '' cubes , and control the quantity of the bad cubes via the wgl assumption .
second , lemma [ l : ds_lem ] , coupled with a `` coding argument '' , will be used to partition the complement of the `` bad '' cubes in @xmath333 into the sets @xmath334 .
the `` good '' cubes @xmath335 are the familiar @xmath336-good cubes defined right above lemma [ l : ds_lem ] .
the class @xmath337 consists of those maximal ( hence disjoint ) cubes in @xmath326 that violate the first goodness condition , i.e. , @xmath338 let @xmath339 be the class of ( all , not maximal ) cubes that violate the second goodness condition : @xmath340 then , clearly , @xmath341 .
it is also clear that the projections of bad cubes from the first class have small measure : for @xmath342 , we have @xmath343 on the other hand , for the second bad class , one can control the measure of the cubes directly by the variant of wgl formulated in : @xmath344 since @xmath345 for all @xmath346 by lemma [ lipschitzlemma ] , the inequality above shows that the @xmath109-projection of @xmath347 has measure no larger than @xmath348 .
this is a little bit too weak for our purposes ; in analogy with , we wish to replace @xmath349 by a small constant . to this end
, we set @xmath350 where @xmath351 is so large that @xmath352 .
this is possible : @xmath353 with this definition of @xmath354 , lemma [ lipschitzlemma ] gives @xmath355 it remains to find subsets @xmath356 such that @xmath357 and for every @xmath358 and every pair @xmath359 it holds that @xmath360 for @xmath298 small enough ( only depending on @xmath325 and the @xmath2-regularity and wgl constants of @xmath57 ) .
this is done via a `` coding argument '' , which goes back to jones , see @xcite .
the argument is also explained briefly in david s book @xcite , p. 8182 , but we present the full details . we start with a brief informal overview .
write @xmath361 .
why do we need a `` coding argument '' ?
maybe we can show , directly , that if @xmath362 then @xmath360 ? pick two distinct points @xmath250 , and pick two disjoint cubes @xmath363 of some common generation such that @xmath364 , @xmath365 , @xmath366 and @xmath243 ( such cubes exist , as discussed in section [ davidcubes ] ) .
now , since @xmath367 , we know that @xmath368 . if and this is the `` big if '' we also knew that @xmath369 for @xmath189 small enough , we could infer from lemma [ l : ds_lem ] that @xmath244 ( assuming also that @xmath298 is small enough ) .
of course , we do not know that @xmath369 for the particular cube @xmath370 we are interested in : even though @xmath371 , there can still be up to @xmath372 `` exceptional '' cubes @xmath373 such that @xmath374 .
the `` coding argument '' is needed to fix this issue . essentially , if we declare that @xmath375 , say , we want to make sure that the following holds : whenever @xmath373 with @xmath374 , then all the points @xmath376 are stored safely away in the other sets @xmath334 , @xmath377 . once that has been accomplished , the argument above works for @xmath378 ( or any @xmath334 ) in place of @xmath379 .
we turn to the details , which are repeated from @xcite nearly verbatim . for each cube @xmath380
, we will associate a certain finite sequence of @xmath381 s and @xmath25 s , denoted by @xmath382 .
the length of such a sequence is denoted by @xmath383 .
we declare @xmath384 to be the empty sequence .
next , assume inductively that the numbers @xmath382 have been defined for the descendants of @xmath380 down to a certain generation , say @xmath49 .
we now aspire to make the definition for cubes @xmath248 of generation @xmath385 .
if @xmath248 is such a cube , and @xmath386 is its parent , we initially set @xmath387 .
assume that @xmath388 , that is , @xmath374 , and assume that there exists at least one other cube @xmath389 of the same generation as @xmath248 such that @xmath390 ( if either of these assumptions fails , we do not alter @xmath382 now ) .
note that @xmath382 and @xmath391 have both been initially defined .
there are two cases to consider .
* case 1 : @xmath392 .
if @xmath393 , we do not alter @xmath382 or @xmath391 .
but if @xmath394 , we re - define @xmath382 by adding a `` @xmath381 '' , and we re - define @xmath391 by adding a `` @xmath25 '' . *
case 2 : @xmath395 .
if , for instance , @xmath396 , then we do not alter @xmath382 .
but we re - define @xmath391 by adding either `` @xmath381 '' or `` @xmath25 '' to it in such a fashion that the new @xmath391 is * not * an initial segment of @xmath382 . finally ,
if @xmath397 , then we repeat the same step with the roles of @xmath248 and @xmath389 reversed . after
this procedure is complete , we pick another cube @xmath398 of generation @xmath385 ( if it exists ) , and perform the previous case chase with the pair @xmath399 replaced by @xmath400 . once all the pairs @xmath401 with @xmath402
have been processed , we move on to other pairs @xmath403 with @xmath404 and @xmath405 , and give them the same treatment as above .
the algorithm terminates eventually ( because there are only finitely many cube - pairs to consider ) , and , at the end , every cube @xmath248 of generation @xmath385 has an associated sequence @xmath382 . if @xmath406 for all cubes @xmath404 of generation @xmath385 , then @xmath382 retains the initial value @xmath407 .
even if @xmath408 for some @xmath404 , this can occur only for a bounded number , say @xmath409 , of alternatives @xmath404 of generation @xmath385 .
consequently , @xmath382 differs from @xmath407 by a sequence of length @xmath410 . by applying the procedure at all generations @xmath49 , every sub - cube of @xmath333
gets associated with a ( finite ) sequence @xmath382 .
next , we wish to extend the definition of these sequences from cubes to points in @xmath411 .
fix @xmath412 , and let @xmath413 be the unique sequence of dyadic cubes converging to @xmath45 . as discussed in the previous paragraph , @xmath414 can differ from @xmath415 only in case @xmath416 is contained in @xmath417 for some @xmath404 of the same generation as @xmath416 . by definition of @xmath371
, there can only be @xmath418 such indices @xmath419 .
in particular , the sequences @xmath415 converge to some finite sequence of @xmath381 s and @xmath25 s , denoted by @xmath420 .
furthermore , for those @xmath418 indices @xmath419 , where @xmath414 possibly differs from @xmath415 , this difference is a sequence of length at most @xmath409 .
consequently , the possible values of @xmath420 form a finite set @xmath421 , whose cardinality can be bounded from above in terms of the constants @xmath372 and @xmath409 . given an element @xmath422 , we now define @xmath423 it remains to check that the sets @xmath424 satisfy the useful property we hinted at in the informal overview .
assume that @xmath250 belong to `` nearby '' two cubes @xmath425 of the same generation , namely with @xmath426 and @xmath366 , and assume that either @xmath427 or @xmath428 .
then , we claim that @xmath45 and @xmath226 belong to two different sets of the form @xmath424 .
consider the sequences @xmath429 and @xmath430 ( which are initial sequences in @xmath420 and @xmath431 , respectively ) .
assume , for instance , that @xmath427 .
this means that @xmath432 , so the pair @xmath433 is considered while defining the sequences @xmath429 and @xmath430 .
then , inspecting case 1 and case 2 , it is clear that neither of the sequences @xmath429 and @xmath430 can be an initial sequence of the other .
this proves that @xmath434 , as claimed .
now , we can quickly prove that if @xmath435 then @xmath360 , if @xmath140 is small enough .
pick @xmath436 , and let @xmath370 and @xmath437 be sub - cubes of @xmath333 , containing @xmath45 and @xmath226 , respectively , with same generation , satisfying @xmath426 and @xmath366 , and with @xmath252 .
it follows from the claim in the previous paragraph that @xmath369 and @xmath438 .
consequently , by lemma [ l : ds_lem ] , we have @xmath244 for small enough @xmath298 .
this completes the proof of the theorem . we are now prepared to prove theorem [ t : main1_bpilg ] .
again , the proof is very similar to the euclidean argument , see theorem 1.14 in @xcite .
let @xmath57 be a @xmath2-regular subset of @xmath0 with bvp , and satisfying the wgl .
the former property ensures that for fixed @xmath82 and @xmath170 , there is a vertical subgroup @xmath32 such that @xmath439 where @xmath440 is a constant depending only on @xmath57 .
then , there exists @xmath441 , depending only on @xmath440 and the @xmath2-regularity constant of @xmath57 , and a david cube @xmath442 such that @xmath443 and @xmath444 now we apply theorem [ t : main2_inv ] to this particular cube @xmath326 and choice of @xmath441 .
it follows that there exist numbers @xmath140 and @xmath324 ( depending only on @xmath325 , and the @xmath2-regularity and wgl constants of @xmath57 ) with the following property : there exist intrinsic @xmath327-lipschitz graphs @xmath328 , @xmath329 , such that @xmath330 this , together with , implies @xmath445 thus there must exist some @xmath329 such that @xmath446 the proof is complete .
in the first half of the paper , we saw that any @xmath2-ahlfors - david regular subset of @xmath0 with big vertical projections ( bvp ) , and satisfying the weak geometric lemma ( wgl ) , has big pieces of intrinsic lipschitz graphs ( bpilg ) .
the second half of the paper is devoted to proving the converse .
the fact that intrinsic lipschitz graphs ( and thus bpilg as well ) have bvp is almost trivial , see remark [ remark0 ] . hence in order to prove theorem [ maincor ] it suffices to prove the wgl for sets which have bpilg . theorem [ wglbp ] will follow using standard arguments , recalled at the end of the paper , once we have at out disposal the wgl for intrinsic lipschitz graphs . therefore the rest of the paper will be devoted to the proof of theorem [ main ] , which is precisely stated below .
the proof of theorem [ wglintlg ] in the euclidean case is relatively straightforward , and can be carried out as follows .
assume that @xmath453 is an ( entire ) @xmath124-lipschitz function , and let @xmath236 be a system of dyadic squares on @xmath454 .
assume that @xmath220 is such that @xmath46 is far from affine in @xmath248 , in the sense that @xmath455 for all affine functions @xmath456 .
then , it is fairly easy to verify that the gradient of @xmath46 must fluctuate significantly near @xmath248 : there exists a fairly large sub - cube @xmath457 such that @xmath458 , where @xmath440 only depends on @xmath55 and @xmath124 , and @xmath459 denotes the average of @xmath460 over @xmath248 .
the wgl follows from this observation , plus the fact that @xmath461 .
after all , for such a function , there is a concept of an _ intrinsic gradient _
@xmath462 ( see section [ ss : intr_grad ] below ) , which is known to exist at almost every point on @xmath32 , and moreover @xmath463 .
so , if it were the case that the local `` non - affinity '' of @xmath38 forces @xmath462 to fluctuate noticeably , one could wrap up the argument in the fashion outlined above .
however , this is simply not true : in any bounded domain @xmath464 , the equation @xmath465 admits ( smooth ) non - affine solutions ! for instance , the function @xmath466 satisfies @xmath467 on its domain .
in fact , even non - smooth continuous solutions are possible : the function @xmath468 , discussed in ( * ? ? ? * remark 4.4.2 ) and defined by @xmath469 satisfies @xmath465 on @xmath470 , but it is not @xmath471 .
the intrinsic graph of @xmath38 over the @xmath470 is depicted in figure [ fig1 ] .
further examples of similar flavour have been discussed in @xcite in connection with minimal surfaces in the heisenberg group .
we emphasise that the graphs of non - affine solutions to @xmath465 have plenty of non - zero @xmath3-numbers , but this behaviour is not registered by the fluctuation of @xmath462 .
what can be done ? we still want to use the intrinsic gradient , so we need to invent a condition to replace `` non - affinity '' , which forces @xmath462 to fluctuate locally .
it turns out that the right notion is `` being far from constant - gradient ( cg ) graphs '' .
the following ( informal ) statement may sound almost tautological , but it requires a fair amount of work to verify : if @xmath38 deviates locally from all ( locally defined ) cg intrinsic lipschitz functions , then @xmath462 must fluctuate noticeably . following the euclidean idea ,
this observation ( made precise in proposition [ dreamprop ] ) allows us to conclude that intrinsic lipschitz graphs satisfy a `` wgl for cg @xmath3-numbers '' , see theorem [ t : cg - wgl ] . up to this point
, the results can be accused of being abstract nonsense ; as far as we know , locally defined cg graphs could be quite wild , and we do not even claim to understand them very well .
what we can understand , however , are * globally defined * cg graphs .
these turn out to be affine ( see proposition [ constantgradientgraphs ] ) ! using this fact , and a compactness argument , we can prove that even locally defined cg graphs have the following key property : if @xmath129 is a cg graph `` defined in the whole ball @xmath70 '' , then @xmath129 is almost flat in all sufficiently small sub - balls of @xmath70 .
this implies almost immediately that the usual wgl holds for cg - graphs , even if we do not record the argument separately .
instead , we use our `` approximation by cg graphs '' result to conclude directly that the wgl holds for all intrinsic lipschitz graphs .
we wish to mention that the proof strategy above was influenced by x. tolsa s proof @xcite of the fact that the _ weak constant density condition _ implies _ uniform rectifiability _ in @xmath4 .
should the reader be familiar with that proof , she may wish to draw the following parallels in her mind : `` weak constant density '' is represented by `` intrinsic lipschitz '' , and `` uniform measure '' is represented by `` constant gradient graph '' .
according to a well known theorem by rademacher , lipschitz maps between euclidean spaces are differentiable almost everywhere .
the same result , appropriately interpreted , holds true for the intrinsic lipschitz functions appearing in this paper .
similarly as in the euclidean setting , a function @xmath157 shall be differentiable at a point @xmath472 , if its graph at @xmath473 can be well approximated by the graph of a `` linear '' function @xmath474 . following the terminology in @xcite , a function @xmath475 between complementary homogeneous subgroups
is said to be _ intrinsic linear _ , if its intrinsic graph @xmath476 is a homogeneous subgroup of @xmath0 .
if @xmath32 is a vertical subgroup and @xmath89 is a complementary orthogonal horizontal subgroup in @xmath0 , a map @xmath477 is intrinsic linear if and only if it is a _ homogeneous homomorphism _ ( see for instance proposition 3.26 in @xcite ) .
the latter means that @xmath124 is a group homomorphism with the additional property that @xmath478 for all @xmath479 and @xmath198 . to give an example , if @xmath32 denotes the @xmath33-plane , it is not difficult to see that all intrinsic linear maps , or equivalently all homogeneous homomorphisms , are of the form @xmath480 for a constant @xmath481 .
we are now ready to state the definition of _ intrinsic differentiability_. again we assume that @xmath32 is a vertical subgroup with complementary horizontal subgroup @xmath89 .
first let us consider a function @xmath482 defined on an open subset of @xmath464 containing the origin , which we assume to be fixed under @xmath38 .
we say that @xmath38 is intrinsic differentiable at the origin , if there exists an intrinsic linear map @xmath474 such that , for all @xmath479 , @xmath483 the map @xmath124 is called the _ intrinsic differential _ of @xmath38 at @xmath381 and it is denoted by @xmath484 .
since the definition of differentiability is supposed to be _ intrinsic _ , we extend it in a left invariant fashion to arbitrary functions and points . to explain the definition , we consider a function @xmath485 on an open set @xmath464 with intrinsic graph @xmath129 . for @xmath486 ,
we write @xmath487 and let @xmath488 be the uniquely defined function @xmath156 , which parametrizes the graph @xmath489 , see remark [ remk : para ] .
note that the definition of intrinsic graph ensures that any left translate of @xmath129 by a point in @xmath129 is an intrinsic graph passing through the origin .
this uniquely determines the function @xmath488 and it ensures that the origin is fixed under this function .
we also formulate an equivalent definition of intrinsic lipschitz functions using the maps @xmath490 . if @xmath126 , a function @xmath491 with intrinsic graph @xmath129 is an intrinsic @xmath124-lipschitz function if and only if for every @xmath492 @xmath493 for all @xmath494 in the domain of @xmath495 .
see also ( * ? ? ? * proposition 4.49 ) for other equivalent algebraic definitions for intrinsic lipschitz functions .
an explicit formula for @xmath496 is given in lemma [ l : transl ] below .
we say that a function @xmath485 , defined on an open set @xmath464 , is _ intrinsic differentiable _ at a point @xmath497 , if the function @xmath488 for @xmath487 is intrinsic differentiable at the origin .
the _ intrinsic differential _ of @xmath38 at @xmath497 is given by @xmath498 intrinsic differentiability can be characterized in various equivalent ways , see for instance the results in @xcite , @xcite , and @xcite .
one can also define intrinsic differentiability in terms of a `` graph distance '' , see definition 1.4 in @xcite , and for intrinsic lipschitz maps this definition is equivalent to the one above .
the intrinsic differential is unique , and its action can be expressed in terms of a `` gradient '' similarly as in the euclidean case . to explain this
, we identify @xmath499 and @xmath500 .
points in @xmath32 are then denoted by @xmath33 .
assume that a map @xmath501 is intrinsic differentiable at @xmath502 .
then , as pointed out earlier , the intrinsic differential @xmath503 of @xmath38 at @xmath504 is a linear mapping of the form @xmath480 for some @xmath505 .
we next derive a few useful formulae for @xmath507 .
let @xmath508 be the point on the graph of @xmath38 .
by definition of intrinsic differentiability , @xmath509 consequently , dividing by @xmath510 , we have @xmath511 moreover , wherever @xmath38 is differentiable in the usual ( euclidean ) sense , the formula above , and @xmath512 ( this follows from the formula for @xmath513 in lemma [ l : transl ] ) yields the following representation for @xmath514 : @xmath515 this was observed in example 5.5 in @xcite . a large class of almost everywhere intrinsic differentiable functions is provided by intrinsic lipschitz functions whose target is a @xmath25-dimensional horizontal subgroup . this result
was first proved by franchi , serapioni and serra cassano in ( * ? ? ?
* theorem 4.29 ) for heisenberg groups , and later by franchi , marchi and serapioni in @xcite for certain more general carnot groups .
we state here the result relevant for the current paper : [ t : intrradem ] let @xmath32 be a vertical subgroup of @xmath0 with complementary horizontal subgroup @xmath89 .
assume that @xmath516 is an open subset of @xmath32 and @xmath485 is intrinsic lipschitz .
then @xmath38 is intrinsic differentiable @xmath93 almost everywhere in @xmath32 .
it follows that the intrinsic gradient @xmath514 of an intrinsic lipschitz function @xmath482 exists almost everywhere in @xmath516 , and it is an @xmath517 function , see for instance proposition 4.4 in @xcite . in our situation
, one can say something more precise : [ l : bound_grad ] let @xmath32 be a vertical subgroup with complementary horizontal subgroup @xmath89 in @xmath0 , and let @xmath518 be an intrinsic @xmath124-lipschitz function on an open set @xmath464 .
then @xmath519 by theorem [ t : intrradem ] , @xmath38 is intrinsic differentiable in almost every point of @xmath516 .
we let @xmath497 be such a point , and we write @xmath520 for the corresponding point on the graph . recall that the function @xmath496 in is defined as the function
whose graph is the left translate of the graph of @xmath38 by @xmath521 .
we denote the domain of @xmath496 by @xmath522 . in order to prove
, it suffices to find a bound for the limit in . to this end , let @xmath510 be small enough such that @xmath523 . then by @xmath524 here we have used
. thus @xmath525 . since @xmath497 was an arbitrary point of intrinsic differentiability , follows by theorem [ t : intrradem ] . in the converse direction ,
proposition 1.8 in @xcite provides local upper bounds for the lipschitz constant in terms of the @xmath517-norm of the intrinsic gradient .
while these results connect @xmath514 to the geometry of intrinsic lipschitz graphs , the intrinsic gradient has a life of its own outside the world of intrinsic lipschitz maps . to see this
, it is best to express @xmath514 as in .
the equation @xmath526 is well known in pde theory as the _ inviscid burgers equation_. this will be discussed further in section [ s : constantgradientgraphs ] below .
in which sense are intrinsic lipschitz functions solutions to an equation of burgers type ?
by theorem [ t : intrradem ] , the intrinsic gradient of an intrinsic lipschitz function exists pointwise almost everywhere . in connection with pde theory ,
it is useful to know that the pointwise intrinsic gradient of an intrinsic lipschitz function is also a _ distributional _ gradient .
this is the content of proposition 4.7 in @xcite .
precisely , if @xmath157 is intrinsic lipschitz with intrinsic gradient @xmath514 , defined almost everywhere by theorem [ t : intrradem ] , then holds in a distributional sense : @xmath527 for all @xmath528 .
we emphasize that the intrinsic lipschitz functions as in definition [ d : intrlip ] coincide with the intrinsic lipschitz functions in the sense of @xcite , see theorem 4.60 in @xcite and remark 3.6 in @xcite .
note further that the formula looks slightly different from @xcite due to a different model for the heisenberg group , see definition 3.1 in @xcite .
while the definition of intrinsic lipschitz continuity is tailored so that the class of intrinsic @xmath124-lipschitz graphs is preserved under dilations and translations in the heisenberg group , the explicit formula for the parametrization of a translated graph becomes in general slightly complicated due to the non - commutativity of the group law . in the case we consider in the present paper : functions from vertical to horizontal subgroups in @xmath529 , the computations are straightforward . for convenience ,
given a point @xmath197 , we define the map @xmath530 we note that @xmath531 is a diffeomorphism with jacobian determinant constant equal to @xmath25 under the obvious identification of @xmath32 with @xmath91 ( see ( * ? ? ?
* lemma 2.20 ) ) , and with inverse map @xmath532 .
the latter claim follows from the fact that @xmath533 [ l : transl ] let @xmath129 be the intrinsic graph of a function @xmath534 on a vertical subgroup @xmath32 , and let @xmath516 be a domain in @xmath0 .
then , for @xmath535 , the set @xmath536 is the intrinsic graph of the function @xmath537 if @xmath38 is intrinsic @xmath124-lipschitz , then so is @xmath538 with @xmath539 analogously , for @xmath198 , the set @xmath540 is the intrinsic graph of the function @xmath541 if @xmath38 is intrinsic @xmath124-lipschitz , then so is @xmath542 with @xmath543 we concentrate on proving the statement for left translations . in (
* proposition 2.7 ) it has been shown that if @xmath129 is the intrinsic graph of a function @xmath38 over a domain in @xmath89 , then @xmath546 is an intrinsic graph parametrized by the function @xmath538 .
the domain of the new function @xmath538 is simply the image of @xmath546 under the projection onto @xmath32 .
the intrinsic lipschitz property of @xmath538 assuming the corresponding property for @xmath38 is the content of ( * ? ? ?
* theorem 3.2 ) . to compute the intrinsic gradient
, we may assume without loss of generality that @xmath547 agrees with the @xmath33-plane and that @xmath38 is defined on the entire plane @xmath547 .
in this case , for @xmath548 , we have that @xmath549 we then use the fact , proved in @xcite , that the intrinsic gradient of an intrinsic lipschitz function is also a distributional gradient ; see the discussion in section [ ss : intr_grad ] .
let now @xmath550 be an arbitrary test function , that is , a compactly supported @xmath15 function on @xmath547 .
since @xmath551 we find that @xmath552 this , together with the facts that @xmath553 and @xmath514 is a distributional gradient , recall , gives @xmath554 as this computation is valid for arbitrary test functions @xmath550 , the claim follows . in this subsection , we prove that `` entire '' intrinsic lipschitz functions with almost surely constant gradient are affine .
as mentioned in section [ ss : intr_grad ] , if @xmath32 is identified with @xmath91 , the differential equation @xmath555 is known as the inviscid burgers equation and it is not difficult to see by the _ method of characteristics _ ( see ( * ? ? ?
* proposition 5.1 ) ) that the only global @xmath15 solutions are constant functions .
if the right - hand side of the equation is replaced by some other constant @xmath255 , one can show in the same vein that the only @xmath471 solutions are affine functions of the form @xmath556 ; see ( * ? ? ?
* remark 4.3 ) .
[ constantgradientgraphs ]
let @xmath157 be an intrinsic lipschitz function .
if there exists a constant @xmath557 such that @xmath558 almost everywhere in @xmath32 , then the graph of @xmath38 is the left translate of some vertical plane @xmath559 . throughout the proof
, we identify @xmath32 with @xmath91 , using coordinates @xmath33 .
we will prove that @xmath560 for some @xmath561 . since @xmath38 is intrinsic lipschitz ,
it is continuous , see for instance ( * ? ? ?
* proposition 3.4 ) .
we start by observing that the almost sure constancy of the intrinsic gradient leads to improved regularity for @xmath38 . by (
* proposition 4.7 ) , the function @xmath38 is a distributional solution to the equation @xmath562 for @xmath563 . since @xmath564 is constant , it is in particular lipschitz continuous in the euclidean sense and ( * ? ? ?
* corollary 1.4 ) implies that @xmath38 is locally lipschitz on @xmath32 with respect to the euclidean metric ; see also theorem 4.2.1 and theorem 4.4.1 in @xcite .
hence , almost every point @xmath494 of @xmath32 is `` good '' in the sense that the function @xmath38 is differentiable at @xmath494 both in the usual euclidean sense and in the intrinsic sense with @xmath565 .
we denote by @xmath566 the set of such good points in @xmath32 , so that @xmath567 .
let us assume for a moment that @xmath568 is such that @xmath572 for almost every @xmath573 . towards a proof of
, we define the function @xmath574 we note that @xmath575 is locally lipschitz continuous , so @xmath576 exists for almost all @xmath72 .
further , by the assumption that @xmath577 for almost every @xmath72 , we have for such points that @xmath578\partial_t \phi(\gamma_t(s))\\ & = -z(s ) \partial_t \phi(\gamma_t(s)).\end{aligned}\ ] ] here we have used that @xmath579 on @xmath566 , see and the subsequent discussion .
thus @xmath575 solves an ode of the form @xmath580 clearly , @xmath581 is a solution , but we have to argue that it is the _ only _ solution .
here we are interested in _ carathodory solutions _
@xmath582 , that is , in functions which are absolutely continuous on every closed interval @xmath583\subset { \mathbb{r}}$ ] and which fullfill the differential equation pointwise almost everywhere ; see for instance ( * ? ? ?
* chapter 1 ) for a thorough discussion of carathodory differential equations .
since @xmath38 is locally lipschitz as a function on the euclidean plane and @xmath584 is a smooth curve with @xmath572 for almost every @xmath573 , the function @xmath585 is locally integrable on every interval @xmath583\subset { \mathbb{r}}$ ] . by theorem 3 in (
* chapter 1 ) , this suffices to ensure that the ode has a unique carathodory solution on @xmath586 .
hence , @xmath581 and follows for this particular choice of @xmath30 .
next , we would like to show that almost every @xmath568 has the crucial property that @xmath572 for almost every @xmath72 .
this is the content of lemma [ avoidance ] below
. the statement would be immediate if we knew that the curves @xmath587 foliated the plane @xmath32 , or even a large portion thereof , but there is no such _ a priori _ information available . in fact , this foliation property is part of the statement we want to prove .
so , we have to work a bit harder , and we are essentially rescued by the local lipschitz regularity of @xmath588 . in the proof of lemma [ avoidance ] we need a sharpened version of the `` easy implication '' in the besicovitch projection theorem .
this result may be known to some experts , but we did not find it in the literature : we start with a series of reductions . without loss of generality we can assume that @xmath592 is bounded .
moreover it is enough to prove the lemma for @xmath602)$ ] for @xmath603 lipschitz and @xmath190 . to see this , first
recall that if @xmath592 is a bounded rectifiable set there exist countably many lipschitz maps @xmath604 such that @xmath605))=0 $ ] for some @xmath190 . applying the lemma to each of the sets @xmath606)$ ]
we obtain sets @xmath607 satisfying ( i ) and ( ii ) .
now let @xmath608 .
trivially @xmath593 and if @xmath594 is any @xmath595-measurable subset with @xmath609 , then there exists some @xmath610 such that @xmath611 . since @xmath610 satisfies ( ii ) , if @xmath612 then @xmath598 .
fix @xmath189 .
it is enough to find a subset @xmath613 such that @xmath614 and the following property holds : if @xmath615 is measurable with @xmath616 and @xmath617 , then @xmath618 .
then , we can complete the proof by setting @xmath619 . by (
* theorem 7.4 ) there exists a compact @xmath471-curve @xmath620 such that @xmath621 .
then , if @xmath615 is measurable with @xmath622 , we have @xmath623 .
thus , it actually suffices to construct @xmath624 so that the following holds : if @xmath625 is measurable with @xmath626 and @xmath627 , then @xmath618 . one final reduction : for fixed @xmath628 , we construct a set @xmath629 with the properties that ( a ) @xmath630 , and ( b ) if @xmath625 is measurable with @xmath626 and @xmath631 , then @xmath618 .
this suffices , since @xmath632 is then the set we are after . to construct @xmath629
, we fix a number @xmath633 , to be specified later , and cover @xmath634 by a a collection @xmath635 of disjoint arcs of length between @xmath636 and @xmath637 .
next , for some @xmath638 depending on @xmath5 , we partition @xmath129 into short , connected sub - curves @xmath639 such that the following holds : for every fixed @xmath640 , the restriction @xmath641 is bi - lipschitz for all @xmath642 , except possibly those @xmath643 in the union of four arcs in @xmath644 ( depending only on @xmath645 ) .
such a partition @xmath646 exists , because @xmath129 is compact and @xmath471 .
consider a bi - partite graph with vertex set @xmath647 and the following edge set @xmath57 : draw an edge between @xmath640 and @xmath648 , if and only if @xmath641 is bi - lipschitz for * all * @xmath649 .
thus , every vertex @xmath646 is adjacent to at least @xmath650 vertices in @xmath644 . for an edge @xmath651 ,
define the weight @xmath652 thus , if @xmath653 is the sum of all the weights of edges in @xmath57 , we have @xmath654 now , write @xmath655 and call a vertex @xmath648 _ light _ , if the total weight of edges emanating from @xmath656 is at most @xmath657 .
other vertices in @xmath644 are _
denoting the light and heavy vertices in @xmath644 by @xmath658 and @xmath659 , respectively , we have @xmath660 .
we now fix @xmath5 so large that @xmath661 , which gives @xmath662
. then , let @xmath663 the set @xmath629 satisfies the correct length estimate : @xmath664 by the choice of @xmath665 . finally , we want to show that @xmath618 , whenever @xmath625 is measurable with @xmath666 , and @xmath667 .
so , fix @xmath668 with @xmath669 , and write @xmath670 . then @xmath671 then , fix @xmath667 , so that @xmath672 for some @xmath659 .
this implies that @xmath656 is adjacent to at least one vertex @xmath673 ; otherwise , recalling that @xmath674 , we have @xmath675 which contradicts @xmath676 .
now , pick @xmath673 such that @xmath677 . by definition of @xmath57
, this means that @xmath641 is bi - lipschitz , and consequently @xmath678 the proof is complete .
with @xmath683 , we may re - write the claim as follows : @xmath684 so , it suffices to show that for almost every @xmath685 , we have @xmath686 , where @xmath687 assume that this claim is false : there exists a positive measure set @xmath421 of parameters @xmath72 such that @xmath688 . observe that @xmath689 for @xmath422 , where @xmath690 next , write @xmath691 for @xmath692 ; then , up to scaling , @xmath693 is the orthogonal projection onto the line spanned by @xmath694 in the @xmath33-plane . recalling remark [ sfinite ] we can apply lemma [ projectionlemma ] to the lipschitz graph @xmath129 and
obtain a set of parameters @xmath695 with @xmath696 , with the property that @xmath697 , whenever @xmath698 and @xmath699 .
in particular , @xmath697 for almost all @xmath422 . observing that @xmath700 for every @xmath72 , this forces @xmath701 for almost all @xmath422 , which contradicts @xmath702 .
the proof of the lemma is complete .
we have now established that holds for almost every @xmath703 , and the rest of the proof of proposition [ constantgradientgraphs ] is easy .
first , notice that @xmath704 now , if @xmath705 for some @xmath706 , then @xmath707 for some @xmath708 .
it follows that for such @xmath706 , the curves @xmath584 and @xmath709 intersect at @xmath710 .
recall that we aim to show that @xmath711 for some @xmath712 .
we first show that @xmath588 is constant .
pick @xmath30 and @xmath713 satisfying .
if @xmath714 , then by the discussion in the previous paragraph , @xmath715 for some @xmath716
. consequently , @xmath717 by , which contradicts @xmath718 .
so , @xmath588 is constant , say @xmath719 , on the set where holds . referring again to
, we find that , for @xmath319 almost all @xmath720 , we have @xmath721 since @xmath38 is continuous , this is in fact true for all pairs @xmath722 , and hence @xmath723 for all @xmath724 .
the proof is complete . in this section ,
we start to implement the plan outlined in section [ outline ] : we define a variant of @xmath3-numbers , the _ constant gradient @xmath3-numbers _ , and prove that intrinsic lipschitz graphs satisfy a weak geometric lemma with respect to this new definition .
if @xmath129 is an intrinsic lipschitz graph over @xmath32 , the constant gradient @xmath3-number ( with parameter @xmath124 ) of a ball @xmath70 is designed to described how well @xmath725 can be approximated by the graph of an intrinsic @xmath124-lipschitz function whose gradient is constant almost everywhere in @xmath726 . here
@xmath727 is a small constant , given by the following lemma , and @xmath70 denotes a closed ball with radius @xmath728 centred at @xmath45 .
[ eq : ball_inclusion ] for every @xmath729 there exists a constant @xmath727 such that if @xmath157 is an intrinsic @xmath124-lipschitz function on a vertical subgroup @xmath32 , then @xmath730 for all @xmath45 on the graph @xmath129 of @xmath38 and for all @xmath198 .
let @xmath731 be an intrinsic graph , where @xmath732 is an intrinsic lipschitz function .
fix a point @xmath452 and a radius @xmath147 . then , for @xmath172 , define @xmath733 the infimum is taken over all intrinsic @xmath124-lipschitz functions @xmath734 which have intrinsic gradient constant almost everywhere on the set @xmath726 .
the class of such `` admissible '' functions @xmath550 will be denoted by @xmath735 if the lipschitz constant @xmath124 is clear from the context , we omit the subscript @xmath124 for the constant @xmath727 . note that @xmath736 for @xmath737 by lemma [ l : bound_grad ] . as explained in section [ outline ] , a large @xmath749 number implies that @xmath462 fluctuates locally .
more precisely , proposition [ dreamprop ] will show that if @xmath750 for some @xmath751 and @xmath752 , then there exists another ball @xmath753 such @xmath754 , @xmath755 and @xmath756 . in section [ ss : wgl_cg ]
, we use this to prove theorem [ t : cg - wgl ] . before stating proposition [ dreamprop ] , we record a few lemmas .
the first one gives an upper bound on how much @xmath757 can change as a function of @xmath728 .
here and in the following , we employ the notation @xmath758 for the average of a function @xmath46 over a set @xmath201 in the plane .
[ annuluscontrol ] assume that @xmath129 is an intrinsic lipschitz graph defined over @xmath32 , and that @xmath759 .
further , let @xmath145 , and @xmath760 .
then , @xmath761 where @xmath762 is the annulus @xmath763 .
write @xmath764 and @xmath765 . in this proof , let @xmath766 for @xmath767
. then , @xmath768 by the triangle inequality .
next , we observe that @xmath769 \cap \gamma ) \subset \pi_{{\mathbb{w}}}(a(x , s_{1},s_{2 } ) \cap \gamma)\ ] ] by the injectivity of @xmath109 restricted to @xmath129 . finally , we use the fact that @xmath770 , recall lemma [ lipschitzlemma ] . [
l : a - small ] let @xmath771 be metric measure space . for any @xmath776 , there exists a constant @xmath777 with the following property : for any ball @xmath70 with @xmath778 and @xmath147 , there exists a radius @xmath779 $ ] such that @xmath780 has @xmath201-thin boundary . for a fixed @xmath781 , let @xmath782 be the mapping @xmath783 , and
consider the push - forward measure @xmath784 $ ] .
let @xmath785 be the usual centred hardy - littlewood maximal operator on @xmath586 , namely @xmath786 we extend the definition from @xmath46 to @xmath787 in the obvious way .
it is well - known ( see for instance ( * ? ? ?
* theorem 2.19 ) ) that @xmath785 is weakly bounded in the sense that @xmath788 in particular , @xmath789 for @xmath790 large enough , this implies that there is some @xmath779 $ ] such that @xmath791 recalling the definition of @xmath787 this means precisely that @xmath792 for all such @xmath793 that @xmath794 .
since @xmath795 , this covers all @xmath796 . for @xmath797 ,
the thin boundaries condition is trivial .
[ dreamprop ] for every @xmath189 and @xmath172 there exist constants @xmath798 and @xmath799 with the following property .
assume that @xmath501 is an intrinsic @xmath124-lipschitz function with graph @xmath129 .
assume that @xmath452 and @xmath147 are such that @xmath800 then , there exists a ball @xmath753 with @xmath201-thin boundary ( with respect to @xmath801 ) such that @xmath802 , @xmath754 , and @xmath803 in particular , @xmath804 [ remark0 ] observe that the last statement is an immediate consequence of @xmath226 being contained in the @xmath805-neighborhood of @xmath129 ; this implies that @xmath806 contains a set of the form @xmath807 with @xmath808 and @xmath809 .
further , for such balls @xmath810 centred on @xmath129 , we can apply lemma [ eq : ball_inclusion ] to find @xmath811 for some @xmath812 .
finally , @xmath813 with @xmath814 , recalling that the mapping @xmath815 , has unit jacobian for any @xmath101 , see .
see also ( * ? ? ?
* lemma 3.14 ) .
* we formulate a counter assumption to the main claim in proposition [ dreamprop ] . assuming the validity of this assumption , we find @xmath816 and a sequence of intrinsic @xmath124-lipschitz functions @xmath817 and associated graphs @xmath818 , such that @xmath819 has large @xmath749 number in a ball @xmath820 centred on @xmath819 , yet @xmath821 does not fluctuate much in that ball .
we use a blow - up procedure to normalize so that we may assume @xmath822 for all @xmath419 . *
we show that a subsequence of @xmath817 converges locally uniformly to an intrinsic @xmath124-lipschitz function @xmath38 with graph @xmath129 such that , roughly speaking , + ( i ) @xmath823 is large , + ( ii ) @xmath824 is independent of @xmath825 , @xmath826 . *
we show that the conditions ( i ) and ( ii ) are incompatible , which concludes the proof of proposition [ dreamprop ] .
denote by @xmath827 the metric @xmath440-neighborhood of a set @xmath57 .
given an intrinsic lipschitz graph @xmath129 , a ball @xmath70 with @xmath452 , and @xmath828 , define the following collection of `` good '' balls @xmath829 .
a ball @xmath830 is in @xmath831 , if [ eq : a ] for every @xmath189 , @xmath172 , there exists @xmath836 with the following property .
if @xmath129 is any intrinsic @xmath124-lipschitz graph and @xmath70 is centred on @xmath129 with @xmath750 , then there exists a ball @xmath837 such that @xmath838 [ l : normalization ] if claim [ eq : a ] fails , then there there exist @xmath840 and @xmath172 such that for every @xmath828 we can find an intrinsic @xmath124-lipschitz graph @xmath841 , parametrised by @xmath842 such that @xmath843 yet @xmath844 by definition , if claim [ eq : a ] fails , then there exist @xmath189 and @xmath172 such that for every @xmath828 we can find an intrinsic @xmath124-lipschitz graph @xmath645 , parametrised by @xmath842 , and _ some ball _ @xmath845 centred on @xmath645 such that @xmath846 yet @xmath847 * is again an intrinsic @xmath124-lipschitz graph ( lemma [ l : transl ] ) , * has @xmath749-number at least @xmath55 on @xmath851 ( lemma [ l : beta ] below ) , * is parametrised by a function @xmath852 so that holds with @xmath853 in place of @xmath854 .
( lemma [ l : average ] below ) . according to lemma [ l : transl ] and remark [ remk : para ] there exists a uniquely defined intrinsic lipschitz function @xmath857 that parametrizes @xmath858 . by definition ,
@xmath859 we first aim to prove that the family @xmath860 is in @xmath25-to-@xmath25 correspondence with the family @xmath861 .
assume that @xmath744 .
define @xmath862 and note that @xmath863 in the notation of lemma [ l : transl ] .
let us prove that @xmath864 . to this end , we observe first that @xmath865 see for instance ( * ? ? ?
* proposition 2.15 ) . by homogeneity and left invariance of the distance @xmath75 , it follows that @xmath866 and @xmath867 lemma [ l : transl ] then implies that @xmath864 .
[ l : average ] assume that @xmath157 is an intrinsic lipschitz function with graph @xmath129 , @xmath45 is a point on @xmath129 , and @xmath198 .
then @xmath871 and @xmath872 here @xmath873 is the graph parametrized by @xmath857 ( defined as in ) .
we start with the first claim .
since the heisenberg distance is left invariant with respect to the group law , and homogeneous with respect to the dilations @xmath874 , if a ball @xmath834 has @xmath201-thin boundary , then so does the ball @xmath875 .
the remaining conditions that one has to verify in order to prove are also immediate .
regarding , we first recall that lemma [ l : transl ] yields that @xmath876 since @xmath877 restricted to @xmath32 has jacobian determinant constant equal to @xmath878 , and @xmath879 has jacobian determinant equal to @xmath25 , it follows by the usual transformation formula for functions on @xmath91 that @xmath880 and @xmath881 this establishes . in this section , we work under the standing ( counter ) assumption to proposition [ dreamprop ] .
in particular , we may assume by lemma [ l : normalization ] that there exists @xmath882 and a sequence @xmath883 of intrinsic @xmath124-lipschitz functions with graphs @xmath818 such that @xmath884 , yet the intrinsic gradient @xmath821 fluctuates only little in @xmath885 as quantified in .
[ eq : limit_beta ] the sequence @xmath817 defined above contains a subsequence that converges locally uniformly on @xmath32 to an intrinsic @xmath124-lipschitz function @xmath157 with graph @xmath129 such that @xmath887 since each graph @xmath645 by construction contains the origin , we have @xmath888 for every @xmath419 . therefore [ intrili2 ] implies that the family @xmath889 is locally equibounded .
hence , by proposition 3.10 in @xcite , there exists a subsequence which converges locally uniformly to an intrinsic @xmath124-lipschitz function @xmath38 on @xmath32 with intrinsic graph @xmath129 . for simplicity
, we also denote this subsequence by @xmath883 .
let @xmath890 .
in order to prove that @xmath891 , it suffices to fix @xmath892 and find a point @xmath893 with @xmath894 . to this end
, the assumption @xmath895 implies that , for each @xmath828 , there exists a point @xmath896 such that @xmath897 write @xmath898 . the sequence @xmath899 has a subsequence @xmath900 convergent to a point @xmath901 .
since @xmath109 is continuous , the points @xmath902 converge to @xmath903 .
finally , @xmath904 by the continuity of @xmath550 , and the locally uniform convergence @xmath905 .
the proof is complete . without loss of generality , we assume in the following that the whole sequence @xmath889 converges locally uniformly to @xmath38 .
our next goal is to prove the following convergence result for the corresponding intrinsic gradients .
[ l : conv_grad ] let @xmath906 be a sequence of intrinsic @xmath124-lipschitz functions converging locally uniformly to an @xmath124-lipschitz function @xmath157 .
then @xmath907 for all balls @xmath908 with @xmath826 , and such that @xmath909 .
the subtle point here is , of course , that the domain of integration is different on both sides of the above equation .
therefore the following auxiliary result will be useful in the proof of lemma [ l : conv_grad ] .
[ annuluscontrol2 ] assume that @xmath645 is a sequence of intrinsic @xmath124-lipschitz graphs , which converges locally in the hausdorff metric in @xmath0 to an intrinsic lipschitz graph @xmath129 .
then , for any @xmath452 and @xmath910 , we have @xmath911 fix @xmath890 .
first , pick @xmath912 and @xmath913 so that @xmath914 then , let @xmath189 be so small that if @xmath915 , then @xmath916 .
let @xmath917 be an @xmath55-net in @xmath918 , and , finally , let @xmath919 be the subset of points @xmath226 in this net with the property that @xmath920 contains a point in @xmath921 .
then @xmath922 by the @xmath2-regularity of intrinsic @xmath124-lipschitz graphs .
for @xmath419 large enough , every ball @xmath923 with @xmath924 also contains a point in @xmath129 , whence @xmath925 , @xmath924 .
since the balls @xmath926 have bounded overlap ( independent of @xmath189 ) , we conclude that @xmath927 for all large enough @xmath419 .
this completes the proof .
we employ the fact proven in ( * ? ? ?
* proposition 4.7 ) that the intrinsic gradient is also a distributional gradient for intrinsic lipschitz functions : if @xmath38 is intrinsic lipschitz and defined on @xmath32 , then @xmath928 \psi \ , d{\mathcal{l}}^{2}\ ] ] for all compactly supported @xmath471-functions @xmath550 on @xmath32 ( here we assume that @xmath32 is the @xmath33-plane , as we may ) . since uniform convergence implies weak convergence , we infer that @xmath929 \psi \ , d{\mathcal{l}}^{2 } = \lim_{j \to \infty } \int_{{\mathbb{w } } } [ \nabla^{\phi_{j}}\phi_{j } ] \psi \ , d{\mathcal{l}}^{2}\ ] ] for all compactly supported @xmath471 functions @xmath930 .
it remains to deduce from this statement the claim . to achieve this
, we recall from lemma [ l : bound_grad ] that @xmath931 then , given @xmath908 with @xmath826 , for @xmath331 , choose an open set @xmath932 in @xmath933 so that @xmath934 and @xmath935 .
in particular , @xmath936 for all test functions @xmath550 with @xmath937 and all @xmath938 .
now , let @xmath550 be a smooth cut - off function with compact support in @xmath932 , @xmath937 on @xmath932 and such that @xmath939 on @xmath940 .
then , by , there exists @xmath941 such that for @xmath942 , one has @xmath943 let us denote @xmath944 . combining and , and recalling that @xmath550 is supported on @xmath932 , we find that @xmath945 for @xmath942 .
this yields @xmath946 to establish , it remains to show that @xmath129 on the right hand side of the above equation can be replaced by @xmath819 .
this follows immediately from the uniform bound for @xmath514 and @xmath821 , see , provided we can show the following : @xmath947 we now prove .
here we need to assume that @xmath909 . since @xmath948 there exists for every @xmath169 a number @xmath949 such that @xmath950 let @xmath951 be such and observe that @xmath952 and @xmath953 for @xmath419 larger than some @xmath954 given by uniform convergence .
so @xmath955 for @xmath956 .
since @xmath957 by the general inequality @xmath958 ( lemma [ lipschitzlemma ] ) and by , we infer that @xmath959 as @xmath960 .
it remains to prove the same with the roles of @xmath129 and @xmath645 reversed . repeating the argument above
, it suffices to estimate the measure of the projection @xmath961 .
this is not altogether trivial , since the assumption on @xmath962 concerns @xmath963 , not @xmath964 .
however , lemma [ annuluscontrol2 ] still implies that @xmath965 which proves that @xmath966 since @xmath890 is arbitrary , , and hence also , follows .
this concludes the proof of lemma [ l : conv_grad ] .
we now return to the specific sequence @xmath883 and limit function @xmath38 ( with graph @xmath129 ) given by lemma [ eq : limit_beta ] .
in particular , we work under the standing assumptions that @xmath967 and @xmath821 fluctuates only little on @xmath968 , as made precise in lemma [ l : normalization ] .
pick two balls @xmath973 and @xmath974 with @xmath975 satisfying the assumption stated in the lemma , and let @xmath976 be arbitrary .
our goal is to show that @xmath977 we start by applying the triangle inequality : @xmath978 where @xmath979 to see that @xmath980 for large enough @xmath419 , writing temporarily @xmath766 for @xmath767 , and @xmath981 we perform the following estimate : @xmath982 the same is true with `` 1 '' replaced by `` 2 '' . by and , and the uniform bound @xmath983 , both terms above tend to zero as @xmath960 .
so , it suffices to deal with @xmath984 .
we first make the subsequent estimate for certain @xmath985 and @xmath986 to be fixed soon : @xmath987 where @xmath988 it follows from the key ( counter ) assumption that @xmath989 if the balls @xmath990 and @xmath991 belong to @xmath992 .
this requires that condition ( a ) is automatically satisfied for large enough @xmath419 , since @xmath998 .
condition ( b ) is also trivially satisfied for large enough @xmath419 , since @xmath985 and @xmath986 . to achieve condition ( c ) , fix an auxiliary parameter @xmath999 , which will depend on @xmath1000 and @xmath124
then , for large enough @xmath419 , lemma [ l : a - small ] guarantees the existence of @xmath1001 $ ] and @xmath1002 $ ] such that ( c ) is satisfied for @xmath990 and @xmath991 .
we choose @xmath1003 accordingly , and then @xmath1004 ( for @xmath1005 ) .
it remains to estimate @xmath1006 and @xmath1007 ; by symmetry , we may concentrate on @xmath1006 .
recalling lemma [ annuluscontrol ] , we first have @xmath1008 since @xmath1009 , and @xmath1010 , see remark [ remark0 ] .
lemma [ annuluscontrol2 ] then tells us that @xmath1011 and now we choose @xmath1012 so small that the right hand side above is smaller than @xmath1013 for some small constant @xmath1014 depending only on @xmath124 , and the same holds with `` @xmath25 '' replaced by `` @xmath92 '' .
if @xmath1014 is small enough , this implies that @xmath1015 . combining the estimates for @xmath1016 and @xmath1017 above ,
we conclude that @xmath1018 since @xmath890 was arbitrary , this proves that @xmath1019 is independent of the ball @xmath908 with @xmath1020 and satisfying @xmath1021 . in this section ,
we conclude the proof of proposition [ dreamprop ] . under the standing counter assumption derived in lemma [ l : normalization ]
, we will infer from lemma [ l : a.s.const ] that @xmath462 is a.e .
constant on @xmath1022 .
this will contradict the conclusion of lemma [ eq : limit_beta ] , which gave @xmath891 .
recall that we have shown in lemma [ l : a.s.const ] that the averages of @xmath514 are the same for a large class of sets in @xmath32 . in order to conclude that then @xmath514 must be constant almost everywhere on @xmath1024 , and in particular on @xmath1025
, we apply a lebesgue differentiation theorem for the measure space @xmath1026 endowed with the _ graph distance _ @xmath1027 .
the latter can be defined for an arbitrary intrinsic lipschitz function @xmath1028 with graph @xmath129 by setting @xmath1029 we apply this theorem for @xmath1033 .
more precisely , we choose for every lebesgue point @xmath497 of @xmath1034 a sequence of radii @xmath1035 such that @xmath1036 and @xmath1037 for @xmath1038 . then we apply to conclude that @xmath1039 without loss of generality we may assume here that @xmath497 has been chosen so that the pointwise intrinsic gradient @xmath514 exists in @xmath497 .
finally , it follows immediately from the definition of the graph distance that @xmath1040 for all @xmath226 on the graph of @xmath38 and @xmath1041 , and hence the claim of the lemma follows from and lemma [ l : a.s.const ] . according to remark [ remark0 ]
, it suffices prove .
assume that this is not true .
then , by lemma [ eq : limit_beta ] and lemma [ l : constancy ] , there exists an intrinsic lipschitz function @xmath501 with graph @xmath129 such that @xmath1042 this contradiction proves proposition [ dreamprop ] . in this section
, we apply proposition [ dreamprop ] to prove the weak geometric lemma for the @xmath749-numbers , theorem [ t : cg - wgl ] . in brief , if the integral appearing in theorem [ t : cg - wgl ] is large , then we can find many balls centered on the graph @xmath129 with large @xmath749-number ( this is quantified in lemma [ discretisation ] ) . then , proposition [ dreamprop ] implies that @xmath514 fluctuates strongly inside ( the projections of ) such balls , which can be used to find a lower bound for @xmath1043 with @xmath1044 and @xmath190 .
since we also have the upper bound @xmath1045 for this quantity , we will eventually find the correct upper bound for the integral in theorem [ t : cg - wgl ] .
a collection @xmath1046 of balls in a metric space @xmath1047 is _ pre - dyadic _ , if @xmath1048 for any @xmath1049 .
a collection of balls ( or sets in general ) in @xmath1047 is _ dyadic _ , if for every pair of sets @xmath1050 in the family , either @xmath1051 , or @xmath1052 , or @xmath1053 .
[ discretisation ] let @xmath129 be an intrinsic lipschitz graph , and let @xmath452 , @xmath450 .
write @xmath1054 then , there exists a pre - dyadic family of balls @xmath1055 with @xmath1056 such that @xmath1057 for every @xmath1058 , every ball @xmath1059 has @xmath1060-thin boundary with respect to @xmath801 for some @xmath1061 , and @xmath1062 write @xmath1063 . for @xmath1064 ,
let @xmath1065\}.\ ] ] let @xmath1066 be a maximal @xmath1067-separated subset , and for each @xmath1068 , choose a radius @xmath1069 $ ] such that @xmath1070 .
then , any pair of balls @xmath1071 with distinct @xmath1072 is disjoint .
further , since the sets @xmath1073 with @xmath1068 cover @xmath1074 , we have @xmath1075 next , observe that for any @xmath1076 , we have @xmath1077 since the left hand side is always bounded by @xmath1078 , and it is evidently zero for @xmath1079 .
it follows that @xmath1080 and consequently @xmath1081 next , observe that if @xmath1082 and @xmath1083 $ ] , then @xmath1084 .
then , for every ball @xmath1085 , apply lemma [ l : a - small ] with @xmath1086 and @xmath1087 to find a radius @xmath1088 $ ] such that @xmath1089 has @xmath1090-thin boundary with respect to @xmath801 , and @xmath1091 .
then the @xmath2-regularity of @xmath129 implies that holds for the balls @xmath1089 as well .
finally , since the balls @xmath1089 with @xmath226 in a fixed set @xmath1092 are disjoint , the collection @xmath1093 is pre - dyadic . indeed ,
if @xmath145 , @xmath1049 and @xmath147 are given , there are clearly at most @xmath1094 indices @xmath419 such that @xmath1095 and @xmath1096 for some @xmath1068 .
the proof is complete .
[ predyadic ] let @xmath771 be a metric measure space , and let @xmath890 and @xmath1097 .
assume that @xmath1098 is a pre - dyadic family of balls with @xmath1099 , and with @xmath1100 .
moreover , assume that every ball @xmath1101 contains a ( possibly empty ) family of disjoint sub - balls @xmath1102 such that that @xmath1103 for @xmath1104 .
finally , assume that all the balls @xmath1105 in the families @xmath1098 and @xmath1102 , @xmath1101 , have @xmath201-thin boundary , and the balls @xmath1101 have the doubling property @xmath1106 then , there exists a dyadic sub - collection @xmath1107 such that split the balls in @xmath1098 into families @xmath1113 , where @xmath1114 . by the assumption of the family @xmath1098 being pre - dyadic , and the doubling property of the balls @xmath1101 , we can ( by applying the @xmath1115-covering lemma to the balls in @xmath1116 ) choose a subfamily of @xmath1116 * disjoint * balls ( still denoted by @xmath1116 ) with @xmath1117 next , observe that either @xmath1118 we assume that the former option holds , and from now on we will only consider balls @xmath1119 , @xmath207 .
note that if @xmath1120 , @xmath1121 , @xmath1122 and @xmath1123 , then @xmath1124 recalling that @xmath1125 .
since @xmath1099 , the collection @xmath1126 is empty for large enough @xmath419 .
let @xmath1127 be the index of the largest non - empty collection @xmath1128 , and set @xmath1129 inductively , for @xmath1130 , let @xmath1131 be the collection of balls in @xmath1126 , which meet the boundary of one of the balls in @xmath1132 , let @xmath1133 and finally @xmath1134 note that if @xmath1135 , @xmath1130 , then there is some ball @xmath1136 such that @xmath1137 ( the second inclusion follows from ) .
consequently , by the disjointness of the balls in @xmath1138 , and the @xmath201-thinness of boundaries of the balls in @xmath1139 , we infer that @xmath1140 finally , since every ball in @xmath1126 either belongs to one of the collections @xmath1131 , with @xmath1130 , or to @xmath1141 , we have @xmath1142 \notag\\ & \lesssim a\sum_{j \leq j_{0 } } \left [ \sum_{j <
i \leq j_{0 } } 2^{j - i } \sum_{b ' \in { \mathcal{g}}_{i } } \mu(b ' ) + \sum_{b \in { \mathcal{g}}_{j } } \mu(b ) \right ] \notag\\ & \label{form13 } \sim a\sum_{j \leq j_{0 } } \sum_{b \in { \mathcal{g}}_{j } } \mu(b ) .
\end{aligned}\ ] ] define @xmath1143 inspecting the definition of @xmath831 , it is clear that @xmath335 can indeed be written as @xmath1144 for a certain sub - collection @xmath1107 .
since @xmath1145 for all balls @xmath1146 , we infer from that @xmath1147 , as required by condition ( i ) .
the fact that @xmath1148 is a dyadic family of balls follows immediately from the inductive definition .
in fact , the construction shows that even the family @xmath335 is dyadic , and this combined with implies condition ( ii ) .
the proof is complete . during the proof , we will abbreviate @xmath1149 and @xmath1150 to simply @xmath1151 and @xmath1152 .
we apply lemma [ discretisation ] to infer that if @xmath1153 then there exists a pre - dyadic family of balls @xmath1148 , contained in @xmath1154 and centred on @xmath129 , with @xmath1155-thin boundaries , with @xmath1156 , and such that @xmath1157 for each such ball @xmath1158 , we use proposition [ dreamprop ] to find another ball @xmath1159 with @xmath1160-thin boundary , such that @xmath1161 , @xmath1162 and @xmath1163 note that every ball in the family @xmath1164 has @xmath1165-thin boundary . hence , using lemma [ predyadic ] on the metric space @xmath1166 , with @xmath1167 and @xmath1168 , we find a dyadic sub - collection @xmath1169 , which still satisfies @xmath1170 . to avoid the double indices
, we assume that the family @xmath1171 is dyadic itself . since @xmath109 is injective on @xmath129 , it follows that the family @xmath1172 with @xmath1173 and @xmath1174 is also dyadic .
moreover , by ( ii ) of lemma [ predyadic ] , the family @xmath1172 even satisfies slightly more , namely if @xmath1110 , then @xmath1175 holding that thought , we now make some remarks of more abstract nature . given two sets @xmath1176 with @xmath1177 ,
let @xmath1178 be the subspace of @xmath1179 consisting of those functions which are zero outside @xmath59 , are constant on both @xmath1180 and @xmath201 , and have integral zero .
let @xmath1181 be the orthogonal projection onto this subspace .
finding an explicit formula for @xmath1182 is simple , as @xmath1178 is one - dimensional , and spanned by the unit vector @xmath1183 here , and for the rest of the proof , we write @xmath1184 for @xmath767 .
it follows that @xmath1185 , and in particular @xmath1186 we apply these observations to the subspaces @xmath1187 with the family @xmath1172 defined above . since this family is dyadic and satisfies , it is easy to check that the subspaces @xmath1187 are pairwise orthogonal .
now , let @xmath1188 and recall that @xmath1189 .
since @xmath1190 for all @xmath828 , it follows from bessel s inequality , , , and ( in this order ) that @xmath1191 \\ & \gtrsim \delta_{\varepsilon , l}^{2 } \sum_{j } |a^{j}| \sim \delta_{\varepsilon , l}^{2 } \sum_{j } { \operatorname{diam}}(b_{j})^{3 } \gtrsim \delta_{\varepsilon , l}^{2 } c. \end{aligned}\ ] ] recalling the definition of @xmath163 from , this completes the proof of theorem [ t : cg - wgl ] .
armed with theorem [ t : cg - wgl ] , we are nearly ready to prove the wgl for intrinsic lipschitz graphs ( theorem [ wglintlg ] ) .
first , we need a few short lemmas .
the first one states that any intrinsic lipschitz graph @xmath129 has a `` big flat piece '' inside any ball @xmath70 with @xmath452 : [ bigflatpiece ] for every @xmath189 and @xmath172 , there exists @xmath1192 with the following property .
if @xmath129 is an intrinsic @xmath124-lipschitz graph , @xmath452 and @xmath147 , then there exists a ball @xmath753 with @xmath1193 and @xmath802 such that @xmath1194 . the proof is based on the compactness of intrinsic @xmath124-lipschitz graphs , and the existence of tangent subgroups .
we make a counter assumption that there exist @xmath189 and @xmath172 with the following properties . for every @xmath828 , there exists an intrinsic lipschitz graph @xmath1195 , and a ball @xmath845 with @xmath1196 such that @xmath1197 for all balls @xmath1198 with @xmath1020 and @xmath833 . without loss of generality ( or recalling a similar reduction from the previous section ) , we may assume that @xmath1199 , and @xmath1200 for all @xmath1201 .
further , by the compactness results for intrinsic @xmath124-lipschitz functions , already employed in the previous section , we may assume without loss of generality that the functions @xmath854 converge locally uniformly to an intrinsic @xmath124-lipschitz function @xmath501 with graph @xmath129 .
one can then easily check ( emulating the argument for lemma [ eq : limit_beta ] ) that @xmath1202 for any ball @xmath908 with @xmath1203 .
but this is absurd : the function @xmath38 is intrinsically differentiable at almost every point of @xmath1204 , which means precisely that @xmath78-almost every point @xmath1203 has a vertical tangent subgroup to @xmath129 ( theorem 4.15 in @xcite ) .
we record that a subgroup @xmath1205 is a tangent group to @xmath129 in @xmath1206 if @xmath1207 with respect to the hausdorff convergence of compact subsets of @xmath0 .
but this is evidently impossible for such @xmath1203 that @xmath1208 for all sufficiently small @xmath1209 .
all the points @xmath1210 have this property by the previous discussion , and we have reached a contradiction .
recall that the graph of an entire intrinsic lipschitz function @xmath157 with constant gradient almost everywhere is a left translate of a vertical plane ( proposition [ constantgradientgraphs ] ) .
the next lemma , essentially a corollary of this fact , gives a substantial improvement over the previous lemma for constant gradient graphs .
[ stability0 ] for every @xmath189 and @xmath172 , there exists @xmath799 with the following property .
assume that @xmath1211 is an intrinsic @xmath124-lipschitz graph , and @xmath452 , @xmath147 are such that @xmath1212 .
then , if @xmath1020 , @xmath1213 , and @xmath1214 , we have @xmath1215 we make a counter assumption : there exist @xmath189 and @xmath172 such that the following holds for every @xmath1216 , @xmath828 .
there exists an intrinsic @xmath124-lipschitz graph @xmath1217 and two balls @xmath1218 with @xmath1219 , such that @xmath1220 we may assume that @xmath1221 .
thus , @xmath1222 and @xmath1223 , and @xmath1224 for a certain point @xmath1225 .
the assumption @xmath1226 implies that the balls @xmath1227)$ ] eventually cover any ball @xmath1228 , @xmath450 .
we infer that a subsequence of the functions @xmath854 converge locally uniformly to an intrinsic @xmath124-lipschitz function @xmath1229 , which has a.e .
constant gradient on @xmath1230 , and such that @xmath1231 ( here @xmath1232 is the graph of @xmath1233 ) . finally , using a diagonal procedure , a further subsequence of the functions @xmath854 converges locally uniformly to an intrinsic @xmath124-lipschitz function @xmath38 , which has a.e .
constant gradient on @xmath32 , and @xmath1234 .
but the existence of such a @xmath38 was ruled out in proposition [ constantgradientgraphs ] , and we have reached a contradiction .
the proof of the lemma is complete .
let @xmath129 be an intrinsic lipschitz graph , and let @xmath236 be a system of david cubes on @xmath129 ; recall that every cube @xmath1235 contains a set of the form @xmath1236 for some point @xmath215 and some constant @xmath1237 . in this section , where our task is to prove the weak geometric lemma for intrinsic @xmath124-lipschitz graphs , we write @xmath1238 where @xmath1239 is a constant depending on @xmath124 only .
the precise requirement for @xmath163 is that @xmath1240 , where @xmath1241 is the constant from the definition of @xmath1242 . with this notation for @xmath247
, we still write @xmath1243 , and @xmath1244 .
the reason for choosing @xmath1245 so large is the following : if @xmath1246 with @xmath1247 , and @xmath890 is any constant , then @xmath1248 as soon as @xmath1249 .
this is easy to check : if @xmath1249 and @xmath1250 , then @xmath1251 we also mention that @xmath1252 implies @xmath1253 which is verified in the similar manner as above . in this section ,
we abbreviate @xmath1254 .
we choose to prove the weak geometric lemma in the following form : for any fixed cube @xmath221 , @xmath1255 where @xmath1256 ; the integral formulation is an easy corollary . to this end
, we note that the following inequality is a simple consequence ( essentially a reformulation ) of theorem [ t : cg - wgl ] : @xmath1257 where @xmath1258 .
let @xmath1259 be a constant , which will depend only on @xmath55 and @xmath124 .
now , by , the weak geometric lemma will follow , if we are able to prove that @xmath1260 the following corollary of lemma [ stability0 ] will be crucial : [ stability2 ] for every @xmath189 and @xmath172 , there exists @xmath1261 with the following property .
assume that @xmath1246 are cubes such that @xmath1247 , @xmath1262 , and @xmath1263 for all @xmath1264 with @xmath1265
. then @xmath1266 .
write @xmath1267 , where @xmath1268 is the constant from lemma [ stability0 ] .
write @xmath1269 for the @xmath1270
dyadic ancestor of @xmath248 , with @xmath1271 .
let @xmath1272 be a large number depending on @xmath55 and @xmath124 ; we will eventually pick @xmath1273 first , and then require that @xmath1274 is sufficiently small depending on @xmath1273 . if @xmath1275 for some @xmath1276 , we have @xmath1277 assuming @xmath1278 .
this is the first restriction we place on @xmath1274 . now , assume that @xmath1275 for some @xmath1279 .
it follows that @xmath1280 satisfies @xmath1265 , hence @xmath1281 . recalling remark [ remk : bcg ]
this means that there exists an intrinsic @xmath124-lipschitz function @xmath1282 with graph @xmath1283 such that @xmath550 has a.e .
constant gradient on @xmath1284 , and @xmath1285 let @xmath1286 be the nearest points to @xmath1287 on @xmath1283 , so that @xmath1288 write @xmath1289 , and observe that @xmath1290 for @xmath1274 small enough . then , write @xmath1291 , and observe that gives the following chain of inclusions , assuming @xmath1292 to be large enough , and @xmath1293 to be small enough : @xmath1294 from we first conclude that @xmath550 has a.e.constant gradient on @xmath1295 , which means that @xmath1296 then , from we infer that lemma [ stability0 ] is applicable to the ball @xmath1297 .
consequently , @xmath1298 thus , there exists a set of the form @xmath51 , where @xmath1299 and @xmath53 is a vertical subgroup , such that @xmath1300 for every @xmath1301 . finally , recalling that we wish to prove @xmath1302 , we fix @xmath1303 by we can pick @xmath1304 with @xmath1305 .
then , for @xmath1259 small enough , we have @xmath1301 , and @xmath1306 .
this proves that @xmath1302 , as required .
let @xmath1312 be a constant depending only on @xmath124 , and let @xmath1313 .
by lemma [ bigflatpiece ] applied to the ball @xmath1314 with @xmath1315 , there exists a point @xmath1316 and a radius @xmath1317 such that @xmath1318 and @xmath1194 .
now , it suffices to note that @xmath971 contains a ball of the form @xmath247 for some @xmath1247 with @xmath1319 .
hence @xmath1320 .
choosing @xmath1321 small enough , the lemma follows .
we are prepared to prove the weak geometric lemma .
as discussed , it remains to verify .
the proof is nearly verbatim the same as on the last pages of @xcite , but we record the details for completeness .
* @xmath1325 has a maximal element @xmath1326 , called the _ root _ of @xmath1325 . * if @xmath1327 with @xmath212 , and @xmath1328 , then also @xmath1329 . * if @xmath1330 , then either all , or none , of the children of @xmath248 belong to @xmath1325 . *
those cubes @xmath1330 with no children in @xmath1325 are called the _ stopping cubes _ of @xmath1325 , and they are denoted by @xmath1331 .
we now partition @xmath1323 into trees by an inductive procedure .
the roots of the initial trees are simply the maximal cubes in @xmath1323 .
then , if @xmath1332 is one of these trees , and @xmath1333 , we add all the children of @xmath248 into @xmath1332 if all of them belong to @xmath1323 ; if not , we declare that @xmath1334 and stop building @xmath1325 along this branch . after the initial trees have been constructed , and in case some cubes in @xmath1323 still remain outside them , we repeat the previous procedure : we pick the maximal cubes in @xmath1323 , which are not , yet , contained in a tree , and we declare these to be the roots of new trees .
these trees are , then , constructed by the rule described above .
iterating this algorithm produces a partition of @xmath1323 into a countable number of trees @xmath1335 .
if @xmath1336 for some ( unique ) @xmath1337 , we set @xmath1338 . in the opposite case , @xmath1339 , it follows from the construction that either the parent or one of the siblings of @xmath1340 , say @xmath1341 is in @xmath1342 . in this case , we set @xmath1343 . with this notation in hand , and observing that any cube @xmath1341 can only serve as @xmath1344 for boundedly many indices @xmath1058 , we are prepared to show that the roots @xmath1340 satisfy a carleson property : @xmath1345 the last inequality is , of course , .
next , recalling our objective , and writing @xmath1346 , we make the following natural splitting : @xmath1347 by , it remains to prove that @xmath1348 for any fixed @xmath1058 . to this end , let @xmath1349 be the family of maximal cubes @xmath1350 with the property that @xmath1351 . by corollary [ stability2 ] , if @xmath1352 is contained in a cube in @xmath1349 , then @xmath1353 , and hence @xmath1354 .
so , in fact the summation in runs over at most those cubes which are not contained in any cube in @xmath1349 ; let those cubes be called @xmath1355 .
fix @xmath1356 , and let @xmath1309 be the maximal cube with the property that @xmath1311 .
lemma [ bigflatcube ] promises that there exists such a cube @xmath248 with @xmath1357 .
if @xmath1350 , then obviously @xmath1358 , and we set @xmath1359 . if , on the other hand , @xmath248 already lies outside @xmath1360 , then there is some cube @xmath1361 with @xmath1362 , and we set @xmath1363 . in either case latexmath:[$|f(r)| \gtrsim_{\eta , l } @xmath1349 and @xmath1365 consist of disjoint cubes , and that each cube @xmath1366 can only be assigned as @xmath1367 for boundedly many cubes @xmath1368 ( with bounds depending only on @xmath1369 ) , we may conclude that @xmath1370 since @xmath1261 only depends on @xmath55 and @xmath124 , we have proven , and the weak geometric lemma ( theorem [ wglintlg ] ) for intrinsic lipschitz graphs .
we will conclude the paper with a brief discussion of theorem [ wglbp ] .
the analogous result in the euclidean setting is due to david and semmes and it follows by theorem 1.8 ( * ? ? ?
* part iv , chapter 1 ) . before restating it in our setting ,
we record a definition .
let @xmath1371 and @xmath1372 { \rightarrow}(0,\infty)$ ] .
we define @xmath1373 to be the collection of @xmath2-regular sets
@xmath56 with regularity constant at most @xmath1374 such that @xmath1375 for @xmath189 , @xmath449 and @xmath450 . [ ds18heis ]
let @xmath56 be a @xmath2-regular set .
suppose that there exist @xmath1376 and @xmath1372 { \rightarrow}(0,\infty)$ ] such that for each @xmath449 and @xmath170 there exists @xmath1377 satisfying @xmath1378 then @xmath57 satisfies the wgl .
as it happens , the proof of theorem [ ds18heis ] follows exactly as the proof of theorem 1.8 ( * ? ? ?
* part iv , chapter 1 ) , modulo notational changes .
therefore we skip the details .
now observe that theorem [ wglbp ] follows from theorem [ wglintlg ] and theorem [ ds18heis ] , since the @xmath2-regularity and wgl constants ( for any fixed @xmath189 ) of an intrinsic @xmath124-lipschitz graph are bounded above by a constants depending only on @xmath55 and @xmath124 .
l. ambrosio , f. serra cassano , d. vittone : _ intrinsic regular hypersurfaces in heisenberg groups _ , j. geom .
16 * ( 2 ) ( 2006 ) , 187232 g. arena and r. serapioni : _ intrinsic regular submanifolds in heisenberg groups are differentiable graphs _ , calc .
* 35 * ( 4 ) ( 2009 ) , 517 536 z. m. balogh , e. durand cartagena , k. fssler , p. mattila and j. t. tyson : _ the effect of projections on dimension in the heisenberg group _ , rev . mat
. iberoam .
* 29 * ( 2013 ) , no .
2 , 381432 f. bigolin : _ intrinsic regular hypersurfaces in heisenberg groups and weak solutions of non linear first - order pdes _ , phd thesis , trento , ( 2009 ) f. bigolin , l. caravenna and f. serra cassano : _ intrinsic lipschitz graphs in heisenberg groups and continuous solutions of a balance equation _ , ann . inst .
h. poincar anal .
non linaire * 32 * ( 5 ) ( 2015 ) , 925963 f. bigolin and f. serra cassano : _ distributional solutions of burgers equation and intrinsic regular graphs in heisenberg groups _ , j. math .
appl . , * 366 * ( 2 ) ( 2010 ) , 561568 a. bonfiglioli , e. lanconelli and f. uguzzoni : _ stratified lie groups and potential theory for their sub - laplacians _ , springer monographs in mathematics 2007 l. capogna , d. danielli , s.d . pauls and j. tyson : _ an introduction to the heisenberg group and the sub - riemannian isoperimetric problem _ , progress in mathematics ( 2007 ) , birkhuser v. chousionis , v. magnani and j. t. tyson _ removable sets for lipschitz harmonic functions on carnot groups _ ,
* 53 * ( 2015 ) , no . 3 - 4 , 755780 g. citti , m. manfredini , a. pinamonti and f. serra cassano : _ smooth approximation for intrinsic lipschitz functions in the heisenberg group _ ,
pde * 49 * ( 3 ) ( 2014 ) , 12791308 g. david : _ morceaux de graphes lipschitziennes et intgrales singulires sur un surface _ , rev . mat .
iberoamericana * 4 * ( 1988 ) , 73114 g. david : _
unrectifiable @xmath25-sets have vanishing analytic capacity_. rev . mat .
iberoamericana * 14 * ( 1998 ) , no .
2 , 369479 : _ removable sets for lipschitz harmonic functions in the plane_. rev . mat .
iberoamericana * 16 * , ( 2000 ) , no .
1 , 137215 g. david and s. semmes : _ quantitative rectifiability and lipschitz mappings _ , trans .
* 337 * ( 2 ) ( 1993 ) , 855889 g. david and s. semmes : _ analysis of and on uniformly rectifiable sets .
_ in : mathematical surveys and monographs 38 , amer .
soc . , providence , ri ( 1993 ) g. david and s. semmes : _ singular integrals and rectifiable sets in @xmath1379 : beyond lipschitz graphs .
_ astrisque * 193 * , 152 pp .
( 1991 ) v. eiderman and a. volberg _ non - homogeneous harmonic analysis : 16 years of development _ , russ .
surv . * 68 * 973 , ( 2013 ) l. erds : _ notes on the burgers equation _ , available at https://www.mathematik.uni-muenchen.de/ lerdos / ss06/pde / burg.pdf f. ferrari , b. franchi , and h. pajot : _ the geometric traveling salesman problem in the heisenberg group _ rev . mat . iberoam .
* 23 * ( 2007 ) , no .
2 , 437480 a. f. filippov : _ differential equations with discontinuous righthand sides : control systems _ , springer 1988 b. franchi , m. marchi and r. serapioni : _ differentiability and approximate differentiability for intrinsic lipschitz functions in carnot groups and a rademacher theorem _ , anal . geom .
spaces * 2 * ( 1 ) ( 2014 ) , 22993274 b. franchi and r. serapioni : _ intrinsic graphs within carnot groups _ , j. geom .
26 * , ( 2016 ) , issue 3 , 19461994 b. franchi , r. serapioni and f. serra cassano : _ rectifiability and perimeter in the heisenberg group _ , math
* 321 * ( 2001 ) , 479531 b. franchi , r. serapioni and f. serra cassano : _ intrinsic lipschitz graphs in heisenberg groups _ , j. nonlinear convex anal .
* 7 * ( 3 ) ( 2006 ) , 423441 b. franchi , r. serapioni and f. serra cassano : _ differentiability of intrinsic lipschitz functions within heisenberg groups _ , j. geom .
, * 21 * ( 2011 ) , 10441084 j. heinonen , p. koskela , n. shanmugalingam and j. tyson : _ sobolev spaces on metric measure spaces _ ,
cambridge university press , 2015 p. jones : _ lipschitz and bi - lipschitz functions _ , rev . mat .
iberoamericana * 4 * ( 1988 ) , 115122 p.
jones : _ square functions , cauchy integrals , analytic capacity , and harmonic measure , _ harmonic analysis and partial differential equations ( el escorial , 1987 ) , lecture notes in math .
1384 , springer , berlin , 1989 , 2468 p.
mattila : _ geometry of sets and measures in euclidean spaces _ , cambridge university press , 1995 p. mattila , r. serapioni and f. serra cassano : _ characterizations of intrinsic rectifiability in heisenberg groups _ ,
ann . scuola norm .
* 5 * vol .
ix ( 2010 ) , 687723 : _ on the uniform rectifiability of ad - regular measures with bounded riesz transform operator : the case of codimension 1 _ , acta math .
* 213:2 * ( 2014 ) , 237321 f. serra cassano : _ some topics of geometric measure theory in carnot groups . _ in geometry , analysis and dynamics on sub - riemannian manifolds , vol .
i. edited by d. barilari , u. boscain , m. sigalotti .
ems ser.lect .
notes in math .
ems , zurich , 2016
|
the purpose of this paper is to introduce and study some basic concepts of quantitative rectifiability in the first heisenberg group @xmath0 . in particular
, we aim to demonstrate that new phenomena arise compared to the euclidean theory , founded by g. david and s. semmes in the 90 s .
the theory in @xmath0 has an apparent connection to certain nonlinear pdes , which do not play a role with similar questions in @xmath1 .
our main object of study are the _ intrinsic lipschitz graphs _ in @xmath0 , introduced by b. franchi , r. serapioni and f. serra cassano in 2006 .
we claim that these @xmath2-dimensional sets in @xmath0 , if any , deserve to be called quantitatively @xmath2-rectifiable .
our main result is that the intrinsic lipschitz graphs satisfy a _
weak geometric lemma _ with respect to _ vertical @xmath3-numbers_. conversely , extending a result of david and semmes from @xmath4 , we prove that a @xmath2-ahlfors - david regular subset in @xmath0 , which satisfies the weak geometric lemma and has _ big vertical projections _
, necessarily has _
big pieces of intrinsic lipschitz graphs_.
|
nearest neighbor methods have been successfully applied in a variety of fields , such as classification ( see @xcite ) , density and regression function estimation ( see @xcite , @xcite ) , and multivariate two - sample testing ( see , e.g. , @xcite , @xcite , and @xcite ) .
moreover , nearest neighbor methods have also been employed in the context of testing the goodness - of - fit of given data with a distributional model ( see @xcite , @xcite and @xcite ) .
this paper is devoted to a class of universally consistent goodness - of - fit tests based on nearest neighbors .
these tests can be applied not only to test for uniformity on a compact domain in @xmath0 , but also to test for a specified density on a @xmath3-dimensional manifold embedded in @xmath0 , where @xmath4 . here
, prominent special cases involve testing for uniformity on a circle or on a sphere . to be specific , let @xmath5 denote a @xmath6 @xmath3-dimensional manifold embedded in @xmath7 , where @xmath8 @xmath5 is endowed with the subset topology and is a closed subset of @xmath7 .
let d@xmath9 be the riemannian volume element on @xmath5 .
a probability density function on @xmath5 is a measurable non - negative real - valued function @xmath10 on @xmath5 satisfying @xmath11 .
the support @xmath12 of @xmath10 is the smallest closed set @xmath13 such that @xmath14 .
let @xmath15 denote the class of bounded probability density functions @xmath10 on @xmath5 , and write @xmath16 for the subset of probability density functions @xmath10 such that @xmath12 is compact and either ( i ) @xmath12 has no boundary or ( ii ) @xmath12 is a @xmath6 submanifold - with - boundary of @xmath5 ; we refer to section 2 of @xcite for details .
notice that @xmath12 could be an @xmath3-sphere ( or any ellipsoid ) embedded in @xmath7 .
let @xmath17 denote those probability density functions @xmath18 which are bounded away from zero on their support . in what follows we
let @xmath19 , @xmath20 , be independent and identically distributed ( i.i.d . )
random variables with density @xmath10 , defined on a common probability space @xmath21 , and we put @xmath22 . given a locally finite subset @xmath23 of @xmath5 and @xmath24 , we write @xmath25 for the @xmath1th nearest neighbor
( with respect to the euclidean norm @xmath26 ) of @xmath9 among @xmath27 .
let @xmath28 be the volume of the unit @xmath3-sphere
. given a fixed @xmath29 and a fixed positive integer @xmath30 , consider the volume score function induced by the @xmath30 nearest neighbor distances : [ volscore ] _
j^()(x , x):= _ k = 1^j ( v_m |x - x^(k)|^m)^ , i.e. , sums of volumes ( to power @xmath2 ) of the @xmath1 nearest neighbor balls around @xmath9 , @xmath31 . when @xmath23 consists of @xmath32 elements in a compact subset of @xmath5 , where @xmath33 , @xmath34 , for some @xmath35 , we study the re - scaled volume scores @xmath36 recalling that @xmath22 , we consider the random measure [ rmeasure ] _
n , j^ ( ) : = _ x_i x_n _ n , j^()(x_i , x_n ) _
x_i , with @xmath37 denoting the dirac point mass at @xmath9 .
if @xmath38 is an arbitrary measurable bounded function on @xmath5 , we write @xmath39 for @xmath40 . given a fixed @xmath41 , this paper considers testing goodness - of - fit of the hypothesis @xmath42 against general alternatives , based on the statistic [ statist ] t_n , j^():= _ n , j^ ( ) , f_0^ : = _ x_i x_n _ n , j^()(x_i , x_n ) ( f_0(x_i))^. notice that for the special case @xmath43 and @xmath44 , this type of statistic has been studied in @xcite and @xcite , but without allowing for lower - dimensional manifolds , and without considering fixed alternatives to @xmath45 . in section [ secmain ]
, we prove the asymptotic normality of @xmath46 as @xmath47 tends to infinity both under @xmath45 and under fixed alternatives to @xmath45 , and we show that @xmath48 has an almost sure limit under a fixed alternative to @xmath45 . in the case
@xmath49 , this limit is , apart from a multiplicative constant , the @xmath2-entropy between @xmath10 and @xmath50 . as a consequence ,
the statistic @xmath46 yields a universally goodness - of - fit test of @xmath45 for each @xmath51 , @xmath52 , and each @xmath30 .
the versatility of this class of tests is demonstrated in section [ secsimul ] , which presents the results of a simulation study comparing our tests with several well - known competitors .
the paper concludes with some remarks and open problems .
the limit theory for the statistic may be deduced from general theorems established in @xcite and goes as follows .
[ lln ] if @xmath53 , @xmath29 , then as @xmath54 we have [ lln1 ] _ k = 1^j _ m f_0(x)^ f(x)^1- dx in @xmath55 and also @xmath56-a.s .
_ remarks .
_ ( i ) notice that the right - hand side of ( [ lln ] ) is distribution - free if @xmath57 .
thus , in view of the testing problem ( [ hyph0 ] ) , it is indispensable to have @xmath52 .
.3 cm ( ii ) if @xmath58 , if the support @xmath12 of @xmath10 is a convex polyhedron , and if @xmath50 is the uniform density over @xmath5 , then the asserted @xmath55 convergence in is given by theorem 2 of @xcite .
that paper , which is based on @xcite , shows that [ wadelln ] e _
n , j^()(x_1 , x_n ) _
k(f ) e _ j^ ( ) ( , _ f(x ) ) f(x ) d x holds in @xmath55 as @xmath54 . here
, @xmath59 denotes a point at the origin of @xmath60 , and @xmath61 stands for a homogeneous poisson process of intensity @xmath62 in @xmath60 , with @xmath60 embedded in @xmath7 so that the random variable @xmath63 is well - defined .
as will be shown in the upcoming proof , the paper @xcite upgrades to give convergence of the measures at , it provides @xmath55 and a.s .
convergence , and also allows @xmath64 to be replaced by a @xmath6 @xmath3-dimensional submanifold of @xmath7 .
.3 cm ( iii ) the statistic @xmath46 may be considered a multivariate analogue of the statistic @xmath65 introduced in @xcite for testing the hypothesis of a uniform distribution in the unit interval @xmath66 $ ] . here , @xmath67 , where @xmath68 are the order statistics of i.i.d .
random variables @xmath69 with common density function @xmath10 . from the asymptotic distribution of @xmath65 ( see , e.g. , @xcite ) , it follows that @xmath70 in probability as @xmath54 .
this result obviously corresponds to ( [ lln1 ] ) for @xmath44 and @xmath50 being the uniform density over @xmath5 , where @xmath43 and @xmath5 has lebesgue measure one .
.3 cm ( iv ) if @xmath71 , the integral @xmath72 figuring on the right - hand side of is known as the @xmath2-entropy between ( the distributions associated with ) @xmath50 and @xmath10 , see @xcite .
notice that @xmath73 , where @xmath74 is the hellinger distance of @xmath50 and @xmath10 . by hlder
s inequality , @xmath75 , with equality if and only if the distributions pertaining to @xmath50 and @xmath10 coincide .
if @xmath76 , put @xmath77 , and recall that @xmath78 has density @xmath10 . then @xmath79,\ ] ] and , by jensen s inequality , @xmath80 .
as above , equality holds if the distributions associated with @xmath50 and @xmath10 are the same .
.2 cm ( v ) it follows from ( iii ) and theorem [ lln ] that , for fixed @xmath71 , a test of fit that rejects the hypothesis @xmath45 figuring in ( [ hyph0 ] ) for _ small _ values of @xmath46 is consistent against each fixed alternative density @xmath10 . if @xmath76 , rejection of @xmath45 is for _ large _ values of @xmath46 , and the resulting test is universally consistent .
.3 cm before stating variance asymptotics and a central limit theorem we introduce more notation from @xcite , especially ( 3.8 ) and ( 3.9 ) of that paper . given @xmath81 , abbreviate @xmath82 by @xmath83 .
we consider an integrated ` covariance ' of scores @xmath84 and an integrated ` add - one cost ' @xmath85 \ , \textrm{d}u.\ ] ] as shown in theorem 3.2 of @xcite , these integrals are finite .
let n@xmath86 denote a mean zero normal random variable with variance @xmath87 .
[ clt ] if @xmath53 is a.e .
continuous and @xmath29 , then @xmath88 moreover , as @xmath54 , @xmath89 .5 cm _ remark . _
theorem 2.1 of @xcite provides variance asymptotics and a central limit theorem for sums of functions of @xmath1th nearest neighbor distances in the special case @xmath43 .
.5 cm _ proof of theorem [ lln]_. we deduce this from theorem 3.1 of @xcite with @xmath90 , especially display ( 3.16 ) of @xcite , with the @xmath10 in ( 3.16 ) of @xcite set to @xmath91 and with the @xmath92 in ( 3.16 ) of @xcite set to @xmath10 .
observe that @xmath93 belongs to the class @xmath94 of that paper , and notice that @xmath95 holds for all @xmath96 , i.e. , the moment condition ( 3.4 ) of @xcite holds for all @xmath97 .
the limit ( 3.16 ) of @xcite tells us that as @xmath54 we have convergence in @xmath55 [ 2lln ] _
m f_0(x)^e _ j^ ( ) ( , _ f(x ) ) f(x ) x , where @xmath98 is defined at . the last assertion in theorem 3.1 of @xcite also gives a.s .
convergence in ( [ 2lln ] ) .
given @xmath99 and @xmath100 , we let @xmath101 be the @xmath1th nearest neighbor to the origin . we compute @xmath102 where the penultimate equality follows by display ( 15 ) of @xcite ( with @xmath2 replaced by @xmath103 , @xmath104 replaced by @xmath3 ) .
we have thus shown [ 3lln ] e _ j^ ( ) ( , _ ) = ^- _ k = 1^j .
letting @xmath62 equal @xmath105 in and applying gives the claimed limit .
.5 cm _ proof of theorem [ clt]_. this is an immediate consequence of theorem 3.2 of @xcite as well as remark ( iv ) on p. 2174 of @xcite . in that remark
we may set the function @xmath10 there to @xmath106 , we set @xmath90 , and we put @xmath107 equal to @xmath108 . keeping @xmath109 set to infinity , it is a straightforward matter to show that @xmath108 satisfies the moment conditions ( 3.5 ) and ( 3.6 ) of @xcite .
since @xmath108 satisfies all the conditions of remark ( iv ) on p. 2174 of @xcite , theorem [ clt ] follows as desired .
by means of a simulation study , this section compares the finite - sample power performance of the test based on @xmath46 with that of several competitors .
all simulations are performed using the statistical computing environment r , see @xcite .
we consider testing for uniformity on the unit square @xmath66 ^ 2 $ ] , on the unit circle @xmath110 , and on the unit sphere @xmath111 .
since , strictly speaking , there is not only one new test , but a whole family of tests that depend on the choice of the power @xmath2 and the number @xmath30 of neighbors taken into account , the impact on finite - sample power of @xmath2 and @xmath30 will be of particular interest . in each scenario
, we consider the sample sizes @xmath112 , @xmath113 and @xmath114 , and the nominal level of significance is set to @xmath115 . throughout ,
critical values for @xmath46 under @xmath45 ( see tables [ crit1 ] [ crit3 ] ) have been simulated with @xmath116 replications , and each entry in a table referring to the power of the test is based on @xmath117 replications . *
the distance to boundary test @xmath118 ( see @xcite ) , which is based on the distance of @xmath69 to the boundary @xmath119 of @xmath120 ^ 2 $ ] .
writing @xmath121 for the distance of @xmath122 to @xmath119 and @xmath123 for the largest of such distances ( which equals @xmath124 in our case ) , the test statistic computes the values @xmath125 under @xmath45 the random variables @xmath126 have a @xmath127-distribution .
the test employs the kolmogorov - smirnov type statistic @xmath128}|g_n(y)-g_0(y)|.\ ] ] here , @xmath129 is the empirical distribution function of @xmath126 , and @xmath130 is the distribution function of the @xmath127-distribution .
rejection of @xmath45 is for large values of @xmath131 , and critical values can be taken from the kolmogorov distribution .
note that this test is not consistent against some easily computable alternatives , e.g. , the uniform distribution on the subset @xmath132 ^ 2 $ ] of @xmath133 .
* the maximal spacing test @xmath134 , see @xcite .
writing @xmath135 for an open circle centered at @xmath9 with radius @xmath136 , this test considers the maximum radius @xmath137 ^ 2\setminus { { \cal x}}_n\}\ ] ] of a circle that does not contain any of @xmath69 as an inner point .
rejection of @xmath45 is for large values of the test statistic @xmath138 .
the limit distribution of @xmath139 under @xmath45 follows from ( 2.5 ) of @xcite , which states that @xmath140 where the random variable @xmath141 follows a gumbel distribution with distribution function @xmath142 , @xmath143 . letting @xmath144 denote the ( @xmath145)-quantile of this distribution
, the test rejects @xmath45 at asymptotic level @xmath2 if @xmath146 since dealing with nearest neighbors in the square involves boundary effects ( see , e.g. , @xcite ) , we initially employed both the euclidean metric and the torus metric , i.e. , the euclidean metric on the 3d - torus , obained as the quotient of the unit square by pasting opposite edges together via the identifications @xmath147 . because the power of the tests was in general somewhat higher for the torus metric than for the euclidean metric
, we decided to use the torus metric .
it should be stressed that this choice conforms to the general set - up adopted in @xcite so that theorem [ lln ] and theorem [ clt ] remain valid .
an empirical study of uniformity tests in several settings including the hypercube can be found in @xcite .
guided by the simulation study in @xcite , we used a contamination and a clustering model as alternatives to the uniform distribution .
the contamination model , denoted by con , for the distribution of @xmath78 is the mixture
@xmath148 ^ 2+\varepsilon_1 { \rm n}_2(c_1,\sigma_1 ^ 2)+\varepsilon_2{\rm n}_2(c_2,\sigma_2 ^ 2),\ ] ] conditionally on @xmath149 ^ 2 $ ] . here , @xmath150 , @xmath151 , @xmath152 , @xmath153 , @xmath154 , @xmath155 , @xmath156 ^ 2 $ ] is the uniform distribution over @xmath66 ^ 2 $ ] , and @xmath157 stands for the bivariate normal distribution with expectation vector @xmath158 and covariance matrix @xmath159 , where @xmath160 is the identity matrix of order 2 .
in other words , this model produces a uniform background noise and two radially symmetric point sources of data , centered at the points @xmath161 and @xmath162 .
the additional specification `` conditionally on @xmath149 ^ 2 $ ] '' means that a realization was discarded whenever the generated point did not fall into the unit square .
the clustering alternative clu ( say ) considers an alternative to @xmath45 in the non i.i.d .
case , using a two step - technique . in a first step ,
one simulates @xmath163 i.i.d .
random points with the uniform distribution u@xmath66 ^ 2 $ ] , which are then discarded but play the role of centers of clusters .
in a second step , one generates , independently of each other , for each of those @xmath164 centers @xmath165 points that are uniformly distributed in a disc of radius @xmath115 , the midpoint being the center .
similar to the con alternative , each point was discarded if it fell outside @xmath66 ^ 2 $ ] , and the point was simulated according to u@xmath66 ^ 2 $ ] to describe a small uniform noise effect .
figure [ contclu ] shows a realization of the con ( left ) and the clu ( right ) model .
|
we present a unified approach to goodness - of - fit testing in @xmath0 and on lower - dimensional manifolds embedded in @xmath0 based on sums of powers of weighted volumes of @xmath1-th nearest neighbor spheres .
we prove asymptotic normality of a class of test statistics under the null hypothesis and under fixed alternatives . under such alternatives , scaled versions of the test statistics converge to the @xmath2-entropy between probability distributions .
a simulation study shows that the procedures are serious competitors to established goodness - of - fit tests .
|
if @xmath3 is a finite set of primes , we define the set of _ @xmath0-units _ to be those integers of the shape @xmath4 , with exponents @xmath5 nonnegative integers .
the arithmetic of such sets has been frequently studied due to its connections to a wide variety of problems in number theory and arithmetic geometry . in the latter direction , equations of the shape @xmath6 where @xmath7 and @xmath8 are @xmath0-units for certain specific sets @xmath0 ,
arise naturally when one wishes to make effective a theorem of shafarevich on the finiteness of isomorphism classes of elliptic curves over a number field @xmath9 with good reduction outside a given finite set of primes . by way of a simple example ,
if we wish to find all elliptic curves @xmath10 with nontrivial rational @xmath11-torsion and good reduction outside @xmath12 , we are led to consider curves @xmath13 of the shape @xmath14 where @xmath15 and @xmath16 are rational integers satisfying @xmath17 for nonnegative integers @xmath5 .
writing @xmath18 , we thus seek to solve equation ( [ two ] ) , with @xmath19 and @xmath20 .
an algorithm for computing all solutions to equations of the shape ( [ two ] ) , over @xmath21 , can be found in chapter 7 of de weger @xcite , where , for instance , one can find a complete characterization of the solutions to equation ( [ two ] ) in case @xmath22 .
this algorithm combines lower bounds for linear forms in complex and @xmath23-adic logarithms with lattice basis reduction for @xmath23-adic lattices . more generally ,
for a given set of primes @xmath0 , we may consider equations of the shape @xmath24 with @xmath25 an integer , @xmath7 and @xmath8 @xmath0-units , and @xmath26 a nonzero integer .
we will call such a quadruple @xmath27 a _ primitive _ solution of ( [ enn ] ) if @xmath28 is @xmath29th - power free .
such equations are the main topic of discussion in chapter 9 of shorey and tijdeman @xcite , due to their connections to the problem of characterizing perfect powers in nondegenerate binary recurrence sequences of algebraic numbers .
if we write @xmath30 in equation ( [ enn ] ) , with @xmath31 @xmath29th - power free ( whereby there are at most @xmath32 choices for @xmath31 ) , then it follows from the theory of thue - mahler equations that , for a fixed value of @xmath33 , the set of primitive solutions @xmath27 of ( [ enn ] ) is finite ( see ( * ? ? ?
* thm . 7.2 ) ) .
a stronger statement still is the following ( essentially theorem 9.2 of @xcite ) : [ shti - thm9.2 ] there are only finitely many coprime @xmath0-units @xmath34 and @xmath35 for which there exist integers @xmath36 and @xmath37 such that @xmath38 . the aim of this paper is to illustrate the use of frey - hellegouarch curves in solving equations like ( [ two ] ) and , more generally , ( [ enn ] ) .
specifically , we will apply such an approach to provide a new proof of theorem [ shti - thm9.2 ] that _ a priori _ avoids the use of lower bounds for linear forms in logarithms , instead combining frey - hellegouarch curves , modularity and level - lowering , with the aforementioned theorem of shafarevich . in fairness , it must be mentioned , that effective versions of the latter result have typically depended fundamentally on linear forms in logarithms ; for recent papers along these lines , see the work of fuchs , von knel and wstholz @xcite and von knel @xcite .
the benefit of our approach is it enables us to , in section [ s : all_solutions ] , explicitly solve equation ( [ enn ] ) for a pair of sets @xmath0 with cardinality @xmath39 .
to the best of our knowledge , this is the first time this has been carried out .
indeed , it is unclear whether the classical approach to theorem [ shti - thm9.2 ] via only lower bounds for linear forms in logarithms can be made practical with current technology , in any nontrivial situations .
the outline of our paper is as follows .
section [ sec2 ] introduces our basic notation . in section [ s : finiteness ] , we show how to obtain various finiteness results currently proved with techniques from diophantine approximation , via frey - hellegouarch curves over @xmath40 .
philosophically , this bears a strong resemblance to recent work of von knel @xcite and of murty and pasten @xcite .
section [ sec4 ] contains explicit details of the connections between frey - hellegouarch curves and modular forms . in sections [ s : n=2 ] and
[ s : n=3 ] , we carry out such a `` modular '' approach quite explicitly for exponents @xmath41 and @xmath42 respectively . as an illustration of our methods
, we completely solve ( [ enn ] ) for @xmath43 and @xmath44 ( hence recovering de weger s aforementioned result ) , and also for @xmath45 and @xmath44 , for every prime @xmath46 .
it should be emphasized that this is not a `` serious '' application of our method , but merely meant as an illustration of a partial converse of the connection between solving equations of the shape ( [ enn ] ) and computing elliptic curves . the reader may wish to omit these sections at first ( and , for that matter , subsequent ) readings . a more interesting result along these lines is due to kim @xcite , where the connection between more general cubic thue - mahler equation and shafarevich s theorem is mapped out .
section [ s : all_solutions ] contains , as previously mentioned , the main result of the paper , namely an explicit solution of equation ( [ enn ] ) for the sets @xmath1 and @xmath47 .
the techniques we employ to prove these results , besides the aforementioned use of frey - hellegouarch curves and their associated modular forms , are local methods and appeal to computer algebra packages for solving thue and thue - mahler equations ; for the last of these , we rely extensively upon the computational number theory packages magma @xcite , pari @xcite and sage @xcite .
we thank rafael von knel and benjamin matschke for pointing out to us a missing solution in a previous version of proposition [ prop : local_obstructions_2 - 3-p - n=2 ] .
in what follows , we will let @xmath23 be a prime number and @xmath48 a nonzero integer .
we denote by @xmath49 the radical of @xmath50 , i.e. the product of distinct primes dividing @xmath48 , and by @xmath51 the largest nonnegative integer @xmath52 such that @xmath53 divides @xmath48 .
we write @xmath54 for the prime - to-@xmath23 part of the radical of @xmath50 , that is , the largest divisor of @xmath49 that is relatively prime to @xmath23 .
we will begin by noting some basic results on modular forms and connections between them and elliptic curves .
suppose that the @xmath55-expansion @xmath56 defines a weight @xmath11 , level @xmath57 ( cuspidal ) newform , with coefficients @xmath58 generating a number field @xmath59 .
further , let @xmath10 be an elliptic curve of conductor @xmath60 and @xmath29 a rational prime .
if @xmath61 is a prime satisfying @xmath62 , we define @xmath63 we say that @xmath13 _ arises modulo @xmath29 from the newform @xmath64 _ and write @xmath65 if there exists a prime ideal @xmath66 of @xmath9 such that , given any prime @xmath67 , we have either @xmath68 or @xmath69
throughout this section , we will let @xmath0 denote a finite set of primes and @xmath15 a positive integer .
the second part of the following lemma is a classical and easy application of the theory of linear forms in logarithms ( see corollary 1.2 of @xcite ) .
we here give a complete proof of the result below using instead frey - hellegouarch curves and shafarevich s theorem ( theorem ix.6.1 of @xcite ) .
[ lem : sum_three_s_units ] there are only finitely many @xmath0-units @xmath70 with @xmath71 for which there exists a nonzero integer @xmath26 such that @xmath72 . in particular , if @xmath73 are @xmath0-units with @xmath71 such that @xmath74 is again a @xmath0-unit , then @xmath75 is bounded by a constant depending on @xmath0 and @xmath15 .
let @xmath34 and @xmath35 be @xmath0-units and let @xmath37 be an integer such that @xmath72 . consider the elliptic curve @xmath76 with discriminant @xmath77 it follows that @xmath13 has good reduction outside @xmath78 . by shafarevich
s theorem , there are only finitely many isomorphism classes of rational elliptic curves having good reduction outside a finite set of primes .
this forces @xmath79 to take only finitely many values when @xmath7 and @xmath8 range over all @xmath0-units .
but then , since @xmath80 , the quotient @xmath81 also takes finitely many different values and therefore @xmath82 is bounded whenever @xmath71 .
we next prove a version of theorem 9.1 of @xcite through appeal to ( various ) frey - hellegouarch curves and level - lowering .
[ thm : st ] let @xmath34 and @xmath35 be @xmath0-units with @xmath71 . let @xmath36 and @xmath83 be integers . if @xmath38 , then @xmath29 is bounded by a constant depending only on @xmath0 and @xmath15 .
assume @xmath84 .
if @xmath26 is an @xmath0-unit , then , by the previous lemma , @xmath82 and thus @xmath29 is bounded by a constant depending only on @xmath0 and @xmath15 .
therefore , one may assume that @xmath26 is not a @xmath0-unit and , in particular , we may choose a prime @xmath85 dividing @xmath26 .
define an elliptic curve @xmath86 as follows .
if @xmath87 , then consider @xmath88 whose discriminant and @xmath89-coefficient are given by @xmath90 if however @xmath91 , we take @xmath92 with discriminant and @xmath89-coefficient given by @xmath93 in both cases , @xmath13 has multiplicative reduction at @xmath55 with @xmath94 divisible by @xmath29 .
let us then further assume that @xmath95 , @xmath96 and that the mod @xmath29 representation attached to @xmath13 is absolutely irreducible . by classical bounds on conductors ( see @xcite for instance ) ,
modularity of elliptic curves over @xmath40 ( @xcite ) and ribet s level lowering theorem ( @xcite ) , the elliptic curve @xmath13 arises modulo @xmath29 from a weight @xmath11 newform @xmath97 of ( trivial nebentypus and ) level @xmath98 such that @xmath57 is coprime to @xmath55 .
therefore , @xmath57 is bounded by a constant depending only on @xmath0 and there exists a prime ideal @xmath99 above @xmath29 in the ring of integers of the coefficient field @xmath9 of @xmath64 such that @xmath100 by deligne s bounds , @xmath101 is a nonzero algebraic integer in @xmath9 whose galois conjugates are all less than @xmath102 in absolute value and whose norm is divisible by @xmath29 .
therefore we have @xmath103}$ ] . since @xmath104 $ ]
is bounded by the dimension of the space of weight-@xmath11 cuspforms of level @xmath57 , it follows that @xmath29 is bounded from above by a constant depending only on @xmath0 , as desired .
combining these results with the finiteness of the number of solutions to thue - mahler equations ( see , for instance , theorem 7.1 of @xcite ) allows us to prove the following .
[ thm : finiteness ] there are only finitely many @xmath0-units @xmath70 with @xmath71 for which there exist integers @xmath36 and @xmath37 such that @xmath38 . by lemma [ lem : sum_three_s_units ]
, the equation @xmath105 has only finitely many solutions in @xmath0-units @xmath34 and @xmath35 , with @xmath71 .
further , if @xmath34 and @xmath35 are @xmath0-units with @xmath71 such that there exist integers @xmath36 and @xmath26 with @xmath83 satisfying @xmath38 , then , by the previous theorem , @xmath29 is bounded by a constant depending only on @xmath0 and @xmath15 .
but , for a fixed value of @xmath36 , the finiteness of solutions to equation @xmath38 in @xmath0-units @xmath70 with @xmath71 and integer @xmath26 follows from lemma [ lem : sum_three_s_units ] and from the finiteness of the set of solutions to thue - mahler equations for @xmath41 and @xmath33 respectively .
given a primitive solution @xmath27 of ( [ enn ] ) , let us denote by @xmath106 the gcd of @xmath7 and @xmath8 .
then @xmath106 divides @xmath107 and , without loss of generality , we may write @xmath108 where @xmath109 is a nonzero integer and @xmath35 is a @xmath29th - power free positive @xmath0-unit . if @xmath110 and @xmath111 , then one has @xmath112 conversely , let @xmath113 be pairwise coprime @xmath0-units with @xmath35 positive satisfying the above equation and let @xmath114 be a positive @xmath0-unit such that @xmath115 .
put @xmath116 , @xmath117 and @xmath118 .
then , @xmath27 is a primitive solution of ( [ enn ] ) .
therefore , all the primitive solutions of equation ( [ enn ] ) can be deduced from the finite set of @xmath0-units satisfying the condition of theorem [ thm : finiteness ] with @xmath119 . moreover , combining this remark with the previous results , we have the following : for a fixed value of @xmath36 , there are only finitely many triples @xmath120 such that @xmath121 with @xmath26 nonzero integer , @xmath73 @xmath0-units and @xmath28 @xmath29th - power free .
moreover , there are only finitely many primitive solutions to equation ( [ enn ] ) if and only if @xmath122 . according to the discussion above , the first part of the corollary is a direct consequence of theorem [ thm : finiteness ] ( with @xmath119 ) .
if , however , @xmath123 , then @xmath124 is a primitive solution to ( [ enn ] ) for any @xmath36 .
conversely , if @xmath122 , consider a primitive solution @xmath27 to ( [ enn ] ) . dividing the equation by @xmath125 leads , as explained earlier , to an equation of the shape @xmath126 where @xmath113 are coprime @xmath0-units and @xmath127 . by assumption
, @xmath113 are odd and therefore @xmath109 is even . in particular , we have @xmath128 and by theorem [ thm : st ] , @xmath29 is bounded independently of @xmath73 and @xmath26 .
the desired finiteness result now follows from the first part of the corollary . in the proof of lemma [ lem : sum_three_s_units ]
we have appealed to @xmath129-frey - hellegouarch curves and , for theorem [ thm : st ] , to @xmath130 and @xmath131-frey - hellegouarch curves .
the main goal of this paper is to make the statements of this section completely explicit in a number of situations . for this purpose
, we will have use of refined information on frey - hellegouarch curves of the above signatures .
we recall some ( by now ) classical but useful results on frey - hellegouarch curves associated with generalized fermat equations of signature @xmath130 , @xmath129 and @xmath131 , and the newforms from which they arise .
let @xmath132 and @xmath133 be @xmath29th - power free pairwise coprime nonzero integers and let @xmath134 and @xmath135 be pairwise coprime nonzero integers such that @xmath136 we make the additional simplifying assumptions ( which are not without loss of generality , but will be satisfied in the cases of interest to us ) that @xmath137 define an elliptic curve @xmath138 via @xmath139 we summarize the properties of @xmath138 that will be useful to us in the following result ( see kraus @xcite ) . [ nnn ] if @xmath140 is prime and @xmath141 , we have that @xmath142 for @xmath64 a weight @xmath11 cuspidal newform of level @xmath143 further , if @xmath61 is prime with @xmath144 , then @xmath145 next let @xmath146 and @xmath133 be nonzero integers such that @xmath147 with @xmath148 and @xmath149 pairwise coprime , @xmath133 squarefree and @xmath150 prime . without loss of generality , we may suppose that @xmath151 and @xmath152 are @xmath29th - power free and that we are in one of the following situations : 1 .
[ item : nn2_case_i ] @xmath153 and @xmath154 ; + 2 .
[ item : nn2_case_ii ] @xmath155 and either @xmath156 or @xmath157 ; + 3 .
[ item : nn2_case_iii ] @xmath155 , @xmath158 and @xmath159 ; + 4 .
[ item : nn2_case_iv ] @xmath155 , @xmath160 and @xmath161 ; + 5 .
[ item : nn2_case_v ] @xmath162 and @xmath161 . in cases ( [ item : nn2_case_i ] ) and ( [ item : nn2_case_ii ] )
, we will consider the curve @xmath163 in cases ( [ item : nn2_case_iii ] ) and ( [ item : nn2_case_iv ] ) , we will instead consider @xmath164 and in case ( [ item : nn2_case_v ] ) , @xmath165 these are all elliptic curves defined over @xmath21 .
the following lemma summarizes some useful facts about these curves .
apart from its ( easy - to - check ) assertion ( [ item : lemma_nn2_2 ] ) and up to some slight differences of notation , this is lemma 2.1 of @xcite .
[ lem : nn2_arith ] let @xmath166 or @xmath167 and @xmath168 . 1 .
the discriminant @xmath169 of the curve @xmath13 is given by @xmath170 2 .
[ item : lemma_nn2_2 ] the @xmath171-invariant @xmath172 of the curve @xmath13 is given by @xmath173 3 .
the conductor @xmath174 of the curve @xmath13 is given by @xmath175 where @xmath176 in particular , @xmath13 has multiplicative reduction at each odd prime @xmath23 dividing @xmath177 . also , @xmath13 has multiplicative reduction at @xmath11 if @xmath178 .
4 . the curve @xmath13 has a @xmath21-rational point of order @xmath11 . for the purposes of our applications ,
we will have need of an analog of proposition [ nnn ] , essentially lemma 3.3 of @xcite .
[ nn2 ] if @xmath95 is prime and @xmath179 , we have , for each @xmath180 , that @xmath181 for @xmath64 a weight @xmath11 cuspidal newform of level @xmath182 where @xmath183 where @xmath184 is as defined in lemma [ lem : nn2_arith ] .
let us suppose that @xmath146 and @xmath133 are nonzero integers such that @xmath185 with @xmath186 prime .
assume @xmath187 , @xmath188 , and @xmath189 are pairwise coprime , and , without loss of generality , that @xmath190 and @xmath191 .
further , suppose that @xmath133 is cubefree and that @xmath151 and @xmath152 are @xmath29th - power free .
we consider the elliptic curve @xmath192 with the above assumptions , we have the following result ( lemma 2.1 of @xcite ) .
[ lem : nn3_arith ] let @xmath193 . 1 .
the discriminant @xmath194 of the curve @xmath13 is given by @xmath195 2 .
the @xmath171-invariant of @xmath13 is given by @xmath196 3 .
the conductor @xmath174 of the curve @xmath13 is @xmath197 where @xmath198 in particular , @xmath13 has split multiplicative reduction at each prime @xmath199 dividing @xmath188 , split multiplicative reduction at each prime dividing @xmath187 congruent to @xmath200 , @xmath201 , @xmath202 , @xmath203 , @xmath204 , @xmath205 or @xmath206 modulo @xmath207 , and non - split multiplicative reduction at all other primes dividing @xmath187 , except @xmath167 .
also , @xmath13 has split multiplicative reduction at @xmath167 if @xmath208 , and good reduction if @xmath209 .
4 . the curve @xmath13 has a @xmath21-rational point of order @xmath167 . in what follows
, we will apply frey - hellegouarch curves of signature @xmath131 more frequently than other signatures ; the following result , essentially proposition 4.2 of @xcite ( though it is worth noting that the value @xmath210 as define in lemma 3.4 of that paper is actually stated incorrectly for the cases where @xmath211 ) , will be of particular use to us .
[ lem : nn3_ll-1 ] if @xmath140 is prime , @xmath212 and @xmath213 does not correspond to the identities @xmath214 then we have that @xmath215 for @xmath216 a weight @xmath11 cuspidal newform of level @xmath217 where @xmath218 with @xmath184 as defined in lemma [ lem : nn3_arith ] unless @xmath219 in which case we have @xmath220 more precisely , if @xmath61 is a prime , coprime to @xmath221 , then @xmath29 divides @xmath222 where @xmath9 is the number field generated by the fourier coefficients of @xmath64 and @xmath223 , with @xmath224 if @xmath225 , and @xmath226 otherwise .
in this section , we consider equation ( [ two ] ) ( as treated by de weger in @xcite and @xcite ) , via an @xmath129 frey - hellegouarch curve approach . according to the discussion of section [ s : finiteness ] , the corresponding equations to treat are of the shape @xmath227 where @xmath34 and @xmath35 are pairwise coprime @xmath0-units .
define @xmath228 , @xmath229 , @xmath230 , @xmath231 and @xmath232 .
then we have @xmath233 and may assume , without loss of generality , that we are in one of the situations ( [ item : nn2_case_i])-([item : nn2_case_v ] ) of [ ss : nn2 ] . consider the associated elliptic curve @xmath86 of lemma [ lem : nn2_arith ] .
it has good reduction outside @xmath234 .
therefore , if we know representatives @xmath235 of all ( the finitely many ) isomorphism classes of rational elliptic curves ( with a nontrivial two - torsion subgroup ) having good reduction outside @xmath236 , all that remains to do to solve ( [ gen ] ) is to check for an equality @xmath237 and @xmath238 denote the @xmath171-invariants of @xmath13 and @xmath239 respectively .
computing such representatives is a classical but challenging problem that has only been achieved for a rather restrictive list of sets , including @xmath240 , @xmath241 , @xmath242 , @xmath243 , @xmath244 and @xmath245 ( see @xcite ) . nevertheless , the precise information on the conductor @xmath174 of @xmath13 provided by lemma [ lem : nn2_arith ] , namely @xmath246 where @xmath247 allows us to sometimes show , for some specific sets @xmath0 , that we have @xmath248 except for some precisely identified quadruples @xmath249 .
in that situation , we can therefore appeal , via the sage package ` database_cremona_ellcurve ` ( @xcite , @xcite ) , to cremona s tables which , at the time of writing ( spring 2015 ) , contain representatives for all rational elliptic curves of conductor less than @xmath250 . by way of example , consider @xmath43 or @xmath251 with @xmath23 prime , @xmath252 .
we stress the fact that , for most of these latter sets , we presently do not know a complete list of representatives of the isomorphism classes of rational elliptic curves having good reduction outside @xmath0
. nevertheless , the strategy outlined above does apply and this allows us to easily recover de weger s result mentioned in the introduction and to solve equations that remained unsolved in a recent paper by terai @xcite ( see corollary [ prop : terai ] ) .
we carry out the details in the next two subsections ; the sage code used for our computations is available at 8ex ` http://www.math.ubc.ca/~bennett/sum_of_two_s-units.pdf ` . specializing the approach above to the case @xmath43 considered by de weger
, we note that _ a priori _ we must consider conductors of the shape @xmath253 where @xmath254 , @xmath255 , and @xmath256 . however , we have the following easy result . [
lem : local_obstructions ] in each case @xmath248 .
further , if @xmath257 , then either @xmath258 or @xmath259 and @xmath260 .
if @xmath261 , we have a solution to an equation of the shape @xmath262 where @xmath52 is nonnegative , @xmath263 and @xmath264 . the solutions to this equation for @xmath265 correspond to those listed in the lemma
moreover , if @xmath266 , none of these equations has a solution modulo @xmath267 unless we have @xmath268 or @xmath269 , that is @xmath259 .
similarly , it is easy to check that @xmath270 , unless we have @xmath271 @xmath272 or @xmath273 of these , only the cases @xmath274 may correspond to solutions to ( [ gen ] ) , namely to equations of the shape @xmath275 and @xmath276 .
each of these equations , however , is insoluble modulo @xmath267 . from lemma [ lem : local_obstructions ] and the above discussion
, we may thus appeal to cremona s sage package to reproduce de weger s results ( with a slightly faster search provided by the restrictions on the exponents given in the above lemma ) .
in particular , if @xmath43 , as noted in ( * ? ? ? * thm .
7.2 ) , there are precisely @xmath277 solutions @xmath120 to ( [ two ] ) with @xmath28 squarefree and @xmath278 and @xmath279 .
we list these solutions in the file ` http://www.math.ubc.ca/~bennett/2-3-5-7.pdf ` . as an obvious byproduct
, we also obtain a complete list of solutions to ( [ two ] ) with @xmath280 and @xmath281 . as we will consider these sets again in section [ s : all_solutions ] , we state these results here .
the only primitive solutions to equation ( [ two ] ) with @xmath280 and with , say , @xmath282 are given by @xmath283 the only primitive solutions to equation ( [ two ] ) with @xmath281 and with , say , @xmath282 are given by @xmath284 we now turn our attention to the case @xmath251 where @xmath23 is a prime in the range @xmath285 . once again , we must _ a priori _ consider conductors of the shape @xmath286 where @xmath254 , @xmath287 , and @xmath256 ; many of these conductors exceed the limits of the current cremona database . however , we have the following result . [ prop : local_obstructions_2 - 3-p - n=2 ] in each case @xmath248 , unless @xmath249 corresponds to one of the following equations @xmath288
@xmath289 it is easy to check that we have @xmath248 unless we are in one of the following situations : @xmath290 of these , only the cases @xmath291 may actually correspond to our equations .
we are therefore led to solve the following equations , where @xmath263 and @xmath52 is a positive integer : @xmath292 @xmath293 and @xmath294 we deal with each of these in turn .
considering first the equation @xmath295 , we readily check that it has no solution modulo @xmath296 for odd values of @xmath52 and @xmath297 .
assume therefore that @xmath52 is even .
if moreover @xmath298 , then by factorization , we end up with an equation of the shape @xmath299 or @xmath300 for some integer @xmath109 . according to ( * ? ? ?
1.1 ) , this leads to a unique solution to our equation , namely @xmath301 .
finally , if @xmath52 is even and @xmath302 , the equation @xmath303 has no solution modulo @xmath304 unless @xmath305 or @xmath306 .
however , for @xmath307 and @xmath306 , it has no solution modulo @xmath308 where @xmath309 and @xmath310 respectively . in the remaining case , namely @xmath311 ,
write @xmath312 with @xmath313 and @xmath314 .
then , @xmath315 is an integral point on the elliptic curve @xmath316 computing its integral points using the aforementioned sage command thus leads to the unique solution @xmath317 .
we now turn our attention to the equation @xmath318 .
we easily check that for @xmath23 in the range @xmath319 , this equation has no solution modulo @xmath304 unless we have @xmath320 for @xmath321 and @xmath306 , the corresponding equation has no solution modulo @xmath308 where @xmath322 and @xmath310 respectively . for the remaining case , that is @xmath323 , reducing modulo @xmath324 , we find that @xmath325 . writing @xmath326 , @xmath327 is thus an integral point on the elliptic curve @xmath328 as before , we compute its integral points on sage and deduce the unique solution @xmath329 .
consider now equation @xmath330 for @xmath23 in the range @xmath331 .
we check that there is no solution modulo @xmath304 unless we have @xmath332 for @xmath333 and @xmath334 , we however have that the corresponding equation has no solution modulo @xmath308 where @xmath335 and @xmath203 respectively .
it remains to deal with the cases @xmath336 and @xmath337 . in the latter case
, we find that @xmath338 by reducing modulo @xmath339 .
writing @xmath340 , @xmath341 is necessarily an integral point on the elliptic curve @xmath342 this again leads to a unique solution , namely @xmath343 . in the former case , that is @xmath336 ,
write @xmath312 with @xmath313 and @xmath314 .
then , @xmath344 is an integral point on the elliptic curve @xmath345 computing its integral points using the command ` integralpoints ` in magma ( @xcite ) leads to a unique solution which is @xmath346 .
we note here that the corresponding routine in sage has marked difficulty with this equation .
we now consider equation @xmath347 .
if @xmath302 , then we have no solution modulo @xmath304 unless @xmath348 or @xmath339 . for @xmath349 or @xmath339 , however , we have that the corresponding equation has no solution modulo @xmath308 where @xmath350 or @xmath351 respectively . if @xmath352 , write @xmath312 with @xmath313 and @xmath314 .
then , @xmath353 is an integral point on the elliptic curve @xmath354 computing its integral points using sage leads to the unique solution @xmath355 .
if now @xmath298 , then reducing @xmath356 modulo @xmath357 shows that @xmath52 is necessarily even .
write @xmath358 and @xmath359 with @xmath313 and @xmath314 .
then , for some divisor @xmath360 of @xmath26 , either @xmath361 or @xmath362 is an integral point on one of the elliptic curves @xmath363 however none of their integral points ( which we have already computed ) corresponds to a solution of our equation .
let us now consider the equation @xmath364 .
if @xmath302 , then the corresponding equation has no solution modulo @xmath304 unless @xmath365 or @xmath334 .
moreover in the former case , we have @xmath366 . if @xmath367 , we check that @xmath368 has no solution modulo @xmath369 .
assume thus that @xmath365 and write @xmath370 .
then @xmath371 is an integral point on the elliptic curve @xmath372 we therefore find that there is only one solution corresponding to @xmath373 .
similarly , if @xmath298 , then the corresponding equation has no solution modulo @xmath304 unless @xmath374 .
write @xmath312 with @xmath314 and @xmath375 .
then , @xmath376 is an integral point on the elliptic curve @xmath377 this gives rise to the unique solution @xmath378 .
we finally deal with the last equation , namely @xmath379 .
if @xmath302 , then the corresponding equation has no solution modulo @xmath304 unless @xmath380 or @xmath306 .
if @xmath380 , we find a local obstruction modulo @xmath381 .
if @xmath382 , we write @xmath312 with @xmath314 and @xmath383 , whereby @xmath384 is an integral point on the elliptic curve @xmath385 we therefore conclude that there is only one solution corresponding to @xmath386 .
similarly , if @xmath298 , then the corresponding equation has no solution modulo @xmath304 unless @xmath349 or @xmath365 . moreover , in that latter case , we have @xmath338 and it is easy to check that @xmath387 has no solution modulo @xmath388 . assume therefore that @xmath365 and write @xmath340 .
then @xmath389 is an integral point on the elliptic curve @xmath390 this gives rise to the unique solution @xmath391 , which completes the proof of the proposition . utilizing proposition [ prop : local_obstructions_2 - 3-p - n=2 ]
, we can again appeal to cremona s sage package to compute primitive solutions to equation ( [ two ] ) .
for each set @xmath251 , we have tabulated these solutions in the file ` http://www.math.ubc.ca/~bennett/2-3-p.pdf ` . by quick examination of this table for @xmath392 and @xmath393
, we immediately deduce the following result about equations that remained unsolved in proposition 3.3 of terai @xcite ( but have been recently solved via rather different methods by deng @xcite ) .
[ prop : terai ] the only solutions to @xmath394 and @xmath395 are @xmath396 and @xmath397 respectively .
we now deal with equation ( [ enn ] ) when @xmath42 using a @xmath398 frey - hellegouarch curve approach . the corresponding equations to treat are of the shape @xmath399 where @xmath34 and @xmath35 are pairwise coprime @xmath0-units .
we may assume , without loss of generality , that @xmath35 is cubefree and positive and that we have @xmath400 and @xmath401 . with the notation of [ ss : nn3 ]
, we attach to such a solution the elliptic curve @xmath402 : @xmath403 as in section [ s : n=2 ] , all that remains to do to solve equation ( [ eq : n=3 ] ) is to check for an equality @xmath404 and @xmath238 denote the @xmath171-invariants of @xmath13 and @xmath239 respectively , with @xmath239 ranging over all representatives of the isomorphism classes of elliptic having good reduction outside @xmath405 ( and a nontrivial @xmath167-torsion subgroup ) . to circumvent the difficulty of computing representatives and make this approach work for a broader list of sets @xmath0 ( including some for which we do not know a complete list of such representatives )
, we also make use of the precise formula for the conductor of @xmath13 given by lemma [ lem : nn3_arith ] , namely @xmath406 where @xmath407 as in the previous section , we now apply this approach to the sets @xmath43 and @xmath251 with @xmath23 prime , @xmath408 . specializing to @xmath43
, we _ a priori _ need to consider all conductors of the shape @xmath409 with @xmath410 , @xmath411 and @xmath412 .
however , we have , as before , the following result . in each case @xmath413 .
it is easy to check that we have the desired inequality for @xmath174 , unless @xmath414 among these possibilities , only the case @xmath415 may correspond to solutions .
since each of the equations @xmath416 is insoluble modulo @xmath267 , we obtain the stated result .
as explained previously , we can therefore appeal to cremona s table of elliptic curves to solve equation ( [ eq : n=3 ] ) and thus ( [ enn ] ) with @xmath42 and @xmath43 .
it turns out that there are exactly @xmath417 triples @xmath120 such that @xmath418 with @xmath7 and @xmath8 @xmath240-units , with @xmath28 cubefree and , say , @xmath278 , and @xmath26 a positive integer .
they are listed in the file ` http://www.math.ubc.ca/~bennett/2-3-5-7.pdf ` .
from there we easily extract the latter solutions to this equation for @xmath280 and @xmath281 .
we list them here for later use .
the only primitive solutions to equation ( [ enn ] ) with @xmath42 , @xmath280 and , say , @xmath282 are given by @xmath419 the only primitive solutions to equation ( [ enn ] ) with @xmath42 , @xmath281 and , say , @xmath282 are given by @xmath420 we now turn our attention to the case @xmath251 where @xmath285 is prime . once again , we _ a priori _ need to consider all conductors of the shape @xmath421 with @xmath422 , @xmath423 and @xmath412 . however , we have the following result , the proof of which , it being quite similar to that of proposition [ prop : local_obstructions_2 - 3-p - n=2 ] , we omit for the sake of concision
. we have @xmath248 unless @xmath249 corresponds to a one of the following equations @xmath424
@xmath425 combining this proposition with cremona s tables of elliptic curves then allows us to compute all the solutions to ( [ enn ] ) with @xmath42 , @xmath251 and @xmath23 as above .
we list them in the file ` http://www.math.ubc.ca/~bennett/2-3-p.pdf ` .
in this last section , we completely solve equation ( [ enn ] ) for two specific sets of primes , namely @xmath280 and @xmath281 , using a variety of frey - hellegouarch curves , level - lowering and heavy computations involving modular forms and thue - mahler equations .
it is only through careful combination of a variety of frey - hellegouarch curves , in conjunction with local arguments , that we are able to reduce these problems to a feasible collection of thue - mahler equations ( in our case , all of degree @xmath202 ) .
this is , in essence , the main feature of our approach that distinguishes it from one purely reliant upon lower bounds for linear forms in logarithms .
this latter method is , in our opinion , at least with current technology , impractical for explicitly solving equation ( [ enn ] ) for any set @xmath0 with at least two elements .
the cases where @xmath428 were covered in the previous two sections .
we may therefore assume , without loss of generality , that @xmath140 is prime .
the corresponding equation to treat is @xmath429 with @xmath34 and @xmath35 coprime @xmath430-units and @xmath35 @xmath29th - power free .
assume first that @xmath431 .
since precisely one of @xmath34 and @xmath432 is even , we may suppose , without loss of generality , that @xmath433 and that @xmath434 we write @xmath435 or @xmath436 if @xmath8 or @xmath35 is even , respectively , and set @xmath437 , for @xmath132 and @xmath133 @xmath29th - power free integers .
with the notation of [ ss : nnn ] , considering the elliptic curve @xmath438 , we thus have that @xmath29 fails to divide @xmath439 and @xmath440 , and hence , in the notation of proposition [ nnn ] , @xmath441 where @xmath64 is a weight @xmath11 newform of level @xmath442 .
the resulting contradiction implies that necessarily @xmath443 .
assume next that @xmath444 .
without loss of generality , we may suppose @xmath400 , @xmath401 and that @xmath445 we define @xmath446 depending on whether @xmath447 or @xmath35 , respectively , where , in either case , @xmath151 and @xmath152 are chosen to be @xmath29th - power free and @xmath133 is cubefree . in both cases , we have @xmath448 and @xmath449 . if @xmath450 , then with the notation of proposition [ lem : nn3_ll-1 ] , we have @xmath451 where @xmath64 is a weight @xmath11 newform of level @xmath452 and hence a contradiction
. we therefore have @xmath453 and we are led to solve @xmath454 with @xmath455 . both equations have no solution for @xmath456 and , by reducing modulo @xmath357 , we see that the equation @xmath457 has no solution for @xmath458 as well . if @xmath459 and @xmath458 , then @xmath52 is necessarily even , whereby both @xmath460 and @xmath461 are powers of @xmath11 .
hence @xmath462 and this is again a contradiction . the following result is the main theorem of the paper . as we shall observe , despite the apparently small size of @xmath0 and its elements , the computations involved here are really approaching the limits of current `` off the shelf '' technology ( though coming refinements in computational tools for modular forms will alleviate this somewhat ) .
[ lem : factorization ] let @xmath7 and @xmath8 be coprime integers and @xmath29 an odd prime number .
if we write @xmath467 then we have that @xmath468 and , moreover , @xmath469 further , @xmath470 and if @xmath471 is a prime dividing @xmath472 , then either @xmath473 or @xmath474 . apart from its very last assertion , this lemma is proved in ( * ? ? ?
* lemma 2.1 ) .
suppose that @xmath471 is a prime dividing @xmath472 .
then , by coprimality , @xmath471 does not divide @xmath8 .
we may thus find an integer @xmath475 such that @xmath476 , whence @xmath477 .
if @xmath478 , it follows that @xmath479 so that @xmath480
. otherwise , since @xmath29 is prime , @xmath481 has order @xmath29 modulo @xmath61 , so that @xmath482 , i.e. @xmath483 .
assume @xmath487 and @xmath488 .
then , with the above notation , we have that @xmath489 , since , if @xmath490 , then necessarily @xmath491 .
it follows that there exists a prime @xmath492 such that @xmath493 and hence , by lemma [ lem : factorization ] , @xmath480 , whereby @xmath494 .
we thus have @xmath495 which again has no solutions with @xmath488 .
this contradiction completes the proof of the lemma .
we first deal with @xmath496 .
since @xmath26 is even , we have that @xmath133 is odd and by factorization ( up to permutation of @xmath23 and @xmath55 ) that either @xmath501 for some odd integer @xmath360 . by a classical result of strmer ( @xcite )
, the first case holds only for @xmath502 .
similarly in the third case , we deduce that @xmath503 for some nonzero integer @xmath504 which in turn contradicts corollary 1.4 of @xcite .
we therefore end up in the situation where @xmath505 for some positive integer @xmath504 . by factorization , @xmath506 and @xmath507 for some positive integers @xmath508 and @xmath509 ,
whereby we have @xmath510 we may thus appeal to theorem 1.1 of @xcite with ( in the notation of that paper ) @xmath511 to conclude that @xmath512 and @xmath513 , whence @xmath514 as claimed .
we now turn our attention to the equation @xmath497 . by factorization , for some positive integers @xmath508 and @xmath509 ,
we have either @xmath515 according to _ loc .
cit . _ applied to @xmath516 , the former equation has no such solutions , whereas the only solutions to the latter are with @xmath512 , @xmath517 and @xmath513 .
this gives rise to @xmath518 , as claimed . with those preliminary results in hands
, we now turn to the actual proof of theorem [ thm : main ] . once again , recall that we have to solve @xmath519 where @xmath70 are coprime @xmath520-units with @xmath521 positive , @xmath186 prime and @xmath26 nonzero .
[ prop : y_pm_1 ] let @xmath7 and @xmath35 be coprime positive @xmath520-units , @xmath522 and let @xmath186 be a prime number such that @xmath38 for some positive integer @xmath26 .
then , @xmath523 , @xmath512 , @xmath524 and @xmath525 according to lemma [ lem : factorization2 ] , we may assume , without loss of generality , that @xmath526 .
similarly by applying ( * ? ? ?
1.1 ) to each subset of @xmath520 of cardinality @xmath11 , we may also assume that @xmath527 .
besides , sieving modulo @xmath528 shows that we necessarily have @xmath523 and that either @xmath7 is a @xmath201th power and @xmath35 has two distinct prime factors ( in @xmath520 ) , or that @xmath7 is a square and @xmath35 has only one prime factor .
we finally conclude using proposition [ prop : stormer ] . according to lemma [ lem : factorization2 ] and proposition [ prop : y_pm_1 ] above , in order to prove theorem [ thm : main ] , it remains to solve each equation of the shape @xmath529 where @xmath530 and @xmath531 are nonnegative integers , @xmath186 is prime and @xmath532 . in the remainder of this section
, we prove the following precise result on these equations . combining it with proposition [ prop : y_pm_1 ] and results from sections [ s : n=2 ] and [ s : n=3 ] completes the proof of theorem [ thm : main ] .
suppose that we have a solution to one of equations ( [ eq6 ] ) ( [ eq18 ] ) in nonnegative integers @xmath530 and @xmath531 , prime @xmath140 and @xmath532 , and rewrite this in the shape ( [ eq - nn3 ] ) for suitable choices of @xmath534 and @xmath135 .
applying proposition [ lem : nn3_ll-1 ] , we thus have that @xmath65 for some weight @xmath11 cuspidal newform of level @xmath57 where , crudely , we have that @xmath535 . since @xmath536 and hence necessarily @xmath537 , we may apply congruence ( [ tech2 ] ) with @xmath538 to conclude that @xmath539 where @xmath540 denotes the number of cuspidal weight-@xmath11 newforms of level @xmath57 . in
what follows , we will in fact show that this upper bound can , through somewhat refined arguments using various frey - hellegouarch curves , local computations and thue(-mahler ) solvers , be replaced by the assertion that @xmath541 , corresponding to the solutions noted in theorem [ boss ] . in the case
@xmath512 , an approach via frey - hellegouarch curves , while theoretically of value , in practice appears to work poorly .
instead , we will use magma code for solving thue - mahler equations due to k. hambrook @xcite ; documentation for this may be found at according to the shapes of the equations and the noted restrictions on @xmath544 and @xmath531 , it suffices to solve the following thue - mahler equations ( some of which may be treated locally ) @xmath545 this is easily achieved using hambrook s code and leads to the solutions mentioned .
it is worth noting , at this point in the proceedings , that , at least as currently implemented , the thue - mahler solver we are using has severe difficulties with equations of degree @xmath203 and higher . in particular
, we will need to work very carefully in order to avoid its use for the remaining cases under consideration .
we will now deal with each of equations ( [ eq6])([eq18 ] ) in turn , assuming @xmath546 and that we have a solution satisfying the conditions mentioned .
we will make repeated use of proposition [ lem : nn3_ll-1 ] , explicitly using magma to calculate newforms of all relevant levels @xmath57 .
the results are as follows .
we list in table [ table : levels ] the triples @xmath547 of interest to us for which the space of weight @xmath11 cuspidal newforms of level @xmath548 is nontrivial , together with the list of all primes @xmath150 for which there exists at least one form @xmath64 of level @xmath57 satisfying both ( [ tech2 ] ) with @xmath538 , and the congruences of proposition [ lem : nn3_ll-1 ] for all primes @xmath549 .
the magma code used for this computation ( when @xmath550 ) is available at ` http://www.math.ubc.ca/~bennett/nn3.m ` . for the larger levels @xmath551 and @xmath552 ,
we have first used wiese s magma function ` decomposition ` ( from his package ` artinalgebras ` available on his homepage ) to compute the characteristic polynomials of the first few fourier coefficients .
the output can be found in by reducing the equation modulo @xmath357 , we see that @xmath184 and @xmath554 have the same parity , and that @xmath184 is odd if and only if we have @xmath555 .
we begin by assuming that @xmath184 and @xmath554 are even , and @xmath546 . if , further , @xmath556 , we consider the curve @xmath557 where @xmath558 , with @xmath559 .
it has good reduction at @xmath202 and is of the shape @xmath560 where @xmath561 with @xmath562 and @xmath563 , whereby @xmath564 .
on the other hand , proposition [ nn2 ] , @xmath565 where @xmath64 is a weight-@xmath11 newform of level @xmath566 .
since there is a unique such newform and it satisfies @xmath567 , we obtain a contradiction from ( [ tech2 ] ) for @xmath568 .
assume next that @xmath184 is even , @xmath569 and @xmath570 .
if @xmath571 , then we have @xmath572 and , since @xmath573 , @xmath574 .
this leads to the conclusion that @xmath575 and hence a contradiction .
if @xmath576 , then we have @xmath577 and the equation to solve is @xmath578 .
however we check that there is no value of @xmath579 such that @xmath580 and hence obtain a contradiction in this case as well .
we therefore may assume that @xmath184 is even , @xmath569 and @xmath581 , and consider the curve @xmath582 where @xmath583 , with @xmath584 and @xmath585 , with @xmath586 .
then , by proposition [ lem : nn3_ll-1 ] , @xmath565 where @xmath64 is a weight @xmath11 newform of level @xmath57 with @xmath587 or @xmath588 if @xmath589 or @xmath590 respectively . since @xmath591
, we therefore reach a contradiction upon appealing to table [ table : levels ] .
next , suppose that @xmath184 and @xmath554 are odd .
if @xmath592 and @xmath593 , we simply note that there is no solution modulo @xmath594 .
if @xmath592 and @xmath576 , then @xmath595 and modulo @xmath596 , we have that @xmath597 and @xmath598 .
we are therefore led to solve the thue equation @xmath599 using pari / gp .
this gives rise to a unique solution to our equation , namely @xmath600 . if , however , we assume that @xmath601 , we begin by considering the following @xmath131 frey - hellegouarch curve @xmath602 where @xmath603 then , by proposition [ lem : nn3_ll-1 ] , @xmath565 where @xmath64 has level @xmath57 with @xmath604 or @xmath605 , where @xmath606 , if @xmath589 or @xmath260 respectively . according to table [ table : levels ] , since @xmath591 , we necessarily have level @xmath607 , whence @xmath608 to treat this remaining case , we now consider the following @xmath130 frey - hellegouarch curve @xmath609 where @xmath610 and @xmath611 with @xmath612 ( recall that @xmath613 ) . according to proposition [ nnn ] , @xmath13 arises from a weight-@xmath11 newform @xmath64 of level @xmath614 and @xmath64 corresponds to an elliptic curve @xmath615 such that @xmath616 this implies that @xmath239 has good reduction at @xmath310 and @xmath617 , or , in other words , that neither @xmath310 nor @xmath617 divides @xmath26
. we may thus sieve locally at these primes to show that there is no triple @xmath618 ( with @xmath554 odd and @xmath555 ) such that @xmath619 simultaneously for @xmath350 and @xmath617 .
this leads to the desired contradiction , whereby we may conclude that the only solution to ( [ eq6 ] ) for @xmath546 prime corresponds to @xmath600 . by reducing modulo @xmath357
, we see that @xmath184 and @xmath554 necessarily have the same parity , and that @xmath184 is odd precisely when we have @xmath555 .
assume first that @xmath184 and @xmath554 are even , and @xmath546 .
let us consider @xmath621 where @xmath583 , @xmath558 , with @xmath622 . since @xmath623 , @xmath13 has good reduction at @xmath202 . moreover ,
if @xmath577 , then it satisfies @xmath624 . from proposition [ nn2 ]
, @xmath13 arises modulo @xmath29 from a weight @xmath11 newform of level @xmath625 ( recall that @xmath626 ) .
but there is a unique newform at level @xmath625 corresponding to an elliptic curve @xmath615 with @xmath627 , a contradiction .
it follows that we have @xmath595 , which in turn implies @xmath531 even .
we may thus consider instead @xmath628 where @xmath629 with @xmath630 .
then , by proposition [ nn2 ] , @xmath13 arises from a weight @xmath11 newform of level @xmath631 .
this is an obvious contradiction .
next , assume that @xmath184 and @xmath554 are odd , and @xmath150 .
we consider the elliptic curve @xmath193 with @xmath632 where @xmath583 , @xmath558 with @xmath622 and @xmath633 with @xmath634 .
then @xmath441 where @xmath64 is a weight @xmath11 newform of level , say , @xmath57 which we may compute using proposition [ lem : nn3_ll-1 ] .
if @xmath589 , we find that @xmath635 and hence reach a contradiction from consideration of table [ table : levels ] .
assume therefore that @xmath260 and that @xmath636 does not correspond to the solution @xmath637 .
then , necessarily , @xmath592 and @xmath638 . in particular
, we have @xmath639 , i.e. , @xmath640 .
we now use local arguments to conclude . indeed , if @xmath593 , then by reducing mod @xmath641 , we obtain that @xmath642 , a contradiction . if , however , @xmath576 , then @xmath595 and we check that there is no value of @xmath554 ( odd ) such that @xmath643 and @xmath644 holds for @xmath645 and @xmath646 simultaneously . by reducing modulo @xmath357 , we see that @xmath184 and @xmath554 have the same parity and that @xmath184 is odd if and only if we have @xmath555 . from the preceding two subsections
, we may suppose that @xmath648 .
if @xmath592 and @xmath649 , then considering the equation modulo @xmath650 , we conclude that @xmath575 , a contradiction .
if , however , @xmath592 and @xmath576 , we can easily check that there is no solution modulo @xmath651 .
we may thus suppose that @xmath601 .
further , if @xmath544 and @xmath531 are all even ( so that @xmath595 ) , we can consider the elliptic curve @xmath652 where @xmath583 with @xmath584 . by proposition [ nn2 ]
, it arises at level @xmath653 , a contradiction .
we may thus suppose that at least one of @xmath544 or @xmath531 is odd .
we now use an @xmath130 frey - hellegouarch curve argument to treat this remaining case .
consider either the elliptic curve @xmath658 or @xmath659 where @xmath660 if @xmath531 is even or odd , respectively .
then , by proposition [ nnn ] , @xmath565 where @xmath64 has level @xmath661 .
it follows that the newform @xmath64 corresponds to an elliptic curve @xmath615 for which there are unique isogeny classes at level @xmath662 and five isogeny classes at level @xmath614 . for @xmath663 , we compute @xmath664 to show that @xmath13 has good reduction at @xmath55 ( i.e. , @xmath665 ) unless , perhaps , if either @xmath666 and @xmath64 corresponds to the isogeny class ` 210c ` ( in cremona s notation ) , or @xmath667 and @xmath64 corresponds to the isogeny class ` 210d ` .
we finally sieve over @xmath544 and @xmath531 ( not all even ) to show that we do not have @xmath668 for @xmath669 and @xmath306 simultaneously if the isogeny class of @xmath239 is not ` 210c ` and for @xmath669 and @xmath670 simultaneously otherwise .
assume now @xmath671 .
we basically follow the same strategy we used for the case @xmath589 applying first an @xmath131 and then an @xmath130 argument , though the details are somewhat simpler . indeed , we first consider @xmath672 or @xmath673 where @xmath674 if @xmath260 or @xmath590 respectively .
then , by proposition [ lem : nn3_ll-1 ] , @xmath565 where @xmath64 has level @xmath675 , where @xmath676 . from appeal
to table [ table : levels ] , we conclude that @xmath677 ( and @xmath678 ) .
we now use an @xmath130 frey - hellegouarch curve argument to deal with the case @xmath677 . as before , we consider the elliptic curve @xmath679 or @xmath680 with @xmath681 as above , if @xmath531 is even or odd , respectively
. then @xmath13 arises mod @xmath205 from an elliptic curve @xmath239 of conductor @xmath661 and for @xmath682 , we check , by computing @xmath683 that @xmath13 has good reduction at @xmath55 ( i.e. , that @xmath665 ) . for @xmath671
, we finally sieve over @xmath554 and @xmath531 to show that @xmath684 does not hold for @xmath685 and @xmath686 simultaneously .
this finishes the proof that equation ( [ eq15 ] ) has no solution for @xmath150 .
considering the equation @xmath357 , we conclude that @xmath184 and @xmath554 have the same parity and that @xmath184 is odd if and only if we have @xmath577 .
if @xmath184 and @xmath554 are odd , then @xmath577 and , since @xmath688 , we have @xmath689 and hence a contradiction .
next , assume that @xmath184 and @xmath554 are even and @xmath150 .
the equation to treat is now @xmath690 we thus consider the elliptic curve @xmath691 where @xmath692 by proposition [ lem : nn3_ll-1 ] , we have @xmath565 where @xmath64 has level @xmath693 with @xmath694 and @xmath695 using table [ table : levels ] , it then follows that @xmath590 and @xmath696 .
we first consider the @xmath131 frey - hellegouarch curve @xmath697 where @xmath629 with @xmath698 . by proposition [ lem : nn3_ll-1 ] , the curve @xmath13 arises modulo @xmath29 from a newform @xmath64 of level @xmath699 , where @xmath676 .
hence , using table [ table : levels ] , we deduce that @xmath700 and , in particular , that @xmath701 .
if , however , @xmath592 , then we use local arguments to get a contradiction .
if @xmath593 , we check that there is no solution modulo @xmath702 .
similarly , if @xmath576 , then there is no solution modulo @xmath703 .
this shows that equation ( [ eq16 ] ) has no solution for @xmath150 prime . by reducing mod @xmath705
, we conclude that @xmath706 , @xmath707 and @xmath595 .
write @xmath708 and @xmath709
. then , there exist nonzero integers @xmath710 with @xmath504 odd such that either @xmath711 adding these equations and recalling that @xmath712 is even yields either @xmath713 if @xmath591 , we consider the @xmath129 frey - hellegouarch curve @xmath714 according to whether we are in the first or second case above , respectively .
then , by proposition [ nn2 ] , the curve @xmath13 arises at level @xmath631 , an immediate contradiction .
we may therefore assume that @xmath592 and sieve over @xmath544 and @xmath531 to show that there is no solution modulo @xmath715 .
this shows that equation ( [ eq17 ] ) has no solution for @xmath150 prime . by reducing mod @xmath717
, we find that @xmath718 , @xmath719 and @xmath595 .
we then consider the @xmath129 frey - hellegouarch curve @xmath720 where @xmath629 with @xmath721 .
it has good reduction at @xmath203 and satisfies @xmath722 . on the other hand ,
the curve @xmath13 arises from the ( unique ) newform @xmath64 at level @xmath723 , which satisfies @xmath724 .
this gives us the desired contradiction and hence proves that equation ( [ eq18 ] ) has no solution for @xmath150 prime .
|
in this paper , we develop a new method for finding all perfect powers which can be expressed as the sum of two rational @xmath0-units , where @xmath0 is a finite set of primes .
our approach is based upon the modularity of galois representations and , for the most part , does not require lower bounds for linear forms in logarithms .
its main virtue is that it enables to carry out such a program explicitly , at least for certain small sets of primes @xmath0 ; we do so for @xmath1 and @xmath2 .
|
simple models for non - fermi liquids , such as the multi - channel kondo models , have attracted recently much theoretical interest .
the predicted thermodynamic properties as well as transport properties are markedly different from the properties of a fermi liquid . for example
the conductance is predicted to approach its asymptotic zero temperature value as the square root of the temperature . in this work
we study a system built of a small quantum dot with even number of electrons coupled to a large dot and free leads .
we explore the possibility to observe two channel kondo physics in it by a fine tuning of the magnetic field and the couplings between the small quantum dot , the leads and the large dot . in the first experimental realization of the _
single channel kondo _ effect in planar quantum dot @xcite transport was measured through a small dot containing odd number of electrons , which may be described by a local spin impurity model . at low temperature
the spin is screened by the conduction electrons ( of the single channel ) , a kondo resonance is formed and the conductance through the dot approaches ( in the symmetric dot case ) to the universal value @xmath0 with a fermi liquid @xmath1 law . soon after the experimental realization of the single channel kondo effect it was understood theoretically that in the presence of magnetic field @xcite a _ single channel kondo _ resonance may be formed in a dot with an even number of electrons .
for example , when there are two electrons in a two - level dot the relevant states are a singlet state @xmath2 with total spin @xmath3 and energy @xmath4 ; and a triplet of states , @xmath5 with @xmath6 and @xmath7 and energies @xmath8 , ( @xmath9 is the zeeman energy ) . if @xmath10 than the application of magnetic field reduces the energy of the triplet state with @xmath11 until at @xmath12 the states @xmath13 , @xmath14 are degenerate .
the doublet of the degenerate state may be considered as isospin with which the conduction electrons form the kondo state . following these theoretical predictions
the even kondo effect was observed experimentally in vertical dots @xcite and nanotubes@xcite .
it is important to appreciate that although the screening of the spin in both the single - channel kondo effect and the single - channel - even kondo effect is a non trivial many body phenomenon , eventually at low temperature , the spin is fully screened and both effects exhibit conventional fermi liquid behavior .
in contrast , when the dot has a doubly degenerate ground - state and there are two independent channels that equally screen the spin , an over - screening occurs and a non trivial many body two channel kondo effect is formed .
this many body state has both thermodynamic and transport properties that do not follow the predictions of the fermi liquid theory . the main difficulty in realizing a physical system that materializes the two channel kondo model is in creating two separate channels that equally screen the spin . in conventional setups
an electron from one channel that hops on the dot may hop to the other channel and causes mixing between the channels .
this mixing lead eventually to twoeigen channels " with one channel coupled stronger than the other one .
the channel with the stronger coupling fully screen the spin and the other channel is decoupled , and we thus have again the single channel kondo case .
it was suggested @xcite to overcome this mixing problem by using a large quantum dot as an additional channel .
then , the free leads form one channel and the large dot forms the second channel ( cf .
fig [ fg : model ] ) .
the channels do not mix as transfer of electrons between them will charge the large dot .
this suggestion for the realization of the _ two channel kondo effect _ was realized experimentally @xcite . in this work
we combine the two channel kondo effect with the even kondo effect
. we will show that it is possible to tune the magnetic field , the coupling strengthes of the small dot to the large dot and to the free leads in such a way that the two channel fixed point is stabilized .
the paper is arranged as follows : in section [ se : model ] we introduce a model that describes a small dot with even number of electrons connected to a large dot and free reservoirs . in subsections [ subse : symmetry ] and [ subse : projection ]
we introduce a few simplifying assumptions concerning the symmetry of the dot
lead couplings , and define the appropriate vector operators @xmath15 which describe the effective spin built from the singlet and one of the triplet states in the presence of accidental degeneracy induced by a magnetic field . then in subsection [ subse : sw ] we integrate out the high energy fields using a generalized schrieffer wolff transformation until we obtain the effective low temperature hamiltonian [ eq . ( [ eq : hanizo ] ) ] , which in general case demonstrates both channel and exchange anisotropy . as a special case
we consider a situation when the small dot is split to two small dots . in section [ se : rg ] we analyze the flow of the parameters using a renormalization group approach and discuss the conditions to observe the two channel kondo physics .
we find that in spite of the complicated form of hamiltonian ( [ eq : hanizo ] ) , the fixed point for two channel kondo effect may be achieved by fine tuning the setup parameters . in section
[ se : experimental ] we discuss possible experimental realizations and observations of the predictions of the model .
some details of the calculations are relegated to appendices that appear after the summary .
generalizing the idea formulated in ref . , we study a system consisting of a small dot @xmath16 with two discrete energy levels labeled by index @xmath17 coupled with left @xmath18 and right @xmath19 leads and a large dot @xmath20 , by means of tunneling amplitudes @xmath21 , respectively .
( we have assumed here that the coupling constants do not depend on the state indices of the leads and the large dot . )
the key feature of the model is that the dot @xmath20 is sufficiently large so that the single particle mean level spacing @xmath22 in this dot is small enough to make its spectrum quasi continuous , @xmath23 , whereas the coulomb blockade energy @xmath24 is large enough to fix the number of electrons @xmath25 , i.e. to prevent occupation of this dot by extra electron injected from the leads or from the small dot @xmath16 . with even number of electrons coupled to a large dot @xmath20 and two leads @xmath18 and @xmath19 .
there are two levels in dot @xmath16 .
the charging energy @xmath26 , in the small dot @xmath16 is larger than the charging energy @xmath24 in the large dot @xmath20.,width=302 ] we model the system ( cf .
[ fg : model ] ) by the hamiltonian @xmath27 , with @xmath28 here @xmath29 are the creation and annihilation operators in the free left and right leads @xmath30 , @xmath31 are the creation and annihilation operators of the large dot @xmath20 , @xmath32 , and the set of states @xmath33 denote the exact many body eigenstates of the isolated small dot @xmath16 . in our notation
the eigen energies include not only the strong coulomb blockade but also the zeeman term , which do not influence the electrons in the free lead and the electrons in the large dot as by assumption they have a constant density of states .
the operators @xmath34 are the hubbard operators@xcite ( see appendix [ app : spinalgebra ] ) . when there is only one level in the dot , it is possible to perform a unitary transformation of the electron creation and annihilation operators in the left and right leads so that only one linear combination is coupled to the small dot , while the other one is decoupled @xcite . in the generic case
it is not possible to perform such rotation for the multilevel dot.@xcite however , if the tunneling matrix elements do not depend on @xmath36 and @xmath37 we can define : @xmath38 \nonumber \\ a_{\phi k \sigma } = e^{i \varphi } \left [ - \sin\theta a_{rk\sigma } + \cos\theta a_{lk\sigma}\right]% \\\end{aligned}\ ] ] which diagonalize the left - right sector of the hamiltonian . to obtain @xmath39 in terms of @xmath40 and @xmath41 we simply have to substitute in eq .
( [ eq : hamiltonian ] ) @xmath42 , but the tunneling hamiltonian is now : @xmath43 with @xmath44 the @xmath45-channel is decoupled from the dot . through out the paper we will assume that the symmetry relation ( [ eq : t - symmetry ] ) holds . the interesting physics will occur in a certain occupation of the small dot @xmath16 . to understand what is the relevant subspace
, we first assume that the strong coulomb energy @xmath26 fixes the number of electrons in the small dot to be even ( say @xmath47 ) .
when the single particle level spacing in the small @xmath16-dot , denoted by @xmath22 , obey the inequality @xmath48 , we may further assume that only two single particle levels @xmath49 in @xmath50 are involved in the tunneling .
so one can model the small dot @xmath16 by a two - level dot with two electrons .
we denote the single particle levels of the dot by @xmath51 and @xmath52 and assume that @xmath53 .
there are six possible states in this system , but we will neglect for simplicity the possibility of double occupation of level 2 and the singlet state of electrons occupying different levels .
we thus remain with a subset of four lowest states : a spin singlet state , @xmath54 , with energy @xmath4 when two electrons occupy level 1 ; and a spin triplet , @xmath55 , with energy @xmath56 , when one electron occupy level 1 and the other level 2 . in terms of the creation operators with @xmath57 of electrons with spin @xmath58 at levels 1 and 2 respectively , the singlet state is given by @xmath59 and the three triplet components by @xmath60 the two - electron levels @xmath61 are the eigenvalues of hamiltonian @xmath62 truncated in the way described above : @xmath63 we further assume that the exchange splitting @xmath64 obeys the equation @xmath65 . in this case
the singlet has a lower energy in the absence of magnetic field .
next , in accordance with the standard approach @xcite , we neglect the finite level spacing @xmath66 in the large dot @xmath20 , but suppose that the coulomb blockade @xmath24 fixes the number of electrons , @xmath25 , in the large dot .
the interval @xmath66 denotes the lower limit of the energy of electron - hole excitation in dot @xmath20 .
the kondo tunneling regime is ineffective in our system unless @xmath67 .
this is a special situation , which may be achieved in some cases by tuning the gate voltage or playing with diamagnetic shift.@xcite we study here a general case of singlet ground state and appeal to the mechanism of accidental degeneracy induced by an in - plane external magnetic field @xmath68 , which occurs when the zeeman effect compensates the singlet
triplet splitting .
@xcite this mechanism is illustrated in fig .
[ fg : ad ] . in our model
the degeneracy between the states @xmath69 and @xmath4 occurs when the condition @xmath70 is satisfied . here
@xmath71 is the zeeman splitting energy .
to find out the possibility of a kondo effect induced by an external magnetic field , one should derive the effective spin hamiltonian containing relevant variables .
various procedures for the description of this phenomenon were discussed in refs . .
we will use the formalism of dynamical symmetry group , offered and elaborated in refs . .
according to this approach briefly described in appendix [ app : spinalgebra ] , the spin degrees of freedom in case of accidental degeneracy of the states @xmath72 and @xmath73 are described by the vector @xmath74 with components @xmath75 the operator @xmath74 obeys conventional commutation relations of momentum operator .
it can be treated as an effective spin operator describing transversal and longitudinal spin excitations in a quantum dot occupied by two electrons under condition of accidental degeneracy illustrated by fig .
[ fg : ad ] .
we therefore may introduce an effective spin operator whose creation and annihilation operators with components @xmath76 are denoted , respectively , by @xmath77 . to obtain the effective low energy hamiltonian one has to integrate in the renormalization group sense the states with high energy .
there are few energy scales here : the charging energy of the small dot , @xmath26 , the charging energy of the large dot @xmath24 and the precise position of the many body levels in the dot .
we will assume that this integration have renormalized the original parameters of hamiltonian ( [ eq : hamiltonian ] ) and will concentrate on a situation when only the state @xmath14 and @xmath78 and single - electron states of the dot are relevant for virtual hopping between @xmath16 and @xmath18 , @xmath19 and @xmath20 .
we will also assume that the scale of the energy considered is smaller than @xmath24 so that transition between @xmath18 and @xmath19 to @xmath20 ( via @xmath16 ) are excluded .
after concentrating on the two states of the dot we are in a position to perform the schrieffer - wolff @xcite transformation .
the schrieffer - wolff transformation leads to terms of the form : @xmath79 in the general case @xmath80 is a tensor of rank @xmath81 . however , due to the symmetry assumption ( [ eq : t - symmetry ] ) it reduces to a rank 4 tensor .
for example the term @xmath82 describes a virtual transition from the triplet state to the large dot @xmath20 and back .
this transition occurs when an electron hops from the large dot @xmath20 on the small dot @xmath16 and back , or when one of the electrons that build the triplet state of dot @xmath16 hops on dot @xmath20 and back .
according to our previous assumptions the former process is negligible and there are two contributions to the latter process . in the first an electron from level 1
hop on dot @xmath20 and back and in the second electron from level 2 hop on dot @xmath20 and back .
similar processes should be taken into account for @xmath83 .
we have therefore : @xmath84 here the exchange integral is estimated for the intermediate state with the electron escaped from the dot @xmath16 to the fermi level which is supposed to be the same for the leads and the dot @xmath20 , @xmath85 .
the sw hamiltonian is derived under the condition : @xmath86 using this degeneracy condition , we derive @xmath87 notice that due to the presence of magnetic field the @xmath88 symmetry is broken and @xmath89 depend on @xmath58 and @xmath90 . in the spirit of usual sw approximation
, one may neglect the small differences in the denominators of above exchange integrals and estimate them as @xmath91 due to the charging energy @xmath24 , transitions between the @xmath92 , and @xmath20 are excluded so that @xmath93 , in addition @xmath94 since it was decoupled from the dot due to the symmetry assumption ( [ eq : t - symmetry ] ) . rearranging the terms in ( [ eq : sw ] ) , we may write the hamiltonain in terms of the components of the vector @xmath15 and the conduction electron spin @xmath95 ( in the general case @xmath96 has a more complicated form ) @xmath97 where @xmath98 , describes potential scattering of conduction electrons without spin flips , @xmath99 and @xmath100 the total effective hamiltonian @xmath101 contains an additional term @xmath102 , which does not exist in the standard two channel kondo hamiltonian .
@xcite in the presence of magnetic field we may write it as : @xmath103 here the first two terms describe an effective _ multichannel _ kondo hamiltonian , which is anisotropic both in spin and channel variables and includes effective magnetic field @xmath104 . this field " is small in comparison with the zeeman field @xmath105 ( see fig .
[ fg : ad ] ) , @xmath106 here @xmath107 is the average electron density in channel @xmath108
. it is seen from ( [ eq : jcc1 ] ) that an exchange anisotropy arises from the fact that there are two tunnel matrix elements @xmath109 , @xmath110 in @xmath111 [ see eq .
( [ htus ] ) ] , which enter the longitudinal and transversal exchange constants in different combinations .
the exchange anisotropy may be eliminated from @xmath96 if the experimental setup is modified in accordance with fig .
[ fg : expsetup ] . in this modification
the small dot is split into two valleys coupled by tunneling element @xmath112 .
each of these valleys is assumed to be occupied by odd number of electrons . if the highest unoccupied levels are equivalent , then the levels @xmath113 , which enter the hamiltonian @xmath114 are @xmath115 the singlet - triplet splitting in eqs .
( [ level ] ) is determined by the indirect exchange @xmath116 instead of direct exchange @xmath64 ; and the sw procedure described above gives an effective hamiltonian in which only one tunnel matrix element , @xmath117 , enters all parameters @xmath118 ( [ eq : jcc1 ] ) . in contrust to the model of fig . [
fg : model ] , with the split dot we obtain the standard two - channel hamiltonian with isotropic exchange coupling ( and no @xmath119 ) in an effective magnetic field , @xmath120 where @xmath121 is the effective zeeman field given by the first term in eq .
( [ eq : be ] ) .
to derive the flow equations for exchange vertices in a two - channel hamiltonian , the conventional scaling procedure is used @xcite and 3-rd order diagrams that contain loops are included.@xcite we start with the setup of fig .
[ fg : expsetup ] which is described by the hamiltonian ( [ be2 ] ) with isotropic exchange . in this case
the system of flow equations has the form @xmath122 where @xmath123 if @xmath124 and vice versa , @xmath125 , @xmath126 is the corresponding density of states on the fermi level and @xmath127 is the current scaling variable with initial value @xmath128 . in accordance with the nozieres - blandin scenario ,
the flow diagram @xmath129 for this system has a weak coupling fixed point , which is unstable against channel anisotropy .
the system ( [ scal3 ] ) is written for the case of @xmath130 .
it is known that the external magnetic field is a relevant parameter for the conventional two - channel kondo effect.@xcite in our case the two - channel regime arises only at certain magnetic field @xmath131 ( fig .
[ fg : ad ] ) and , similar to the effect of an external magnetic field in conventional two channel kondo system , the deviation @xmath132 from this critical value is detrimental for the two - channel physics .
the system of flow equations for the anisotropic hamiltonian ( [ eq : hanizo ] ) with dimensionless coupling constants @xmath133 is more complicated . in this case
the third - order corrections in the r.h.s . of scaling equations contains the sums @xmath134 , where @xmath135 are the channel indices and the indices @xmath136 mark the cartesian components @xmath137 in spin space . besides , additional scaling equations should be written for the integrals @xmath138 .
it is easily seen that the vertices @xmath139 remain unrenormalized in 2-nd order , and the third - order corrections enter with plus sign in the r.h.s . of corresponding scaling equations .
thus , the parameters @xmath140 scale into zero . as to the anisotropic third order terms in scaling equations for @xmath141 , it is known from predictions of conformal field theory @xcite , that the exchange anisotropy is irrelevant for two - channel regime .
so we conclude that the flow diagram for the two - channel hamiltonian ( [ eq : hanizo ] ) with anisotropic exchange coupling retains the finite isotropic fixed point and only the channel anisotropy is relevant .
the even charge kondo effect can be realized experimentally in a setup similar to the one that was suggested to the odd two - channel kondo effect @xcite . in the ordinary setup (
that was realized recently experimentally @xcite ) two noninteracting leads ( l and r ) and a large dot c are attached to a single - level small dot d. in the setup considered here we need more then one level on the small dot and require also to fulfil certain relations between the coupling constants ( [ eq : t - symmetry ] ) .
in addition , even when these symmetries are fulfilled , we still need to tune the magnetic field to be at a degeneracy point between the singlet and the triplet , and fine tune the dot - channel coupling so that @xmath142 . in practice
this would be rather complected as many parameters should be tuned . in a setup with two - valley double dot shown in fig .
[ fg : expsetup ] only one tunnel constant enters all exchange integrals . in this case
the two - level dot @xmath16 is formed by two single particle dots .
two dot levels @xmath113 are given by eq .
( [ ddot ] ) and the single particle states of the small dot @xmath16 are the symmetric and antisymmetric wave functions of single dots . as a result the coupling constants to the two levels are equal and the ratio ( [ eq : t - symmetry ] ) is fulfilled authomatically . this condition facilitates the fine tuning procedure . in order to identify the even channel kondo effect one has first to identify the kondo case when the finite reservoir @xmath20 is open .
this can be found by tuning the magnetic field , and the levels of the small dot so that @xmath143
. next when dot @xmath20 is formed and tunneling of electrons between @xmath18 and @xmath19 to @xmath20 is forbidden , one has to tune the coupling constants so that @xmath144
. we will assume in the following the the coupling to lead @xmath18 is much bigger than the coupling to lead @xmath19 .
then as in the regular two channel kondo effect @xcite we expect that when @xmath145 differential conductance between the left and right leads @xmath146 ( with @xmath147 the voltage and @xmath148 the temperature ) will follow the 2ck scaling law @xcite : @xmath149 with @xmath150 and the scaling function @xmath151 is given in appendix [ app : y ] . while @xmath152 , the scaling law is : @xmath153 with @xmath154 and @xmath155 .
unlike the single channel case , the magnetic field is a relevant parameter in the two channel case .
we expect that the effective magnetic field in the even two channel kondo case will operate in a way very similar to the operation of the ordinary magnetic field in the two channel kondo case @xcite .
notice though that the effective magnetic field @xmath156 [ see eq .
( [ eq : be ] ) ] is controlled by the real external magnetic field , by the position of the two levels and by the coupling constants .
namely for @xmath157 we have , for the linear conductance @xmath158 the relation : @xmath159 for @xmath160 the conductance is given by the bethe - ansatz solution @xcite @xmath161 with @xmath162 and @xmath163 of @xmath164 .
we studied the possibility to observe the two channel kondo effect in multiple dots hetero - structures with even occupation in the presence of magnetic field . like in case of single channel kondo tunneling @xcite , we have found that magnetic field , which is as a rule detrimental for odd occupation kondo regime , may induce the two - channel kondo effect provided the zeeman splitting compensates the exchange gap between singlet ground state and the state with spin projection oriented parallel to magnetic field .
the effective spin of the small quantum dot in this case is 1/2 despite of the even occupation , so that the two channels provided by the large dot and the leads are sufficient to overscreen the dot spin .
the exchange anisotropy which arises in presence of magnetic field disappears because of additional degeneracy in case when the small dot is split into two identical wells ( fig .
[ fg : expsetup ] ) .
the two channel kondo quantum critical physics may be stabilized by fine tuning of the parameters ( i.e. coupling constants and external magnetic field ) .
we acknowledge useful discussion with eran sela , ileana rau and david goldhaber - gordon . the research was supported by dip , isf and bsf foundations .
in case of strong coulomb blockade , the low - energy part of the spectrum of complex quantum dots with even occupation consists of a system of singlet / triplet pairs ( in the absence of the orbital symmetry , which may impose the hund scheme of level occupation ) .
these levels may be tuned by external magnetic field and gate voltages , and the accidental degeneracy arises as a result of level crossing . under these conditions
the notion of dynamical symmetry is extremely useful.@xcite this symmetry characterizes not only the ground state of a system , but also the transitions between the eigenstates of quantum dot within a given energy interval ( in our case this interval is determined by the kondo temperature @xmath165 ) .
the simplest object , which possesses a dynamical symmetry is the s / t multiplet .
it is convenient to describe the eigenstates of a given hamiltonian and the intra - multiplet transitions by the hubbard operators @xmath166 .
these operators obey the obvious multiplication rule @xmath167 , from which the commutation relations may be easily derived.@xcite the dynamical symmetry of s / t multiplet is characterized by two vectors @xmath168 and @xmath169 defined as @xmath170 these two vectors obeying @xmath171 spin algebra generate @xmath172 symmetry group of spin rotator.@xcite the dynamical symmetry of spin rotator is realized in kondo tunneling near the point of accidental s / t degeneracy .
as is known , @xcite the zeeman splitting @xmath71 reduces @xmath172 symmetry of cqd , when the external magnetic field induces the resonance @xmath173 in this case one deals with a system of two quasi degenerate levels @xmath174 , whereas two other components of triplet are quenched at low energy @xmath175 .
the dynamical symmetry of this two - level system is described by a set of operators @xmath176 , which form a closed @xmath88 algebra . to show this ,
one may introduce a pair of vectors @xmath74 , @xmath177 instead of ( [ m.1 ] ) . the components of these vectors are defined as @xmath178 these two vectors realize the mathematical possibility of factorization of the spin rotator symmetry group into a direct product , @xmath179 , so that each vector transforms as an effective spin 1/2 .
only first of these vectors is relevant in the resonance condition @xmath180 , where @xmath165 is the kondo temperature , which characterizes cotunneling involving effective `` spin '' @xmath74 with usual commutation relations @xmath181= i\varepsilon_{ijk}p^k$ ] , where @xmath182 are cartesian components of this vector .
the function @xmath151 is given by : @xmath183,\end{aligned}\ ] ] @xmath184 with @xmath185 the complete elliptic function .
|
we study the possibility to observe the two channel kondo physics in multiple quantum dot heterostructures in the presence of magnetic field .
we show that a fine tuning of the coupling parameters of the system and an external magnetic field may stabilize the two channel kondo critical point .
we make predictions for behavior of the scaling of the differential conductance in the vicinity of the quantum critical point , as a function of magnetic field , temperature and source - drain potential .
|
symmetries impose powerful constraints in physics , leading wick _ et al . _ to suggest that the associated conserved quantities lead to additional restrictions on quantum theory , the so - called superselection rules ( ssrs ) @xcite .
they conjectured that superselection prevents the existence of coherent superpositions of charge , for example .
however , aharonov and susskind @xcite showed that the ability to observe superpositions depends on having a shared reference frame relative to which the system can be prepared and measured .
more generally , elements of quantum theory require reformulation in the presence of ssrs . quantum entanglement and various forms of non - locality
are particular examples of phenomena that are affected by the presence of ssrs , and a vast body of literature already exists on these topics @xcite-@xcite . here , we focus on bell inequality violation in the presence of ssrs .
we concentrate on the bi - partite case where _ both _ the entangled principal system and any ancilla / reference frame are subject to particle - number superselection . in particular , we examine the role of the measurement apparatus or reference frame used by the two observers ( our prototypical alice and bob ) .
we find that reference frames prepared using only local operations which satisfy ssrs and classical communication ( ssr - locc ) _ can not _ reveal the non - locality of an entangled system .
however , jointly prepared but separable reference frames can be used to violate a bell inequality with an entangled principal system . by imposing separability of the reference frame
, we deduce that violation is due to the measured entangled state ; the reference allows the observers to carry out measurements that lead to a violation of the bell inequality . in such cases , the reference is said to activate bell violation .
previous work has explored the issue of bell inequality violation in the presence of ssrs , given suitable reference frames @xcite-@xcite .
in contrast to some previous works , here the reference frame is both explicitly separable and compliant with the ssrs .
we show in general that all references prepared jointly , and only such references , can activate bell violation .
we find minimal reference frames and relate the degree of bell violation to the ` non - locality ' in the reference as measured by superselection - induced variance @xcite .
this holds in particular for measurements of single particle states and is a clear proof that such states can exhibit non - locality @xcite-@xcite .
we also discuss related single - photon experiments @xcite and conclude that there is still the need for new experiments .
we begin with the description of a bell experiment in the presence of particle - number superselection .
consider the situation as shown in fig .
[ fig_bell ] .
a source distributes to alice and bob an entangled pure state of @xmath0-particles , @xmath1 where @xmath2 is the probability that alice finds she has @xmath3 particles . under particle - number superselection ,
all states and measurements commute with the particle - number operator @xmath4 .
ordinarily , all alice and bob can do is simply count the number of particles each receives . in order to do more than this
, they share in advance a joint reference frame in the state @xmath5 , assumed to be separable and also obeying particle - number superselection .
therefore , before state ( [ principal ] ) is distributed to alice and bob they share no entanglement . by making joint ssr respecting measurements on each of their respective halves of the entangled system and reference frame ( @xmath6 for alice , @xmath7 for bob ) , they hope to be able to demonstrate a bell inequality violation . .
the experiment begins with an emission of a principal system in a state @xmath8 . according to the ssr
both @xmath5 and @xmath8 are incoherent mixtures of states with well - defined total number of particles .
alice ( on the left ) has now access to subsystems @xmath9 and @xmath10 and similarly bob ( on the right ) has access to @xmath11 and @xmath12 . both local subsystems are next measured by superselection - constrained observables , described by projections onto states with well - defined total number of particles in the subsystems . in a given experimental run , one of many observables is measured at each site , the choice of which is depicted by a tuneable knob ( arrow ) on the measuring device . finally , one of many measurement results is obtained as depicted by a number on the yellow screen ( e.g. @xmath13 for alice , @xmath14 for bob ) . in short , for this scenario a violation of a bell inequality can be observed if and only if the reference state @xmath5 is prepared jointly , and that the more particles in the reference state the larger the violation . ]
we first show that reference frames prepared using only local operations satisfying ssrs and classical communication ( ssr - locc ) can not activate violation of any bell inequality .
the proof is straightforward .
all such reference states commute with _ local _ particle - number operators and therefore are of the form @xmath15 where @xmath16 ( @xmath17 ) counts particles in the reference frame of alice ( bob ) .
these states contain only classical correlations between fixed local number of particles as measured by quantum discord and similar quantities @xcite-@xcite .
consider for the moment that the reference frame is in the pure state @xmath18 .
we can express the initial joint state of the system and reference frame as @xmath19 grouping subsystems accessible to alice and bob , respectively .
note that every term of this superposition contains a different number of local particles , i.e. @xmath20 for alice and @xmath21 for bob . as local ssr observant measurements project onto states with a well - defined number of local particles , only one term in the superposition can contribute to the probability of a corresponding result .
this , however , is exactly the same as making the measurements on a state in a mixture of @xmath22 with probability @xmath2 , and this separable state clearly admits a local hidden variable model . if one replaces @xmath2 in this model with @xmath23 all measurement results obtained with a general mixed reference state ( [ ssr - locc - ref ] )
are reproduced .
therefore , no bell inequality can be violated .
we now show that all reference frames which can not be prepared via ssr - locc are useful for bell violation .
we begin with the characterization of such references .
we can express all reference states in the general form @xmath24 where @xmath25 is the probability of @xmath26 particles in the reference frame and @xmath27 is any state with a fixed total number of particles @xmath26 , i.e. @xmath5 is an arbitrary mixture of pure states of the form @xmath28 . since we assume that @xmath5 can not be prepared via ssr - locc , it necessarily contains off - diagonal elements in the particle number basis .
all such states have a non - vanishing expectation of @xmath29 , \label{coherence}\ ] ] where @xmath30 and @xmath31 for some @xmath32 . to see this , note that @xmath33 is proportional to the average value of the sum @xmath34 over the pure states @xmath35 in the decomposition of @xmath5 .
therefore , @xmath33 vanishes if the sums vanish for all the pure states . for states
@xmath35 that have coherences in the particle number basis , this only occurs if the signs of the products @xmath36 alternate for some values of @xmath37 leading to cancellation in the sum .
in such a case however , we can always choose a larger @xmath38 to skip the terms which lead to the cancellation .
thus , a state @xmath5 can be prepared via ssr - locc if and only if @xmath39 for all values of @xmath32 .
as we now show , all reference states with non - vanishing coherence parameter @xmath33 enable observers to choose measurements that lead to a violation of the clauser horne - shimony - holt ( chsh ) inequality @xcite .
consider an entangled principal system in the state @xmath40 we show in appendix a that there always exist dichotomic local measurements compatible with the ssr whose outcomes on the joint state of the principal system and reference frame are correlated as @xmath41 where the angle @xmath42 ( @xmath43 ) parameterizes the @xmath16th ( @xmath17th ) setting of alice ( bob ) .
we insert this expression into the chsh parameter @xmath44 and find values of @xmath42 and @xmath43 for which @xmath45 is higher than the local realism bound @xmath46 .
namely , we choose @xmath47 , @xmath48 , and parameterize the settings of bob by a single angle @xmath49 , leading to @xmath50 to find its maximum , note that @xmath45 has a form of a scalar product between the vector @xmath51 and an arbitrary normalized vector @xmath52 .
therefore , there always exists an angle @xmath53 such that @xmath54 is parallel to @xmath55 and the maximum of the scalar product is given by the length of @xmath55 : @xmath56 to summarize , all reference frames that can not be prepared via ssr - locc have a non - vanishing coherence parameter @xmath33 and consequently allow observers to carry out measurements on entangled states that lead to a violation of the chsh inequality : @xmath57 an identical conclusion holds for single - particle entangled principal system .
the calculations are the same as long as the reference frame does not contain any vacuum .
we have derived conditions for violating the chsh inequality in the presence of particle - number ssr and we further study the properties of reference frames activating the violation .
we show here that the minimal _ separable _ reference frame allowing violation contains two particles in total . relaxing for a moment the separability requirement , eq .
( [ coherence ] ) shows that the minimal reference has only one particle in total .
namely , any state @xmath58 with @xmath59 has a non - vanishing parameter @xmath33 .
however , the application of the ppt criterion for entanglement @xcite reveals that @xmath35 is entangled for all @xmath60 .
when entangled states are used as references , it is unclear whether the violation of a bell inequality is due to the entanglement of the principal system or the reference frame . for this reason we consider a reference frame with at most two particles : @xmath61 where as before @xmath62 . since the definition of eq .
( [ coherence ] ) involves only the real part of the off - diagonal elements , we choose the coefficients @xmath63 and @xmath64 to be real , i.e. @xmath65 , and we have used @xmath66 in eq .
( [ coherence ] ) .
the application of the ppt criterion reveals that the state @xmath5 is separable if and only if @xmath67 therefore , for all separable reference frames activating the violation , @xmath68 , there must be some mixture of the two - particle state ( @xmath69 ) . note that the same argument applies to @xmath70 and
one concludes that separable reference states of the form ( [ minimal ] ) enabling the violation must contain some vacuum .
this is a consequence of the fact that an arbitrary mixture of any pure two - qubit entangled state with ` colored noise ' @xmath71 is always entangled @xcite .
a useful mathematical tool which illustrates and generalizes the results presented so far is the twirling operation .
twirling @xmath72 eliminates the coherences that are not compatible with ssr : @xmath73 where @xmath74 is a projector on a subspace with a fixed number of particles @xmath3 .
this operation describes the lack of a reference frame enabling access to the phase information of the probability amplitudes .
the usefulness of twirling is best illustrated by considering states that can be prepared via ssr - locc .
a simple proof demonstrates that they can not activate chsh violation .
a ssr - locc reference frame commutes with local particle number operators and therefore it is invariant under the action of local twirlings , @xmath75 .
this implies for the coherence parameter @xmath76 where the last equality follows from the fact that @xmath77 because @xmath78 contain only off - diagonal elements in the particle - number basis .
note that the operator @xmath79 conserves total particle number and therefore it is invariant under global twirling , @xmath80 .
all states satisfying ssr are of the form @xmath81 , where now @xmath5 need not have a fixed number of particles . therefore , @xmath82 where we have used the cyclic property of trace .
this means that in order to calculate @xmath33 for a ssr respecting reference frame @xmath83 , we can use in eq .
( [ coherence ] ) any state @xmath5 whose twirling gives @xmath83 .
twirling allows further study of separable reference frames .
note that the minimal reference frame of two particles we have derived in eq .
( [ minimal ] ) activates the violation but does not lead to maximal violation .
indeed , the highest value of @xmath33 for entangled reference states ( [ minimal ] ) is @xmath84 and for separable reference states eq .
( [ ppt_v ] ) implies @xmath85 , whereas the maximal violation of the chsh inequality allowed by quantum theory @xmath86 @xcite requires @xmath87 .
note also that the tsirelson bound of @xmath86 implies that @xmath88 . here
we show that there are separable reference frames allowing maximal violation of the chsh inequality with an entangle state .
first note that all separable states satisfying ssr are of the form @xmath89 , with @xmath90 where @xmath91 need not have a fixed number of particles .
this follows from the fact that global twirling is an locc operation ( but not ssr - locc ) and as such can not produce entanglement .
we now derive the separable reference frames that maximize the coherence parameter @xmath33 .
( [ coherence ] ) is linear in @xmath5 , @xmath33 is maximal for a pure product state @xmath92 .
moreover , due to the fact that only the real part enters ( [ coherence ] ) , it is sufficient to consider states with real coefficients @xmath93 and @xmath94 with @xmath95 . for such pure states @xmath96 where @xmath97 with @xmath0 and @xmath98 denoting the dimensionality of the reference of alice and bob respectively , i.e. the maximal number of particles in their reference frames , and we put @xmath66 . to find the maximum of @xmath33 ,
it is now sufficient to optimize @xmath99 , because @xmath100 has the same form and optimization over bob s state is independent of that over alice s .
note that for states with real coefficients @xmath101 and therefore @xmath102 .
the only non - vanishing elements of matrix @xmath103 are a strip of identities above and below its diagonal , and therefore it is a hermitian matrix .
the maximal value of @xmath99 is attained for @xmath104 being the eigenvector of @xmath103 with the highest eigenvalue .
the amplitudes of the optimal state read @xmath105 and its maximal eigenvalue gives @xmath106 for small references containing at most one particle on both sides @xmath107 we find @xmath108 in agreement with the results of section @xmath109 on minimal reference frames .
if the references of alice and bob are both unbounded , eq . ( [ max_fn ] ) implies that @xmath110 and the violation of the chsh inequality approaches its maximum . practically , for @xmath111 particles in each reference frame , @xmath112 .
the references for violation of bell inequality were also studied in @xcite in the context of directional reference frames , finding that in the limit of unbounded reference frame , maximal violation can be achieved .
we stress that in our case the corresponding limit is twofold : to maximally violate chsh inequality the reference has to be prepared jointly , and it should be unbounded .
let us now discuss the relation between @xmath113 and the non - locality of reference frames as captured by siv @xcite . we show that violation of the chsh inequality is a witness of non - zero siv in the reference frame and that the amount of siv in small references gives an upper bound on the chsh violation .
the siv of a pure state @xmath114 is defined as the variance of local number of particles @xmath115 the factor of @xmath116 is introduced for normalization : one unit of siv is defined for the state @xmath117 . since siv is symmetric with respect to permutation of the parties , one can equally consider the variance of the local particle number on bob s side ( @xmath118 ) .
pure states that have non - zero siv can not be prepared via ssr - locc and this is the property of reference frames we are interested in .
however , such pure states are always entangled , whereas we insist on separability of the reference frame .
therefore we must consider mixed states . just like entanglement , for mixed states another measure of siv is required .
we shall use the variance of formation defined as @xcite @xmath119 where the ensembles of pure states @xmath120 obey ssrs . as a measure of the off - diagonal terms in the density matrix
, we can consider @xmath33 , or equivalently bell inequality violation , as a witness of non - zero siv .
moreover , it was shown in ref .
@xcite that for states ( [ minimal ] ) the variance of formation reads @xmath121 with real @xmath63 and @xmath64 .
accordingly , one can directly relate siv to @xmath122 as @xmath123 therefore , the corresponding states with vanishing siv have also vanishing @xmath33 .
our last topic is the experimental verification of bell inequality violations under ssrs and the need for new experiments .
we relate this by commenting on current experiments related to the bell inequality and single - photon non - locality @xcite .
although not intended to violate a bell inequality under an ssr , these experiments may be seen as such for ( an induced ) photon - number ssr @xcite . in the proposal of ref .
@xcite , a single photon is directed onto a balanced beam - splitter producing ( it is hoped ) a non - local state of one photon . in each output port of this first beam - splitter there is another balanced beam - splitter with a ( reference ) coherent state directed at its second input port ( see fig . 1 of ref
. @xcite ) .
one considers correlations between the number of photons registered in detectors placed in the output ports of the second set of beam - splitters .
the experimental realizations @xcite differ from the proposal @xcite in that the secondary beam - splitters may be unbalanced .
note that in principle the possible measurement results are unbounded , and therefore the chsh inequality can not be applied .
it turns out that the correlation functions violate the chsh inequality only for small mean number of photons in the coherent states , in which case the events of having many photons in the detectors are rare and the chsh inequality becomes applicable .
this , however , opens up an effective detection loophole which allows for a local hidden variable model .
let us denote by @xmath124 ( @xmath125 ) reflectivity of the first ( second ) beam - splitter supplied with a coherent state .
the corresponding transmittances are : @xmath126 with @xmath127 .
it is assumed that both coherent states have the same mean number of photons @xmath128 and relative phase @xmath129 .
the correlation function between the number of photons measured behind the two beam - splitters reads @xmath130 using this expression in the chsh parameter , one finds that the proposal of ref .
@xcite is optimal in the sense that it is best to choose balanced beam - splitters @xmath131 .
any other values of @xmath132 and @xmath125 lead to smaller values of the chsh parameter .
in particular , the assumption of ref .
@xcite that after the beam - splitter a photon may have equal likelihood to have come from a single photon ` beam ' or a coherent state , i.e. @xmath133 leads to no violation for all values of @xmath128 .
for the balanced beam - splitters the inequality is violated if @xmath134 , which translates into the critical probability of vacuum in the coherent state @xmath135 .
using such coherent states it is quite rare to measure two photons in a setup and one may utilize this effective detection loophole to explain the observed results with , e.g. , the model of gisin @xcite .
we therefore hope this research will stimulate further experiments testing bell violation in the presence of ssrs .
ideally one would use systems with natural ssrs such as massive particles or charges .
however , studies of partial superselection can also be performed through controlled decoherence , as decoherence between different subspaces can be seen as a type of ssr .
we have studied the effects of restrictions imposed by ssrs on bell inequality violation .
we found that the violation primarily depends on how a reference frame is prepared and only secondarily on its size .
even unbounded reference frames do not lead to bell violation if they are prepared via ssr - locc and therefore are strictly classically correlated according to quantum discord and similar measures .
this condition was shown to be necessary and sufficient for the violation ; that is reference frames enable the violation of a bell inequality if and only if they can not be prepared via ssr - locc .
the violation can be achieved with separable reference frames explicitly consistent with particle - number superselection , the minimal such reference containing up to two particles .
we linked the violation to the amount of non - locality in the reference frame as captured by siv .
it would be interesting to study how other subfields of quantum theory , e.g. quantum tomography , are modified in the presence of superselection .
we acknowledge discussions with .
brukner , b. hessmo and j. vaccaro .
we are grateful to an anonymous referee for pointing out a problem in the previous version of the manuscript which eventually led us to stronger results .
this research is supported by the national research foundation and ministry of education in singapore .
dklo acknowledges the support of the quantum information scotland network ( quisco ) .
assume for the moment that the reference system is in a pure state @xmath136 since it can not be prepared via ssr - locc , we have @xmath137 for some @xmath37 and @xmath32 .
consider principal system in a state @xmath138 such that the initial state of the principal system and reference together reads : @xmath139 where we grouped in kets subsystems accessible to alice and bob respectively .
we present local dichotomic measurements compatible with particle - number ssr which lead to the correlation function ( [ corr ] ) of the main text .
alice measures a local observable parameterized by angle @xmath140 : @xmath141 where the eigenstates are defined as follows : @xmath142 for @xmath143 ; @xmath144 for @xmath145 ; @xmath146 for @xmath147 .
these observables are compatible with the ssr because all the eigenstates contain a fixed total number of particles @xmath148 .
the reason behind the three different cases is that the reference subsystem can not contain more than @xmath26 particles and less than zero .
they are also chosen to form an orthonormal set of vectors . to obtain the observables of bob one
just needs to replace @xmath149 , @xmath150 , @xmath151 and @xmath152 .
we reverse the equations for the eigenvectors and write the initial state of the principal system and the reference as @xmath153 the probabilities of the results corresponding to different eigenvectors are : @xmath154 note that for every @xmath155 there is only one @xmath156 for which the corresponding probability does not vanish , and it is easy to verify that indeed @xmath157 .
finally , the correlation function is the average of the product of dichotomic local results @xmath158 plugging in the formulae for the probabilities , we obtain @xmath159 where @xmath160 .
alternatively @xmath33 can be expressed using operators @xmath161 and @xmath162 with the help of which @xmath163 .
this calculation holds for _ arbitrary _ pure state @xmath35 of the reference .
therefore , for the reference in an arbitrary mixed state @xmath164 the correlations formula reads @xmath165 and therefore it is of the same form as eq .
( [ app_corr ] ) , but with the modified coherence parameter @xmath166.\ ] ]
|
superselection rules ( ssrs ) constrain the allowed states and operations in quantum theory . they limit preparations and measurements hence impact upon our ability to observe non - locality , in particular the violation of bell inequalities .
we show that a reference frame compatible with a particle number ssr does not allow observers to violate a bell inequality if and only if it is prepared using only local operations and classical communication . in particular , jointly prepared separable reference frames are sufficient for obtaining violations of a bell inequality .
we study the size and non - local properties of such reference frames using superselection - induced variance .
these results suggest the need for experimental bell tests in the presence of superselection .
|
the dynamical properties of isolated many - body quantum systems have long been of interest , due to their role in the fundamentals of quantum statistical mechanics .
more recently , experiments approximating this ideal of isolated many - body quantum systems have become feasible in systems of trapped atoms @xcite and ions @xcite , and as a consequence this topic has received renewed attention .
it appears that a broad class of such systems obey the eigenstate thermalization hypothesis ( eth ) @xcite .
the eth asserts that each exact many - body eigenstate of a system s hamiltonian is , all by itself , a proper microcanonical ensemble in the thermodynamic limit , in which any small subsystem is thermally equilibrated , with the remainder of the system acting as a reservoir . in the present paper
we present some interesting results for an infinite - range transverse - field ising model that obeys the eth and also has spontaneous symmetry - breaking .
quantum many - body systems with static randomness may fail to obey the eth due to many - body anderson localization stopping thermalization @xcite .
the interesting interplay of many - body localization and discrete symmetry - breaking was recently explored in refs . @xcite ; the present paper instead explores an example of the interplay of the eth and ising symmetry breaking .
we start with the infinite - range transverse - field ising model : @xmath0 where @xmath1 , and @xmath2 are the pauli operators for the spin-1/2 at ` site ' @xmath3 .
we choose to set @xmath4 .
this model has been extensively studied recently , particularly as an example for exploring quantum information issues where it is known as the ` lipkin - meshkov - glick model ' ; see , e.g. , @xcite and references therein .
one can determine and use many properties of the exact eigenstates of this hamiltonian , but there are extensive degeneracies in its spectrum due to its symmetry under any permutation of the @xmath5 spins , which give it eigenstates that do not obey the eth . thus we add static random ising interactions to the hamiltonian to break the permutation symmetry and lift all the degeneracies ( with probability one ) , so the full hamiltonian of the system we consider is @xmath6 where @xmath7 , the @xmath8 are independent gaussian random numbers of mean zero and variance one , and the power @xmath9 satisfies @xmath10 .
as we argue below , the eigenstates of this hamiltonian should obey the eth , but the randomness is weak enough so that many of the properties of @xmath11 , such as the thermodynamics , are unchanged and can be used in our analysis .
this randomness is too weak to produce any localization .
we have chosen to put the randomness on the interactions , since we have found in exact diagonalizations that this produces much better thermalization at numerically accessible system sizes as compared to , e.g. , only making the local transverse fields random .
we now briefly summarize our results , before deriving and discussing them in more detail below .
the phase diagram of this spin model as a function of the energy per spin @xmath12 and the transverse field @xmath13 is shown in fig .
1 . there are the usual two thermodynamic phases of a ferromagnet : the paramagnetic phase ( para ) at high energy and/or high @xmath14 , and the ferromagnetic phase ( f ) when both @xmath14 and the energy are low enough . in the ferromagnetic phase , for any finite @xmath5
, we can ask about the dynamics of the system s order parameter .
there are three regimes of behavior that are sharply distinguished from one another in the thermodynamic limit @xmath15 : at the highest energies within region f3 of the ferromagnetic phase the system is a sufficiently large thermal reservoir for itself so that the most probable path by which it flips from ` up ' to ` down ' magnetization under unitary time evolution is by thermally activating itself over the free energy barrier between the two ordered states . at lower energies ( f1 and f2 ) the barrier is higher and wider and as a result the reservoir is inadequate , so the system quantum tunnels through the barrier when it flips the ising order parameter . at the lowest energies in region f1 one can in principle
prepare a state that is a linear combination of two schrdinger cat eigenstates of @xmath16 that will coherently oscillate via macroscopic quantum tunneling between up and down magnetizations . in the intermediate energy regime ( f2 )
the magnetization dynamics due to quantum tunneling is always overdamped . throughout the ferromagnetic phase ,
the exact eigenstates of @xmath16 for any finite @xmath5 are schrdinger cat states that are superpositions of up and down magnetized states , and the properties of these cats differ in the three regimes of the ferromagnetic phase that are indicated in fig .
1 . thus the two phase transitions between these three dynamically distinct ferromagnetic phases are not only dynamical phase transitions but also ` eigenstate phase transitions ' @xcite , while the equilibrium thermodynamic properties are perfectly analytic through these two phase transitions . we consider the infinite - range model not only because this allows a controlled calculation of this novel physics within the ferromagnetic phases , but also because finite - range , finite - dimensional models obeying eth do not show these features . in the latter models the free energy needed to flip the magnetization by making a domain wall and sweeping it across the system is sub - extensive , while at any nonzero temperature
the system is a reservoir of extensive size , so a large system will always flip via the thermal process without macroscopic quantum tunneling through the energy barrier ; phases f1 and f2 thus do not exist for such models .
one can also consider intermediate cases of transverse - field ising models with interactions that fall off as a power of the distance between spins .
when this power is small enough , the resulting free energy barrier to flip the magnetization in the ferromagnetic phase is extensive , and we thus expect phases f1 and f2 to also occur in those models , although we do not see a way to simply calculate the locations of the phase boundaries as we can for the infinite - range model .
first we examine the the unperturbed hamiltonian @xmath11 , which has the same thermodynamics as our full model @xmath16 .
the hamiltonian @xmath11 commutes with all permutations of the spins , as does the total spin operator : @xmath17 the magnitude @xmath18 of the total spin squared also commutes with @xmath11 , so we can choose a set of eigenstates of @xmath11 that are also eigenstates of @xmath18 .
this unperturbed hamiltonian only depends on the total spin and thus can be written as @xmath19 the magnitude @xmath20 of the total spin is of order @xmath5 and ranges from zero up to @xmath21 .
for the thermodynamics ( but not the dynamics ) of this system in the limit of large @xmath5 we can treat the components of @xmath22 classically and ignore their nonzero commutators when obtaining the extensive thermodynamic properties ( energies , entropies , magnetizations ) .
the ground state of @xmath11 always has the maximum value of @xmath21 . for @xmath23 ,
the ground state is paramagnetic with the spins polarized along the @xmath24-direction : @xmath25 , @xmath26 and @xmath27 . for @xmath28 ,
the two nearly - degenerate ground states are ferromagnetic , with @xmath29 , @xmath30 and @xmath31 . for each value of @xmath20 the spectrum of @xmath11
has @xmath32 eigenenergies .
these eigenenergies and the corresponding eigenstates can be approximated for large @xmath20 using a discrete version of the wkb method @xcite , as we discuss below .
each energy level in the spectrum of @xmath11 with total spin @xmath33 has entropy per spin @xmath34 ~,\ ] ] due to all the different ways one can add together
@xmath5 spin-1/2 s to get total spin @xmath20
. since we are interested here in eigenstates , which are at a given energy @xmath35 , we will do the statistical mechanics in the microcanonical ensemble . for a given transverse field @xmath13 and energy @xmath36 ,
the equilibrium ( most probable ) state of the system is the one that maximizes the entropy , which means minimizing the total spin @xmath20 . to minimize @xmath20 for a given @xmath36 , clearly we set @xmath37 , since @xmath38 does not appear in the hamiltonian . in the paramagnetic phase
the total spin points along the @xmath24-direction and the equilibrium value of the total spin is thus @xmath39 . in the ferromagnetic phase
the system can go to higher entropy ( lower total spin ) for a given @xmath36 by making @xmath40 .
some algebra shows that in the ferromagnetic phase , which is @xmath28 and @xmath41 , the equilibrium is at @xmath30 , @xmath42 and @xmath43 .
the line of critical points separating the para- and ferromagnetic phases is @xmath44 , for @xmath28 , as indicated in fig .
1 ; this critical line ends at the quantum critical points at @xmath45 , @xmath46 .
due to its full symmetry under all permutations of the @xmath5 spins , the hamiltonian @xmath11 is integrable , with all the good quantum numbers associated with this permutation symmetry , including the magnitude @xmath20 of the total spin .
we want to study a more generic system , so we add to the hamiltonian the small term @xmath47 ( see eq .
( 2 ) ) to break the permutation symmetry , lift all degeneracies , and make the eigenstates thermal .
the only symmetry that remains in our full @xmath16 is the ising ( @xmath48 ) symmetry under a global rotation of all spins by angle @xmath49 about their @xmath24 axes . for a given @xmath36 and @xmath13 ,
the eigenstates of @xmath11 have total spin ranging from the minimum and equilibrium value @xmath50 up to the maximum value of @xmath21 . to make these in to thermal eigenstates
we need @xmath47 to perturb the system enough so that the eigenstates of the full @xmath16 are linear combinations of all these total spin values , weighted as at thermal equilibrium . at first order
, the perturbation we are adding , @xmath47 , flips at most two spins , so it can change the total spin by at most @xmath51 .
the spectrum of @xmath11 at each value of total spin @xmath20 contains @xmath32 eigenenergies spread over a range of energy that is of order @xmath20 .
thus the level - spacing in the spectrum of @xmath11 at a given @xmath20 remains of order one in the limit of large @xmath5 .
this is reflected in the dynamics under @xmath11 , which is spin precession about the mean field , and the mean field is of order one , so the rate of precession is also of order one . for the eigenstates of @xmath16 to strongly and thermally mix the different values of @xmath20
we thus need the matrix elements of @xmath47 between states at different @xmath20 to be large compared to the ( order - one ) level spacing of @xmath11 in the large @xmath5 limit .
this is why we require that the exponent @xmath9 in the definition of @xmath47 satisfies @xmath52 , since this is the condition for these matrix elements to diverge in the large @xmath5 limit .
this should be sufficient to make all the eigenstates of @xmath16 thermal in that limit .
we have not yet found a way to actually _ prove _ that this is sufficient to make all the eigenstates of @xmath16 satisfy the eth , but below we provide some numerical evidence for this from exact diagonalization of finite - size systems .
in addition to making sure that @xmath47 is strong enough to thermalize the system , we also want it to be weak enough so that we can use the well - understood dynamics of @xmath11 in our analysis . by restricting the exponent @xmath9 to be greater than @xmath53 , in the large @xmath5 limit the effective field that each spin is precessing about
is the mean field from @xmath11 , with only a small correction from @xmath47 that vanishes as @xmath15 .
this small correction is enough to thermalize the system for @xmath10 , which is the range where the perturbation due to @xmath47 on a single spin s dynamics vanishes for @xmath15 , while the perturbation to the dynamics of the full many - body system diverges . in this regime ,
the system s primary dynamics is the @xmath20-conserving dynamics due to @xmath11 , which is spin precession at a rate of order one , and , assuming we are in the ferromagnetic phase , ` attempts ' at rate of order one to tunnel through the energy barrier between total @xmath54 up and down . at a rate that is slower by a power of @xmath5 ,
the dynamics due to @xmath47 allow ` hopping ' between different values of the total spin @xmath20 , and thus thermalization to the equilibrium probability distribution of @xmath20 dictated by the entropy @xmath55 . and at a rate that is slower still , exponentially slow in @xmath5 , the system succeeds in crossing the free energy barrier between up and down magnetizations .
it is the separation between these three time scales that allows us to systematically understand the dynamics of this system for large @xmath5 .
since our full hamiltonian @xmath16 has ising symmetry under a global spin flip , and the randomness in @xmath47 means there are no exact degeneracies in the spectrum of @xmath16 , for finite @xmath5 any eigenstate of @xmath16 is either even or odd under this ising symmetry ( with probability one ) . in the ferromagnetic phase , this means the exact eigenstates of @xmath16 are all schrdinger cat states that are either even or odd linear combinations of states with total @xmath54 up and down .
these two equal and opposite values of @xmath54 are extensive , thus ` macroscopically ' different , which is why it is appropriate to call these eigenstates ` schrdinger cats ' .
next we examine the rate at which this system , in its ferromagnetic phase , will flip from the up state with positive total @xmath54 to the down state with negative @xmath54 under the unitary time evolution due to its hamiltonian @xmath16 .
there are two steps to this process : first the system gets ` excited ' from its usual ( high entropy ) total spin @xmath50 ` up ' to a larger total spin @xmath20 with lower entropy , with a probability @xmath56 given by the resulting decrease of the entropy .
as @xmath20 is increased , the energy barrier , whose top is at energy @xmath57 , decreases . for @xmath58 , one way the system can flip is to increase @xmath20 enough so that @xmath59 and then it will simply cross over the top of the barrier without quantum tunneling . in the higher - energy part ( f3 ) of the ferromagnetic phase , in the limit of large @xmath5 this is the dominant process that flips the magnetization : the system ` thermally activates ' itself ( via its unitary time - evolution ) to a low - entropy , high - total - spin state where the energy barrier can be crossed without quantum tunneling .
the ` height ' of the _ entropy _ barrier it must cross to do this is extensive : @xmath60 where @xmath61 is the reduction in entropy per site needed to go over the barrier .
if the system does not or can not go over the energy barrier by increasing the total spin @xmath20 , then in order to flip the magnetization it must quantum tunnel through the barrier . for large @xmath5 , this tunneling probability can be estimated using a version of the wkb method @xcite .
we will summarize how this is done and its results , but not present all the details , which follow ref .
the total @xmath54 serves as the ` position ' , while the operator @xmath62 serves as the ` kinetic energy ' .
what we need to calculate is the probability of the system tunneling between positive and negative total @xmath54 for a given total spin @xmath63 satisfying @xmath64 .
this probability behaves as @xmath65 , with @xmath66 being an intensive quantity .
if we define a scaled ` position ' @xmath67 , the wkb ` turning points ' adjacent to the barrier are at @xmath68 with @xmath69 then the intensive factor in the exponent of the tunneling probability is given by the wkb tunneling integral @xmath70 since the probabilities of being ` excited ' to total spin @xmath20 and of quantum tunneling through the barrier with total spin @xmath20 are both exponentially small in @xmath5 , in the limit of large @xmath5 the dominant process by which the magnetization flips is given by a standard ` saddle point ' condition .
the total spin @xmath63 at which the system tunnels is the value that maximizes the product of these two probabilities and thus minimizes the quantity @xmath71 we have located the saddle point numerically at many points within the ferromagnetic phase and it appears to always be unique , without any discontinuities as the parameters @xmath72 and @xmath13 are varied .
some straightforward analysis shows that in the higher - energy part ( f3 ) of the ferromagnetic phase where @xmath73 the saddle point is ` thermal ' : the ` entropy cost ' of going to higher @xmath20 is less than the ` tunneling cost ' , and the system goes over the barrier without any quantum tunneling .
we call the eigenstates in this regime ` thermal cats ' , since these schrdinger cat states flip by thermally activating themselves over the barrier . in regions f1 and f2 this inequality is instead false , and the system quantum tunnels through the barrier at a value of @xmath20 satisfying @xmath64 , so the eigenstates are instead ` quantum cats ' .
the location of the dynamical phase transition between phases f2 and f3 is given by converting the above inequality ( 10 ) to an equality .
near the quantum critical point ( @xmath74 ) , this transition line becomes exponentially adjacent to ( and above ) the straight line @xmath75 : @xmath76 ; near @xmath77 , it behaves as a power law : @xmath78 . within the lower - energy phases ( f1 and f2 ) of ` quantum cats '
, there is a second dynamical phase transition within the ferromagnetic phase .
this is also a ` spectral phase transition ' @xcite in the level - spacing statistics of the eigenenergies .
it one starts with a state ( not an eigenstate ) that is magnetized up , the rate at which the system crosses the barrier to down is @xmath79 , and as a result the uncertainty of the energy of this initial ` up ' state must be at least this large , by the time - energy uncertainty relation . compare this minimum energy uncertainty to the typical many - body level spacing @xmath80 of the eigenstates of @xmath16 at that energy .
there is clearly a sharp change at the phase transition line , which is the line where @xmath81 .
this transition line between phases f1 and f2 is shown in fig 1 .
the location of this transition was obtained numerically , since we do not have a simple closed - form expression for @xmath82 . near the quantum critical point ( @xmath74 ) , this transition line also becomes exponentially adjacent to ( but now below ) the straight line @xmath75 : @xmath83 ; near @xmath77 , it is logarithmically tangent to the u - axis : @xmath84 . in the lowest - energy phase ( f1 )
where @xmath85 the many - body eigenstates come in almost - degenerate pairs that are well separated in energy from other pairs of eigenstates .
these eigenstates are still thermal in their fluctuations near equilibrium , so the inter - pair eigenenergy spacings have the level statistics of the gaussian orthogonal ensemble ( goe ) .
each pair of eigenstates within phase f1 is well - approximated by the two states @xmath86 where @xmath87 is a thermal state that is magnetized up , while @xmath88 is its opposite under the global spin flip symmetry ; here the label @xmath89 refers to such a pair of eigenstates . in this regime ( f1 )
, one can make an up - magnetized initial state using a simple linear combination of only these two eigenstates , and this linear combination will then oscillate with a frequency @xmath79 .
thus in this phase the schrdinger cats can in principle be made to oscillate between ` alive ' ( up ) and ` dead ' ( down ) . in the higher - energy parts of the ferromagnetic phase ( f2 and f3 in fig .
1 ) where @xmath90 the energy associated with the tunneling is large compared to the many - body level spacing , so there are no closely degenerate pairs of states from which we can make a coherently oscillating cat . instead
, the eigenstates are of the form @xmath91 with each eigenstate being either even ( @xmath92 ) or odd ( @xmath93 ) under the global spin flip ; here the label @xmath89 refers to just one eigenstate .
states of opposite symmetry ( @xmath94 ) that are nearly degenerate are not made out of the same thermal states in each
well , so @xmath95 for @xmath15 . to make
a state that is initially magnetized up requires coherently adding together exponentially many eigenstates in order to destructively cancel all the amplitudes for down magnetizations .
under the unitary time evolution this special initial linear combination will dephase and the probability of down magnetization will increase , but presumably in an overdamped fashion as the state relaxes to equal probability of up and down magnetization .
thus we expect the dynamics of the average magnetization to always be an overdamped relaxation in phases f2 and f3 . in phase f1 ,
on the other hand , one can in principle prepare an initial state whose macroscopic magnetization will oscillate , as discussed above .
however , even in phase f1 , if one starts in a generic state that has a given energy density and is magnetized up , this will also be a linear combination of exponentially many eigenstates and also presumable show an overdamped relaxation of the average magnetization .
we now present some numerical results about the thermalization properties and the distinction between phases f1 and f3 ( phase f2 is too narrow to clearly see in the size systems we can diagonalize ) .
we exactly diagonalize hamiltonians with up to @xmath96 spins .
we first put the hamiltonian into a block diagonal form using basis states that are even and odd with respect to the system s @xmath48 ising symmetry , and then diagonalize each sector numerically to obtain all of the eigenstates within that sector . for finite @xmath5 , the coefficient @xmath97 in the disorder term @xmath47 in eq .
( 2 ) matters .
if @xmath97 is too large the system will become a spin glass rather than a ferromagnet .
thus @xmath97 needs to be carefully chosen to be large enough for our finite systems to show thermalization , but small enough to avoid the spin glass regime . after some exploration
, we chose to use the parameters @xmath98 , @xmath99 and @xmath100 for our exact diagonalizations .
first , we show the level - spacing statistics , which should be goe if the eigenstates are thermal .
this must be done within one @xmath48 symmetry sector , since there is no level - repulsion between states in different sectors .
we look at each set of three consecutive levels in one sector and denote @xmath101 as the smaller level spacing and @xmath102 as the larger level spacing .
then the histogram of the ratio @xmath103 can be compared to goe level statistics @xcite .
the even sector results in the f1 phase for 100 realizations at @xmath96 are shown in fig .
2 . we see the expected strong level repulsion , consistent with the thermalization . all other phases and symmetry sectors
were also examined and the results are also thermal , since phase f1 is the lowest - energy phase and thus the most difficult to thermalize .
next , we examine a ` distance ' between two eigenstates that are adjacent in the energy spectrum by comparing their probability distributions for the total spin @xmath54 .
we define this distance between eigenstate 1 and eigenstate 2 as @xmath104 where @xmath105 and @xmath106 are the probability distributions of @xmath54 in eigenstates 1 and 2 , respectively .
we tested three different distances : @xmath107 the distance between an even parity state ( eigenstate 1 ) and the nearest - energy odd parity state ( eigenstate 2 ) , and similarly for @xmath108 and @xmath109 .
if a system thermalizes , each eigenstate is equivalent to a microcanonical ensemble characterized by its energy . for two eigenstates that are adjacent in energy , the energy difference @xmath110
, therefore we expect the eigenstate distances @xmath107 , @xmath108 and @xmath109 should decrease exponentially with @xmath5 . in phase f1 , since the spectrum consists of nearly - degenerate pairs of states , the energy differences satisfy @xmath111 .
thus we expect in phase f1 , @xmath112 .
in addition , if we choose the upper bound of the energy window we average over to be well within the f1 phase , we would also expect that eq . ( 11 ) holds , so the exponential decay rate of @xmath107 would be greater than those of @xmath108 and @xmath109 .
meanwhile , in phases other than f1 , we expect all three @xmath113 s are well coincident . the numerical results are shown in fig .
3 . as we expected , all these eigenstate distances decay exponentially with @xmath5 . in addition , @xmath107 in phase f1 is much smaller and decreasing much faster than the other two distances , and in phase f3 all three @xmath113 s are well coincident .
this demonstrates the clear distinction between phases f1 and f3 , and further tests the thermalization in phase f3 . in the ferromagnetic phases , the system spontaneously flips between magnetization up and down at a rate that behaves as @xmath114 , where the quantity @xmath82 is defined in eq . ( 9 ) .
the probability of the system having total magnetization zero ( or 1/2 for systems with odd @xmath5 ) also behaves as @xmath114 .
thus from a single many - body eigenstate @xmath115 we can obtain an estimate of the quantity @xmath116 as @xmath117 where @xmath118 is the probability distribution of the total magnetization @xmath54 in this eigenstate .
if the eth is true , the magnetization thermalizes and these estimates @xmath119 will converge to @xmath82 in the limit of large @xmath5 . in fig .
4 we show the mean ( @xmath120 ) and the standard deviation ( @xmath121 ) of @xmath119 within energy bins for systems of size @xmath122 and @xmath96 . the standard deviation decreases with increasing @xmath5 , as expected for these schrdinger cat states that obey the eth . in the limit of large @xmath5
every eigenstate at a given energy density will have the same probability distribution of the total magnetization . that distribution is a thermal distribution for the magnetizations that are accessed by thermal excitation and ` tails ' due to quantum tunneling in the remainder of the distribution .
the model we are studying here is a ferromagnet ( @xmath11 ) with a small added spin - glass term ( @xmath47 ) in its hamiltonian . in terms of its affect on the system s thermodynamics , the relative strength of the spin - glass term scales as @xmath123 for the case
@xmath99 that we have here .
thus it is perhaps reasonable to expect the mean value of @xmath119 to exhibit a finite - size correction that vanishes for large @xmath5 as @xmath123 .
the spin - glass term weakly frustrates the ferromagnetism , so will cause a reduction in the apparent @xmath116 .
given the very modest range of @xmath5 for which we can do exact diagonalizations , this @xmath123 is a very slow convergence towards the thermodynamic limit . in fig .
4 we also show the expected value of @xmath116 in the limit of large @xmath5 . as @xmath5
is increased from 11 to 15 , @xmath120 does indeed increase slowly toward @xmath82 , as expected .
thus the results of these exact diagonalizations provide some numerical support for the theoretical results derived above assuming this system obeys the eth
. of course , given the very small size systems that can be diagonalized , there are strong finite - size effects that make a detailed demonstration of thermalization in the large system limit not possible .
this is the situation with all numerical tests of thermalization .
in addition to the model discussed above , we tested two other ways of adding disorder to @xmath11 .
the other hamiltonians that we diagonalized are @xmath124 and @xmath125 where in @xmath126 , the @xmath127 are also independent gaussian random variables of mean zero and variance one , and the power @xmath128 here satisfies @xmath129 ; specifically we looked at @xmath130 . in @xmath131 ,
the @xmath132 are independent random variables uniformly distributed in the interval @xmath133 $ ] .
these latter two models only have @xmath5 random parameters , one per spin , unlike ( 2 ) which has one per pair of spins : @xmath134 random couplings .
this difference appears to be quantitatively important , as neither of these latter two models showed good evidence of thermalization under the tests illustrated above for the sizes that can be exactly diagonalized .
but we expect that in the limit of large @xmath5 this difference should go away , with all of these models thermalizing .
we thank mike kolodrubetz ( mk ) and hyungwon kim for discussions , including preliminary numerical work by mk on a related model .
this work was supported in part by the national science foundation under dmr0819860 , and by the darpa ole program .
|
we explore a transverse - field ising model that exhibits both spontaneous symmetry - breaking and eigenstate thermalization . within its ferromagnetic phase , the exact eigenstates of the hamiltonian of any large but finite - sized system are all schrdinger cat states : superpositions of states with ` up ' and ` down ' spontaneous magnetization .
this model exhibits two dynamical phase transitions _ within _ its ferromagnetic phase : in the lowest - temperature phase the magnetization can macroscopically oscillate between up and down .
the relaxation of the magnetization is always overdamped in the remainder of the ferromagnetic phase , which is divided in to phases where the system thermally activates itself _ over _ the barrier between the up and down states , and where it quantum tunnels .
|
the @xmath1 meson plays a special role in understanding low
energy quantum chromo dynamics ( qcd ) .
chiral symmetry , its realization in hadron physics at low energies and the role of explicit chiral symmetry breaking due to the masses of the light quarks @xmath5 can be investigated using @xmath1 decays .
this work focuses on the anomalous sector of qcd , which is manifested in the radiative decays of the @xmath1 meson .
the radiative decay @xmath0 is the fourth strongest decay mode of the @xmath1 meson with a branching ratio of 4.60 @xmath3 0.16% @xcite .
conservation of charge conjugation and angular momentum including parity constrain the dynamics of the decay products .
the photon and the @xmath1 meson are eigenstates of the charge symmetry transformation with the eigenvalues c = 1 and c = + 1 , respectively .
therefore , due to c invariance the @xmath4 system must have c = 1 . to ensure c invariance
, the orbital angular momentum @xmath6 between the two pions must be odd .
all involved particles have negative parity .
consequently , parity invariance demands that the orbital angular momentum @xmath7 between the photon and the two - pion system must also be odd .
finally , total angular momentum conservation incorporating the fact that the intrinsic spin of the photon is unity leads to the requirement @xmath8 .
thus , the lowest partial waves which contribute are _ p_-waves
. presumably , higher partial waves with @xmath9 are not very important .
radiative decays of the @xmath1 meson are driven by the chiral anomaly of qcd .
the effects of the anomaly have been summarized by wess and zumino in an effective lagrangian @xcite . as shown by witten ,
this lagrangian is an essential part of effective field theories , because it is necessary in order to correctly incorporate the parity transformation of qcd @xcite . at the chiral limit of zero momentum and massless
quarks the decay @xmath0 is determined by the box anomaly term of the wess - zumino - witten lagrangian , which describes the direct coupling of three pseudoscalar mesons and a photon .
the dynamic range of the decay is limited by two pion rest masses and the @xmath1 mass , @xmath10 , and is , thus , far from the chiral limit . as a consequence ,
the decay rate calculated from the box anomaly term at the tree level is smaller by a factor of two compared to the measured value .
higher order terms of the chiral lagrangian have to be taken into account to achieve a correct description of the decay @xmath0 .
calculations at the one - loop level show an improved agreement between experiment and theory @xcite . but there remains , however , a significant difference .
several efforts have been made to include final state interactions by unitarized extensions to the box - anomaly term , e.g. a momentum dependent vector meson dominance ( vmd ) model @xcite , the hidden local symmetry model @xcite , an omnes - function which accounts for the _ p_-wave pion scattering phase shift @xcite , and a chiral unitary approach @xcite . to test the validity of the different models , not only the decay rate but also differential distributions of the dalitz plot variables need to be compared with experimental data .
for this purpose it is useful to parameterize the dalitz plot in terms of the photon energy @xmath11 in the rest frame of the @xmath1 meson and the angle @xmath12 of the @xmath13 relative to the photon in the pion - pion rest frame .
e@xmath14 is related to the squared invariant mass of the pion pair @xmath15 according to @xmath16 the photon energy distribution has been subject of only a few measurements forty years ago @xcite .
the results of the two statistically most significant publications @xcite are presented without acceptance corrections . instead
, the models which have been used for the interpretation of the data were folded with the acceptance and the results seem to be inconsistent @xcite . due to the limitations of the currently available experimental data the potential of the decay @xmath0 to provide insight to the anomalous sector of qcd can not be fully exploited . in order to perform compelling tests of the box anomaly and its higher order terms in chiral perturbation theory as well as of the non - perturbative extensions of the box - anomaly terms a new measurement with higher precision
is called for .
the results presented in this paper are based on a measurement with the wasa detection system @xcite installed at the cooler synchrotron cosy @xcite at the forschungszentrum jlich in germany .
a pellet target system produces small spheres of frozen hydrogen or deuterium which interact with the ion beam of the accelerator .
the interaction region is surrounded by a central detector covering scattering angles from 20 to 169 degrees .
it consists of a straw tube drift chamber , which is operated in the magnetic field of 0.85 t provided by a superconducting solenoid for the momentum reconstruction of charged particles , an electromagnetic calorimeter to measure energies of charged as well as neutral particles and thin plastic scintillators to discriminate charged and neutral particles already at the trigger level .
energy loss patterns in the calorimeter and the plastic scintillators allow to identify charged particles . for the identification and reconstruction of particles emitted at polar angles from 3 to 18 degrees
a forward detector is used . while track coordinates are measured precisely by a straw tube drift chamber ,
the kinetic energies of the ejectiles are reconstructed from the signals in plastic scintillators of different thickness , using the @xmath17e - e method .
the @xmath1 mesons have been produced in the fusion reaction pd@xmath2he x at a proton beam momentum of 1.7 gev / c .
this corresponds to an excess energy of 60 mev in the center of mass for the reaction pd@xmath2he @xmath1 at a cross section of 0.412 @xmath18b @xcite . at the trigger level events with one track in the forward detector with a high energy deposit in the thin scintillator detectors close to the exit window of the scattering chamber
have been accepted .
this condition selects @xmath19he ions without bias on the @xmath1 decay system .
( left ) @xmath17e - e plot for tracks emitted at scattering angles between 3 and 18 degree .
@xmath19he is well separated from protons and deuterons .
( right ) missing mass of @xmath19he for all events in the @xmath19he band .
the number of events in the peak above the background is 1.1@xmath20.,title="fig : " ] ( left ) @xmath17e - e plot for tracks emitted at scattering angles between 3 and 18 degree .
@xmath19he is well separated from protons and deuterons .
( right ) missing mass of @xmath19he for all events in the @xmath19he band .
the number of events in the peak above the background is 1.1@xmath20.,title="fig : " ] fig .
[ fig : hemm ] shows the energy loss correlations of all tracks in the forward detector .
the structure attributed to @xmath19he is clearly visible and can be easily selected .
it is well separated from the structures of protons and deuterons , which are strongly suppressed due to the trigger conditions .
the right panel of fig .
[ fig : hemm ] shows the inclusive missing mass distribution of @xmath19he .
a pronounced peak at the @xmath1 meson mass is visible .
it contains 1.1@xmath21 tagged mesons on top of a continuous background which originates from direct multi - pion production .
the cross sections of the relevant background processes and are unknown at the center - of - mass energy of this measurement .
however , they can be estimated , by extrapolation of results at lower energies @xcite , to be about 5 @xmath18b and 0.4 @xmath18b , respectively .
the missing mass spectrum in the right panel of fig .
[ fig : hemm ] illustrates that the direct reactions contribute to the background for @xmath1 decays only in a range of a few mev / c@xmath22 around the @xmath1 meson mass .
this remaining background is subtracted in a model independent way , bin by bin from the final spectra .
for the selection of the final state , one charged particle , which is identified as @xmath19he , is required in the forward detector . in the central detector ,
two charged particle tracks of opposite curvature and one neutral particle track with an energy deposit in the calorimeter of at least 20 mev are required in addition .
both charged particles are identified as pions .
background events which are picked up by the selection rules stem mainly from charged multi - pion production and other @xmath1 decay modes .
correlation plot for laboratory photon energy vs charged track photon opening angle ( monte carlo ) ( left : ) signal , ( right : ) @xmath23he@xmath4 reaction .
the graphical cut to suppress reaction is shown.,title="fig : " ] correlation plot for laboratory photon energy vs charged track photon opening angle ( monte carlo ) ( left : ) signal , ( right : ) @xmath23he@xmath4 reaction .
the graphical cut to suppress reaction is shown.,title="fig : " ] two - pion production is expected to form the largest background contribution in the selected data due to the large cross section and the possibility of cluster splitoffs in the calorimeter , which would fake the signal of a photon candidate in the actually photon free final state .
splitoffs are characterized by a small energy deposit and a small distance to either of the pion candidates in the calorimeter .
therefore , this background contributes predominantly to the low end of the reconstructed @xmath11 spectrum . in order to increase the signal to background ratio in this region a correlated condition
is imposed on the energy deposits of the photon candidates and the distance between the shower positions of photon and pion candidates in the calorimeter .
the latter is measured by the opening angle between the reconstructed cluster positions . in fig .
[ fig : split ] the condition is shown as graphical cut with a dashed curve .
a strong enhancement is observed at low photon energies and small opening angles between photon and pion candidates from the contribution of the reaction .
the cut used is the best compromise between a high signal - to - background ratio and a large reconstruction efficiency of the signal channel .
the contribution from two - pion production is reduced by more than 55% by rejecting all events below the curve .
final states with three pions contribute to the background only if one photon of the @xmath24 decay remains undetected .
the @xmath24 can be identified from the distribution of the squared missing mass of the system , as demonstrated in fig .
[ fig : back ] . here
, the experimental missing mass spectrum is reproduced with monte carlo distributions of the the signal ( @xmath0 ) and background ( @xmath25 ) contributions from @xmath1 decays as well as direct two- and three - pion production .
the cross sections of the direct pion processes were fitted using simultaneously the missing mass distribution and the missing mass for the selected data sample .
the result is in agreement with the expectations @xcite .
the contributions of the different @xmath1 meson decay channels are fixed by the known branching ratios @xcite .
the discrepancy between monte carlo and experiment for negative masses might be attributed to the unknown production mechanisms of the direct processes . in the simulations isotropic phase space population has been assumed .
the discrepancy disappears when the contribution of the direct production is subtracted bin - by - bin , as discussed later . by rejecting events with a squared missing mass value larger than 0.0125 gev@xmath22/c@xmath26 , as indicated with a vertical dashed line in fig .
[ fig : back ] , @xmath27 of the background from the final states in the mass region of the @xmath1 meson is removed .
the @xmath24 signal in the squared missing mass of the @xmath19he @xmath4 system .
the dashed vertical line corresponds to a squared missing mass of 0.0125 gev@xmath22/c@xmath26 . ]
additional suppression of background is achieved by a kinematic fit of the complete final state to the reaction hypothesis , using four - momentum conservation as the only fit constraint .
the uncertainties of the kinematic variables at the input to the fit have been extracted depending on energy and angle from a geant monte carlo simulation , tuned to match the experimental resolutions . after the fit , all events with a probability of less than 10% are rejected .
the invariant mass distribution of the fitted @xmath28 system is shown for the remaining events in fig .
[ fig : kfit ] .
the condition on the probability distribution additionally suppresses background .
in particular a comparison with monte carlo shows a reduction of the @xmath19he@xmath29 background to the level of @xmath30 in the selected events . the invariant mass spectrum of the fitted @xmath28 system with a probability above 10% .
the monte carlo cocktail consists of the signal ( @xmath0 ) and background contributions ( @xmath25 ) from @xmath1 decays ( diagonally hatched ) as well as direct two - pion ( vertically hatched ) and three - pion ( horizontally hatched ) production . ] the remaining background is subtracted bin by bin from the @xmath11 and @xmath31 distributions .
the bin size of 5 mev for the photon energy and 0.1 for the pion angular distributions is chosen to reflect the resolution achieved in the respective observables .
the bin width of the angular distribution is similar to previous measurements @xcite the bin width of the @xmath11 spectrum is smaller by about a factor of two .
the left panel in fig .
[ fig : bgsub ] shows the correlation between the photon energy and the invariant mass of the @xmath28 system .
two structures can be seen in the plot .
one at the mass of the @xmath1 meson containing the signal events and another structure , showing a correlation of photon energy and invariant mass , caused mainly by background from two - pion production .
the invariant mass of the @xmath28 system is calculated for the events in each bin of the @xmath11 and @xmath31 distributions .
this is illustrated for one bin of the @xmath11 distribution in the right panel of fig .
[ fig : bgsub ] . since both ,
cross sections and differential distributions are unknown for multi - pion production at the energy of this measurement , the individual mass spectra are fitted in two ways in order to determine the amount of continuous background in the @xmath1 mass region . in the first approach a function that is the sum of a lorentzian for the signal and an exponential for the background
is used . in the second method
the background is fitted with an exponential function without making an assumption on the signal shape by excluding the range of the signal peak from the fit .
the excluded region was determined as the @xmath32 region of a gaussian fit to the peak of the @xmath1 meson .
the individual results of the fits using both methods are shown in the right panel of fig .
[ fig : bgsub ] with solid and dashed curves , respectively .
the average number of the background events from both fits is used to calculate the number of events from the @xmath1 meson decay in the corresponding bin of the @xmath11 or @xmath31 distribution .
background subtraction from the e@xmath14 distribution . *
left : * correlation of e@xmath14 and the invariant mass of the @xmath28 system .
* right : * illustration of the background subtraction methods with the invariant mass spectrum corresponding to the photon energy region : determination of the background shape by fitting signal and background ( solid curve ) and by excluding the signal range from the fit of the background ( dashed curve).,title="fig : " ] background subtraction from the e@xmath14 distribution . * left : * correlation of e@xmath14 and the invariant mass of the @xmath28 system . *
right : * illustration of the background subtraction methods with the invariant mass spectrum corresponding to the photon energy region : determination of the background shape by fitting signal and background ( solid curve ) and by excluding the signal range from the fit of the background ( dashed curve).,title="fig : " ] the remaining background from the decay in the signal region is subtracted using scaled monte carlo distributions .
the scaling factors are determined by a fit of the experimental spectrum of the squared missing mass of the system with monte carlo distributions of the relevant@xmath1 decay modes , @xmath0 ( signal ) , @xmath33 and @xmath34 ( background ) , after subtraction of the continuous background from multi - pion production . the contribution of determined to be 7% .
this background is isotropic in @xmath31 and contributes to the e@xmath14 distribution in the energy region above 50 mev with a maximum at 120 mev .
the total statistics in the final distributions is @xmath35 events of the decay .
this is the largest number of events from an exclusive measurement of this decay mode . for acceptance corrections the general form of the squared matrix element for the @xmath0 decay @xmath36 with @xmath37 being the pion momentum in the pion - pion rest frame , and the form factor @xmath38 according to the vmd calculations in ref .
@xcite has been used .
using instead monte carlo distributions based on the simplest gauge invariant matrix element ( @xmath39 ) does not alter the experimental result significantly .
thus , it can be concluded that systematic effects due to the applied form factors are negligible .
the acceptance varies smoothly as a function of @xmath11 and @xmath31 . for photon energies less than 10 mev the acceptance becomes vanishingly small . in case of the angular distribution a reduced acceptance
is observed for small opening angles between each of the pions and the photon .
the reduction of the acceptance in both variables is found to be correlated .
it is caused by the method of two - pion suppression presented in fig .
[ fig : back ] , where a condition on the correlation of photon energy and opening angle between pion and photon candidates is used .
[ fig : final ] shows the background subtracted and acceptance corrected photon energy and pion angular distributions with the statistical errors . the e@xmath14 distribution is also given numerically in tab . [ tab : egfinal ] . in the upper panel of fig .
[ fig : final ] the final distribution of @xmath31 is shown .
it can be described by , as indicated by the dashed curve .
thus , the measurement is consistent with the relative _
p_-wave assumed in eq .
[ eq : sgime ] . the background subtracted and acceptance corrected angular distribution of the pions ( top ) and the photon energy distribution ( bottom ) , with error bars indicating the statistical uncertainties .
the angular distribution is compared with a relative _ p_-wave of the pions ( dashed curve ) .
the shape of the photon energy distribution is confronted with predictions of the square of the simplest gauge invariant matrix element , eq .
[ eq : sgime ] ( dashed curve ) , multiplied by the squared modulus of the pion vector form factor @xmath40 ( dotted curve ) and further multiplied by @xmath41 , the square of a real polynomial of first order , with its coefficient fitted to the data ( solid curve ) .
all curves are normalized to the same integral . ]
the background subtracted and acceptance corrected angular distribution of the pions ( top ) and the photon energy distribution ( bottom ) , with error bars indicating the statistical uncertainties .
the angular distribution is compared with a relative _ p_-wave of the pions ( dashed curve ) .
the shape of the photon energy distribution is confronted with predictions of the square of the simplest gauge invariant matrix element , eq .
[ eq : sgime ] ( dashed curve ) , multiplied by the squared modulus of the pion vector form factor @xmath40 ( dotted curve ) and further multiplied by @xmath41 , the square of a real polynomial of first order , with its coefficient fitted to the data ( solid curve ) .
all curves are normalized to the same integral . ] the photon energy distribution in the @xmath1 rest frame is shown in the lower panel of fig .
[ fig : final ] . the line shape obtained from eq .
[ eq : sgime ] , normalized to the integrated rate , is given by the dashed curve .
it does not describe the experimental data , which in comparison are shifted significantly towards lower energies .
the observed disagreement confirms the findings of the previous measurements @xcite . in order to achieve a correct description of the photon energy spectrum in addition to the already properly described angular distribution , eq .
[ eq : sgime ] can be multiplied by an energy dependent form factor @xmath42 .
the origin of the deviation in the e@xmath14 distribution is predominantly given by the @xmath43 final state interaction in the vector channel .
unitarity and analyticity dictate that this effect should be given by the pion vector form factor @xmath44 multiplied by a polynomial @xmath45 that parameterizes contributions that do not contain the @xmath43 unitarity cut . for a detailed discussion about the multiplier @xmath46 of eq .
[ eq : sgime ] see _ e.g. _ refs .
the pion vector form factor is experimentally directly accessible via @xmath47 or may be derived using the omnes representation from the @xmath43 elastic phase shifts in the vector channel here the representation for @xmath48 derived in ref .
@xcite is applied .
the uncertainty in this form factor is negligible compared to the experimental uncertainties , which are presented in detail in the next paragraph .
furthermore , following refs .
@xcite , we parameterize the term @xmath45 as a real polynomial of first order : @xmath49 the parameter @xmath50 can then be determined from a fit to the data , which is shown with the solid curve in the lower panel of fig .
[ fig : final ] .
the result for @xmath51 is shown as the dotted curve .
tests for systematic uncertainties of @xmath50 have been performed by varying in the analysis chain one by one the conditions for the suppression of splitoffs , the cut on the missing mass of the @xmath19he@xmath4 system , the condition on the probability of the kinematic fit , the method of subtracting the background from direct multi - pion production , the @xmath33 contribution and the model used for the acceptance correction .
additionally , subsets of the data , collected with different experimental settings , allowed to cross check the influences of luminosity variations and different rf settings during the measurement . using different methods to subtract the continuous multi - pion background , different models to perform the acceptance correction or using different rf settings does not cause statistically significant deviations from the original result .
the contributions to the systematic uncertainty of @xmath50 derived from the other tests are listed in tab .
[ tab : alpha_sys ] .
the overall systematic error is obtained by adding the contributions quadratically .
one of the largest uncertainties results from the fluctuations of the luminosity during data taking .
the variation of the final result with the chosen luminosity can be explained by the accuracy of the simulations concerning pile - up effects and beam - target overlap parameters , which have not been included in a systematic way .
.[tab : alpha_sys]summary of contributions to the systematic uncertainty of the parameter @xmath50 .
[ cols="^,^",options="header " , ]
for the first time , background subtracted and acceptance corrected differential distributions of the decay @xmath0 have been extracted in the analysis of exclusive data .
the distributions clearly show the importance of final state interactions .
the shape of the @xmath11 spectrum can be very well described by a parameterization that includes the factors required by gauge invariance and the centrifugal barrier as well as the pion vector form factor times a first - order polynomial written as @xmath52 .
a fit to the data gives . in order to shed further light on the anomalous sector of qcd , future theoretical studies will have to explain simultaneously both the value of @xmath50 as well as the branching ratio for @xmath0 . in recent production runs of the wasa facility at cosy
further data on @xmath1 decays have been taken with high statistics . from a preliminary analysis
at least an order of magnitude more fully reconstructed @xmath0 events is expected .
the analysis of the acquired data will significantly decrease not only the statistical but also the systematic uncertainties by an improved understanding of background contributions .
the data will also be used to determine the branching ratio of the decay @xmath0 .
a recent measurement of the cleo collaboration @xcite shows a relative branching ratio which differs by more than three standard deviations from the results of previous measurements @xcite . due to the unbiased tagging of @xmath1 mesons in the reaction it is not only possible to extract relative but also absolute branching ratios at the wasa facility
this will help to resolve the discrepancy .
we thank f .- k .
guo for providing us with the code to calculate the pion vector form factor .
this work was in part supported by : the forschungszentrum jlich including the jchp - ffe program , the european commission under the 7th framework programme through the research infrastructures action of the capacities programme .
call : fp7-infrastructures-2008 - 1 , grant agreement n. 227431 , the german bmbf , the german - indian daad - dst exchange program , viqcd ( vh - vi-231 ) , the german research foundation ( dfg ) , and the polish national science centre and foundation for polish science - mpd program , co - financed by the european union within the european regional development fund .
we gratefully acknowledge the financial support given by the knut and alice wallenberg foundation , the swedish research council , the gran gustafsson foundation , the polish ministry of science and higher education under the grant pbs 7p - p6 - 2/07 .
|
an exclusive measurement of the decay @xmath0 has been performed at the wasa facility at cosy .
the @xmath1 mesons were produced in the fusion reaction pd@xmath2he x at a proton beam momentum of 1.7 gev / c .
efficiency corrected differential distributions have been extracted based on 13960@xmath3140 events after background subtraction . the measured pion angular distribution is consistent with a relative _ p_-wave of the two - pion system , whereas the measured photon energy spectrum was found at variance with the simplest gauge invariant matrix element of @xmath0 .
a parameterization of the data can be achieved by the additional inclusion of the empirical pion vector form factor multiplied by a first - order polynomial in the squared invariant mass of the @xmath4 system .
@xmath1 meson , box anomaly , exclusive measurement 14.40.be , 13.20.jf , 12.39.fe
|
hilbert s program was launched , in 1922 , with the specific goal of demonstrating the consistency of modern , set - theoretic methods , using only finitary means .
but the program can be viewed more broadly as a response to the radical methodological changes that had been introduced to mathematics in the late nineteenth century .
central to these changes was a shift in mathematical thought whereby the goal of mathematics was no longer viewed as that of developing powerful methods of calculation , but , rather , that of characterizing abstract , possibly infinite , mathematical structures , often in ways that could not easily be reconciled with a computational understanding .
grisha s work over the years has touched on almost every aspect of proof theory , both of the reductive ( foundational ) and structural sort , involving a wide range of logical frameworks . but
much of his work addresses the core proof - theoretic concern just raised , and has served to provide us with a deep and satisfying understanding of the computational content of nonconstructive , infinitary reasoning .
such work includes his characterization of the provably total computable functions of @xmath0 as exactly the primitive recursive functions ( @xcite ) ; his method of continuous cut elimination , which provides a finitary interpretation of infinitary cut - elimination methods ( @xcite ) ; and his work on the epsilon substitution method ( for example , @xcite ) .
grisha has also been a friend and mentor to me throughout my career .
the characterization of the provably total computable functions of @xmath0 just mentioned was , in fact , also discovered by charles parsons and gaisi takeuti , all independently .
i shudder to recall that at a meeting at oberwolfach in 1998 , when i was just two - and - a - half years out of graduate school , i referred to the result as `` parsons theorem '' in a talk before an audience that , unfortunately , included only the other two .
grisha asked the first question after the talk was over , and nothing in his manner or tone even hinted that i had made a _
( it did nt even occur to me until much later ) .
in fact , i still vividly remember his encouraging and insightful comments , then and in later discussion .
( for the record , gaisi was equally gracious and supportive . ) in this essay , i will discuss methods of interpreting classical first - order arithmetic , often called _ peano arithmetic _
( @xmath1 ) , in computational terms .
although the study of @xmath1 was central to hilbert s program , it may initially seem like a toy theory , or an artificially simple case study .
after all , mathematics deals with much more than the natural numbers , and there is a lot more to mathematical argumentation than the principle of induction .
but experience has shown that the simplicity of the theory is deceptive : via direct interpretation or more elaborate forms of proof - theoretic reduction , vast portions of mathematical reasoning can be understood in terms of @xmath1 @xcite . here
, i will be concerned with the @xmath2 , or `` computational , '' consequences of @xmath1 .
suppose @xmath1 proves @xmath3 , where @xmath4 and @xmath5 range over the natural numbers and @xmath6 is a decidable ( say , primitive recursive ) relation .
we would like to understand how and to what extent such a proof provides an _ algorithm _ for producing such a @xmath5 from a given @xmath4 , one that is more informative than blind search .
there are four methods that are commonly used to extract such an algorithm : 1 .
gdel s _ dialectica _ interpretation @xcite , in conjunction with a double - negation interpretation that interprets @xmath1 in its intuitionistic counterpart , heyting arithmetic ( @xmath7 ) 2 .
realizability @xcite , again in conjunction with a double - negation translation , and either the friedman @xmath8-translation @xcite ( often also attributed to dragalin and leivant , independently ) or a method due to coquand and hofmann ( @xcite ) to `` repair '' translated @xmath2 assertions 3 .
cut elimination ( @xcite ; see , for example , @xcite ) 4 .
the epsilon substitution method ( @xcite ) these four approaches really come in two pairs : the dialectica interpretation and realizability have much in common , and , indeed , paulo oliva @xcite has recently shown that one can interpolate a range of methods between the two ; and , similarly , cut elimination and the epsilon substitution method have a lot in common , as work by grisha ( e.g. @xcite ) shows . that is not to say that there are nt significant differences between the methods in each pairing , but the differences between the two pairs are much more pronounced .
for one thing , they produce two distinct sorts of `` algorithms . ''
the dialectica interpretation , and kreisel s `` modified '' version of kleene s realizability , extract terms in gdel s calculus of primitive recursive functionals of finite type , denoted @xmath9 in section [ preliminaries : section ] below .
in contrast , cut elimination and the epsilon substitution method provide iterative procedures , whose termination can be proved by ordinal analysis . specifically , one assigns ( a notation for ) an ordinal less than @xmath10 to each stage of the computation in such a way that the ordinals decrease as the computation proceeds .
terms in @xmath9 and @xmath11-recursive algorithms both have computational meaning , and there are various ways to `` see '' that the computations terminate ; but , of course , any means of proving termination formally for all such terms and algorithms has to go beyond the means of @xmath1 .
second , as indicated above , the first two methods involve an intermediate translation to @xmath7 , while the second two do not .
it is true that the dialectica interpretation and realizability can be applied to classical calculi directly ( see @xcite for the dialectica interpretation , and , for example , @xcite for realizability ) ; but i know of no such interpretation that can not be understood in terms of a passage through intuitionistic arithmetic @xcite .
in contrast , cut elimination and the epsilon substitution method apply to classical logic directly .
that is not to deny that one can apply cut elimination methods to intuitionistic logic ( see , for example , @xcite ) ; but the arguments tend to be easier and more natural in the classical setting .
finally , there is the issue of canonicity .
algorithms extracted from proofs in intuitionistic arithmetic tend to produce canonical witnesses to @xmath2 assertions ; work by grisha @xcite shows , for example , that algorithms extracted by various methods yield the same results .
in contrast , different ways of extracting witnesses from classical proofs yield different results , conveying the impression that there is something `` nondeterministic '' about classical logic .
( there is a very nice discussion of this in @xcite .
see also the discussion in section [ conclusions : section ] below . )
insofar as one has a natural translation from classical arithmetic to intuitionistic arithmetic , some of the canonicity of the associated computation is transferred to the former theory . in this essay
, i will discuss realizability and the dialectica interpretation , as they apply to classical arithmetic , via translations to intuitionistic arithmetic .
after reviewing some preliminaries in section [ preliminaries : section ] , i will discuss variations of the double - negation interpretation in section [ double : negation : section ] .
one , in particular , is very efficient when it comes to introducing negations ; in section [ realizability : section ] , i will show that , when combined with realizability or the dialectica interpretation , this yields computational interpretations of classical arithmetic that are efficient in their use of higher types . in section [ herbrand : section
] , i will consider another curious double - negation interpretation , and diagnose an unfortunate aspect of its behavior .
this work has been partially supported by nsf grant dms-0700174 and a grant from the john templeton foundation .
i am grateful to philipp gerhardy , thomas streicher , and an anonymous referee for helpful comments and corrections .
somewhat imprecisely , one can think of intuitionistic logic as classical logic without the law of the excluded middle ; and one can think of minimal logic as intuitionistic logic without the rule _
ex falso sequitur quodlibet _
, that is , from @xmath12 conclude anything .
computational interpretations of classical logic often pass through minimal logic , which has the nicest computational interpretation .
( one can interpret intuitionistic logic in minimal logic by replacing every atomic formula @xmath8 by @xmath13 , so the difference between these two is small . ) to have a uniform basic to compare the different logics , it is useful to take the first - order logical symbols to be @xmath14 , @xmath15 , @xmath16 , @xmath17 , @xmath18 , and @xmath12 , with @xmath19 defined to be @xmath20 .
however , when it comes to classical logic , it is often natural to restrict one s attention to formulas in _ negation - normal form _ , where formulas are built up from atomic and negated atomic formulas using @xmath16 , @xmath17 , @xmath14 , and @xmath15 . a negation operator , @xmath21 , can be defined for such formulas ; @xmath21 is what you get if , in @xmath22 , you exchange @xmath16 with @xmath17 , @xmath14 with @xmath15 , and atomic formulas with their negations . note that @xmath23 is just @xmath22 .
classically , every formula @xmath22 has a negation - normal form equivalent , @xmath24 , obtained by defining @xmath25 to be @xmath26 , and treating the other connectives in the obvious way .
this has the slightly awkward consequence that @xmath27 translates to @xmath28 , but simplifying @xmath29 to @xmath30 and @xmath31 to @xmath12 easily remedies this .
there are a number of reasons why negation - normal form is so natural for classical logic .
first of all , it is easy to keep track of polarities : if @xmath22 is in negation - normal form , then every subformula is a positive subformula , except for , perhaps , atomic formulas ; an atomic formula @xmath8 occurs positively in @xmath22 if it occurs un - negated , and negatively if it occurs with a negation sign before it .
second , the representation accords well with practice : any classically - minded mathematician would not hesitate to prove `` if @xmath22 then @xmath32 '' by assuming @xmath33 and deriving @xmath19 , or by assuming @xmath22 and @xmath33 and deriving a contradiction ; so it is convenient that @xmath34 , @xmath35 , and @xmath36 have the same negation - normal form representation .
finally , proof systems for formulas in negation - normal form tend to be remarkably simple ( see , for example , @xcite ) .
it was gdel @xcite who first showed that the provably total computable functions of arithmetic can be characterized in terms of the primitive recursive functionals of finite type ( see @xcite ) .
the set of finite types can be defined to be the smallest set containing the symbol @xmath37 , and closed under an operation which takes types @xmath38 and @xmath39 to a new type @xmath40 . in the intended ( `` full '' ) interpretation
, @xmath37 denotes the set of natural numbers , and @xmath41 denotes the set of all functions from @xmath38 to @xmath39 .
a set of terms , @xmath9 , is defined inductively as follows : 1 . for each type @xmath38
, there is a stock of variables @xmath42 of type @xmath38 .
2 . @xmath43 is a term of type @xmath37 .
3 . @xmath44 ( successor ) is a term of type @xmath45 .
4 . if @xmath46 is a term of type @xmath47 and @xmath48 is a term of type @xmath39 , then @xmath49 is a term of type @xmath38 .
5 . if @xmath46 is a term of type @xmath38 and @xmath4 is a variable of type @xmath39 , then @xmath50 is a term of type @xmath51 .
if @xmath46 is a term of type @xmath38 , and @xmath48 is a term of type @xmath52 , then @xmath53 is a term of type @xmath54 .
intuitively , @xmath49 denotes the result of applying @xmath46 to @xmath48 , @xmath55 denotes the function which takes any value of @xmath4 to @xmath46 , and @xmath53 denotes the function defined from @xmath46 and @xmath48 by primitive recursion , with @xmath56 and @xmath57 for every @xmath4 . in this last equation ,
i have adopted the convention of writing @xmath58 instead of @xmath59 .
it will be convenient below to augment the finite types with products @xmath60 , associated pairing operations @xmath61 , and projections @xmath62 and @xmath63 .
product types can be eliminated in the usual way by currying and replacing terms @xmath48 with sequences of terms @xmath64 .
it will also be convenient to have disjoint union types @xmath65 , an element of which is either an element of @xmath38 or an element of @xmath39 , tagged to indicate which is the case .
that is , for each such type we have insertion operations , @xmath66 and @xmath67 , which convert elements of type @xmath38 and @xmath39 respectively to an element of type @xmath65 ; predicates @xmath68 and @xmath69 , which indicate whether @xmath70 is tagged to be of type @xmath38 or @xmath39 ; and functions @xmath71 and @xmath72 , which interpret @xmath70 as an element of type @xmath38 and @xmath39 , respectively .
references to such sum types can be eliminated by taking @xmath65 to be @xmath73 , defining @xmath74 , defining @xmath75 , where @xmath76 and @xmath77 are constant zero functionals of type @xmath38 , @xmath39 respectively , and so on . in the next section
, i will describe various double - negation interpretations that serve to reduce classical arithmetic , @xmath1 , to intuitionistic arithmetic , @xmath7 in fact , to @xmath7 taken over _
minimal logic_. these show that if @xmath1 proves a @xmath2 formula @xmath78 , then @xmath7 proves @xmath79 ; in fact , a variant @xmath80 of @xmath7 based on minimal logic suffices .
this reduces the problem to extracting computational information from the latter proof .
one method of doing so involves using kreisel s notion of modified realizability , combined with the friedman a - translation .
one can extend @xmath7 to a higher - type version , @xmath81 , which has variables ranging over arbitrary types , and terms of all the primitive recursive functionals .
fix any primitive recursive relation @xmath82 ; then to each formula @xmath83 in the language of arithmetic , one inductively assigns a formula `` @xmath84 , '' as follows .
@xmath85 now , suppose classical arithmetic proves @xmath78 , for some primitive recursive relation @xmath86 .
then , using a double - negation translation , @xmath80 proves @xmath87 , and hence it proves @xmath88 for a fresh constant @xmath89 .
fix @xmath82 in the realizability relation above to be the formula @xmath90 .
inductively , one can then extract from the proof of term @xmath48 of @xmath9 such that @xmath81 proves that @xmath48 realizes @xmath91 .
now notice that the identity function , @xmath92 , realizes @xmath93 , since a realizer to @xmath94 is simply a value of @xmath70 satisfying @xmath95 .
thus if @xmath70 realizes @xmath96 , then @xmath97 satisfies @xmath98 .
viewing @xmath70 , now , as a function of @xmath89 , yields the following conclusions : [ main : witnessing : thm ] if classical arithmetic proves @xmath87 , there is a term @xmath99 of @xmath9 of type @xmath37 to @xmath37 such that @xmath81 proves @xmath100 .
see @xcite for more about realizability , and @xcite for the a - translation .
gdel s dialectica interpretation provides an alternative route to this result .
in fact , one obtains a stronger conclusion , namely that the correctness of the witnessing term can be proved in a quantifier - free fragment @xmath101 of @xmath81 .
to each formula @xmath22 in the language of arithmetic , one inductively assigns a formula @xmath102 of the form @xmath103 , where @xmath4 and @xmath5 are now tuples of variables of appropriate types .
assuming @xmath104 , the assignment is defined as follows : @xmath105 the clause for implication is the most interesting among these , and can be understood as follows : from a witness , @xmath4 , to the hypothesis , @xmath106 is supposed to return a witness to the conclusion ; and given a purported counterexample , @xmath107 , to the conclusion , @xmath108 is supposed to return a counterexample to the hypothesis .
since we have defined @xmath109 to be @xmath110 , notice that @xmath111 is @xmath112 .
the dialectica interpretation of @xmath113 is @xmath114 , which is intuitionistically equivalent to @xmath115 , given the decidability of primitive recursive relations .
one can show that from a proof of @xmath22 in @xmath7 , one can extract a term @xmath99 such that for every @xmath4 , @xmath101 proves @xmath116 , once again yielding theorem [ main : witnessing : thm ] .
we have seen that one can use modified realizability or the dialectica interpretation to extract an algorithm from a proof of a @xmath2 statement in classical arithmetic , modulo a method of reducing classical arithmetic to intuitionistic arithmetic .
double negation translations provide the latter .
a formula is said to be _ negative _ if it does not involve @xmath15 or @xmath17 and each atomic formula @xmath8 occurs in the form @xmath117 ; in other words , the formula is built up from negated atomic formula using @xmath14 , @xmath16 , @xmath18 , and @xmath12 . over minimal logic ,
negative formulas are stable under double negation , which is to say , if @xmath22 is any negative formula , then @xmath7 proves that @xmath118 is equivalent to @xmath22 ( see , for example , @xcite ) .
the gdel - gentzen double - negation translation maps an arbitrary first - order formula @xmath22 to a negative formula , @xmath119 : @xmath120 the translation has the following properties : for any formula @xmath22 and set of sentences
@xmath121 : 1 .
classical logic proves @xmath122 2 .
if @xmath22 is provable from @xmath121 in classical logic , then @xmath119 is provable from @xmath123 in minimal logic .
since the @xmath7 proves the @xmath124 translations of its own axioms , we have as a corollary : if @xmath1 proves @xmath22 , then @xmath7 proves @xmath119 .
in fact , one can strengthen the corollary in three ways : 1 .
since @xmath7 proves @xmath125 for atomic formulas @xmath30 , one can define @xmath126 to be @xmath30 .
2 . assuming the language of @xmath7 includes , say , symbols denoting the primitive recursive functions , every negated atomic formula , @xmath127 , has an atomic equivalent , @xmath128 ; so one can define @xmath129 to be @xmath128 .
the theorem remains true if one replaces @xmath7 by a suitable variant , @xmath80 , based on minimal logic .
these considerations hold in the theorems that follow as well .
the reason to be concerned about negations is that they are undesirable with respect to the two computational interpretations given in the last section , since they lead to the use of more complicated types in the resulting terms of @xmath9 .
there is a variant of the double - negation translation known as the _ kuroda translation _ that fares slightly better in this regard : for any formula @xmath22 , let @xmath130 denote the result of doubly - negating atomic formulas , and adding a double negation _ after _ each universal quantifier , and , finally , adding a double - negation to the front of the formula .
then we have : for every formula @xmath22 , @xmath131 is provable in minimal logic
. hence @xmath1 proves @xmath22 if and only if @xmath7 proves @xmath130 .
note that intuitionistic logic , rather than minimal logic , is required in the conclusion .
late in 2005 , grisha asked whether a version of the dialectica interpretation designed by shoenfield @xcite , for classical arithmetic , could be understood as a composition of the usual dialectica interpretation together with a double - negation translation .
i set the question aside and solved it a few months later @xcite , only to find that ulrich kohlenbach and thomas streicher had solved it more quickly @xcite . in a way that can be made precise
, the shoenfield translation corresponds to the following version of the double - negation interpretation ( itself a variant of a translation due to krivine ) , expressed for a basis involving the connectives @xmath132 , @xmath16 , @xmath17 , and @xmath14 .
we define @xmath133 to be @xmath134 , where @xmath135 is defined recursively by clauses below .
it helps to keep in mind that @xmath135 is supposed to represent the _ negation _ of @xmath22 : @xmath136 note that we can eliminate either @xmath137 or @xmath138 and retain a complete set of connectives , but including them both is more efficient
. formulas of the form @xmath139 , however , have to be expressed as @xmath140 to apply the translation . for every formula @xmath22 , @xmath141 is provable in minimal logic
. hence @xmath1 proves @xmath22 if and only if @xmath80 proves @xmath133 .
the @xmath142-translation is particularly good when it comes to formulas in negation - normal form ; it only adds two quantifiers for each existential quantifier , as well as one at the beginning . but one can do even better @xcite . taking advantage of the classical negation operator ,
now @xmath143 is defined to be @xmath144 , where the map @xmath145 is defined recursively as follows : @xmath146 once again , we have for every formula @xmath22 in negation - normal form , @xmath147 is provable in minimal logic
. hence @xmath1 proves @xmath22 if and only if @xmath80 proves @xmath143 .
the @xmath148-translation is extremely efficient with respect to negations , introducing , roughly , one at the beginning of the formula , and one for every quantifier alternation after an initial block of universal quantifiers .
alternatively , can define @xmath149 in analogy to @xmath150 , as @xmath151 .
this gives the translation the nice property that given the formulas @xmath143 and @xmath152 , one is the negation of the other .
but there is a lot to be said for keeping negations to a minimum .
of course , the @xmath148-translation extends to all classical formulas by identifying them with their canonical negation - normal form equivalents .
since the translation relies on the negation - normal form representation of classical formulas , it shares many nice properties with a more complicated double - negation translation due to girard @xcite .
it is this translation that i will use , in the next section , to provide an efficient computational interpretation of classical arithmetic .
let me close with one more translation , found in @xcite , which is interesting in its own right . for reasons that will become clear later on , i will call it `` the awkward translation '' :
if @xmath22 is any formula in negation - normal form , let @xmath153 denote @xmath154 .
[ awk : prop ] for any formula @xmath22 , @xmath155 is provable in minimal logic .
hence , @xmath1 proves @xmath22 if and only if @xmath80 proves @xmath153 . once can show by induction that if @xmath32 is any formula in negation - normal form , then @xmath156 is provable in minimal logic .
so minimal logic proves that @xmath21 implies @xmath157 , and hence that @xmath158 implies @xmath153 .
but since @xmath21 is classically equivalent to @xmath19 , @xmath158 is equivalent to @xmath159 , which is implied by @xmath119 .
the @xmath160-translation is almost absurdly efficient with respect to negations : the one classical negation on the inside adds no negations at all ( recall that in arithmetic , negated atomic formulas have atomic equivalents ) , and the translation adds only one negation on the outside .
but the attentive reader will have noticed that the first assertion in theorem [ awk : prop ] is slightly weaker than the corresponding assertions in the the theorems that precede it : only one direction of the equivalence is minimally valid .
we will see , in section [ herbrand : section ] , that this means that the translation fares very poorly with respect to modus ponens , making it impossible to translate ordinary proofs piece by piece .
for pure first - order logic , an alternative proof of theorem [ awk : prop ] can be found in @xcite .
benno van den berg ( personal communication ) later hit upon this same translation , independently .
in @xcite , i claimed that with intuitionistic logic in place of minimal logic , the result is a consequence of a characterization of glivenko formulas due to orevkov , described in a very nice survey ( * ? ? ? * section 3.2.5 ) of russian proof theory by grisha . ;
the last class seven lines from the bottom of the page should be @xmath161 ; and the first class at the bottom line should be @xmath162 . ] that seems to be incorrect ; but van den berg and streicher have pointed out to me that in that case the result follows a theorem due to mints and orevkov ( * ? ? ? * , paragraph 4 ) .
we now obtain direct computational interpretations of classical arithmetic simply by combining the @xmath148-translation of section [ double : negation : section ] with the computational interpretations of @xmath80 given in section [ preliminaries : section ]
. one annoying consequence of the use of the classical negation operator in the @xmath148-translation is that it is impossible to carry out the translation of a formula @xmath22 from the inside out : depending on the context in which a subformula @xmath32 occurs , the computational interpretation of the full formula may depend on either the computational interpretation of @xmath32 or the computational interpretation of @xmath163 . in practice ,
then , it is often more convenient to carry out the interpretation in two steps , applying the @xmath148-translation first , and then one of the two computational interpretations described in section [ preliminaries : section ] . nonetheless , it is interesting to see what happens when the steps are composed , which is what i will do here .
both translations apply to formulas in negation - normal form , and we can assume that negated atomic subformulas are replaced by their atomic equivalents . as in section
[ preliminaries : section ] , the appropriate version of classical realizability is defined relative to a fixed primitive recursive predicate @xmath82 .
most of the clauses look just like ordinary modified realizability : @xmath164 one can then straightforwardly extract , from any proof of a formula @xmath22 in classical arithmetic , a term @xmath70 that refutes @xmath21 . but now notice that the identity function realizes @xmath165 , where @xmath166 is the negation of @xmath8 ; so , from a proof of @xmath167 in classical arithmetic , since the identity function realizes @xmath165 , one obtains a term @xmath70 satisfying @xmath168 .
this provides a direct proof of theorem [ main : witnessing : thm ] .
details can be found in @xcite .
a more elaborate realizability relation , based on the a - translation , can be found in @xcite .
the corresponding variant of the dialectica translation is similarly straightforward . as with the shoenfield variant @xcite , each formula @xmath22
is mapped to a formula @xmath169 of the form @xmath170 , where @xmath4 and @xmath5 are sequences of variables .
assuming @xmath171 is @xmath172 , the translation is defined recursively , as follows : @xmath173 on can interpret this as saying that for any function @xmath174 that purports to witness @xmath175 , there are a @xmath176 and an @xmath177 denying that claim .
once again , one can straightforwardly extract , from any proof of a formula @xmath22 in classical arithmetic , a term @xmath70 satisfying @xmath178 .
this provides another direct proof of theorem [ main : witnessing : thm ] .
i would now like to come back to the `` awkward translation , '' discussed at the end of section [ double : negation : section ] .
i will do this via what at first might seem to be a digression through kreisel s _ no - counterexample interpretation_. let @xmath22 be a formula in prenex form , for example , @xmath179 the _ herbrand normal form _
@xmath180 of @xmath22 is obtained by replacing the universally quantified variables of @xmath22 by function symbols that depend on the preceding existential variables , to obtain @xmath181 it is not hard to check that , in classical logic , @xmath22 implies @xmath180 .
thus , by a slight variant of theorem [ main : witnessing : thm ] ( relativizing it to function symbols ) , if classical arithmetic proves @xmath22 , there will be terms @xmath182 and @xmath183 of @xmath9 such that @xmath81 proves @xmath184 think of @xmath185 and @xmath186 as providing purported counterexamples to the truth of @xmath22 , so that @xmath187 and @xmath188 effectively foil such counterexamples .
the no - counterexample interpretation is simply the generalization of this transformation to arbitrary prenex formulas .
one need not invoke herbrand normal form to arrive at the previous conclusion .
one can check that if @xmath22 is a prenex formula , the no - counterexample interpretation of @xmath22 is essentially just the dialectica interpretation of @xmath153 , so the result follows from theorem [ awk : prop ] as well .
the no - counterexample interpretation can be viewed as a computational interpretation of arithmetic .
but , in a remarkable article @xcite , kohlenbach has shown that it is not a very _ modular _ computational interpretation , in the sense that it does not have nice behavior with respect to modus ponens . to make this claim precise , note that the set of ( terms denoting ) primitive recursive functionals , @xmath9 , can be stratified into increasing subsets @xmath189 , in such a way that any finite fragment of @xmath7 has a dialectica interpretation ( or modified realizability interpretation ) using only terms in that set .
kohlenbach ( * ? ? ?
* proposition 2.2 ) shows : [ kohlenbach : theorem ] for every @xmath190 there are sentences @xmath22 and @xmath32 of arithmetic such that : 1 .
@xmath22 is prenex .
2 . @xmath32 is a @xmath2 sentence , that is , of the form @xmath191 for some primitive recursive relation @xmath86 .
3 . primitive recursive arithmetic proves @xmath22 .
4 . @xmath1 proves @xmath34 .
5 . @xmath22 and every prenexation of @xmath34 has a no - counterexample interpretation with functionals in @xmath192 . but : 1 .
there is no term @xmath99 of @xmath189 which satisfies the no - counterexample interpretation of @xmath32 ; that is , there is no term @xmath99 such that @xmath100 is true in the standard model of arithmetic .
theorem [ kohlenbach : theorem ] shows that there is no straightforward way to combine witnesses to the no - counterexample interpretations of @xmath22 and @xmath34 , respectively , to obtain a witness to the no - counterexample interpretation of @xmath32 .
the problem with the awkward translation is that , similarly , it may behave poorly with respect to modus ponens .
consider pure first - order logic with a single predicate symbol , @xmath193 .
then there are formulas @xmath22 and @xmath32 such that @xmath194 does nt follow from @xmath153 and @xmath195 in minimal logic : just take @xmath22 to be the formula @xmath196 and @xmath32 to be @xmath12 . in that case , @xmath197 is equivalent , over minimal logic , to the double - negation shift , @xmath198 , which is not even provable in intuitionistic logic . when i showed the awkward translation to grisha , he remarked right away that its behavior has something to do with kohlenbach s result . at the time , i had no idea what he meant ; but writing this paper finally prodded me to sort it out .
grisha was right : theorem [ kohlenbach : theorem ] can , in fact , be used to show that the awkward translation does not provide a modular translation of peano arithmetic to heyting arithmetic , in the following sense .
[ awk : theorem ] for any fragment @xmath199 of @xmath7 , there are formulas @xmath22 and @xmath32 such that the following hold : 1 .
@xmath1 proves @xmath22 and @xmath34 , but 2 . @xmath200
together with @xmath153 and @xmath195 does not prove @xmath194 .
this shows that modus ponens fails under the @xmath160-translation , in a strong way .
since @xmath1 proves @xmath22 and @xmath34 , it also proves @xmath32 , and so by theorem [ awk : prop ] , @xmath7 proves @xmath194 . but having the translation @xmath153 and @xmath195 may not help much in obtaining such a proof of @xmath194 ; indeed , obtaining a proof of @xmath194 from @xmath153 and @xmath195 may be no easier than simply proving @xmath194 outright .
given @xmath199 , first let @xmath190 be large enough so that the dialectica interpretation of @xmath199 uses only terms in @xmath189 , and then let @xmath22 and @xmath32 be as in theorem [ kohlenbach : theorem ] . if there were a proof of @xmath194 from @xmath153 and @xmath195 in @xmath199 , applying the dialectica interpretation , one would obtain terms witnessing the dialectica interpretation of @xmath194 from terms witnessing the dialectica interpretations of @xmath153 and @xmath195 .
but the dialectica interpretation of @xmath153 is the no - counterexample interpretation of @xmath22 , and it is not hard to check that the dialectica interpretation of @xmath195 is the no - counterexample interpretation of one of the prenexations of @xmath34 .
thus there would be a witness to the no - counterexample interpretation of @xmath32 in @xmath189 , contrary to the choice of @xmath22 and @xmath32 .
as noted in the introduction , conventional wisdom holds that classical logic is `` nondeterministic , '' in that different ways of extracting algorithms from classical proofs can yield different results . sometimes , however , nondeterminacy is unavoidable .
for example , even minimal logic can prove @xmath201 , and any computational interpretation of this formula will have to choose either @xmath4 or @xmath5 to witness the conclusion .
the difference is that this sentence is typically not taken to be an _ axiom _ of minimal logic ; rather , there are two axioms , @xmath202 and @xmath203 , and any proof of the sentence has to choose one or the other .
in contrast , standard calculi for classical logic provide cases where there are multiple choices of witnesses , with no principled reason to choose one over the other .
for example , starting from canonical proofs of @xmath204 and @xmath205 , one can weaken the conclusions to obtain proofs of @xmath206 and @xmath207 , for an arbitrary formula , @xmath22 . then , using the law of the excluded middle , @xmath208 , one can combine these to obtain a proof of @xmath209 where there is little reason to favor @xmath70 or @xmath210 as the implicit witness to the existential quantifier .
( this example is essentially that given by lafont @xcite . )
this shows that standard classical calculi are nondeterministic in a way that proofs in intuitionistic and minimal logic are not .
girard @xcite has neatly diagnosed the source of the nondeterminacy , and has provided a calculus for classical logic that eliminates it by forcing the prover to make an explicit choice in exactly those situations where an ambiguity would otherwise arise . from @xmath118 and @xmath211 , and this inference is needed to verify the classical axiom @xmath125 under the double - negation translation .
another way is to notice that since , in classical logic , @xmath22 and @xmath118 are equivalent , there is little reason to favor @xmath22 or @xmath19 in situations where intuitionistic logic treats them differently .
i am grateful to thomas streicher for these insights . ] but note that the realizability interpretation of section [ realizability : section ] also avoids this nondeterminacy ; the translation procedure described in @xcite is fully explicit and unambiguous . with respect to the example discussed in the last paragraph , the interpretation chooses @xmath70 or @xmath210 based on the logical form of @xmath22 .
and @xmath118 have the same negation - normal form , which is distinct from ( and dual to ) that of @xmath19 . ] indeed , the results of @xcite show that the witnesses obtained in this way coincide with those obtained using a natural class of cut - elimination procedures .
yet another response to the example above is that of urban and bierman @xcite , who simply embrace the nondeterminism as an inherent part of the computational interpretation of classical logic .
in fact , urban @xcite has provided a nondeterministic programming language to interpret classical logic in a natural way . passing through a double - negation interpretation , as we have done here ,
amounts to making specific choices to resolve the nondeterminism .
it is an open - ended conceptual problem to understand which deterministic instances of the general nondeterministic algorithms can be realized in such a way . at this point , however , we should be clearer as to the goals of our analysis .
ordinary mathematical proofs are not written in formal languages , and so the process of extracting an algorithm from even a rather constructive mathematical argument can involve nondeterminism of sorts . and , despite some interesting explorations in this direction @xcite , it is far from clear that classical arithmetic can be used as an effective programming language in its own right .
but formal methods are actively being developed in support of software verification @xcite , and a better understanding of the computational content of classical logic may support the development of better logical frameworks for that purpose @xcite .
formal translations like the ones described here have also been effective in `` proof mining , '' the practice of using logical methods to extract mathematically useful information from nonconstructive proofs @xcite .
grisha s work has , primarily , addressed the general foundational question as to the computational content of classical methods . in that respect ,
the general metatheorems described here provide a satisfying answer : for the most part , classical mathematical reasoning _ does _ have computational content , which is to say , algorithms can be extracted from classical proofs ; but by suppressing computational detail , the proofs often leave algorithmic detail underspecified , rendering them amenable to different implementations .
grisha s work has thus contributed to an understanding of the computational content of classical arithmetic that is mathematically and philosophically satisfying , providing a solid basis for further scientific research .
solomon feferman .
infinity in mathematics : is cantor necessary ? in g. toraldo di francia , editor , _ linfinito nella scienza _ , pages 151209 .
istituto della enciclopedia italiana , 1987 . reprinted in solomon feferman , _ in the light of logic _ , oxford university press , new york , 1998 , pages 2873 and 229248 .
gerhard gentzen . . ,
112:493465 , 1936 .
translated as `` the consistency of elementary number theory '' in gerhard gentzen , _ collected works _
( edited by m. e. szabo ) , north - holland , amsterdam , 1969 , pages 132213 .
kurt gdel .
ber eine bisher noch nicht bentzte erweiterung des finiten standpunktes . , 12:280287 , 1958 . reprinted with english translation in feferman et al . , eds . , _ kurt gdel : collected works _ ,
volume 2 , oxford university press , new york , 1990 , pages 241251 .
georg kreisel .
interpretation of analysis by means of constructive functionals of finite type . in arendt
heyting , editor , _ constructivity in mathematics _ , north - holland , amsterdam , 1959 , pages 101128 .
grigori mints and sergei tupailo .
epsilon subsitution method for the ramified language and @xmath212-comprehension rule . in andrea
cantini et al . , editor , _ logic and foundations of mathematics _ , pages 107130 .
kluwer , the netherlands , 1999 .
|
almost from the inception of hilbert s program , foundational and structural efforts in proof theory have been directed towards the goal of clarifying the computational content of modern mathematical methods .
this essay surveys various methods of extracting computational information from proofs in classical first - order arithmetic , and reflects on some of the relationships between them .
variants of the gdel - gentzen double - negation translation , some not so well known , serve to provide canonical and efficient computational interpretations of that theory .
|
the galaxy cluster macs j1149.5 + 2223 ( @xmath7 , * ? ? ?
* hereafter ) is one of the most interesting and best studied cosmic lenses .
its strong lensing power is demonstrated by many sets of multiply - imaged galaxies , including the largest known lensed images of a face - on spiral galaxy at @xmath8 , whose highly magnified images display little distortion @xcite .
the first multiply - imaged supernova was discovered in the lensed images of this spiral , forming an einstein cross @xcite .
the supernova reappeared a year later in another counter image of the spiral as predicted by lensing models @xcite . because of its high magnification power ,
was selected for the clash program ( the cluster lensing and supernova survey with hubble , * ? ? ?
* ) , which obtained images of 25 galaxy clusters in 16 broad bands between @xmath9 @xmath10 m to a depth of ab magnitude @xmath11 .
a relatively bright , young galaxy , m1149-jd at @xmath1 and magnitude @xmath12 , was discovered approximately one arc minute north of the cluster center ( * ? ? ?
* z12 hereafter ) .
lensing models based on early data @xcite suggested that the source was magnified by approximately a factor of 15 , and that no counter images were expected ( although of note , the models lack sufficient constraints in this region ) .
is also one of the six targets of the _ hubble frontier fields _ program ( hff ; * ? ? ?
* ) , whose chief goal is reaching the deepest universe with the aid of gravitational lensing .
this is particularly important for galaxies at @xmath13 as they are believed to be the main source for the ionization of the intergalactic medium ( igm ) during this period @xcite , but most of them are too faint to be detected in random deep fields . in cycle 21 , abell 2744 and macs j0416.1@xmath142403 ( macs0416 ) were observed , and in cycle 22 macs j0717.5 + 3745 ( macs0717 ) and were observed . in cycle 23 , abell s1063 and abell 370 were observed .
thanks to the additional , intensive monitoring observations of the multiply - imaged supernova ( pis : kelly , rodney ) , the /wfc3-ir data of are the deepest among all clusters . the hff clusters have therefore been the subject of many studies searching for high - redshift lyman - break galaxies ( lbg ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
an important result is the faint - end luminosity function ( lf ) of galaxies down to an intrinsic absolute magnitude of @xmath15 , as estimated around the rest - frame wavelength of 1500 .
this is considerably deeper than the results from the hubble deep field @xcite , manifesting the uniqueness of the hff program .
our group has been carrying out a systematic study of high - redshift lbgs in the hff clusters , reported in a series of papers to date ( @xcite on abell 2744 ; @xcite on macs0416 ; and @xcite on macs0717 ) . here , we extend our search to and report 11 candidates at @xmath0 in the field of and 11 in its parallel field .
the hff observations of ( go / dd 13504 , pi : lotz ) were carried out between 2013 november 2 and 2015 may 19 .
additional archival images were obtained from three groups of resource : ( 1 ) the early images ( go 9722 , pi : ebeling ; go 10493 , pi : gal - yam ) ; ( 2 ) the clash multi - band images ( go 12068 , pi : postman ) ; and ( 3 ) recent wfc3/ir images ( go 14041 , 14199 , pi : kelly ; go 13790 , pi : rodney ; and go 13459 , pi : treu ) .
table [ tbl - sum ] lists the exposure times and limiting magnitudes for the imaging data used in our analysis .
we processed the data using aplus @xcite .
the calibrated images from the instrument pipelines , namely the _ flc _
images for acs ( corrected for the detector charge transfer efficiency ) and _ flt _ images for wfc3/ir , were retrieved .
the pixel scale for the final mosaic images is 0065 .
figure [ fig - fov ] displays a composite color image of the cluster field , and figure [ fig - par ] the parallel field .
the source catalogs are generated with sextractor @xcite in dual mode , using the summed wfc3/ir image as a reference . as a comparison , we ran sextractor with the public hff images @xcite and checked with the measurements in table [ tbl - cls ] .
the results confirm that the photometry of our candidates agrees within the stated errors . as part of the hff campaign ,
deep /irac images of were obtained between 2014 mar . and 2015 mar . in channels 1 and 2 , using director s discretionary time ( program 90260 , pi : soifer ) .
the effective exposure time in each channel , including that of the archival data ( program 60034 , pi : egami and 90009 , pi : brada ) obtained in 2010 , 2011 and 2013 , is @xmath16 ks .
the irac corrected basic calibrated data ( cbcd ) images were processed with mopex @xcite and sampled to a final pixel scale of @xmath17 .
the estimated @xmath18 limiting magnitude is @xmath19 for irac channel 1 ( irac1 , 3.6 ) and @xmath20 for channel 2 ( irac2 , 4.5 ) .
more details of the processing of and irac data can be found in @xcite .
our selection consists of two steps : a color selection and a photometric redshift selection .
we first searched for lbgs using their distinct color around @xmath21 . at @xmath22 ,
these sources are optical dropouts . at @xmath23 ,
the ly@xmath24 break is at @xmath25 , between the f814w and f125w bands .
our selection criteria for @xmath23 , in units of magnitude , are as follows : @xmath26@xmath27@xmath28 for @xmath29 , the break is at @xmath30 , between the f105w and f140w bands : @xmath31@xmath32@xmath33 for candidates at @xmath34 the break is between the f125w and f160w bands : @xmath35 .
we require that a candidate is not detected above @xmath18 in a summed image blueward of the selection bands defined above . for objects at @xmath36 ,
this requires a non - detection in a summed image of the f606w and f435w bands , while for candidates at @xmath37 this requires a non - detection in the stacked optical detection image . for candidates at @xmath34
, a non - detection is also required in the f105w or even the f125w band .
in addition to the color selection criteria described above , we excluded candidates lying within one arcsecond of the detector edges , in order to mitigate potentially spurious detections .
approximately 150 candidates were initially selected in the cluster field , and we visually inspected them .
a large portion of these candidates were excluded as artifacts following a visual inspection , and often also due to an incomplete image coverage in certain bands .
some others are excluded as they are near stellar diffraction spikes , which are difficult to remove because the hff wfc3/ir exposures were obtained at the same position angle . through these steps
the number of candidates was reduced to approximately 30 , enabling us to carry out irac photometry manually with reasonable effort .
the candidates with a color decrement of f160w - irac1 @xmath38 were rejected , as they are most likely extremely red objects at lower redshifts @xcite .
a similar procedure was performed in the parallel field .
the irac images of our candidates suffer from crowding due to the instrument s large point spread function ( psf , fwhm @xmath39 ) , such that simple aperture photometry might result in inaccurate fluxes due to contamination from nearby sources . to address this issue
, we used a deblending technique @xcite whereby contaminating neighbors are subtracted , by performing galfit @xcite fits to the objects of interest and all their close neighbors simultaneously in a fitting window of @xmath40 around the source of interest .
the irac psfs were created from the isolated stars in the same image . for each source , we chose the psf from nearby stars with a balance between brightness and isolation .
positions and radial profiles of neighboring sources in the region around a candidate were derived from the f160w - band mosaic and fixed ( except for those of the bright and extended galaxies ) after adjusting a small offset between the _ hst _ and irac images .
all other parameters were allowed to vary within preset ranges .
the initial input magnitudes were obtained by running sextractor on the irac images and partitioned by the flux ratios of sources in the images . if one sersic model did not yield a good fit to an extended source
, an additional psf or sersic component was added in fitting .
if a source is dominantly bright , we masked out several central pixels to avoid galfit being overwhelmed by these bright pixels .
the typical @xmath41-square values of our fitting results are around 10 and can be as good as 2 when nearby sources are well subtracted .
the photometric uncertainties were derived based on the fluctuations of residual images .
we performed aperture photometry in the residual images with a diameter of 25 and an aperture correction of a factor of 2 and measured the flux variations as our photometric uncertainties .
we then carried out our second - step selection , on the initial list of color - selected objects . using the photometry in seven bands and two irac bands
, we calculated photometric redshifts with the code bpz ( bayesian photometric redshifts ; * ? ? ?
* ; * ? ? ?
* ) , adopting the same template library used by the clash collaboration @xcite .
we assumed flat priors on both galaxy type and redshift in the range @xmath42 .
we required that all the candidates have a best - fit redshift solution of @xmath43 or higher , a @xmath18 redshift range greater than 5 , and that the probability of @xmath44 is less than 10% of that for the high - redshift solutions . in figures [ fig - jd ] and [ fig - stamp1 ]
we show cutout images of the @xmath45 candidates in the cluster field , and in figure 5 those of the parallel field . using bpz
, we identify 25 candidates in the two fields that satisfy our color selection criteria and whose photometric redshifts ( best values ) place them at @xmath45 . in tables
[ tbl - cls ] and [ tbl - par ] we list their coordinates , photometric redshifts , photometry , and magnifications .
four of these 25 are considered as the multiple images of a single galaxy at @xmath46 ( see m1 and m2 in table [ tbl - cls ] ) .
therefore , we consider 22 galaxy candidates in these two fields .
our irac analysis yields results for 17 sources , for which photometry or upper limits from galfit were possible .
as part of the hff initiative , seven strong - lensing models and one weak - lensing model were developed for and publicly released through the mast archive .
we used the strong - lensing models for as they were made to higher spatial resolutions . to estimate the systematic uncertainty in the magnification of each of our high - redshift candidates
, we calculated the median magnification factor for each candidate from the seven models ( table [ tbl - cls ] ) and its corresponding range for the five middle - ranked values , namely excluding the highest and lowest magnification factors .
this procedure is intended to mitigate potential extremes in the model predictions , and to properly reflect the underlying systematic uncertainties .
figure [ fig - fov ] shows the composite color image of the field , overlaid with the critical curves from the recent zitrin ltm ( light - tracing - mass ) model and identification numbers for our candidates . for more details on the zitrin ltm model
see the hff webpage and @xcite .
note that the critical curves in this best - fit model suggest that counter images for at @xmath47 should appear , although we do not secure such an identification in the data , and the prediction is weak given the lack of constraints in that region of the cluster .
in other words , the exact position of the critical curves in this northern region is uncertain ( see also * ? ? ? * k16 hereafter ) .
we discuss this further in [ sec : dis ] .
additionally , the predicted positions of some counter images are close to a member galaxy near the critical curve , making predictions in that area difficult and sensitive to small changes in the mass of this member galaxy . the objects in the parallel field are far away from the lensing field , but still experience weak - lensing effects . using the weak - lensing model of j. merten
, we derived magnification factors for the sources in table [ tbl - par ] between 0.95 and 1.2 .
the @xmath48 dropout macs1149-jd , detected by z12 using shallower clash imaging and shallower irac data , remains robust .
now that considerably deeper and data are available , we revisit the properties of this object .
as shown in table [ tbl - cls ] , it remains undetected in the f105w , f606w , and f435w bands , but weakly detected in the f814w band .
we inspected the f814w image and found that the residual flux is not centered at the source .
there appears to be a region of extended emission of approximately 1 arcsec in size toward the west direction .
this extended emission appears to be associated with a source approximately one arcsecond south - west away ( marked with a yellow arrow in figure [ fig - jd ] ) .
it is so weak that it can be seen only in the f814w and f105w bands .
even with this possible contamination , the color decrement between f140w and f814w is at least 3.5 mag , further reducing the possibility of a low - redshift nature .
we ran bpz with the new data , and found a slightly updated redshift of @xmath2 , in good agreement with our previous estimate of @xmath49 .
the probability of being a low - redshift object is less than @xmath50 .
is therefore photometrically reconfirmed within an accurate redshift range .
the source is well detected in both the irac1 and 2 bands , allowing an improved estimate of its age and star formation rate ( sfr ) .
the seven predicted magnification factors of are : 9.18 for cats @xcite ; 12.73 for lenstool @xcite ; 8.48 for zitrin - ltm ; 11.56 for zitrin ltm - gauss @xcite ; 17.0 for glafic @xcite ; 4.8 for grale @xcite ; and 14.9 for brada & hoag @xcite .
the new models yield estimates of its magnification factor as @xmath51 .
was not detected in irac1 in z12 because of the shallow irac data that were available . with deeper irac data , @xcite and @xcite found the source s irac1 magnitude as @xmath52 , close to our measurement .
the latest irac data enable an accurate estimate of the source s intrinsic properties in the optical bands . the flux in irac2 is brighter than irac1 by @xmath53 mag , likely the result of a balmer break . in order to derive the galaxy s physical properties
, we used the modeling code isedfit @xcite to fit the source s spectral energy distribution ( sed ) . with a monte carlo technique
, we generated @xmath54 model seds with a broad range of star formation histories , ages , stellar metallicities , dust contents , and nebular emission - line strengths .
our models in particular included nebular emission lines whose luminosity is tied self - consistently to the number of hydrogen - ionizing photons . for more details of isedfit
see z12 , @xcite and @xcite .
figure [ fig - sed ] displays the results for , with a fixed bpz value of @xmath55 .
the derived sfr is @xmath56 per year , with stellar mass of @xmath57 , where @xmath10 is the magnification factor .
the most interesting parameter is probably the galaxy age of @xmath58 myr , implying a formation redshift of @xmath59 .
the significant decrement between irac1 and 2 are largely explained by a balmer decrement .
the color between irac1 and f160w represents the uv continuum slope , which appears to be flat in @xmath60 and shows no sign for dust extinction
. object 594 in the vicinity is consistent with @xmath61 .
if physically related , they are separated by about one arcsecond ( 4.4 kpc in the source plane ) .
while it is possible that these are counter images of the same object ( see k16 ) due to lensing , it is unlikely : if the critical curve at @xmath62 is located between these two objects , they should be of roughly similar brightness and , in any case , the surface brightness must be the same .
pairs of multiple images should also have been found for nearby background galaxies at lower redshifts .
the ltm model predicts counterpart images , but none were found .
some other lensing models , including the previous versions of zitrin ltm model , place outside the @xmath63 critical curve so that no multiple images are predicted .
it should be noted that there is lack of multiple image constraints in the northern part of the critical curves and thus its extent in this region is poorly constrained .
given the depth of the hff data , it becomes unlikely that is multiply imaged .
to help corroborate the high - redshift nature of our candidates in the cluster field , we searched for potential counter images near the locations predicted by the zitrin ltm gravitational lensing model as well as that of @xcite .
objects m1 and m2 ( see figure [ fig - fov ] ) display similar redshifts and are at the positions accurately predicted by the lensing models .
both images are binary , as shown in the lower two rows of figure [ fig - stamp1 ] .
because of the high magnification , the intrinsic separation between the components is about 02 , or 0.8 kpc .
each of these components is unresolved even under significant magnification .
we carried out image deconvolution of m2 using the lucy - richardson algorithm in 20 iterations , as the sources are in a region clear of contaminations .
the reference psf image was derived from a field star .
the resultant image sizes are @xmath64 pixel or 01 in half - light radius . from a magnification factor of 7
, we derived an intrinsic magnitude of 29.8 and @xmath65 .
the result , together with that for ( z12 ) , is consistent with the expectation from a size - luminosity relation @xmath66 ( figure 9 , * ? ? ?
the lensing models also predict a third image , but its position is uncertain , and no counterpart source is found with confidence . for faint counter images predicted near bright cluster galaxies ( bcg ) , we used bcg - subtracted images to carry out our search .
several potential candidate dropouts are seen nearby , albeit none of these pass our selection criteria .
our lensing models predict that object 3315 is highly magnified and has at least two counterpart images .
one of the potential candidates is at r.a.=177.39724 and decl.=22.39125 .
while it displays a color similar to that at @xmath67 , its faint magnitude of 29 does not yield a high bpz value .
another image is predicted in the northeast part of the field ( near object 1069 ) but likely undetected because of the low magnification . among the 14 sources listed in table [ tbl - cls ] , 11
have been reported in k16 .
object 4385 is the faintest candidate , therefore it may fall below their detection threshold . for candidates 4267 and 2811 , the photometry of the public hff images confirmed
independently the bpz values @xmath68 for these two candidates .
the pairs m1 and m2 were also identified in k16 as individual components at @xmath69 .
since they are undetected or in the f814w band , their redshifts should be similar and close to 7 .
object yj3 in k16 was not confirmed in our analysis because a weak detection in the f814w band places the source at redshift between 6 and 7 .
it is close to a red diffuse source and may be subject to contamination .
we searched for high - redshift dropout candidates in the hff cluster macs1149 .
we found 22 lbg candidates at @xmath45 in the cluster field and the parallel field , reaching an intrinsic uv magnitude of @xmath70 .
two of the candidates are image pairs of a single galaxy .
three of them are detected in the /irac images .
we also used the new , deep data to revise the fit to the previously reported @xmath71 candidate ( z12 ) .
the deeper data support the high - redshift solution for ( it is not detected in the deeper optical images ) , and the updated fitting suggests a photometric redshift of @xmath72 , stellar mass of @xmath73 , star - formation rate of @xmath74 per year , and a formation redshift of @xmath75 ( age of @xmath5 myr ) .
our results show once more the usefulness of using gravitational lenses for accessing the faint , early universe . aided by gravitational lensing
, we have found galaxies as faint as @xmath76 , a remarkable depth for observing galaxies at the heart of the reionization era , in particular , in the advent of the _ james webb space telescope_. the work presented in this paper is based on observations made with the nasa / esa _ hubble space telescope _
, obtained from the data archive at the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas 5 - 26555 .
it is also based on data obtained with the _
spitzer space telescope _ , which is operated by the jet propulsion laboratory , california institute of technology under a contract with nasa .
this work utilizes gravitational lensing models produced by the teams led by brada , ishigaki , kneib & natarajan , sharon , williams , merten & zitrin , respectively .
support for a.z . is provided by nasa through hubble fellowship grant hst - hf-51334.01-a awarded by stsci . n.l .
acknowledges support from a european research council advanced grant fp7/669253 .
, f.e.b . and p.t.i .
are in part supported by conicyt - chile grants basal - cata pfb-06/2007 , 3140542 and conicyt - pia - act 1417 .
also thanks conicyt - chile grant fondecyt regular 1141218 and the ministry of economy , development , and tourism s millennium science initiative through grant ic120009 , awarded to the millennium institute of astrophysics , mas .
j.m.d . acknowledges support of the consolider projects csd2010 - 00064 , aya2012 - 39475-c02 - 01 and aya2015 - 64508-p ( mineco / feder , ue ) .
a.m. acknowledges the financial support of the brazilian funding agency fapesp ( post - doc fellowship 2014/11806 - 9 ) .
x.x.h . and j.x.w .
acknowledge support from the national science foundation of china ( grants 11233002,11421303 ) and the cas frontier science key research program ( qyzdj - sswslh006 ) .
atek , h. , richard , j. , jauzac , m. et al .
2015 , , 814 , 69 atek , h. , richard , j. , kneib , j .-
2014 , , 786 , 60 bentez , n , 2000 , , 536 , 571 bertin , e. & arnouts , s. 1996 , , 117 , 393 bouwens , r. j. , illingworth , g. d. , franx , m. , & ford , h. 2008 , , 686 , 230 bouwens , r. j. , illingworth , g. d. , oesch , p. a. et al .
2010 , , 708 , 69 2012 , , 752 , l5 2015 , , 803 , 34 2016 , arxiv:1608.00966 brada , m , russell , r. , stefano , c. et al .
2014 , , 785 , 108 brada , m , treu , t. , applegate , d. et al .
2009 , , 706 , 1201 coe , d. , bentez , n. , snchez , s. f. , jee , m. , bouwens , r. & ford , h. 2006 , , 132 , 926 diego , j. m. , broadhurst , t. , chen , c. et al .
2016 , , 456 , 356 ebeling , h. , barrett , e. , donovan , d. , ma , c .- j . , edge , a. c. , & van speybroeck , i. 2007 , , 661 , l33 eyles , l. , bunker , a. , stanway , e. , lacy , m. , ellis , r. , & doherty , m. 2005 , , 364 , 443 grillo , c. , suyu , s. h. , rosati , p. et al .
2015 , , 800 , 38 huang , k .- h .
, brada , m , lemaux , b. c. et al .
2016 , , 817 , 11 infante , l. , zheng , w. , laporte , n. et al .
2015 , , 815 , 18 ishigaki , m. , kawamata , r. , ouchi , m. , oguri , m. , shimasaku , k. , & ono , y. 2015 , , 799 , 12 jouvel , s. , hst , o. , lehav , o. et al .
2014 , , 562 , a86 johnson , t. l. , sharon , k. , bayliss , m. b. et al .
2014 , , 797 , 48 kawamata , r. , oguri , m. , ishigaki , m. , shimasaku , k. , & ouchi , m. 2016 , , 819 , 114 ( k16 ) kelly , p. l. , rodney , s. a. , treu , t. et al .
2015 , science , 347 , 1123 2016 , , 819 , l8 koekemoer , a. m. , mack , j. , lotz , jennifer m. et al .
2016 , 228 , 315.02 laporte , n. , infante , l. , troncoso iribarren , p. et al .
2016 , , 820 , 98 livermore , r. c. , feikelstein , s. l. & lotz , j. m. 2016 , arxiv:1604:06799 loeb , a. , & barkana , r. 2001 , araa , 39 , 19 lotz , j. m. , koekemoer , a. , coe , d. et al .
2016 , arxiv:1605.06567 makovoz , d. & khan , i. 2005 , in astronomical data analysis software and systems vi , asp conf .
132 , eds .
p. l. shopbell , m. c. britton , & r. ebert ( san francisco : asp ) , 81 mcleod , d. j. , mclure , r. j. , dunlop , j. s. et al .
2015 , , 450 , 3032 moustakas , j. , coil , a. , aird , j. et al .
2013 , , 767 , 50 overzier , r. a. , shu , x. , zheng , w. et al .
2009 , , 704 , 548 peng , c. y. , ho , l. c. , impey , c. d. & rix , h .- w .
2010 , , 139 , 2097 postman , m. , coe , d. , bentez , n. et al .
2012 , , 199 , 25 richard , j. , jauzac , m. , limousin , m. et al .
2014 , , 444 , 268 robertson , b. e. , ellis , r. s. , furlanetto , s. r. & dunlop , j. s. 2015 , , 902 , l19 rodney , s. a. , strolger , l .- g .
kelly , p. l. et al .
2016 , , 820 , 50 smith , g. p. , ebeling , h. , limousin , m. et al .
2009 , , 707 , l163 stark , d. p. , bunker , a. j. , ellis , r. s. , eyles , l. p. , & lacy , m. 2007 , , 659 , 84 treu , t. , brammer , g. , diego , j. m. , et al .
2016 , , 817 , 60 wang , w .- h . , barger , a. j. , & cowie , l. l. 2012 , , 744 , 155 zheng , w. , bradley , l. d. , saraff , a. et al .
2012a , in seventh conference on astronomical data analysis , cargse , corsica , france http://ada7.cosmostat.org/ada7_proceeding_zheng.pdf zheng , w. , postman , m. , zitrin , a. et al . 2012b , nature , 489 , 406 ( z12 ) zheng , w. , shu , x. , moustakas , j. et al .
2014 , , 795 , 126 zitrin , a. & broadhurst , t. 2009 , , 703 , 132 zitrin , a. , broadhurst , t. , barkana , r. , rephaeli , y. , & bentez , n. 2011 , , 410 , 1939 zitrin , a. , fabris , a. , merten , j. et al .
2015 , , 801 , 44 zitrin , a. , meneghetti , m. , umetsu , k. et al .
2013 , , 762 , 30 zitrin , a. , zheng , w. , broadhurst , t. et al .
2014 , , 793 , l12 cccccc & f160w & & 107.6 & 28.9 + & f140w & & 30.6 & 28.6 + & f125w & & 66.9 & 28.9 + & f105w & & 85.5 & 29.1 + & f814w & & 104.2 & 29.3 + & f606w & & 26.9 & 29.0 + & f435w & & 44.7 & 29.2 + & irac1 & & 357.7 & 25.5 + & irac2 & & 357.7 & 25.6 + lcccccccccccl 611 & @xmath77 & 177.41769 & 22.41369 & @xmath78 & @xmath79 & @xmath80 & @xmath81 & @xmath82 & & & @xmath83 & k16 ( y2 ) + jd ( 663 ) & @xmath84 & 177.38995 & 22.41271 & @xmath85 & @xmath86 & @xmath87 & @xmath88 & @xmath89 & @xmath90 & @xmath91 & @xmath51 & z12 , k16 ( yj1 ) + 594 & @xmath92 & 177.39055 & 22.41341 & @xmath93 & @xmath94 & @xmath95 & @xmath96 & @xmath82 & & & @xmath97 & k16 ( yj4 ) + 3315 & @xmath98 & 177.40073 & 22.39244 & @xmath99 & @xmath100 & @xmath101 & @xmath96 & @xmath82 & & & @xmath102 & k16 ( yj3 ) + 4267 & @xmath103 & 177.39453 & 22.38231 & @xmath104 & @xmath105 & @xmath106 & @xmath107 & @xmath108 & @xmath109 & @xmath110 & @xmath111 & + 2811 & @xmath112 & 177.38284 & 22.39600 & @xmath113 & @xmath114 & @xmath115 & @xmath116 & @xmath108 & & & @xmath117 + 763 & @xmath118 & 177.38298 & 22.41203 & @xmath119 & @xmath120 & @xmath121 & @xmath122 & @xmath108 & @xmath123 & @xmath123 & @xmath124 & k16 ( i19 ) + 4385 & @xmath125 & 177.40151 & 22.38434 & @xmath126 & @xmath127 & @xmath128 & @xmath129 & @xmath82 & & & @xmath130 & + 187 & @xmath131 & 177.41776 & 22.41744 & @xmath132 & @xmath133 & @xmath134 & @xmath135 & @xmath136 & & & @xmath137 & k16 ( i1 ) + 1069 & @xmath138 & 177.41278 & 22.40902 & @xmath139 & @xmath140 & @xmath141 & @xmath142 & @xmath108 & & & @xmath143 & k16 ( i11 ) + m1 - 4012 & @xmath144 & 177.40451 & 22.38688 & @xmath145 & @xmath146 & @xmath147 & @xmath148 & @xmath108 & @xmath149 & @xmath150 & @xmath151 & k16 ( 39.2 ) + m1 - 4023 & @xmath152 & 177.40460 & 22.38668 & @xmath153 & @xmath154 & @xmath155 & @xmath156 & @xmath157 & @xmath149 & @xmath150 & @xmath158 & k16 ( 21.1 ) + m2 - 2952 & @xmath159 & 177.41226 & 22.39497 & @xmath160 & @xmath161 & @xmath162 & @xmath163 & @xmath164 & @xmath165 & @xmath166 & @xmath167 & k16 ( i24 ) + m2 - 2963 & @xmath168 & 177.41220 & 22.39489 & @xmath169 & @xmath170 & @xmath171 & @xmath172 & @xmath164 & @xmath165 & @xmath173 & @xmath174 & k16 ( i26 ) + lcccccccccc 2556 & @xmath175 & 177.41644 & 22.29354 & @xmath176 & @xmath177 & @xmath178 & @xmath96 & @xmath108 & @xmath179 & @xmath180 + 2591 & @xmath181 & 177.40555 & 22.29315 & @xmath182 & @xmath183 & @xmath184 & @xmath96 & @xmath108 & @xmath179 & @xmath185 + 1587 & @xmath186 & 177.42131 & 22.30224 & @xmath187 & @xmath188 & @xmath189 & @xmath88 & @xmath82 & @xmath190 & @xmath191 + 168 & @xmath192 & 177.41658 & 22.31556 & @xmath193 & @xmath194 & @xmath195 & @xmath196 & @xmath108 & @xmath197 & @xmath198 + 2249 & @xmath199 & 177.43588 & 22.29614 & @xmath200 & @xmath201 & @xmath202 & @xmath203 & @xmath108 & @xmath204 & @xmath205 + 418 & @xmath206 & 177.40706 & 22.31268 & @xmath207 & @xmath208 & @xmath209 & @xmath210 & @xmath164 & @xmath211&@xmath212 + 2417 & @xmath213 & 177.40326 & 22.29470 & @xmath214 & @xmath215 & @xmath216 & @xmath217 & @xmath108 & @xmath218&@xmath219 + 1245 & @xmath220 & 177.41138 & 22.30530 & @xmath221 & @xmath222 & @xmath223 & @xmath224 & @xmath108 & @xmath225&@xmath185 + 2714 & @xmath226 & 177.41692 & 22.29187 & @xmath227 & @xmath78 & @xmath228 & @xmath229 & @xmath108 & @xmath230&@xmath231 + 1657 & @xmath232 & 177.43742 & 22.30168 & @xmath233 & @xmath234 & @xmath235 & @xmath236 & @xmath108 & @xmath237&@xmath231 + 164 & @xmath238 & 177.43379 & 22.31566 & @xmath239 & @xmath240 & @xmath241 & @xmath242 & @xmath243 & @xmath231&@xmath231
|
we search for high - redshift dropout galaxies behind the hubble frontier fields ( hff ) galaxy cluster macs j1149.5 + 2223 , a powerful cosmic lens that has revealed a number of unique objects in its field . using the deep images from the _ hubble _ and _ spitzer _ space telescopes
, we find 11 galaxies at @xmath0 in the macs j1149.5 + 2223 cluster field , and 11 in its parallel field .
the high - redshift nature of the bright @xmath1 galaxy macs1149-jd , previously reported by zheng et al . , is further supported by non - detection in the extremely deep optical images from the hff campaign . with the new photometry ,
the best photometric redshift solution for macs1149-jd reduces slightly to @xmath2 .
the young galaxy has an estimated stellar mass of @xmath3 , and was formed at @xmath4 when the universe was @xmath5 myr old .
data available for the first four hff clusters have already enabled us to find faint galaxies to an intrinsic magnitude of @xmath6 , approximately a factor of 10 deeper than the parallel fields .
|
one of the most important definitions in euclidean geometry concerns the norm of a vector and the distance between two points : they are defined using the pythagoras theorem . in the xix century , riemann generalized ideas of gauss , introduced the concept of manifold and constructed the riemannian geometry , in which pythagoras theorem would be only valid at a point or along a line , but would no longer be the way of measuring distances in an open set in the manifold ; instead , such measure would be performed using a general metric field .
euclidean geometry was the background of newtonian mechanics and was invariant by galilean boosts and rotations , but the advent of special relativity and a new kind of invariance group inspired the use of a geometry that unified space and time in a way that could keep invariant the interval between two events : such was minkowski geometry , in which the interval represents a distance in such spacetime , that is measured by a formula similar to the euclidean one , however allowing for positive , null or negative norms of vectors : this is a pseudo - euclidean space .
the next generalization , that is responsible for a relativistic description of gravity was performed by einstein by writing general relativity using a generalization of the minkowski space by the same terms of the generalization from euclidean to riemannian geometry : spacetime was defined as a differential manifold endowed with a pseudo - riemannian metric .
in the geometries just described , one uses a metric to define the norm of vectors and co - vectors and , then defines the length of curves and the distance between points .
an important feature of special relativity that gains a geometrical interpretation is the dispersion relation of particles , _ i d est _ the equation that relates energy , momentum and mass , it is defined as the norm of the four - momentum in minkowski spacetime
. therefore the pseudo - euclidean nature of the minkowski metric is what allows us to have a dispersion relation of the form @xmath0 .
such relation can be generalized to a curved spacetime , where the dispersion relation is defined as the norm with the curved metric , and the minkowskian form is achieved in normal coordinates .
on the other hand , some approaches to quantum gravity phenomenology refers to the possibility of having a modified dispersion relation ( mdr ) as a property of the semi - classical limit of quantum gravity theories , for example @xmath1 , where @xmath2 is a deformation parameter that depends on the model ( which in some cases can be assumed to be proportional to the planck length @xcite ) .
so , in this context , an immediate question arises : may a mdr be a manifestation of a departure of the riemannian nature of spacetime at a scale sensible for quantum gravity ?
if the situation was the spacetime being non - riemannian in a sense that could afford for a non - quadratic norm of vectors , then we would have a mdr , but the inverse is not necessarily true , therefore we can only speculate about it .
a known branch of differential geometry that fits perfectly into this property is the so called _ finsler geometry _ @xcite . basically , it is a rigorous formalization for possible non - quadratic norms of vectors , which is achieved through a four - velocity - dependent metric , but still preserving the parametrization invariance of the arc - length of curves in the manifold through the requirement of having a homogeneous in the velocities metric .
the relations between finsler geometry and some approaches to the quantum gravity problem have been widely investigated @xcite , with special interest in models introducing some modification to particles dispersion relation ( see for instance ref.@xcite and references therein ) .
in particular , in @xcite girelli , et al . pointed out that phenomenological models that implement some kind of departure from lorentz symmetries can be systematically formalized within the framework of finsler geometry , hypothesizing that such a formalism may be also the correct mathematical framework to describe the so called rainbow metrics " approach @xcite .
it is still unclear whether the introduction of momentum - dependent rainbow " metrics imposes some sort of breakdown of lorentz symmetry or otherwise they may be suitable for a scenario with a deformation of ( local ) spacetime symmetries through a deformed poincar group as in the deformed special relativity approach @xcite .
this issue was preliminarily treated by the authors of ref.@xcite for what concerns the fate of spacetime symmetries in mdr - inspired finsler geometries and was explicitly considered in ref.@xcite , in which it was shown that it is possible to obtain a description of modified relativistic particle kinematics satisfying both finsler geometry and deformed special relativity prescriptions within the so called @xmath3-poincar framework @xcite , at least at first order in the deformation parameter @xmath4 , where ( in units such as @xmath5 ) we can expect @xmath6 to be of the order of planck scale @xmath7 .
however satisfactory for what concerns the description of particles kinematics in the flat - spacetime limit , ref.@xcite leaves open the question whether this approach can be generalized to describe deformed relativity particles phenomenology @xcite within a curved background .
we would like here to implement this aspect , using as guidance some previous approaches @xcite to @xmath2-deformed particles kinematics in a totally - symmetric curved spacetime , where we will denote the parameter of spacetime curvature as @xmath8 .
a clear aspect presented in @xcite is the lack of a finsler metric that would be probed through spacetime inferences with massless particles , because the massless limit of the presented metric is ill - behaved . for example , this behavior prevents us from having a metric probed by neutrinos , since , even being massive , for most practical reasons they are considered massless due to their tiny rest mass . in this paper , besides generalizing some results of ref .
@xcite , we propose a different way of calculating a four - velocity - dependent ( and still partially homogeneous in the velocities ) metric , that can reproduce the main features of the standard finsler case , like the dispersion relation , the geodesics as worldlines , the presence of dsr symmetries and , furthermore , presents a well - behaved massless limit , thus allowing for a _
generalized finsler metric _ for any type of particle .
such generalization will allow us to propose an expression that deepens the one of jacob and piran @xcite for the mdr - induced time delay for particles with different energies , and to analyze the finsler nature of spacetime using photons / neutrinos as probes .
furthermore , we will study in the last section the law of interaction between elementary particles compatible with dsr models , furnishing a description in terms of the tangent space , thus being compatible with the finsler formalism . in this paper , in order to connect with the previous literature @xcite and express the mathematical and physical concepts in an effective and comprehensible way
, we will need to rely on different notations and coordinate sets .
+ the first one is the comoving relative locality coordinate set for momentum - space @xmath9 and spacetime flat slicing coordinates @xmath10 , which satisfy the relation @xmath11 in this coordinate set the hamiltonian , which formalizes the physics of particles embedded in a de sitter - like curved spacetime , with @xmath8 the parameter of curvature and @xmath2 the deformation parameter due to planck - scale effects , can be written as @xmath12 we stress that for sake of simplicity we are working in @xmath13 dimensions at first order in @xmath2 ( since the same kinematical results are preserved in a higher number of dimensions ) . in order to describe physics in an intuitive way
, we will also consider the so called _ conformal time coordinatization _ expressed by coordinates @xmath14 and @xmath15 .
the relation between the previous coordinate set and this latter one is @xmath16 one can verify the effectiveness of expressing physics in conformal - time coordinates by noticing that the spacetime line - element in this coordinatization can be expressed simply as @xmath17 .
previous works analysing such deformed physical frameworks ( see ref.@xcite ) , express this same hamiltonian ( [ flathamilt ] ) depending on canonical variables @xmath18 , deformation parameters @xmath19 and @xmath20 , deformation parameter @xmath21 and cosmological constant @xmath8 .
we define the energy and momentum respectively as @xmath22 and spacetime coordinates as @xmath23 .
therefore , in conformal coordinates we have @xmath24 it is also easy to verify that @xmath25 and @xmath26 ( as well as @xmath27 and @xmath28 ) still are conjugate variables . + the last couple of variables which will have some importance along this article is the one composed by the so called natural momenta , _ i d est _ those that have the same functional dependence of the charges of the translation generators of our spacetime : @xmath29 using those variables the expression of the hamiltonian ( [ flathamilt ] ) becomes : @xmath30 the physics of particles described by the deformed hamiltonians defined in this section was widely explored in ref.@xcite , however many aspects of the formalization of those phenomena seem to lack in such approaches to the study of deformed relativistic frameworks , such as metric formalism , killing vectors and particle interactions .
we would like here to provide a first exploration on how the integration of elements borrowed by the so called deformed momentum - space framework and finsler geometry could enrich this kind of approaches to the study of deformed relativity theories , providing solid theoretical foundations to the formalism and also new suggestions on the phenomenological side .
finsler geometry has been introduced in previous articles @xcite to formalize deformed relativistic frameworks defining a so called finsler norm , a homogenous function @xmath31 on tangent space . starting from the particle action in the hamiltonian formalism in flat spacetime , performing a legendre transform in @xmath32 where @xmath33 is hamiltonian ( [ flathamilt ] ) in the limit @xmath34 , and @xmath35 some lagrange multiplier . in general a semi - classical metric structure
is effectively encoded in finsler geometry . on the other hand
it turns out that @xmath3-poincar group ( that we are here using as foundation of a planck - scale deformed particle dynamics in a maximally symmetric spacetime ) does describe the symmetries of particles living on a de sitter curved momentum - space @xcite and flat spacetime ( see appendix for more details on momentum - space dynamics ) .
the two different metric formalisms finsler metric @xmath36 and momentum - space metric @xmath37 could lead to confusion on the metric structures we refer to when we will introduce concepts such as geodesics and invariant line - element . therefore we may need here to briefly review deformed - relativistic lagrangian formalism and the role that those different metrics play in it , in the massless case for sake of simplicity .
+ the minimization of ( [ genaction ] ) provide the relation between momentum and four - velocity ( here in 1 + 1d ) : @xmath38 then at first order in @xmath2 is not hard to invert this relation , finding some @xmath39 , which makes possible to re - express the casimir in terms of four - velocities : @xmath40 where @xmath41 is the inverse of momentum - space metric expressed in terms od @xmath42 .
it is also possible now to find an explicit solution for @xmath43 , by imposing @xmath44 . in the massive case this leads to @xmath45 in the massless case , in general , it is not possible to solve @xmath44 with respect to @xmath35 , however taking into account ( [ cxpunto ] ) and ( [ lambxpunto ] ) we can find very naturally that in the limit @xmath46 , @xmath47 . in this limit then ( [ hameq ] ) in 1 + 1d become @xmath48 and their inversion gives @xmath49 using those last results we find that the lagrangian of our theory can be expressed as @xmath50 where the lagrangian metric @xmath51 is not univocally determined , as we will better discuss later , and should be chosen according to the requests of the theory .
as we can see from ( [ genlag ] ) in this kind of deformed - symmetry theories we loose the uniqueness of the metric for spacetime and momentum - space .
those different metrics play a different role in the theory and one should be always careful to use the right one in the right formula .
for instance for euler - lagrange derived relations , such as geodesic equations or killing equations we are referring to metric @xmath51 .
the light cone structure is instead characterized in terms of momentum - space metric , since the on - shell relation is @xmath52 .
intriguingly , this defines the invariant ( see refs@xcite ) element for flat spacetime as @xmath53 .
+ it may be disappointing to have lost the possibility to express our theory with a unique metric , however from ( [ genlag ] ) we can still identify a couple of interesting relations : @xmath54 which highlight the duality between spacetime and momentum - space of this theory , widely discussed in the literature ( see for instance @xcite ) .
+ in order to fully understand whether deformed relativistic frameworks and finsler geometry are physically equivalent , in @xcite were introduced symmetry transformations derived within the finsler framework with the ones generated by a deformed - poincar group .
we want to show here how those generators can be obtained _ a priori _ using finsler killing equations with metric @xmath51 . for the sake of simplicity
, we will here find the boost @xmath55 representation in the limit @xmath34 and then generalize this result to the case at all orders in @xmath8 .
hamiltonian ( [ flathamilt ] ) at order zero in @xmath8 is of course invariant under translations since @xmath56 however if we try to find the representation of the boost imposing @xmath57 we obtain two possible solutions : @xmath58 or maybe some combination of the two .
which one should we choose ?
a solution to this problem can be provided by the powerful finsler formalism already explored in the formalization of deformed relativistic frameworks in ref.@xcite . in finsler geometry
it is , in fact , possible to define a deformed killing equation , which for flat spacetime has the form @xmath59 we can solve those equations for a generic four - velocity - dependent metric such as @xmath60 obtaining ( see appendix for further details ) a generic representation of the boost generator of the form @xmath61 whose hamiltonian can be obtained though the contraction of momenta with the metric @xmath62 : @xmath63 the deformed poincar algebra is therefore characterized by the following poisson brackets @xmath64 in equation ( [ gencasimir12 ] ) we can recognize the limit for @xmath34 of ( [ flathamilt ] ) , under the condition @xmath65 , which introduces a freedom that we will have to constrain based on the requests of the theory . in general
an approach based on curved momentum - spaces work perfectly fine with a diagonal metric , _ i d est _ @xmath66 as well as with other choices . on the other hand , in order to formalize our theory ( [ hamiltonian ] ) as a finsler geometry we will need a homogeneous - like metric , which will force us to the choice @xmath67 , as we will discuss further in this article . now that we have obtained the representation of the boost generator in the @xmath34 limit , knowing the form of the classical de sitter generator @xmath68 it is not so difficult to find its generalization al first order in @xmath2 , _ i d est _ : @xmath69 therefore the algebra of spacetime symmetry generators @xmath70 at all orders in @xmath8 is characterized by the following poisson brackets : @xmath71 of course those relations are , by definition , coherent with the poisson brackets between generators and hamiltonian ( [ hamtrasl ] ) : @xmath72 it should be noticed that the algebra described in this section is compatible with the well - known q - de sitter algebra @xcite , at first order in @xmath2 , under the choice @xmath73 .
we started with a non - trivial local structure of spacetime that could be modeled by a curved momentum - space , which is responsible for the definition of a modified dispersion relation and a deformed picture of interactions ( that will be discussed in section [ comp_rule ] ) .
now , we can analyze how a spacetime can emerge from these considerations , _ i d est _ how the local structure can interfere on the geometry of the effective manifold that these high - energetic particles probe .
we will follow a procedure that resembles the one presented in refs .
@xcite . in these references
, the authors perform a legendre transformation that links the action of a particle from the hamiltonian to the lagrangian formalism .
since in special and general relativity one identifies the action of a massive particle with the arc - length of its trajectory , the authors generalized this assumption to the case of mdrs , defining a non - quadratic norm of the particle s four - velocity vector . which can be modeled by the well studied finsler geometry , that is defined by a four - velocity - dependent norm .
the finsler norm satisfies the usual properties : @xmath74 the second one is rather important , since it expresses the homogeneity property of the norm , which in general implies @xmath75 in @xcite it was formalized that a mdr theory can be well expressed in terms of finsler geometry , identifying the finsler norm from the action integral as @xmath76 where @xmath77 is of course the finsler metric .
as we will see , this metric depends on the four - velocity and mass of the particle , however , it will not have a smooth limit for massless particles .
therefore , we propose an intermediate procedure that allows to define an action that serves both for the massive and massless cases , thus allowing a well - defined , unique metric that may describe a generalized finsler spacetime , that is probed by high - energetic particles . in order to generalize
the procedure just described to fully symmetric curved spacetimes , our starting point is the particle action in conformal coordinates with modified dispersion relation @xmath78 , @xmath79=\int d\tau [ \dot{\eta}\omega+\dot{x}\pi-\lambda ( { \cal h}_\omega - m^2)].\ ] ] this is the simple generalization of covariant mechanics in de sitter spacetime .
the equations of motion are @xmath80 following the same procedures of @xcite , _ i d est _ performing the legendre transformation , substituting @xmath81 , the lagrangian is @xmath82 minimizing this lagrangian with respect to @xmath35 we get @xmath83 and thus we can identify the finsler norm related to hamiltonian ( [ hamiltonian ] ) as @xmath84 at this point we can make contact with geometry , identifying the semi - classical metric structure effectively encoded in the finsler formalism .
the finsler metric is defined through a homogenous function on tangent space @xmath85 so @xmath86 it furnishes the de sitter metric if @xmath87 and furnishes the metric of paper @xcite for @xmath88 , @xmath89 and @xmath90 .
the analysis done so far is valid for the case of massive particles , @xmath91 , since the limit @xmath46 corresponds to a singularity , implying that observables measured with such metric involving observations with particles with tiny mass , like neutrinos , would not be well defined .
despite this fact , it is possible to solve the geodesic equation for massless particles by making @xmath92 in the geodesic solution just like was done in refs .
@xcite . in fact , now that we have a metric for a massive particle , we can minimize the arc - length to find its world line . this way , just as it is done in @xcite , even though the massless limit does not exist for the definition of the metric , it is well - behaved for the expression of the geodesics .
in fact , one could solve the geodesic equation that arises by extremizing the action , which furnishes : @xmath93=\frac{\dot{{\cal f}}}{{\cal f}}\left[\dot{q}^{\alpha}+\frac{1}{2}g^{\mu\alpha}\left(\frac{\partial}{\partial\dot{q}^{\mu}}g_{\gamma\beta}\right)\dot{q}^{\gamma}\dot{q}^{\beta}\right],\ ] ] for @xmath94 using the known expressions from the finsler geometry literature , that follow from euler theorem for homogenous functions @xmath95 we can write the geodesic equation as @xmath96 as this finsler function does not depend on the coordinate @xmath28 , we conclude that @xmath97 is a first integral to this problem , this is equivalent to equation ( [ p1 ] ) . as this formalism is invariant by reparametrization of the solution , we can choose the gauge @xmath98 .
doing it and substituting in the first constrain , we can solve the equation @xmath99 integrating this equation with initial condition @xmath100 we obtain its simple solution in which incoming photons have signature @xmath101 : @xmath102 the same worldline with spatial origin in @xmath103 can as well be obtained upon integration from @xmath103 to @xmath104 , imposing the on - shell relation to the finsler norm . for a massless , on - shell particle , @xmath92 , @xmath105 , the above equation is @xmath106
this is exactly the worldline found in the literature from the hamiltonian ( [ hamiltonian ] ) , see ref .
@xcite . as this is the straightforward generalization of refs .
@xcite , it also possesses the same kind of formal difficulties with respect to the massless metric .
however , it has been argued , for example in ref @xcite , that physical observables calculated by a macroscopic observer should be performed using the classical , riemannian metric , and in this case , there would be no problem in having such ill - defined metric , because its effects ( apart from its geodesics ) would be unobservable .
but we raise the possibility that the natural metric seen " by a particle , like the one we found with a finsler structure , could be the one considered for the calculus of observables in the new geometrical framework . which allows us to a deeper conclusion that observations performed with particles with a non - trivial dispersion relation ( no matter the origin of this deformation ) would indicate a possible non - riemannian structure of the spacetime in which they propagate .
thereby , by generalizing the geometrical structure with which one describes a spacetime , the physical observables would change reciprocally . in the previous section we performed a legendre transformation on the action of a particle with a mdr until we found a version of the action that we could identify with the arc - length of the particle s trajectory in a certain geometry .
since this was a non - quadratic function , but still invariant under reparametrizations , we could identify a finsler function and derive a metric from it . for the massive case
, there was no problem in the definition of the metric , but for the massless one , the metric was ill - defined .
such behaviour is due to the fact that the arc - length is the nambu - goto - like version of the action and it is well - known that it can only properly describe and only makes sense its extremization for the massive case and since the metric that would be found should depend on the mass of the particle , it is natural that such kind of approach would be problematic for @xmath46 ( which was not the case in riemannian geometry .
a possible solution for such impasse is the use of an unique action for both cases .
therefore , we propose to read a metric from the _ polyakov - like action _ of a point particle @xcite .
such formalism was initially explored in ref.@xcite .
in fact , by performing the legendre transformation in ( [ ham_action ] ) , we ended up with the lagrangian ( [ lagr1 ] ) .
as the lagrangian is an analytic function , the action can be uniquely expressed by a taylor expansion in the four - velocities @xmath107 = \int d\tau\left[{\cal l}{\big |}_{\dot{q } = 0 } + \frac{\partial \cal l}{\partial \dot{q}^{\mu}}{\bigg |}_{\dot{q } = 0}\dot{q}^{\mu } + \frac{1}{2!}\frac{\partial^2 \cal l}{\partial \dot{q}^{\mu}\partial \dot{q}^{\nu}}\bigg|_{\dot{q } = 0}\dot{q}^{\mu}\dot{q}^{\nu } + \frac{1}{3!}\frac{\partial ^3 \cal l}{\partial \dot{q}^{\mu}\partial\dot{q}^{\nu}\partial\dot{q}^{\gamma}}\bigg|_{\dot{q } = 0}\dot{q}^{\mu}\dot{q}^{\nu}\dot{q}^{\gamma } + ... +
\lambda m^2 \right],\ ] ] where the zeroth and first order terms vanish as well as those of higher than the third order
. therefore the action can be expressed as @xmath107=\int d\tau \left[\frac{1}{4\lambda}\ , g_{\mu\nu}(q,\dot{q},\lambda)\ , \dot{q}^{\mu } \dot{q}^{\nu } + \lambda m^2\right],\ ] ] where @xmath108 and we have identified @xmath109 this defines a metric that depends on the point of the manifold and the four - velocity . as we will see , the field @xmath35 allows to distiguish between the massive and massless cases . if @xmath91 , @xmath35 is found to be ( [ lambda ] ) .
if we substitute it into the action , we find @xmath110=m\int d\tau \sqrt{g_{\mu\nu}(q,\dot{q})\ , \dot{q}^{\mu } \dot{q}^{\nu}},\ ] ] where the metric is @xmath111 from this expression , one can already realize that such metric is well - defined in the massless limit , since the extra terms behave like @xmath112 , whereas in the standard finsler case we had @xmath113 , which of course explodes in the massless regime .
the extremization of this action furnishes the expected geodesics ( which is necessary for consistency ) , given by eq .
( [ solmassive ] ) , which coincide with those obtained using hamilton equations for the hamiltonian subject to the mass - shell condition @xmath114 , in the same parametrization . for @xmath92
, the lagrange multiplier @xmath35 can not be solved
. however , we can absorb it into the definition of the parameter and define the usual _ affine parameter _ as @xmath115 . with this definition
, the action assumes the familiar form @xmath116=\frac{1}{2}\int ds\ , g_{\mu\nu}(q , q')q'^{\mu}q'^{\nu},\ ] ] where @xmath117 , generalizing the result obtained in ( [ genlag ] ) to curved spacetimes , for the metric ( [ beta_metric ] ) and @xmath67 .
moreover , the extremization of this functional , taking into account the massless condition as a fundamental one , _
i d est _ the on - shell condition @xmath118 written in terms of the four - velocities @xmath119 furnishes the geodesic equation ( [ solmassless ] ) .
we have shown that a natural geometrical formalism that describes the spacetime probed by particles with a modified dispersion relation consists in the use of a four - velocity - dependent metric , such that the riemannian structure is recovered in the low - energy limit . to give another appearance to this approach
we can use the definition of the conjugate momenta @xmath120 and substitute the four - velocities that appear in the metric by the energy - momentum of the particle .
in fact , we end up with @xmath121 where @xmath122 .
this is how our formalism takes the form of a
rainbow metric " , where the metric that an observer assigns to the spacetime by using high - energetic particles depends on its energy - momentum . as can be seen , once the metric is written this way ,
it is evident the smooth limit between the massless and massive cases , since the dependence on the mass has disappeared , remaining just the momenta one .
this was not the case for the same substitution in the standard finsler case , as can be verified in ref .
@xcite already in the flat case , which is still true by adding curvature to the spacetime .
the finsler geometry is a formulation in which the arc - length functional plays a central role , and from it , one can derive a metric tensor that depends homogeneously on tangent - space elements . in this paper
, we review a way of defining a four - velocity - dependent metric that shares many properties of the finsler one , but is derived from a polyakov - like action for a particle that obeys a modified dispersion relation and generates the same basic features , like the dispersion relation from a norm and geodesics .
one of our achievements is the smooth limit that the metric presents when passing from the massive to the massless case .
we would like to probe the effects of such metric structure observing the travel - time through cosmological distances of a particle sufficiently energetic .
for the sake of measurements , since they are performed using a macroscopic apparatus , usually it is considered the classical riemannian metric for the definition of observables @xcite , for example , the proper time of an observer on earth , which is used to derive the expression for the time - delay of high energetic photons emitted from a grb @xcite . since the metric is riemannian , its components
do not depend on the energy / frequency of the probed particle .
the formalism of finsler geometry that we just described allows to relax this hypothesis , as we see in this section .
the finsler metric is a function that depends on spacetime points and vector fields , _ i d est _ to calculate the inner product of two vectors at a given point , it is also necessary to determine a direction or a third vector at that point . denoting @xmath123 the metric tensor that depends on @xmath124 , then the inner product of vectors @xmath125 and @xmath126 is @xmath127 , it is then , necessary to specify the three vectors @xmath128 to measure the inner product .
_ however , for each fixed @xmath129 , we have that @xmath130 is a riemannian metric . _
therefore , it is possible to calculate a vector - directed riemannian scalar product at a given point for a fixed @xmath131 , if this vector is defined at that point .
see for instance ref.@xcite for an analysis of vector dependent tensor calculus . as aforesaid , in formalisms with energy - dependent or
rainbow " metrics , this metric is considered as the one that a particle probes " , but the one that an observer assigns to the spacetime needs to be the classic one , because if it also was rainbow - like , the observer s metric would depend on the product between the observer s mass and the planck length , which would be in obvious contradiction with the current observations , where @xmath132 is the number of elementary particles that constitute the observer , which gives a negligible contribution . ] .
the formalism we describe here allows to circumvent the impossibility of the last paragraph by showing that it is possible to have a riemannian metric , induced by a finsler one by fixing the vector which the metric depends on as being the integral line of the particle that probes the rainbow " metric , _
@xmath133 , where @xmath131 is the particle s four - velocity , is the _
riemannian metric constructed by an observer by means of measurements with a particle with four - velocity @xmath131 .
_ to visualize the reasonability of this proposal , consider the following example .
suppose that the planck scale was not so far from our experience , such that spacetime inferences could be always performed with particles that manifest modified dispersion relations .
since we are hypothesizing that the spacetime probed by such particles is deformed with respect to the standard riemannian one by means of a momenta - dependent metric , if an observer alice that makes inferences using these particles intends to preserve the equivalence principle , she should assign such metric structure to the spacetime manifold .
for example , the geometrical locus of the photon sphere of a black hole ( bh ) is a property of the photon s worldlines , and if she wants to preserve the equivalence principle by representing them as geodesics , they would need to be compatible with the metric structure of the spacetime surrounding a bh , which would need to be of rainbow nature . in other words ,
the map of a spacetime constructed by such observer would need to inform such energy - dependent behavior .
so , following this principle , another example consists in the measurement of the proper - time , elapsed on earth from the emission of a particle from a cosmological source up to its arrival at a terrestrial detector . as the spacetime geometry inferred using a particle with the same energy
is determined by the deformed metric , the proper - time is calculated as @xmath134 , which in the comoving frame is @xmath135 , where @xmath136 is the rainbow metric that depends on the energy of the detected particle . and considering the standard riemannian metric . ] therefore , for the measurement of the time delay between massless particles from grbs @xcite ( which is one of the standard sources of constrains for mdrs ) , we define it analogously to the shapiro delay @xcite . in this case , the delay was calculated as the difference in the proper time of earth between the time taken for a radar signal to reach a target and return depending if there is a massive object that deforms the spacetime geometry probed by the photon and the observer , _ i d est _ in the sending and receiving of the signal one compares the proper times calculated in first and zeroth order in the mass parameter , for example in a schwarzschild spacetime .
equivalently , in this model , depending on the energy of the probed particle , the induced metric will be the ordinary de sitter one induced by a standard dispersion relation or will manifest a first order correction if the particle is sufficiently energetic , induced by the corresponding mdr , which is similar to the first order correction in the mass that deforms the flat metric in the shapiro delay . in our case ,
the parameter is not the mass of the schwarzschild metric , but it is the planck length @xmath2 .
what we define as a time delay for the arrival of photons is the difference between the proper times ( measured on earth ) that record the time elapsed from the emission of massless particles at @xmath137 , to their arrival at the detector , labeled as @xmath138 . for each photon ,
the proper time elapsed is @xmath139 in this case , @xmath140 is the metric s component in the deformed riemannian geometry induced by a finsler metric that the observer associates to the spacetime from eq .
( [ metricem ] ) , where the photon s four - velocity / momenta is the fixed vector that induces the deformed riemannian metric . in the coordinates that we have been using @xmath141 .
integrating the proper time , @xmath142 the redshift @xmath143 is defined as @xmath144 now , consider that we have a hard and a soft massless particle , for which @xmath145 , @xmath146,\\ x_s(\eta)&=&\eta.\end{aligned}\ ] ] suppose that both particles are emitted simultaneously , _
@xmath147 , at the same spatial point @xmath148 ( see fig.[fig : timedelay ] ) .
let @xmath138 be the point where the detector on earth is .
the elapsed parameter necessary for the soft particle to reach the detector , which is located at the origin is @xmath149 .
and @xmath19 is respectively @xmath150 or @xmath151.,title="fig : " ] [ fig : timedelay ] to determine the elapsed parameter , denoted by @xmath152 , necessary for the hard trajectory to reach the same spacial point , we consider @xmath153 since we normalize the scale - factor for @xmath154 , and the redshift @xmath143 is an experimental parameter determined by soft particles , it is natural to define it as @xmath155 so , the proper time elapsed on earth until the arrival of the hard particle can be calculated using equations ( [ x0hb ] ) and ( [ x0sb ] ) as @xmath156.\end{aligned}\ ] ] the time delay is defined as the difference between the proper times @xmath157 and @xmath158 , therefore it is @xmath159.\ ] ] such expression contains the usual one used for the calculus of time - delays presented in @xcite , added to an extra contribution due to the deformed metric that the observer assigns to spacetime when performing measurements with high - energetic particles .
the previous approach could not distinguish between the parameters @xmath19 and @xmath20 , since the effect appeared as a factor of @xmath160 .
our new proposal allows to distinguish them , and introduces a phenomenological rule for the time - delay in a more flexible shape @xmath161 in our case , @xmath162 and @xmath163 .
another interesting correction is the redshift of the frequency of photons in this geometry .
different from the previous case in which the metric played a central role , for calculating this quantity it is only necessary to consider the equations of motion of the photon .
in fact , in comoving coordinates , the equations of motion for spacetime coordinates and the energy for a massless particle are ( [ uxoux1 ] ) @xmath164 the solution for the energy is @xmath165 where @xmath166 is the energy of the photon for @xmath167 . , while the dotted - blue one for @xmath150 .
the red line represents the classical undeformed cosmological redshift effect , for a low - energetic particle.,title="fig : " ] [ fig : lateshift ] for on - shell massless particles we have @xmath118 , which is simply @xmath168.\ ] ] substituting equation ( [ energy_massless ] ) and ( [ on - shell - massless ] ) in the first equation of ( [ uxoux2 ] ) , we find @xmath169,\ ] ] which can be solved for the parameter @xmath170 as @xmath171.\ ] ] substituting this expression in the dependence of the energy with the parameter @xmath170 given by ( [ energy_massless ] ) , we derive the cosmological redshift of the frequency of a photon that obeys a modified dispersion relation ( see fig . [
fig : lateshift ] ) @xmath172
it was explained in @xcite , using an argument already introduced in @xcite , that if one assumes deformed spacetime symmetries , then if we assume that whenever two particles interact , different observers must agree on the existence of the interaction vertex , this latter may be deformed too .
this feature is formalized in the @xmath3-poincar framework , introducing a deformed composition law for momenta @xmath173 which we can imagine as describing the decaying of a particle with momentum @xmath174 in two different particles with momenta @xmath175 and @xmath176 . at first order in the deformation parameter @xmath2
we can express all the possible composition rules of momenta ( requiring invariance under the parity transformation @xmath177 , @xmath178 , see @xcite ) relying just on four parameters @xmath179 as @xmath180 such deformed composition rules are known to be related to the coproduct of translation generators .
then , in order to determine which coalgebric sector can be compatible with our deformed boost generator ( [ boostrepresentation ] ) , we need to find some condition which would allow us to find some relations between the composition rule parameters .
those relations can be found by requiring the composition of the boosted momenta to be invariant under the action of the boost @xmath181 however , before checking whether different ( boosted ) observers agree on the existence of the particles vertex , we should pay attention to the possibility to have the so called _ backreaction _ effect @xcite , since the rapidity parameter @xmath182 can in general change in a momentum - dependent way , compatible with the coproducts of momenta and the action of lorentz transformations on momenta themselves .
this effect is defined as the right action @xmath183 , that we can parametrize as @xmath184 therefore , in general we will have @xmath185 now , imposing the complete parametrization of vertex transformation ( [ rightsumrule ] ) with ( [ boostaction ] ) and ( [ rightactions ] ) , we are able to find four relations between our eight parameters @xmath186 , which ensure the existence od the vertex for any observer : @xmath187 those relations are perfectly compatible with the hamiltonian invariance under boost : @xmath188 which shows _ a posteriori _ the coherence of those composition laws with the curved momentum - space framework . the problem is if also finsler geometry can be compatible to such deformed sum " . in finsler geometry , in fact , the particle trajectory depends on spacetime coordinates @xmath189 and four - velocities @xmath190 , which live in a flat tangent space to the curved spacetime . on the other hand , in relative locality
@xcite momenta are the fundamental observables and spacetime is defined as the cotangent bundle to the curved momentum - space .
if the deformation of momenta composition law fits very well with the curved momentum - space framework , one should not give for granted for this to happen also in the finsler case .
however in this section we showed that non trivial composition laws are imposed by the deformation of spacetime symmetries that arise from our theory .
the impossibility to express this feature also for finsler formalism would then jeopardise our entire argument . from our perspective
, the problem of the tangent space flatness is not a real problem .
in fact every local observer is flat , and curvature arises only when confronting observations between different reference frames .
assuming then ( [ nontrivcomp ] ) , we are treating different particles as different reference frames in momentum - space , which can be characterized also as a non - trivial composition for four - velocities , formalizing the relations between the different locally - flat reference frames .
this feature can be obtained taking into account a two - parameters family of coproducts and momenta compositions that we can obtain from ( [ rappparamgener ] ) : @xmath191 and @xmath192 then , using the expression of four - velocities with respect to momenta ( [ uxoux1 ] ) with ( [ composition0 ] ) and ( [ composition1 ] ) , in the massless case , we can easily obtain the composition laws for @xmath42 : @xmath193 those laws satisfy the relation @xmath194 where the backreaction has the exact same form of the one described in the previous section .
a nice property that we can find _ a posteriori _ in the massless case is @xmath195 in which @xmath196 is the velocity that we already found in our massless worldlines ( [ solmassless ] ) .
this structure of relations between different tangent spaces allows us to identify the phase - space invariant element as : @xmath197 the doubt we had at the beginning of this section was whether it is possible to express departures from poincar symmetries , often formalized in the literature as a deformed hamiltonian framework ( @xmath2-deformed phase space ) , within the context of the lagrange - finsler geometry ( tangent space ) .
as commented in @xcite finsler geometry could be used to describe both breakdown and deformation of spacetime symmetries .
however we have showed here that this symmetry deformation implies the necessity to characterize velocities as living in a tangent space with non - trivial translation laws .
those laws moreover result to be compatible with the phase - space deformation of the theory ( [ flathamilt ] ) .
the question of what may be the nature of the spacetime emerging from a semi - classical limit of a quantum description of the gravitational degrees of freedom is still open .
possibly , we may have a dimensional reduction coming from cdt , hoava - lifshitz gravity and others ( see @xcite and references therein ) , non - local metrics @xcite , rainbow - like metrics coming from a quantization of geometrical and matter - like degrees of freedom @xcite and , as we have shown , a possible finsler nature coming from a mdr
. in all these cases there exists a common point : the non - riemannian nature of spacetime . which is not a complete surprise , after all there is no _ a priori _
request for nature of spacetime to be limited to riemannian geometry at all scales , despite the abundance of well - defined geometrical structures ( not to mention those that are still to be discovered ) . in this paper , we analyzed the possibility of having a deformed dispersion relation within a curved background , investigating what kind of manifold could be the most natural to `` host '' them .
we generalized the results presented in @xcite and we proposed a solution for the massless limit problem , _ i d est _ we found a candidate partially - homogeneous metric using , which are possible to formalize , the most important properties that a proper geometrical model for the description of elementary particles should provide , such as the geodesic equation and the dispersion relation .
the presence of such unified finsler structure that an observer would assign to spacetime , when performing measurements observing high - energetic particles , allowed us to formalize the usual phenomenology of time delays of photons from grbs due to the energy - dependent speed of light presented in @xcite , plus a correction of the order of the planck length originated from the non - riemannian structure of spacetime , with these both contributions having the same geometrical origin .
we also estimated the correction that such a model could bring to the cosmological redshift effect . in this context
, it would be extremely interesting to test such effect using cosmological data , however unfortunately the best observations on some kind of energy dependence for the cosmological redshift yet published @xcite do not have the sensitivity to discriminate the kind of effect we found here .
the authors found no energy dependence with a precision of @xmath198 , whereas in the visible range of electromagnetic radiation ( @xmath199 ) the effect we take into account would be of the order of @xmath200 on the redshift magnitude . in any case
, we need phenomenological analysis with more energetic particles .
the last open issue we explored is the deformation of velocities composition law in the finsler tangent space that symmetrical @xmath2-deformed theories require ( as already pointed out in @xcite ) , in order to allow to different observers to agree on the occurrence ( or not ) of some interaction event .
we found that introducing a slightly more complicated formalism inspired by hopf algebras literature it is still possible to obtain a full relativistic description for particles interactions .
nl acknowledges support by the european union seventh framework programme ( fp7 2007 - 2013 ) under grant agreement 291823 marie curie fp7-people-2011-cofund ( the new international fellowship mobility programme for experienced researchers in croatia - newfelpro ) , and also partial support by the h2020 twinning project n@xmath201 692194 , rbi - twinning .
ipl is supported by the international cooperation program capes - icranet financed by capes - brazilian federal agency for support and evaluation of graduate education within the ministry of education of brazil grant bex 14632/13 - 6 .
fn acknowledges support from conacyt grant no .
first of all let us define a generic momentum - space metric depending on our two parameters @xmath202 , at first order in the deformation parameter @xmath2 : @xmath203 as well established in literature @xcite we can find the hamiltonian by integrating the momentum - space invariant line - element @xmath204 where the momentum - space geodesics @xmath205 are defined by the geodesic equation : @xmath206 in which the connections are our usual momentum - space christoffel symbols : @xmath207 given our generic metric , then , we just have to solve the following equations : @xmath208 since @xmath209 , we observe that in our case the geodesic can be expressed as @xmath210 let us now impose the initial values for the geodesic as @xmath211 and @xmath212 .
it is easy to notice that they imply @xmath213 and @xmath214 .
we have now all we need in order to solve our equations , _ i d est _ : @xmath215 and also @xmath216 therefore we are now able to solve integral ( [ intcasimir ] ) , and find the generic hamiltonian expression : @xmath217
in the introductory section we have shown that the lagrangian - hamiltonian legendre transformation defines the relation @xmath218 since eq .
( [ cxpunto ] ) basically states that the first term of [ b1 ] must be invariant under boost , then we have that @xmath219 therefore @xmath220 however as already noticed in @xcite , being spacetime flat we have @xmath221 , and then the second term of last equation is just a total derivative : @xmath222 which can be eliminated during the integration procedure .
the killing equation of our theory is then equivalent to ( [ killfinslflat ] ) , being @xmath223 excepted total derivative terms .
+ given this , from killing equation ( [ killfinslflat ] ) we find the system of differential equations we need to solve in order to find our symmetry generators : @xmath224 we look for perturbative solutions at first order in @xmath2 such as @xmath225 in which the 0th order solutions are the minkowskian ones : @xmath226 this leads to @xmath227 from system ( [ eq : killsysparam ] ) first and second equation we just obtain that the generalized killing vector components should have the form : @xmath228 now using the results obtained ( [ kill1 ] ) with the third equation , re - deriving this one we get @xmath229 then , using ( [ xi1imperfetta ] ) again with the third equation of system ( [ eq : killsysparam ] ) we obtain the relation between @xmath230 and @xmath231 which is : @xmath232 we can now express the complete solutions for the @xmath233 as @xmath234 the natural choice @xmath235 gives us the total charge , in fact it is easy to verify that in this case : @xmath236 the choice @xmath237 works fine in the diagonal case metric .
however finsler formalism relies on homogeneous metrics , which in general have non zero off - diagonal terms .
the total charge general expression imposing no particular condition to our parameters at first order in @xmath2 has the form @xmath238 in which @xmath239 , @xmath240 and @xmath55 are the deformed - poincar algebra generators in 1 + 1d . in order to obtain a coherent physical framework ( _ i d est _ invariant casimir and particles worldlines ) also in the finsler case , we need then to impose @xmath241 .
this choice eliminates any direct sign of the metrics off - diagonal terms in the boost representation ( [ boostrepresentation ] ) .
g. amelino - camelia , `` quantum - spacetime phenomenology , '' living rev .
* 16 * , 5 ( 2013 ) doi:10.12942/lrr-2013 - 5 [ arxiv:0806.0339 [ gr - qc ] ] . h. rund , _ the differential geometry of finsler spaces .
_ springer , berlin ( 1959 ) .
n. e. mavromatos , `` lorentz invariance violation from string theory , '' pos qg * -ph * , 027 ( 2007 ) [ arxiv:0708.2250 [ hep - th ] ] . s. i. vacaru , `` finsler branes and quantum gravity phenomenology with lorentz symmetry violations , '' class .
* 28 * , 215001 ( 2011 ) doi:10.1088/0264 - 9381/28/21/215001 [ arxiv:1008.4912 [ math - ph ] ] .
f. girelli , s. liberati and l. sindoni , `` planck - scale modified dispersion relations and finsler geometry , '' phys . rev .
d * 75 * , 064015 ( 2007 ) doi:10.1103/physrevd.75.064015 [ gr - qc/0611024 ] .
j. magueijo and l. smolin , `` gravity s rainbow , '' class .
* 21 * , 1725 ( 2004 ) doi:10.1088/0264 - 9381/21/7/001 [ gr - qc/0305055 ] .
g. amelino - camelia , `` relativity in space - times with short distance structure governed by an observer independent ( planckian ) length scale , '' int .
j. mod .
d * 11 * , 35 ( 2002 ) doi:10.1142/s0218271802001330 [ gr - qc/0012051 ] .
g. amelino - camelia , l. barcaroli , g. gubitosi , s. liberati and n. loret , `` realization of doubly special relativistic symmetries in finsler geometries , '' phys .
d * 90 * , no .
12 , 125030 ( 2014 ) doi:10.1103/physrevd.90.125030 [ arxiv:1407.8143 [ gr - qc ] ] . j. lukierski , h. ruegg , a. nowicki and v. n. tolstoy , `` q - deformation of poincar algebra , '' phys .
b * 264 * , 331 ( 1991 ) ; j. lukierski , a. nowicki and h. ruegg , `` new quantum poincar algebra and @xmath3-deformed field theory , '' phys .
b * 293 * , 344 ( 1992 ) .
doi:10.1016/0370 - 2693(92)90894-a .
j. lukierski and h. ruegg , `` quantum @xmath3-poincar in any dimension , '' phys .
b * 329 * , 189 ( 1994 ) doi:10.1016/0370 - 2693(94)90759 - 5 [ hep - th/9310117 ] ; j. lukierski , h. ruegg and w. j. zakrzewski , `` classical quantum mechanics of free @xmath3-relativistic systems , '' annals phys . *
243 * , 90 ( 1995 ) doi:10.1006/aphy.1995.1092 [ hep - th/9312153 ] . g. amelino - camelia , `` are we at the dawn of quantum gravity phenomenology ? , '' lect .
notes phys .
* 541 * , 1 ( 2000 ) [ gr - qc/9910089 ] .
t. jacobson , s. liberati and d. mattingly , `` a strong astrophysical constraint on the violation of special relativity by quantum gravity , '' nature * 424 * , 1019 ( 2003 )
doi:10.1038/nature01882 [ astro - ph/0212190 ] . g. amelino - camelia , g. gubitosi , n. loret , f. mercati and g. rosati , `` weakness of accelerator bounds on departures from lorentz symmetry for the electron , '' europhys .
lett . * 99 * , 21001 ( 2012 ) doi:10.1209/0295 - 5075/99/21001 [ arxiv:1111.0993 [ hep - ph ] ] .
g. amelino - camelia , l. barcaroli , g. damico , n. loret and g. rosati , `` icecube and grb neutrinos propagating in quantum spacetime , '' phys .
b * 761 * , 318 ( 2016 ) doi:10.1016/j.physletb.2016.07.075 [ arxiv:1605.00496 [ gr - qc ] ] . g. amelino - camelia , a. marciano , m. matassa and g. rosati , `` deformed lorentz symmetry and relative locality in a curved / expanding spacetime , '' phys . rev .
d * 86 * , 124035 ( 2012 ) doi:10.1103/physrevd.86.124035 [ arxiv:1206.5315 [ hep - th ] ] ; g. rosati , g. amelino - camelia , a. marciano and m. matassa , `` planck - scale - modified dispersion relations in frw spacetime , '' phys
. rev .
d * 92 * , no .
12 , 124042 ( 2015 ) doi:10.1103/physrevd.92.124042 [ arxiv:1507.02056 [ hep - th ] ] .
u. jacob and t. piran , `` lorentz - violation - induced arrival delays of cosmological particles , '' jcap * 0801 * , 031 ( 2008 ) doi:10.1088/1475 - 7516/2008/01/031 [ arxiv:0712.2170 [ astro - ph ] ] .
g. amelino - camelia , m. matassa , f. mercati and g. rosati , `` taming nonlocality in theories with planck - scale deformed lorentz symmetry , '' phys .
* 106 * , 071301 ( 2011 ) doi:10.1103/physrevlett.106.071301 [ arxiv:1006.2126 [ gr - qc ] ] . g. amelino - camelia , l. freidel , j. kowalski - glikman and l. smolin , `` the principle of relative locality , '' phys .
d * 84 * , 084010 ( 2011 ) doi:10.1103/physrevd.84.084010 [ arxiv:1101.0931 [ hep - th ] ] .
g. amelino - camelia , n. loret and g. rosati , `` speed of particles and a relativity of locality in @xmath3-minkowski quantum spacetime , '' phys .
b * 700 * , 150 ( 2011 ) doi:10.1016/j.physletb.2011.04.054 [ arxiv:1102.4637 [ hep - th ] ] . g. amelino - camelia , m. arzano , j. kowalski - glikman , g. rosati and g. trevisan , `` relative - locality distant observers and the phenomenology of momentum - space geometry , '' class . quant .
* 29 * , 075007 ( 2012 ) doi:10.1088/0264 - 9381/29/7/075007 [ arxiv:1107.1724 [ hep - th ] ] .
g. amelino - camelia , l. barcaroli , g. gubitosi and n. loret , `` dual redshift on planck - scale - curved momentum spaces , '' class .
grav . * 30 * , 235002 ( 2013 ) doi:10.1088/0264 - 9381/30/23/235002 [ arxiv:1305.5062 [ gr - qc ] ] .
n. loret , `` exploring special relative locality with de sitter momentum - space , '' phys .
d * 90 * , no .
12 , 124013 ( 2014 ) doi:10.1103/physrevd.90.124013 [ arxiv:1404.5093 [ hep - th ] ] . g. gubitosi and f. mercati , `` relative locality in @xmath3-poincar , '' class .
grav . * 30 * , 145002 ( 2013 ) doi:10.1088/0264 - 9381/30/14/145002 [ arxiv:1106.5710 [ gr - qc ] ] .
a. ballesteros , n. r. bruno and f. j. herranz , `` on ( 3 + 1 ) anti - de sitter and de sitter lie bialgebras with dimensionful deformation parameters , '' czech .
j. phys .
* 54 * , 1321 ( 2004 ) doi:10.1007/s10582 - 004 - 9795-x [ hep - th/0408196 ] ; j. lukierski , a. nowicki and h. ruegg , `` real forms of complex quantum anti - de sitter algebra u - q(sp(4:c ) ) and their contraction schemes , '' phys .
b * 271 * , 321 ( 1991 ) doi:10.1016/0370 - 2693(91)90094 - 7 [ hep - th/9108018 ] .
a. marciano , g. amelino - camelia , n. r. bruno , g. gubitosi , g. mandanici and a. melchiorri , `` interplay between curvature and planck - scale effects in astrophysics and cosmology , '' jcap * 1006 * , 030 ( 2010 ) doi:10.1088/1475 - 7516/2010/06/030 [ arxiv:1004.1110 [ gr - qc ] ] .
l. barcaroli and g. gubitosi , `` kinematics of particles with quantum - de sitter - inspired symmetries , '' phys . rev .
d * 93 * , no .
12 , 124063 ( 2016 ) doi:10.1103/physrevd.93.124063 [ arxiv:1512.03462 [ gr - qc ] ] .
m. assanioussi , a. dapor and j. lewandowski , `` rainbow metric from quantum gravity , '' phys
. lett .
b * 751 * , 302 ( 2015 ) doi:10.1016/j.physletb.2015.10.043 [ arxiv:1412.6000 [ gr - qc ] ]
. j. polchinski , _ string theory , vol . 1 : introduction to the bosonic string _ cambridge university press ( 1998 ) . i. p. lobo , n. loret and f. nettel , `` rainbows without unicorns : metric structures in theories with modified dispersion relations , '' arxiv:1610.04277 [ gr - qc ]
. g. g. carvalho , i. p. lobo and e. bittencourt , `` extended disformal approach in the scenario of rainbow gravity , '' phys .
d * 93 * , no . 4 , 044005 ( 2016 ) doi:10.1103/physrevd.93.044005 [ arxiv:1511.00495 [ gr - qc ] ] .
m. a. javaloyes , anisotropic tensor calculus , " arxiv:1602.05492 [ math.dg ] .
j. m. carmona , j. l. cortes and f. mercati , `` relativistic kinematics beyond special relativity , '' phys .
d * 86 * , 084032 ( 2012 ) doi:10.1103/physrevd.86.084032 [ arxiv:1206.5961 [ hep - th ] ] . s. majid , `` algebraic approach to quantum gravity .
noncommutative spacetime , '' in * oriti , d. ( ed . ) : approaches to quantum gravity * 466 - 492 [ hep - th/0604130 ] .
g. amelino - camelia , m. arzano , g. gubitosi and j. magueijo , `` dimensional reduction in the sky , '' phys .
d * 87 * , no .
12 , 123532 ( 2013 ) doi:10.1103/physrevd.87.123532 [ arxiv:1305.3153 [ gr - qc ] ] .
t. padmanabhan , s. chakraborty and d. kothawala , `` spacetime with zero point length is two - dimensional at the planck scale , '' gen
. rel .
grav . * 48 * , no . 5 , 55 ( 2016 ) doi:10.1007/s10714 - 016 - 2053 - 2 [ arxiv:1507.05669 [ gr - qc ] ] .
i. ferreras and i. trujillo , `` testing the wavelength dependence of cosmological redshift down to @xmath242 , '' astrophys . j. * 825 * , no . 2 , 115 ( 2016 ) doi:10.3847/0004 - 637x/825/2/115 [ arxiv:1605.09380 [ astro-ph.co ] ] .
|
in recent years , planck - scale modifications to particles dispersion relation have been deeply studied for the possibility to formulate some phenomenology of planckian effects in astrophysical and cosmological frameworks .
there are some indications [ arxiv : gr - qc/0611024 ] that finsler geometry can provide some generalization of riemannian geometry which may allow to account for non - trivial ( planckian ) structure of relativistic particles configuration space .
we investigate the possibility to formalize planck - scale deformations to relativistic models in curved spacetime , within the framework of finsler geometry .
we take into account the general strategy of analysis of modifications of dispersion relations in curved spacetimes proposed in [ arxiv:1507.02056 ] , generalizing to the de sitter case the results obtained in [ arxiv:1407.8143 ] , for deformed relativistic particle kinematics in flat spacetime using finsler formalism .
|
in 1982 feynman pointed out that it appears to be impossible to efficiently simulate the behaviour of a quantum mechanical system with a computer@xcite .
this problem arises because the quantum system is not confined to its eigenstates , but can exist in any superposition of them , and so the space needed to describe the system is very large . to take a simple example , a system comprising @xmath0 two level sub - systems , such as @xmath0 spin - particles , inhabits a hilbert space of dimension @xmath1 , and evolves under a series of transformations described by matrices containing @xmath2 elements . for this reason it is impractical to simulate the behaviour of spin systems containing more than about a dozen spins .
the difficulty of simulating quantum systems using classical computers suggests that quantum systems have an information processing capability much greater than that of corresponding classical systems .
thus it might be possible to build quantum mechanical computers@xcite which utilise this information processing capability in an effective way to achieve a computing power well beyond that of a classical computer . such a quantum computer could be used to efficiently simulate other quantum mechanical systems@xcite , or to solve conventional mathematical problems@xcite which suffer from a similar exponential growth in complexity , such as factoring@xcite .
considerable progress in this direction has been made in recent years .
the basic logic elements necessary to carry out quantum computing are well understood , and quantum algorithms have been developed , both for simple demonstration problems@xcite and for more substantial problems such as factoring@xcite .
experimental implementation of a quantum computer has , however , proved difficult .
much effort has been directed towards implementing quantum computers using ions trapped by electric and magnetic fields@xcite , and while this approach has shown some success@xcite , it has proved difficult to progress beyond computers containing a single two - level system ( corresponding to a single quantum bit , or qubit ) .
recently two separate approaches have been described@xcite for the implementation of a quantum computer using nuclear magnetic resonance@xcite ( nmr ) .
these approaches shows great promise as it has proved relatively simple to investigate quantum systems containing two or three qubits@xcite .
here we describe our implementation of a simple quantum algorithm for solving deutsch s problem,@xcite on a two qubit nmr quantum computer .
all current implementations of quantum computers are built up from a small number of basic elements .
the first of these is the qubit , which plays the same role as that of the bit in a classical computer .
a classical bit can be in one of two states , @xmath3 or @xmath4 , and similarly a qubit can be represented by any two level system with eigenstates labelled and .
one obvious implementation is to use the two zeeman levels of a spin- particle in a magnetic field , and we shall assume this implementation throughout the rest of this paper . unlike a bit , however , a qubit is not confined to these two eigenstates , but can in general exist in some superposition of the two states .
it is this ability to exist in superpositions which makes quantum systems so difficult to simulate and which gives quantum computers their power .
the second requirement is a set of logic gates , corresponding to gates such as and , or and not in conventional computers.@xcite quantum gates differ from their classical counterparts in one very important way : they must be reversible@xcite .
this is because the evolution of any quantum system can be described by a series of unitary transformations , which are themselves reversible .
this need for reversibility has many consequences for the design of quantum gates .
clearly for a gate to be reversible it must be possible to reconstruct the input bits knowing only the design of the gate and the output bits , and so every input bit must be in some sense preserved in the outputs .
one trivial consequence of this is that the gate must have exactly as many outputs as inputs .
for this reason it is obvious that gates such as and and or are not reversible .
it is , however , possible to construct reversible equivalents of and and or , in which the input bits are preserved .
just as it can be shown that one or more gates ( such as the nand gate ) are universal for classical computing@xcite ( that is , any classical gate can be constructed using only wires and nand gates ) , it can be shown that certain gates or combinations of gates are universal for quantum computing .
in particular it can be shown@xcite that the combination of a general single qubit rotation with the two bit `` controlled - not '' gate ( cnot ) is universal .
furthermore it is possible to build a reversible equivalent of the nand gate , and thus to implement any classical logic operation using reversible logic .
single qubit rotations are easily implemented in nmr as they correspond to rotations within the subspace corresponding to a single spin , and such rotations can be achieved using radio frequency ( rf ) fields .
one particularly important single bit gate is the hadamard gate , which performs the rotational transformation @xmath5 the hadamard operator can thus be used to convert eigenstates into superpositions of states .
similarly , as the hadamard is self - inverse , it can be used to convert superpositions of states back into eigenstates for later analysis .
two - bit gates correspond to rotations within subspaces corresponding to two spins , and thus require some kind of spin
spin interaction for their implementation . in nmr
the scalar spin
spin coupling ( j - coupling ) has the correct form , and is ideally suited for the construction of controlled gates , such as cnot .
this gate operates to invert the value of one qubit when another qubit ( the control qubit ) has some specified value , usually ; its truth table is shown in table [ table : cnot ] .
.the truth table for the cnot gate .
the first qubit ( the control qubit ) is unchanged by the gate , while the second qubit is flipped if the control qubit is in state 1 , effectively implementing an xor gate .
[ cols= " < , < , < , < " , ] these four functions can be divided into two groups : the two `` constant '' functions , for which @xmath6 is independent of @xmath7 ( @xmath8 and @xmath9 ) , and the two `` balanced '' functions , for which @xmath6 is zero for one value of @xmath7 and unity for the other ( @xmath10 and @xmath11 ) .
given some unknown function @xmath12 ( known to be one of these four functions ) , it is possible to determine which of the four functions it is by applying @xmath12 to two known inputs : @xmath3 and @xmath4 .
this procedure also provides enough information to determine whether the function is constant or balanced .
however knowing whether the function is constant or balanced corresponds to only one bit of information , and so it might be possible to answer this question using only one evaluation of the function @xmath12 .
equivalently , it might be possible to determine the value of @xmath13 using only one evaluation of @xmath12 .
( the symbol @xmath14 indicates addition modulo @xmath15 , and for two one bit numbers , @xmath16 and @xmath17 , @xmath18 equals @xmath3 if @xmath16 and @xmath17 are the same , and @xmath4 if they are different . ) in fact this can be achieved as long as the calculation is performed using a quantum computer rather than a classical one . quantum computers of necessity use reversible logic , and
so it is not possible to implement the binary function @xmath12 directly .
it is , however , possible to design a propagator , @xmath19 , which captures @xmath12 within a reversible transformation by using a system with two input qubits and two output qubits as follows @xmath20 the two input bits are preserved ( @xmath7 is preserved directly , while @xmath21 is preserved by combining it with @xmath6 , the desired result ) , and so @xmath19 corresponds to a reversible transformation . note that for any one bit number @xmath16 , @xmath22 , and so values of @xmath6 can be determined by setting the second input bit to @xmath3 . using this propagator and appropriate input states it is possible to evaluate @xmath23 and @xmath24 using @xmath25 and @xmath26
the approach outlined above , in which the state of a quantum computer is described explicitly , swiftly becomes unwieldy , and it is useful to use more compact notations .
one particularly simple approach is to use quantum circuits,@xcite which may be drawn by analogy with classical electronic circuits . in this approach lines
are used to represent `` wires '' down which qubits `` flow '' , while boxes represent quantum gates which perform appropriate unitary transformations .
for example , the analysis of @xmath12 can be summarised by the circuit shown in figure [ fig : cf ] .
( 200,100 ) ( 15,70)(0,0)[r ] ( 20,70)(1,0)50 ( 15,30)(0,0)[r ] ( 20,30)(1,0)50 ( 70,20)(20,60)@xmath19 ( 90,70)(1,0)50 ( 145,70)(0,0)[l ] ( 90,30)(1,0)50 ( 145,30)(0,0)[l ] so far this is simply using a quantum computer to simulate a classical computer implementing classical algorithms . with a quantum computer ,
however , it is not necessary to start with the system in some eigenstate ; instead it is possible to begin with a superposition of states .
suppose the calculation begins with the second qubit in the superposition @xmath27 .
then @xmath28 ( we have used the fact that @xmath22 , as before , while @xmath29 if @xmath30 and @xmath3 if @xmath31 ) .
the value of @xmath6 is now encoded in the overall phase of the result , with the qubits left otherwise unchanged . while this is not particularly useful , suppose the calculation begins with the first qubit also in a superposition of states , namely @xmath32 .
then@xcite @xmath33 with the first qubit ending up in the superposition @xmath34 , with the desired answer ( @xmath13 ) encoded as the _ relative _ phase of the two states contributing to the superposition .
this relative phase can be measured , and so the value of @xmath13 ( that is , whether @xmath12 is constant or balanced ) has been determined using only one application of the propagator @xmath19 , that is only one evaluation of the function @xmath12 .
this approach can be easily implemented using a quantum circuit , as shown in figure [ fig : cd ] .
( 210,100 ) ( 15,70)(0,0)[r ] ( 20,70)(1,0)20 ( 40,60)(20,20)@xmath35 ( 60,70)(1,0)20 ( 15,30)(0,0)[r ] ( 20,30)(1,0)20 ( 40,20)(20,20)@xmath35 ( 60,30)(1,0)20 ( 80,20)(20,60)@xmath19 ( 100,70)(1,0)20 ( 120,60)(20,20)@xmath35 ( 140,70)(1,0)20 ( 165,70)(0,0)[l ] ( 100,30)(1,0)20 ( 120,20)(20,20)@xmath35 ( 140,30)(1,0)20 ( 165,30)(0,0)[l ] this circuit starts off from appropriate eigenstates , uses hadamard transformations to convert these into superpositions , applies the propagator @xmath19 to these superpositions , and finally use another pair of hadamard transforms to convert the superpositions back into eigenstates which encode the desired result .
the deutsch algorithm can be implemented on a quantum computer with two qubits , such as an nmr quantum computer based on two coupled spins .
first it is necessary to show that the individual components of the quantum circuit can be built .
it is convenient to begin by writing down the necessary states and operators using the product operator basis set@xcite normally used in describing nmr experiments ( this basis is formed by taking outer products between pauli matrices describing the individual spins , together with the scaled unit matrix , @xmath36 ) .
the initial state , @xmath37 , can be written as a vector in hilbert space , @xmath38 but this description is not really appropriate .
unlike other implementations , an nmr quantum computer comprises not just a single set of spins but rather an ensemble of spins in a statistical mixture of states .
such a system is most conveniently treated using a density matrix , which can describe either a mixture or a pure state ; for example @xmath39 this density matrix can be decomposed in the product operator basis as @xmath40 . ignoring multiples of the unit matrix ( which give rise to no observable effects in any nmr experiment )
this can be reached from the thermal equilibrium density matrix ( @xmath41 ) by a series of rf and field gradient pulses@xcite .
the unitary transformation matrix corresponding to the hadamard operator on a single spin can be written as @xmath42 this corresponds to a @xmath43 rotation around an axis tilted at @xmath44 between the @xmath45 and @xmath7-axes .
such a rotation can be achieved directly using an off resonance pulse,@xcite or using a three pulse sandwich@xcite such as @xmath46 .
even more simply the hadamard can be approximated by a @xmath47 pulse .
while this is clearly not a true hadamard operator ( for example , it is not self inverse ) , its behaviour is similar and it can be used in some cases : for example , it is possible to replace the first pair of hadamard gates in the circuit for the deutsch algorithm ( figure [ fig : cd ] ) by @xmath47 pulses and the second pair of gates by @xmath48 pulses .
clearly it is possible to apply the hadamard operator either to just one of the two spins ( using selective soft rf pulses@xcite ) or to both spins simultaneously ( using non - selective hard pulses ) . the unitary transformations corresponding to the four possible propagators @xmath19 are also easily derived .
each propagator corresponds to flipping the state of the second qubit under certain conditions as follows : @xmath49 , never flip the second qubit ; @xmath50 , flip the second qubit when the first qubit is in state one ; @xmath51 , flip the second qubit when the first qubit is in state zero ; @xmath52 , always flip the second qubit .
the first and last cases are particularly simple , as @xmath49 corresponds to doing nothing ( the identity operation ) , while @xmath52 corresponds to inverting the second spin ( a conventional not gate , or , equivalently , a @xmath43 pulse ) .
the second and third propagators correspond to controlled - not gates , which can be implemented using spin spin couplings .
for example @xmath50 is described by the matrix @xmath53 which can be achieved using the pulse sequence @xmath54 where @xmath55 indicates a @xmath56 pulse on the second spin , _ couple _ indicates evolution under the scalar coupling hamiltonian , @xmath57 , for a time @xmath58 , and @xmath59 and @xmath60 indicate either periods of free precession under zeeman hamiltonians or the application of composite @xmath45-pulses.@xcite similarly @xmath51 can be achieved using the pulse sequence @xmath61 the pulse sequences described above can be implemented in many different ways , as different composite @xmath45-pulses can be used , the order of some of the pulses can be varied , and in some cases different pulses can be combined together .
we chose to use the implementation @xmath62 where pulses not marked as either @xmath63 or @xmath64 were applied to both nuclei .
the phase of the final pulse distinguishes @xmath50 ( for which the final pulse is @xmath65 ) from @xmath51 ( for which it is @xmath66 ) .
finally it is necessary to consider analysis of the final state , which could in general be one of the four states @xmath67 , @xmath68 , @xmath69 , or @xmath70 . in order to distinguish these states
it is necessary to apply a @xmath47 pulse and observe the nmr spectrum .
the final nmr signal observed from spin @xmath63 is @xmath71 if the spin is in state @xmath3 , and @xmath72 if it is in state @xmath4 . for a computer implementing the deutsch algorithm
the final detection @xmath47 pulses cancel out the two final pseudo - hadamard @xmath48 pulses , and thus all four pulses can be omitted ( see figure [ fig : cdnmr ] ) .
( 230,100 ) ( 0,90)(0,0)(a ) ( 15,70)(0,0)[r ] ( 20,70)(1,0)80 ( 15,30)(0,0)[r ] ( 20,30)(1,0)80 ( 100,20)(20,60)@xmath19 ( 120,70)(1,0)25 ( 145,60)(30,20)@xmath73 ( 175,70)(1,0)25 ( 205,70)(0,0)[l]@xmath74 ( 120,30)(1,0)25 ( 145,20)(30,20)@xmath73 ( 175,30)(1,0)25 ( 205,30)(0,0)[l]@xmath75 ( 230,100 ) ( 0,90)(0,0)(b ) ( 15,70)(0,0)[r ] ( 20,70)(1,0)25 ( 45,60)(30,20)@xmath47 ( 75,70)(1,0)25 ( 15,30)(0,0)[r ] ( 20,30)(1,0)25 ( 45,20)(30,20)@xmath47 ( 75,30)(1,0)25 ( 100,20)(20,60)@xmath19 ( 120,70)(1,0)80 ( 205,70)(0,0)[l]@xmath75 ( 120,30)(1,0)80 ( 205,30)(0,0)[l]@xmath76 the final nmr signal observed is either @xmath77 ( corresponding to @xmath68 ) or @xmath78 ( corresponding to @xmath70 ) .
hence it is simple to determine the value of @xmath13 ( that is , determine whether the function is constant or balanced ) by determining the relative phase of the signals from the two spins .
in order to demonstrate the results described above , we have constructed an nmr quantum computer capable of implementing the deutsch algorithm . for our two - spin system
we chose to use a @xmath79 solution of the pyrimidine base cytosine in @xmath80 ; rapid exchange of the two amine protons and the single amide proton with the deuterated solvent leaves two remaining protons forming an isolated two - spin system .
all nmr experiments were conducted at @xmath81 and @xmath82 on a home - built nmr spectrometer at the oxford centre for molecular sciences , with a operating frequency of .
the observed j - coupling between the two protons was , while the difference in resonance frequencies was .
selective excitation was achieved using gaussian@xcite soft pulses incorporating a phase ramp@xcite to allow excitation away from the transmitter frequency . during a selective pulse
the other ( unexcited ) spin continues to experience the main zeeman interaction , resulting in a rotation around the @xmath45-axis , but the length of the selective pulses can be chosen such that the net rotation experienced by the other spin is zero .
the residual hod resonance was suppressed by low power saturation during the relaxation delays .
this system can be used both for the implementation of classical algorithms to analyse @xmath23 and @xmath24 and for the implementation of the deutsch algorithm ; as shown in figure [ fig : cdnmr ] the pulse sequences differ only in the placement of the @xmath47 pulses .
the results for the classical algorithm to determine @xmath23 are shown in figure [ fig : rf0 ] .
the lefthand pair of signals corresponds to the first spin ( @xmath63 ) , while the pair on the right hand side correspond to the second spin ( @xmath64 ) ; the ( barely visible ) splitting in each pair arises from the scalar coupling @xmath83 . in this experiment
the value of @xmath23 is determined by setting both spins @xmath63 and @xmath64 into state , performing the calculation , and then measuring the final state of spin @xmath64 ; spin @xmath63 should not be affected , and so should remain in state .
the phase of the reference spectrum ( a ) was adjusted so that signals from spin @xmath63 appear in absorption , and the same phase correction was applied to the other three spectra .
the state of a spin after a calculation can then be determined by determining whether the corresponding signals in the spectrum are in absorption ( state ) or emission ( state ) .
as expected spin @xmath63 does indeed remain in state , while the value of @xmath23 ( determined from spin @xmath64 ) is @xmath3 for @xmath84 and @xmath85 but @xmath4 for @xmath86 and @xmath87 .
clearly our nmr quantum computer is capable of implementing this classical algorithm , as it is simple to determine @xmath23 .
the other value , @xmath24 , can be determined in a very similar way ( see figure [ fig : rf1 ] ) .
in this case spin @xmath63 remains in state , while @xmath24 equals @xmath3 for @xmath84 and @xmath86 and equals @xmath4 for @xmath85 and @xmath87 .
there are , however , several imperfections visible in the results . first , the signals are not perfectly phased : rather than exhibiting pure absorption or pure emission lineshapes the signals have more complex shapes , including dispersive components .
these arise from the difficulty of implementing perfect selective pulses , which effect the desired rotation at one spin while leaving the other spin entirely unaffected .
similarly the selective pulses will not perfectly suppress j - couplings during the excitation , leading to the appearance of antiphase contributions to the lineshape .
any practical selective pulse will be imperfect , and so will result in systematic distortions in the final result .
note that these distortions are most severe in cases ( b ) and ( c ) , where the propagator is complex , containing a large number of selective pulses .
interestingly the distortions are also more severe for the measurement of @xmath23 ( figure [ fig : rf0 ] ) than for @xmath24 ( figure [ fig : rf1 ] ) ; there is no simple explanation for this effect , which is due to the complex interplay of many selective pulses .
we are currently seeking ways to minimise these effects .
secondly , the signal intensities vary in different cases ; as before the signal loss is most severe in cases ( b ) and ( c ) , corresponding to complex propagators .
this is in part a consequence of imperfect selective pulses , as discussed above , but may also indicate the effects of spin relaxation , that is decoherence of the states involved in the calculation .
decoherence is a fundamental problem , and may ultimately limit the size of practical quantum computers,@xcite although a variety of error correction techniques@xcite have been devised to overcome it .
these imperfections are not a major problem in our nmr quantum computer , as it is still easy to determine the state of a spin .
however our computer is small , and the programs run on it are short ( that is , they contain a small number of logic gates ) ; if more complex programs are to be run on larger computers then these imperfections must be addressed .
the results of implementing the deutsch quantum algorithm are shown in figure [ fig : rd ] . in this case
the result ( @xmath13 ) can be read from the final state of spin @xmath63 , while spin @xmath64 remains in state .
as expected spin @xmath63 is in state for the two constant functions ( @xmath8 and @xmath9 ) , but in state for the two balanced functions ( @xmath10 and @xmath11 ) .
once again a number of imperfections are visible , though in this case they appear to be most severe in the case of @xmath87 .
we have demonstrated that the isolated pair of nuclei in partially deuterated cytosine can be used to implement a two qubit nmr quantum computer .
this computer can be used to run both classical algorithms and quantum algorithms , such as that for solving deutsch s problem ( distinguishing between constant and balanced functions ) .
this is the first successful implementation of a quantum algorithm on any physical system@xcite .
this result confirms that nmr shows great promise as a technology for the implementation of small quantum computers .
difficulties do exist , largely as a result of the large number of selective pulses involved in the implementation of quantum gates , but we are currently seeking ways to overcome these problems . even with the current level of errors it should be possible to build a three qubit computer capable of implementing more complex logic gates and algorithms .
we are indebted to r. h. hansen ( clarendon laboratory ) for invaluable advice and assistance .
we thank n. soffe and j. boyd ( ocms ) for assistance with implementing the nmr pulse sequences .
we are grateful to a. ekert ( clarendon laboratory ) and r. jozsa ( university of plymouth ) for helpful conversations .
jaj thanks c. m. dobson ( ocms ) for his encouragement and support .
this is a contribution from the oxford centre for molecular sciences which is supported by the uk epsrc , bbsrc and mrc .
mm thanks cesg ( uk ) for their support .
since initial submission of this manuscript there has been considerable progress in this field , including another implementation of an algorithm to solve deutsch s problem@xcite , and two implementations of grover s quantum search algorithm@xcite .
|
quantum computing shows great promise for the solution of many difficult problems , such as the simulation of quantum systems and the factorisation of large numbers .
while the theory of quantum computing is well understood , it has proved difficult to implement quantum computers in real physical systems .
it has recently been shown that nuclear magnetic resonance ( nmr ) can be used to implement small quantum computers using the spin states of nuclei in carefully chosen small molecules . here
we demonstrate the use of an nmr quantum computer based on the pyrimidine base cytosine , and the implementation of a quantum algorithm to solve deutsch s problem ( distinguishing between constant and balanced functions ) .
this is the first successful implementation of a quantum algorithm on any physical system .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ + j. a. jones + _ oxford centre for molecular sciences , new chemistry laboratory , south parks road , oxford ox1 3qt , uk , _ and _ centre for quantum computing , clarendon laboratory , parks road , oxford ox1 3pu , uk _ + m. mosca + _ centre for quantum computing , clarendon laboratory , parks road , oxford ox1 3pu , uk , _ and _ mathematical institute , 2429 st giles , oxford , ox1 3 lb , uk _ + correspondence should be addressed to j. a. jones at the new chemistry laboratory . + email : [email protected] + _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
|
hot spots and knots in large - scale jets have been observed in many active galactic nuclei ( agns ) .
hot spots are often found near the outmost boundaries of radio lobes .
they are regarded as jet termination ( fnanaroff & riley 1974 ; blandford & rees 1974 ; begelman et al .
1984 ; bicknell 1985 ; meisenheimer et al .
knots are usually thought to be a part of a well - collimated jet ( e.g. , harris & krawczynski 2006 ) .
some of them are also detected in the optical and x - ray bands . with high spatial resolution and sensitivity ,
the _ chandra _ x - ray observatory opened a new era to study the x - ray emission of hot spots and knots .
it revealed that bright radio knots and hot spots in radio galaxies and quasars are also often detected in the x - ray band ( wilson et al .
2001 ; hardcastle et al . 2004
; kataoka & stawarz 2005 ; tavecchio et al .
. the x - ray radiation mechanisms of hot spots and knots are highly debated .
it is generally believed that synchrotron radiation of relativistic electrons is responsible for the radio emission .
the detection of high polarization in the optical emission indicates that the optical emission is also from synchrotron radiation ( roser & meisenheimer 1987 ; lhteenmki & valtaoja 1999 ) .
it is uncertain whether the x - ray emission is a simple extrapolation of the synchrotron radiation for the radio - to - optical emission .
indeed , the x - rays of some hot spots ( such as hot spots n1 and n2 in 3c 33 ; kraft et al .
2007 ) and knots ( such as the knots in 3c 371 and m87 ; sambruna et al . 2007
; liu & shen 2007 ) can be interpreted as synchrotron radiation from the same electron population responsible for the radio and optical emission . however , the x - ray emission of some hot spots and knots is apparently not a simple extrapolation of the radio - to - optical component ( kataoka & stawarz 2005 ) .
an inverse compton scattering ( ic ) component is necessary to model the x - rays . as shown by stawarz et al .
( 2007 ) , the x - ray emission of hot spots in cygnus a is well interpreted with the synchrotron - self - compton ( ssc ) scattering model .
the ic scattering of cosmic microwave background ( ic / cmb ) may also significantly contribute to the observed x - ray emission , if the bulk flow of the material in hot spots and knots is relativistic ( georganopoulos & kazanas 2003 ) .
broadband spectral energy distribution ( sed ) places strong constraints on the radiation mechanism models .
systematical analysis on the seds in the radio and x - ray bands of hot spots , knots and lobes were present by hardcastle et al .
( 2004 ) and kataoka & stawarz ( 2005 ) .
note that the optical data are critical to characterize the seds , and may give more constraints on the models . in this paper
, we compile a large sample of the observed seds in the radio - optical - x - ray band of hot spots and knots from literature , and fit them with various models in order to study the radiation mechanisms of the x - rays and to reveal the differences of the two kinds of sources .
the magnetic field and doppler boosting effect are crucial ingredients in sed modeling .
the energy equipartition assumption between magnetic field and radiating electrons is usually accepted in discussion of the energetics and dynamics of radio sources .
hardcastle et al . (
2004 ) reported that most radio - luminous hot spots can be explained with the ssc model under this assumption . with the equipartition condition ,
the derived magnetic field of lobes is @xmath16 g ( kataoka & stawarz 2005 ) , being comparable to the strength of the magnetic field in the interstellar medium . since
hot spots are believed to be the terminal of a relativistic jet , the doppler boosting effect would be less prominent than that in knots , but their magnetic fields may be magnified up to @xmath17 g by strong external shocks produced by interaction of a relativistic jet with circum medium ( kataoka & stawarz 2005 ) .
therefore , the doppler boosting effect and co - moving magnetic field are essential to discriminate the two kinds of sources , if they are physically different . with our detailed sed fits ,
we compare their doppler factor ( @xmath10 ) and co - moving magnetic field ( @xmath18 ) between the knots and hot spots in our sample under the equipartition assumption .
the observed seds and our data analysis are present in 2 .
models and our sed fits are shown in 3 .
conclusions and discussion are present in 4 . throughout
, @xmath19=70 km s@xmath20 mpc@xmath20 , @xmath21 and @xmath22 are adopted .
twenty - two hot spots and 45 knots from 35 agns ( 15 radio galaxies , 16 quasars , 3 bl lac objects , and 1 seyfert galaxy ; see table 1 ) are included in our sample .
most of them are taken from xjet home page .
observations for these hot spots and knots are summarized in table 2 , and their seds are displayed in figure 1 .
we show the distributions and the correlations of the spectral indices in the radio and x - ray bands ( @xmath23 and @xmath24 ) in figure 2 . it is found that @xmath23 is not correlated with @xmath24 .
the @xmath23 of both the knots and hot spots are smaller than 1 .
the @xmath24 narrowly clusters at @xmath25 for the hot spots , but it ranges in @xmath26 for the knots .
we test whether @xmath23 and @xmath24 distributions show statistical differences between the two kinds of sources with the kolmogorov - smirnov test ( k - s test ) , which yields a chance probability @xmath27 . a k - s test probability larger than 0.1 would strongly suggest no statistical difference between two distributions .
we get @xmath28 and @xmath29 for the radio spectral indices and the x - ray spectral indices between the two samples , respectively , indicating that no statistical difference is found for the @xmath23 and @xmath24 distributions of two kinds of sources .
figure [ fig:3](a ) shows the correlations between the observed luminosities at 5 ghz and 1 kev ( @xmath30 and @xmath1 ) for the hot spots and the knots .
it is found that @xmath1 is tightly correlated with @xmath0 .
we measure the correlations with the spearman correlation analysis , which yields @xmath31 with a correlation coefficient @xmath32 and a chance probability @xmath33 for the hot spots and @xmath34 with @xmath35 and @xmath33 for the knots .
the slopes of the two correlations are the same in the error scopes , but averagely speaking , the x - ray luminosity of the knots is larger than that of the hot spots with @xmath36 order of magnitude , indicating systematical difference between the two kinds of sources . as seen in 3.3 ,
this correlation is much tighter by correcting with the doppler boosting effect ( see figure [ fig:3](b ) and details in 3.3 ) .
the knots and the hot spots in the @xmath30-@xmath1 plane are roughly separated with a division line of @xmath37 .
therefore , the ratio of @xmath38 is a characteristic to distinguish the hot spots and the knots .
this ratio should be an intrinsic parameter independent of the doppler boosting effect and the cosmological effect .
it may reflect the properties of the radiation regions .
the tight @xmath30-@xmath1 correlation indicates that the radiations in the two energy bands may be produced by the same electron population . for few cases , such as hot spots n1 and n2 in 3c 33 and knots in m87 ( kraft et al .
2007 ; liu & shen 2007 ) , their x - ray spectra are soft with @xmath39 , and smoothly connect to the spectrum in the radio and optical band .
the x - rays of these sources may be the high energy tail of the synchrotron radiation by the same electron population for the radio and optical emission .
the synchrotron radiation model is preferred to fit the x - rays of these sources .
some well - sampled seds in figure 1 roughly show two bumps similar to that observed in blazars .
the two - bump feature is generally interpreted with the synchrotron radiation and ic scattering by the relativistic electrons .
therefore , we fit these seds with single - zone synchrotron + ic scatting models . the ic seed photons may be originated from the synchrotron radiation itself ( ssc ) or from the cmb .
the photon field energy density of synchrotron radiation in the co - moving frame is given by @xmath40 erg @xmath41 , where @xmath42 in cgs units , @xmath10 the beaming factor , @xmath43 the radius of the radiation region , and @xmath44 the speed of light .
the energy density of the cmb is @xmath45 erg @xmath41 , which dramatically increases with the redshift @xmath46 of the sources and the bulk lorentz factor @xmath47 ( taking @xmath48 ) of the radiation site . without considering the doppler boosting effect , the ic component should be dominated by the ssc process , but it may be dominated by the ic / cmb , if the source is relativistic motion .
we take the two scenarios into account . in our models ,
the radiation region is assumed to be a homogeneous sphere with radius @xmath43 .
the radius is derived from the angular radius @xmath49 ( see table 2 ) , which is obtained from the optical or the x - ray observations .
considering the beaming effect , _ r _ and volume _ v _ of the emitting region are needed to take a relativistic transformation .
we simply assume that the emitting region still is a sphere with @xmath50 and derive the radius of emitting region in co - moving frame by @xmath51 .
the electron distribution as a function of electron energy ( @xmath52 ) is taken as a single power - law or a broken power law , @xmath53 where @xmath54 are the energy indices of electrons below and above the break energy @xmath55 , and @xmath56 are the observed spectral indices . in our calculations , the klein - nishina effect for the radiation in the gev - tev band is considered , but the absorption in the gev - tev band by the infrared background light and by cmb photon during the gamma - ray photons propagating to the earth ( stecker et al .
2006 ) is not taken into account . as mentioned in 1 , the equipartition condition , which assumes that the magnetic field energy density @xmath57 is equal to the electron energy density @xmath58 ,
is usually adopted in discussion of the x - ray origin .
we first derive the magnetic field strength @xmath5 ( see appendix a ) under this condition for the hot spots and knots in our sample without considering the beaming effect ( @xmath3 ) .
the calculation of @xmath5 depends on @xmath59 ( see eqs .
a5 and a10 ) .
the @xmath59 is quite uncertain ( e.g. , harris & krawczynski 2006 ) .
the @xmath59 values of 12 hot spots and all the knots in our sample are constrained with the observed seds via a method reported by tavecchio et al .
the average of @xmath59 for the 12 hot spots is @xmath60 . for those hot spots that their @xmath59 lost constraints from the observed seds , we take @xmath61 in our calculation .
we fit the observed seds in the radio - optical band with the synchrotron radiation model to derive @xmath5 .
our results are reported in table 2 .
they are roughly consistent with the results derived from the formulae given by brunetti et al .
the distributions of @xmath5 for both the hot spots and knots are shown in figure [ fig:4](a ) .
they range in @xmath62 @xmath63 g .
no systematical difference of @xmath5 is found between the two kinds of sources .
the observed x - ray spectra of the knots in 4c 73.18 ( k - a ) , pks 0521 ( k ) , m87 ( k - a , b , c1 , d , e , f ) , 3c 31 ( k ) , 3c 66b ( k - b ) , and 3c 120 ( k - k7 ) smoothly connect to the spectra in the radio and optical band .
they are well fit with the synchrotron radiation model ( the thin solid line in figure [ fig:1 ] ) , indicating that the x - rays of these sources are the high energy tail of the synchrotron radiation by the same electron population for the radio and optical emission .
we do not take these sources into account in our following analysis .
the observed x - ray spectral indices and seds shown in table 2 and figure [ fig:1 ] indicate that the x - rays of most hot spots and knots should be contributed by ic scattering .
we model the seds with the synchrotron + ic model assuming that the sources are at rest . in this scenario
, the ssc process should dominate the ic process , as mentioned above . although the contribution of ic / cmb to the x - ray emission is not negligible for some sources at high redshift
, we only consider the ssc component in this section .
we first calculate the x - ray flux density at 1 kev ( @xmath64 ) with the synchrotron + ssc model under the equipartition assumption , i.e. , @xmath65 .
our results are reported in table 2 .
we measure the consistency between @xmath64 and @xmath66 with ratio @xmath67 , where @xmath68 is the observed flux density at 1 kev .
it is found that @xmath69 is much larger than 1 for almost all the hot spots and knots in our sample , indicating that the observed x - ray flux density is much larger than the model prediction .
the distributions of @xmath69 , and @xmath69 as a function of @xmath0 , @xmath1 , and @xmath14 for the hot spots and knots are shown in figure [ fig:5 ] .
a tentative correlation presented in the @xmath70 plane shows that the brighter sources in the radio band tend to be more consistent with the equipartition condition ( see figure [ fig:5](b ) ) .
however , no similar feature is seen for the x - ray bright sources ( see figure [ fig:5](c ) ) .
it is interesting that @xmath69 is anti - correlated with @xmath14 , and both the hot spots and knots shape a well sequence ( see figure [ fig:5](d ) ) .
the hot spots are at the lower end of the sequence and they tend to be closer to the equipartition condition than the knots . as shown above , the derived x - ray flux densities from the ssc model under the equipartition condition significantly deviate the observations , especially for the knots . in order to model the observed seds with the synchrotron + ssc model , we have to get rid of this assumption . keeping the model parameters the same
as that used above , we fit the seds with this model and derive the magnetic field strengths ( @xmath71 ) .
the fits are shown in figure [ fig:1 ] ( the thick solid line ) .
although the ssc model can represent the observed flux at 1 kev for the knots in 4c 73.18 ( k - a ) , pks 0521 , m87 ( k - a , b , c1 , d , e , f ) , 3c 31 , 3c 66b ( k - b ) , 3c 120 ( k - k7 ) , 3c 273 ( k - c1 , c2 , d1 , d2h3 ) , the bow ties defined by the errors of the x - ray spectral indices clearly rule out this model for these knots .
as discussed in section 3.1 , the x - rays of these knots are well fitted by the synchrotron radiation model except knots in 3c 273 .
we do not include these knots in our following statistics .
the derived @xmath71 are listed in table 2 , and their distributions are shown in figure [ fig:4](b ) .
it is found that the @xmath71 of the knots are much smaller than the hot spots , with medians of @xmath72 g and @xmath73 @xmath63 g for the knots and hot spots , respectively . comparing @xmath71 with @xmath74
, it is found that they are roughly consistent for the hot spots , indicating that the x - rays of the hot spots can be roughly fitted with the ssc model under the equipartition condition . however , the @xmath71 of the knots are much smaller than @xmath74 , even unreasonably smaller than the magnetic field strength of the interstellar medium for some knots .
therefore , the x - rays of these knots may not be dominated by the ssc component . to investigate the deviation of @xmath71 to @xmath75 for individual source , we define ratio @xmath76 , which is physically the same as @xmath69 .
similar to @xmath69 , @xmath8 is much larger than 1 for almost all the sources , especially for the knots .
the distributions of @xmath8 and @xmath8 as a function of @xmath0 , @xmath1 , and @xmath14 are shown in figure [ fig:6 ] .
we find the same features as shown in figure [ fig:5 ] .
it is generally believed that the knots should have relativistic motion .
the hot spots may also be relativistic ( dennett - thorpe et al .
1997 ; tavecchio et al . 2005 ; harris & krawczynski 2006 ) . in this section
, we fit the seds with the synchrotron + ic model by considering the beaming effect under the equipartition condition . in this scenario ,
the ic / cmb process should dominate the ic component .
although the contribution of ssc is negligible comparing with ic / cmb in this case , we still take ssc process into account in our calculation .
our fits are shown in figure [ fig:1 ] ( the dashed line ) . the distributions of @xmath10 for the knots and hot spots are shown in figure [ fig:7 ] .
it is found that , averagely , @xmath77 for most of the knots and @xmath78 for most of the hot spots . some sources in our sample
are included in kataoka & stawarz ( 2005 ) .
we compare our results of @xmath10 for these sources with that reported by kataoka & stawarz ( 2005 ) ( @xmath79 ) in figure [ fig:8 ] .
they are roughly consistent .
the @xmath11 distributions with comparison to @xmath5 and @xmath80 are shown in figure [ fig:4](c ) .
the @xmath12 distributions are more consistent with @xmath5 than @xmath80 .
both @xmath11 and @xmath5 distributions approximately span an order of magnitude , much narrower than that of @xmath80 , especially for the knots .
all @xmath11 are larger than the magnetic filed strength of the interstellar medium , implying that the magnetic filed of the interstellar medium would be amplified in the knots and hot spots by the turbulence of the relativistic shocks .
the @xmath11 of the knots are smaller than that of the hot spots , with typical values of 10 @xmath63 g and 40 @xmath63 g for the knots and the hot spots , respectively , favoring the idea of different origins of the shocks ( internal vs. external ) for the two kinds of sources ( e.g. , harris & krawczynski 2006 ) .
as our discussed in 2 , @xmath14 is an intrinsic characteristic for the sources .
it may be a representative of the co - moving magnetic field and the doppler boosting effect of the sources .
we show @xmath14 as a function of @xmath11 and @xmath10 in figure [ fig:9 ] . it is found that @xmath14 is correlated with @xmath11 for the knots , with a linear coefficient of @xmath81 and chance probability @xmath33 .
we obtain @xmath82 .
no significant correlation between @xmath14 and @xmath11 is found for the hot spots .
however , both the knots and hot spots form a sequence in the @xmath83 plane , with a best linear fit @xmath84 .
the hot spots locate at the higher end of the sequence .
similar feature is also observed in the @xmath85 plane , as shown in figure 9(b ) .
these results imply that @xmath14 would be determined by both @xmath11 and @xmath10 .
the strong anti - correlation of @xmath86 ( or @xmath87 ) shown in figure [ fig:5 ] ( or figure [ fig:6 ] ) may be due to neglect of the beaming effect .
this effect plays important role on the observed flux since the observed flux is proportional to @xmath88 .
correcting by the doppler boosting effect , we show the @xmath89 relations in figure 3(b ) .
the relations are tighter than the observed ones , with a linear coefficient of @xmath90 and @xmath91 for the knots and hot spots , respectively .
we have present extensive analysis and sed fits for 22 hot spots and 45 knots in 35 agn jets .
we find that @xmath30 and @xmath1 are tightly correlated .
the two kinds of sources can be roughly separated with a division of @xmath92 .
our sed fits show that the mechanisms of the x - rays are diverse .
while the x - ray emission of a small fraction of the sources is a simple extrapolation of the synchrotron radiation for the radio - to - optical emission , the ic component may dominate the observed x - rays for most of the sources . without considering the relativistic bulk motion , the ssc model can explain the x - rays for some hot spots with @xmath4 being consistent with the equipartition magnetic field @xmath5 in one order of magnitude , but an unreasonably low magnetic field strength is required in modeling the x - rays for all knots with this model .
considering relativistic bulk motion for the sources , the ic / cmb dominated model well explains the x - ray emission for most sources under the equipartition condition . although the derived @xmath11 and @xmath10 for some hot spots are comparable to that of the knots , the @xmath11 value for the knots tends to be smaller than that of the hot spots and the @xmath10 tends to be larger , favoring the idea that the hot spots are jet termination and knots are a part of a well - collimated jet .
corrected by the beaming effect , the @xmath93 relations for the two kinds of sources are even tighter than the observed ones , indicating that the correlations are intrinsic .
the ratio @xmath14 is correlated with @xmath13 and @xmath10 .
these facts suggest that , under the equipartition condition , the differences on the x - ray observations for the knots and hot spots would be mainly due to the differences of the doppler boosting effect and the co - moving magnetic field , although some hot spots have similar feature to the knots
. the @xmath14 may be an indicator of @xmath13 and @xmath10 .
it is an intrinsic parameter independent of the doppler boosting and the cosmological effects .
our results show that the x - rays of a hot spot with larger @xmath14 are better to be fitted with the ssc model under equipartition condition without considering the beaming effect .
this is consistent with that reported by hardcastle et al .
( 2004 ) , who found that the x - rays of the radio - bright hot spots can be explained with the ssc model under equipartition condition .
the strong anti - correlation between @xmath8 ( or @xmath69 ) and @xmath14 may also offer a tool to discriminate the two kinds of sources .
we find in the figures [ fig:3 ] , [ fig:5 ] and [ fig:6 ] that the hot spot h - a and the knot k - ba , [ fig:5 ] and [ fig:6 ] ] in pks b1421 - 490 are significant outliers . they were reported as knots by gelbord et al .
the knot k - b is very peculiar for its extreme optical output , with a ratio of knot / core optical flux @xmath94 .
gelbord et al .
( 2005 ) suspected that it is a core between components a and c. we find that k - b looks is a significant outlier in the figures and k - a resembles a hot spot .
most recently , godfrey et al . (
2009 ) confirmed that the k - b is a core and the k - a is a hot spot with vlbi observations .
the @xmath11 value for the knots tends to be smaller than that of the hot spots and the @xmath10 tends to be larger , favoring the idea that the hot spots are jet termination and knots are a part of a well - collimated jet .
however , the @xmath11 and @xmath10 for some hot spots are comparable to that of the knots , making uncertainty on identifying a component as a knot or a hot spot . for example , the north - east double hot spots in 3c 351 , which locate at the outer boundary of a lobe , have relativistic motion feature , hence may be identified as knots of the jet ( harris & krawczynski 2006 ) .
the synchrotron radiation in the optical band indicates that there are relativistic electrons existed in these extended regions ( roser & meisenheimer 1987 ; l@xmath95hteenmki & valtaoja 1999 ) .
moreover , the x - ray emission of some sources may be also synchrotron radiation as mentioned in 3.1 . assuming a magnetic field strength @xmath96 g
, one can estimate the energy of relativistic electrons is @xmath97 , which contribute to the optical emission by synchrotron process .
these electrons may interact with the synchrotron photons and external field photons to produce very high energy @xmath52-ray photons by ic scattering . as shown in figure 1 ,
both the ssc and ic / cmb models predict a prominent gev - tev component in the seds of some sources .
we check if the predicted gev - tev emission can be detectable with h.e.s.s . and _
fermi_/lat , and also show the sensitivity curves of h.e.s.s . and _
fermi_/lat in figure 1 for these sources .
the detections of these high energy emission would place much stronger constraints on the radiation mechanisms and on the physical parameters of these sources .
the origin of the high energy tev gamma - ray emission is also a debating issue , and detections of these high energy emission would drastically improved our view of the universe ( see cui 2009 for a review ) .
note that our one - single zone lepton models can not explain the observed seds for the four knots in 3c 273 ( k - c1 , k - c2 , k - d1 , and k - d2h3 ) .
jester et al . (
2006 ) had reported that the x - ray spectra rule out the single - zone model of x - ray emission for some jet knots in 3c 273 .
it is possible that these sources may have a complex structure as the western hot spot in pictor a ( zhang et al . 2009 ) .
we thank the anonymous referee for his / her valuable suggestions .
this work was supported by the national natural science foundation of china ( grants 10778702 , 10533050 , 10873002 ) , the national basic research program ( 973 " program ) of china ( 2009cb824800 ) , and the west phd project of the training programme for the talents of west light foundation of the cas .
jmb thanks supports of the bai - ren - ji - hua and the zhong - yao - fang - xiang ( grant kjcx2-yw - t21 ) projects of the cas .
99 begelman , m. c. , blandford , r. d. , & rees , m. j. 1984 , reviews of modern physics , 56 , 255 bicknell , g. v. 1985 , proceedings of the astronomical society of australia , 6 , 130 blandford , r. d. , & rees , m. j. 1974 , , 169 , 395 brunetti , g. , setti , g. , & comastri , a. 1997 , , 325 , 898 cui , w. 2009 , research in astronomy and astrophysics , 9 , 841 dennett - thorpe , j. , bridle , a. h. , scheuer , p. a. g. , laing , r. a. , & leahy , j. p. 1997
, , 289 , 753 donahue , m. , daly , r. a. , & horner , d. j. 2003 , , 584 , 643 falomo , r. , et al .
2009 , , 501 , 907 fanaroff , b. l. , & riley , j. m. 1974 , , 167 , 31p gelbord , j. m. , et al .
2005 , , 632 , l75 georganopoulos , m. , & kazanas , d. 2003 , , 589 , l5 godfrey , l. e. h. , et al .
2009 , , 695 , 707 harris , d. e. , leighly , k. m. , & leahy , j. p. 1998
, , 499 , l149 harris , d. e. , et al . 2000 , , 530 , l81 harris , d. e. , mossman , a. e. , & walker , r. c. 2004 , , 615 , 161 harris , d. e. , & krawczynski , h. 2006 , , 44 , 463 hardcastle , m. j. , birkinshaw , m. , cameron , r. a. , harris , d. e. , looney , l. w. , & worrall , d. m. 2002 , , 581 , 948 hardcastle , m. j. , harris , d. e. , worrall , d. m. , & birkinshaw , m. 2004 , , 612 , 729 jester , s. , harris , d. e. , marshall , h. l. , & meisenheimer , k. 2006 , , 648 , 900 jester , s. , meisenheimer , k. , martel , a. r. , perlman , e. s. , & sparks , w. b. 2007 , , 380 , 828 kataoka , j. , leahy , j. p. , edwards , p. g. , kino , m. , takahara , f. , serino , y. , kawai , n. , & martel , a. r. 2003a , , 410 , 833 kataoka , j. , edwards , p. , georganopoulos , m. , takahara , f. , & wagner , s. 2003b , , 399 , 91 kataoka , j. , & stawarz , .
2005 , , 622 , 797 kraft , r. p. , hardcastle , m. j. , worrall , d. m. , & murray , s. s. 2005 , , 622 , 149 kraft , r. p. , birkinshaw , m. , hardcastle , m. j. , evans , d. a. , croston , j. h. , worrall , d. m. , & murray , s. s. 2007 , , 659 , 1008 lhteenmki , a. , & valtaoja , e. 1999 , , 117 , 1168 liu , w .- p . , &
shen , z .- q . 2007 , , 668 , l23 meisenheimer , k. , roser , h .- j . , hiltner , p. r. , yates , m. g. , longair , m. s. , chini , r. , & perley , r. a. 1989 , , 219 , 63 meisenheimer , k. , yates , m. g. , & roeser , h .- j .
1997 , , 325 , 57 perlman , e. s. , biretta , j. a. , sparks , w. b. , macchetto , f. d. , & leahy , j. p. 2001
, , 551 , 206 roeser , h .- j . , & meisenheimer , k. 1987 , , 314 , 70 sambruna , r. m. , gambill , j. k. , maraschi , l. , tavecchio , f. , cerutti , r. , cheung , c. c. , urry , c. m. , & chartas , g. 2004 , , 608 , 698 sambruna , r. m. , gliozzi , m. , donato , d. , maraschi , l. , tavecchio , f. , cheung , c. c. , urry , c. m. , & wardle , j. f. c. 2006 , , 641 , 717 sambruna , r. m. , donato , d. , tavecchio , f. , maraschi , l. , cheung , c. c. , & urry , c. m. 2007 , , 670 , 74 stawarz , . , sikora , m. , & ostrowski , m. 2003 , , 597 , 186 stawarz , . , cheung , c. c. , harris , d. e. , & ostrowski , m. 2007 , , 662 , 213 stecker , f. w. , malkan , m. a. , & scully , s. t. 2006 , , 648 , 774 tavecchio , f. , maraschi , l. , sambruna , r. m. , & urry , c. m. 2000 , , 544 , l23 tavecchio , f. , cerutti , r. , maraschi , l. , sambruna , r. m. , gambill , j. k. , cheung , c. c. , & urry , c. m. 2005 , , 630 , 721 tavecchio , f. , maraschi , l. , wolter , a. , cheung , c. c. , sambruna , r. m. , & urry , c. m. 2007 , , 662 , 900 wilson , a. s. , young , a. j. , & shopbell , p. l. 2001 , , 547 , 740 worrall , d. m. , & birkinshaw , m. 2005 , , 360 , 926 zhang , j. , bai , j .-
m . , chen , l. , & yang , x. 2009 , , 701 , 423 lllll 3c 15 & 0.073 & 329.9 & rg ( fr i / ii ) & 1 + 3c 31 & 0.0169 & 73.3 & rg ( fr i ) & 2 + 3c 33 & 0.0597 & 267.3 & rg ( fr ii ) & 3 + 3c 228 & 0.5524 & 3194 & rg & 4 + 3c 245 & 1.029 & 6845.4 & q & 5 + 3c 263 & 0.6563 & 3937 & ldq & 6 + 3c 275.1 & 0.557 & 3226.2 & ldq & 4 + 3c 280 & 0.996 & 6575 & rg ( fr ii ) & 7 + 3c 351 & 0.371 & 1988 & ldq & 6 + 3c 390.3 & 0.0561 & 250.5 & rg ( fr ii ) & 8 + 3c 295 & 0.461 & 2570.6 & rg ( fr ii ) & 9 + 3c 303 & 0.141 & 666.6 & rg ( fr ii ) & 10 + 3c 66b & 0.0215 & 93.6 & rg ( fri ) & 2 + 3c 120 & 0.033 & 144.9 & sy i & 12 + 3c 273 & 0.1583 & 756.5 & cdq & 13 + 3c 454.3 & 0.86 & 5485.1 & cdq & 14 + 3c 207 & 0.684 & 4141 & ldq & 5 + 3c 345 & 0.594 & 3487.4 & cdq & 5 + 3c 346 & 0.161 & 770.7 & rg
( fr i ) & 15 + 3c 371 & 0.051 & 226.9 & bl & 16 + 3c 403 & 0.059 & 264 & rg ( fr ii ) & 17 + m87 & 0.0043 & 18.5 & rg ( fr i ) & 20 + cygnus a & 0.0562 & 251 & rg ( fr ii ) & 11 + pictor a & 0.035 & 153.9 & rg ( fr ii ) & 21 + pks 0405 - 123 & 0.574 & 3345.6 & q & 5 + pks 0521 - 365 & 0.0554 & 247.3 & bl & 22 + pks 0637 - 752 & 0.653 & 3913.1 & cdq & 2 + pks 1136 - 135 & 0.554 & 3205.2 & ldq & 18 + pks 1229 - 021 & 1.045 & 6977.3 & cdq & 14 + pks 1421 - 490 & 0.663 & 3986.3 & q & 19 + pks 2201 + 044 ( ( 4c 04.77 ) ) & 0.027 & 118 & bl & 16 + pks 1928 + 738 ( 4c + 73.18 ) & 0.302 & 1564.7 & cdq & 5 + pks 1354 + 195 ( 4c + 19.44 ) & 0.72 & 4409.1 & cdq & 5 + pks 1150 + 497 ( 4c + 49.22 ) & 0.334 & 1758.4 & cdq & 5 + pks 0836 + 299 ( 4c + 29.30 ) & 0.064 & 287.4 & rg ( fr i ) & 5 + llllllllllllllllll 3c 33&h - s1&0.75&[email protected]&[email protected]&0.5&200&158&0.086&120&&3.3&66.1&&ssc&0.63&2.4&4.4 + & h - s2&0.98&[email protected]&[email protected]&1.5&200&64.6&0.036&20&&2.5&32.9&&ssc&0.71&2.56&3.98 + & h - n1&0.88&[email protected]&[email protected]&1.25&200&36.4&0.0016&2.1&&4&13.2&&syn&1.85&2.4&3.6 + & h - n2&0.9&[email protected]&[email protected]&1.25&200&38.7&0.0012&2.3&&3.8&14.6&&syn&1.41&2.38&3.8 + 3c 263&h - k&0.84&[email protected]&[email protected]&0.39&300&144&0.71&118&&3.1&70&&ssc&0.8&2.62&4.04 + cygnus a&h - a&0.5&[email protected]&[email protected]&1&200&214&19.4&155&&7&52.4&&ssc&0.8&1.84&3.86 + & h - d&0.38&[email protected]&[email protected]&1&100&164&47.8&160&&7&40.7&&ssc&0.77&1.5&3.24 + & h - b&0.59&[email protected]&[email protected]&1&200&330&0.55&85&&7.5&78.9&&ssc&0.77&2.02&3.96 + 3c 351&h - l&0.93&[email protected]&[email protected]&0.8&300&72.4&0.13&12.5&&5&28&&ssc&0.78&2.36&3.3 + & h - j&0.76&[email protected]&[email protected]&0.16&200&186&0.13&29&&10&33.1&&ic / cmb&0.34&2.44&3.1 + 3c 303&h - w&0.84&[email protected]&4&1&200&68.9&0.03&5&&5.9&18.6&&ic / cmb&0.5&2.64&3.7 + 3c 295&h - nw&0.94&[email protected]&3.8&0.1&500&460&0.63&200&&12&102.9&&ssc&0.94&1.9&4.4 + 3c 390.3&h - b&0.71&[email protected]&[email protected]&1&200&38.1&0.004&0.85&&8&9.45&&ic / cmb&0.78&2.56&3.1 + 3c 275.1&h - n&&&1.78&0.6&100&119&0.061&22&&4.3&40.3&&ssc&0.84&2.79 & + 3c 228&h - s&&&1.3&0.27&200&131&0.063&28&&5&40&&ssc&0.87&2.62&3.1 + 3c 245&h - d&&&[email protected]&0.8&600&67.7&0.064&18&&2.8&32&&ssc&0.66&1.96&3.92 + 3c 280&h - w&0.8&[email protected]&0.79&0.3&200&147&0.052&35&&4.2&51.2&&syn&1.22&2.6&3.3 + & h - e&0.8&1.2&0.34&0.3&200&124&0.014&23&&3.8&46.5&&syn&1.13&2.6&3.1 + pks 0405&h - n&&&[email protected]&0.7&400&81.8&0.061&16.5&&3.5&37.2&&ssc&0.91&2.8&3.22 + pks 0836&h - b&&&[email protected]&0.9&100&33.1&6.9e-4&0.5&&7.5&7.45&&syn&1.04&2.68&3.06 + pictor a&h - w&0.74&[email protected]&45&0.3&110&392&0.76&48&&19.3&45.3&&syn&1.32&2.38&3.66 + pks 1421&h - a&0.67&[email protected]&[email protected]&0.24&200&404&13.7&400&&8.5&86.2&&ssc&0.6&1.9&4.06 + m87&k - d&&[email protected]&[email protected]&0.4&150&296&0.015&2.8&&&&&syn&1.34&2.36&3.68 + & k - a&&[email protected]&[email protected]&0.9&150&266&0.15&5&&&&&syn&1.55&2.28&4.1 + & k - e&&[email protected]&[email protected]&0.9&150&101&0.0016&0.38&&&&&syn&1.38&2.42&3.76 + & k - f&&[email protected]&[email protected]&1.2&150&118&0.0049&0.9&&&&&syn&1.53&2.3&4.06 + & k - b&&[email protected]&[email protected]&1.3&150&184&0.066&5.5&&&&&syn&1.77&2.32&4.54 + & k - c1&&[email protected]&[email protected]&0.5&100&380&0.067&20&&&&&syn&1.7&2.36&4.4 + pks 1136&k - a&0.67&[email protected]&[email protected]&0.85&50&25.7&3.6e-4&0.18&&8&5.68&&ic / cmb&0.66&2.32 & + & k - b&0.81&[email protected]&[email protected]&0.6&100&57.6&9.5e-4&0.67&&9&11.4&&ic / cmb&0.8&2.61 & + & k-@xmath99&0.75&[email protected]&[email protected]&0.75&60&34&1.3e-4&0.16&&10&6.95&&ic / cmb&0.71&2.42 & + & k- d&0.71&[email protected]&[email protected]&0.6&100&94.7&0.0061&7.2&&4.7&30&&ic / cmb&0.9&2.8 & + pks 1150&k- b&0.72&[email protected]&[email protected]&0.76&100&41.8&8.8e-4&0.28&&12.2&6.7&&ic / cmb&0.72&2.46&4.6 + & k - c&0.71&[email protected]&[email protected]&0.91&300&27&5.1e-4&0.2&&8.9&6.18&&ic / cmb&0.72&2.42&3.82 + & k - d&0.68&[email protected]&[email protected]&0.91&50&28.4&1.1e-4&0.15&&9&6.7&&ic / cmb&0.69&2.38 & + & k - e&0.67&[email protected]&[email protected]&0.71&800&27.6&7e-4&0.35&&8&7.43&&ic / cmb&0.68&2.36&3.4 + & k - ij&0.81&[email protected]&[email protected]&0.71&200&50.6&0.0021&2.5&&5&17.2&&ic / cmb&0.8&2.6&4.28 + pks 2201&k - a&0.71&[email protected]&5.6&0.5&100&37.2&1.5e-4&0.06&&24&4.08&&syn&1.1&2.2&3.12 + & k-@xmath100&0.59&[email protected]&3.8&0.3&50&67.4&1.5e-5&0.031&&45&4.34&&ic / cmb&0.53&2.15&5 + 3c 371&k - a&0.69&[email protected]&7&0.7&50&32.5&7.5e-4&0.11&&17&4.14&&syn&1.09&2.34&3.12 + pks 1928&k - a&&[email protected]&[email protected]&0.8&100&36.4&5.2e-4&0.18&&&&&syn&0.88&2.6&4.06 + pks 1354&k - a&&[email protected]&[email protected]&1.9&300&27.2&0.0041&0.18&&8.5&5.92&&ic / cmb&0.41&2.12&4 + & k - b&&&[email protected]&1.4&300&23.5&0.0027&0.9&&3.2&10.1&&ic / cmb&0.5&2.32&3.4 + pks 1229&k - a&&&[email protected]&1&250&60.3&0.038&3&&5.4&17.6&&ic / cmb&0.63&2.4&3.6 + pks 0637&k&0.8&[email protected]&6.2&0.4&100&108&0.012&3.8&&9.5&20.6&&ic / cmb&0.8&2.6 & + pks 0521&k&0.89&[email protected]&14&0.4&100&118&0.014&2.5&&&&&syn&1.25&2.4&3.5 + 3c 454.3&k - a&&&[email protected]&1&300&51.8&0.008&1.5&&6&13.8&&ic / cmb&0.3&2.6&5 + & k - b&&&[email protected]&1&40&95.1&0.079&8.5&&6.1&34.6&&ic / cmb&0.72&2.44&5 + 3c 403&k - f1&&[email protected]&[email protected]&0.75&150&56.3&5.2e-4&1.1&&8&13&&ic / cmb&0.79&2.58 & + & k - f6&&[email protected]&[email protected]&0.75&100&60.8&0.0012&0.8&&11&10.6&&ic / cmb&0.68&2.4 & + 3c 346&k - c&&[email protected]&[email protected]&0.9&100&72.3&0.017&6.1&&5.4&20.9&&ic / cmb&0.8&2.6 & + 3c 345&k - a&&[email protected]&[email protected]&0.6&50&131&0.089&17&&6.4&36.6&&ic / cmb&0.76&2.52 & + 3c 207&k - a&&[email protected]&[email protected]&0.5&100&62.7&0.007&1&&11&11.2&&ic / cmb&0.4&1.8&3.6 + 3c 66b&k - a&0.75&[email protected]&[email protected]&0.7&60&45.5&2.2e-5&0.045&&24&4.47&&syn&1.16&2.46&3.4 + & k - b&0.6&[email protected]&[email protected]&0.6&50&73.2&8.5e-4&0.33&&&&&syn&1.27&2.22&3.54 + 3c 31&k - k&0.55&[email protected]&7.3&0.6&150&88.2&5.3e-4&0.2&&&&&syn&0.98&2.72&3.4 + 3c 15&k - c&0.9&[email protected]&[email protected]&0.4&30&162&0.0021&7.2&&9&31.6&&ic / cmb&0.9&2.8 & + 3c 273&k - a&0.85&[email protected]&[email protected]&0.8&20&91.8&0.0085&0.77&&22.5&12.2&&ic / cmb&0.75&2.5&5 + & k - c1&0.73&[email protected]&[email protected]&0.6&20&110&0.016&4.5&&10.5&19.6&&ic / cmb&0.74&2.48&4 + & k - c2&0.75&[email protected]&[email protected]&0.7&20&118&0.039&6.8&&9.5&22.7&&ic / cmb&0.75&2.5&3.82 + & k - b1&0.82&[email protected]&[email protected]&0.6&20&126&0.0056&2.2&&17&21.8&&ic / cmb&0.82&2.64&6 + & k - d1&0.77&[email protected]&[email protected]&0.7&20&149&0.057&13&&8.5&31&&ic / cmb&0.79&2.58&4.74 + & k - dh&0.85&[email protected]&[email protected]&1&20&174&0.19&25.5&&6.8&41.9&&ic / cmb&0.86&2.72&4.4 + 3c 120&k - k4&0.74&[email protected]&10@xmath982&0.7&100&76.2&0.0018&0.58&&18&9.24&&ic / cmb&0.58&2.4 & + & k - s2&0.67&[email protected]&0.882&1.6&100&17.1&6.9e-5&0.085&&8&3.8&&ic / cmb&0.62&2.24 & + & k - s3&0.69&&[email protected]&1.6&100&16.1&3.4e-5&0.055&&8&3.5&&syn&1.16&2.36&3.28 + & k - k7&0.68&[email protected]&[email protected]&1.5&100&35.3&&&&&&&syn&1.2&2.6&7 + 2
under the equipartition condition , we have @xmath101 the peak frequency of the synchrotron radiation is given by @xmath102 where @xmath103 hz is the larmor frequency in magnetic field @xmath104 .
the luminosity of the synchrotron radiation is derived from @xmath105 where @xmath106 is the radiation power of single electron , @xmath107 erg s@xmath20 , and @xmath108 is the volume of radiation region . without considering the beaming effect ( @xmath3 ) and assuming @xmath109 , @xmath110
is expressed as @xmath111 where @xmath112 , @xmath113 , and @xmath114 are given by @xmath115 @xmath116 and @xmath117 for the case of @xmath118 and @xmath119 , @xmath110 can be calculated by @xmath120 where @xmath121 for the case of @xmath122 , @xmath112 in eq .
( a4 ) is @xmath123 if the beaming effect is considered , we have @xmath124 , @xmath50 .
the equipartition magnetic field hance is obtained with @xmath125 this is consistent with the result presented by stawarz et al .
( 2003 ) , @xmath126 for @xmath122 .
it is generally believed the radio emission is produced by synchrotron radiation .
we obtain the values of @xmath56 , @xmath127 , and @xmath128 by fitting the observed radio ( and optical ) data using synchrotron radiation and calculate @xmath110 with eqs .
( a4 ) , ( a8 ) , and ( a11 ) .
|
the observed radio - optical - x - ray spectral energy distributions ( seds ) of 22 hot spots and 45 knots in the jets of 35 active galactic nuclei are complied from literature and modeled with single - zone lepton models .
it is found that the observed luminosities at 5 ghz ( @xmath0 ) and at 1 kev ( @xmath1 ) are tightly correlated , and the two kinds of sources can be roughly separated with a division of @xmath2 .
our sed fits show that the mechanisms of the x - rays are diverse . while the x - ray emission of a small fraction of the sources is a simple extrapolation of the synchrotron radiation for the radio - to - optical emission , an inverse compton ( ic ) scattering component is necessary to model the x - rays for most of the sources . considering the sources at rest ( the doppler factor @xmath3 )
, the synchrotron - self - compton ( ssc ) scattering would dominate the ic process .
this model can interpret the x - rays of some hot spots with a magnetic field strength ( @xmath4 ) being consistent with the equipartition magnetic field ( @xmath5 ) in one order of magnitude , but an unreasonably low @xmath4 is required to model the x - rays for all knots . measuring the deviation between @xmath4 and @xmath6 with ratio @xmath7 , we find that @xmath8 is greater than 1 and it is tightly anti - correlated with ratio @xmath9 for both the knots and the hot spots .
we propose that the deviation may be due to the neglect of the relativistic bulk motion for these sources .
considering this effect , the ic / cmb component would dominate the ic process .
we show that the ic / cmb model well explains the x - ray emission for most sources under the equipartition condition . although the derived beaming factor ( @xmath10 ) and co - moving equipartition magnetic field ( @xmath11 ) of some hot spots are comparable to the knots , the @xmath10 values of the hot spots tend to be smaller and their @xmath12 values tend to be larger than that of the knots , favoring the idea that the hot spots are jet termination and knots are a part of a well - collimated jet .
both @xmath13 and @xmath10 are tentatively correlated with @xmath14 .
corrected by the beaming effect , the @xmath15 relations for the two kinds of sources are even tighter than the observed ones .
these facts suggest that , under the equipartition condition , the observational differences of the x - rays from the knots and hot spots may be mainly due the differences on the doppler boosting effect and the co - moving magnetic field of the two kinds of sources .
our ic scattering models predict a prominent gev - tev component in the seds for some sources , which are detectable with h.e.s.s . and
_ fermi_/lat .
|
radio and x - ray radiation from solar flares have been used for many years to study energy release , plasma heating , and particle acceleration and transport .
joint studies of radio and x - ray radiation are particularly powerful because they probe populations of electrons at different energies , ranging from hot thermal electrons to nonthermal electrons of tens of kev emitting at soft x - ray ( sxr ) and hard x - ray ( hxr ) wavelengths , and nonthermal electrons with energies ranging from hundreds of kev to several mev emitting @xmath1-rays and centimeter- and millimeter - wavelength radiation ( e.g. , takakura et al .
1983 ; klein et al . 1986 ; bruggmann et al . 1994 ; ramaty et al .
1994 , trottet et al .
1998 , 2002 ) .
flares are conveniently divided into two broad classes , impulsive ( or compact ) and gradual ( or extended ) flares ( pallavicini et al .
1977 ) although more complex classification schemes have been proposed ( e.g. , bai & sturrock 1989 ) in response to more detailed and comprehensive observations .
impulsive flares tend to occur in relatively high density , compact , coronal magnetic loops , and display a good correlation between the microwave and hxr emission ( kai et al .
1985 , kosugi et al . 1988 ) .
gradual events tend to occur in lower density , larger scale coronal loops , emitting for significantly longer times .
they tend to be microwave rich " ( bai & dennis 1985 ) by a factor of @xmath2 compared to the microwave - hxr correlation of impulsive events ( kosugi et al .
the spectral and temporal profiles of these two classes also follow a predictable evolution ( e.g. silva et al . 2000 ) .
impulsive bursts show a soft - hard - soft spectral evolution in hard x - rays and a radio peak - frequency evolution that evolves from low to high , and often back to low ( melnikov et al .
gradual bursts , in contrast , show a soft - hard - harder evolution in x - rays and a radio spectral peak that is often constant or slowly increasing in frequency with time .
bursts that do not obey these general rules are often worthy of detailed study , especially when well observed , in order to test our understanding of the underlying physical mechanisms .
the impulsive flare of 2001 october 24 at 23:10 ut is an example of such a burst , which was well observed by a broad suite of instrumentation operating at radio and x - ray wavelengths and therefore offers an opportunity to study a wide range of electron energies .
the event stands out paradoxically as an extremely x - ray - poor impulsive flare , producing a moderately intense radio burst , yet very little euv , sxr , or hxr emission .
it is similar in several respects to a radio burst studied by white et al .
( 1992 ) , an event for which available radio data suggested that energy release occurred in a dense plasma ( @xmath3 @xmath4 ) .
veronig & brown ( 2004 ) have identified a class of hxr sources that also occur in a relatively dense plasma ( @xmath5 @xmath4 ) . as we show here , the oct .
24 event is another example of an event in which energy release occurs is a dense plasma .
however , unlike the events studied by veronig & brown , the oct .
24 event is highly impulsive ; and unlike the event discussed by white et al . , the oct .
24 event shows significant radio spectral evolution .
although the physical conditions and the evolution of the burst are unusual , we believe that the physical processes are not . the excellent observational coverage at radio and x - ray wavelengths provides us with the opportunity to address electron acceleration in a dense plasma environment in some detail .
in particular , the excellent time - resolved radio spectroscopic data available from ovsa and the norp has allowed us to fit the data to a source model as a function of time and thereby deduce the time variation of source parameters .
these , in turn , have yielded insights into the nature and consequences of energy release in a dense plasma .
the paper is organized as follows : in 2 we present the radio , euv , and x - ray observations of the flare . in
3 we argue that the data may be understood in terms of fast electrons injected into a relatively cool , dense magnetic loop .
we explicitly fit a time series of radio spectroscopic data to a simple model that embodies these assumptions using a nonlinear @xmath0 minimization scheme . the parametric trends to emerge
show that the time series of spectra is well fit if the temperature of the ambient plasma increases in time .
we discuss implications of the model fits for plasma heating , electron acceleration , and electron transport in 4 .
we suggest that the lack of significant x - ray emission can be understood if the magnetic mirror ratio is large , thereby suppressing electron precipitation and reducing conduction at the foot points .
the relatively high plasma density in the source renders it collisionally thick to electrons with energies of some 10s of kev .
we argue that collisional heating yields the systematic plasma heating inferred from model fitting .
furthermore , electrons are accelerated progressively to energies exceeding mev energies over over a time scale @xmath6 s. we suggest that stochastic processes may play a significant role in the acceleration and transport of electrons .
we conclude in 5 .
the flare on 2001 october 24 occurred in noaa active region 9672 at a heliocentric position of s18w13 from approximately 23:10 - 23:14 ut .
ar 9672 was in a @xmath7 magnetic configuration and had produced m and x class soft x - ray flares in previous days . while the relatively strong radio event from ar9672 was noted by the noaa space environment center , it was incorrectly associated with an h@xmath8 subflare that occurred nearly concurrently , from 23:07 - 23:12 ut , in ar 9678 ( n06e33 ) .
no h@xmath8 observations of ar 9672 are available at the precise time of the event .
while no soft x - ray emission was noted by noaa / sec from either the h@xmath8 subflare in ar9678 or the radio burst in ar9672 , weak extreme - ultraviolet ( euv ) , soft x - ray ( sxr ) , and hard x - ray ( hxr ) emissions were in fact detected from the flare in ar9672 by the _ transition region and coronal explorer _ ( trace ; handy et al .
1999 ) , the _ yohkoh _ soft x - ray telescope ( sxt ) , and hard x - ray telescope ( hxt ) , respectively ( kosugi et al . 1992 ) , as was a weak sxr enhancement by the goes 10 satellite .
the radio burst was well - observed in total intensity by the owens valley solar array ( ovsa ; gary & hurford 1994 ) , the nobeyama radio polarimeters ( norp ) , and by the nobeyama radioheliograph ( norh ; nakajima et al .
1994 ) .
1a provides an overview of selected x - ray and radio observations .
the hxr and radio light curves show a simple , highly impulsive rise and exponential decay over a duration of approximately 4 min whereas the goes sxr bands decay to half their peak flux on a time scale @xmath6 min . the horizontal line at the top of fig .
1a indicates the time range shown in fig .
, the hxr light curve is shown with the norp frequencies .
we now discuss the euv , x - ray , and radio observations in greater detail .
to place the 2001 october 24 event in the broader context of events jointly observed in the radio and hxr bands , fig .
2 shows a scatter plot of the 135 flares observed jointly by the norp at 17 ghz and the hxt l band during the lifetime of the _ yohkoh _ mission .
analogous figures have compared the 17 ghz peak flux with the hxr counts from the hard x - ray spectrometer on board _ hinotori
_ ( kai et al . 1985 ) and from the hxrbs experiment on board the _ solar maximum mission _ ( kosugi et al . 1988 ) .
as in these previous examples , there is a clear correlation between the peak 17 ghz flux and the peak hxt l band count rate .
the spearman ( rank ) correlation coefficient is 0.57 , and the two - sided significance of the correlation i.e. , the probability that the correlation coefficient is larger than this value in the null hypothesis
is essentially zero , indicating the correlation is highly significant .
the flare of 2001 october 24 has a peak flux of 425 sfu w m@xmath9 hz@xmath10 at 17 ghz .
the regression curve implies that an hxt l band count rate of more than 200 cts s@xmath11 sc@xmath11 is expected , yet a peak count rate of only 7 cts s@xmath11 sc@xmath11 is measured ( including a background of @xmath12 ct s@xmath11 sc@xmath11 ) .
the flare is x - ray poor by a factor of @xmath13 relative to the regression curve .
we use the term x - ray poor " rather than microwave rich " because , as noted in the introduction , the latter term is associated with gradual events showing excess microwave emission ( bai & dennis 1985 , cliver et al . 1986 , kosugi et al . 1988 ) .
while the oct 24 event is very x - ray poor it was nevertheless detected by several instruments operating at euv and x - ray wavelengths .
it was imaged by trace in the 171 band with a cadence of @xmath14 s and an angular resolution of @xmath15 .
an image was obtained just prior to the impulsive rise at 23:10:46 ut and another was obtained at 23:11:26 ut ( fig 1b ) . a difference image formed from the latter image and one obtained at 23:10:05 ut reveals discrete kernels of emission ( fig .
3a ) . in soft x - rays
, the goes 10 background level was approximately c2.3 and the enhancement above background was approximately b2 .
the x - ray emission was too weak to trigger flare mode observing by the sxt .
as a result , only one low - resolution ( 4.92 " ) full - disk sxr image ( almg filter ) is available during this brief impulsive flare , at 23:11:46 ut ( fig 1b , fig .
the _ yohkoh _ hxt typically observed flares in four photon energy bands
l ( 13.9 - 22.7 kev ) , m1 ( 22.7 - 32.7 kev ) , m2 ( 32.7 - 52.7 kev ) , and h ( 52.7 - 92.8 kev ) . in this case , however , the hxt flare mode was not triggered and counts were therefore recorded in l band only . the l band light curve ( fig .
1 ) shows a simple impulsive spike with a maximum background - subtracted count rate of only 6 cts s@xmath11 sc@xmath11 and a fwhm duration of just 18 s. integrating over the full duration of the event yields sufficient counts to image the hxr source with an angular resolution of @xmath16 ( fig .
the _ yohkoh _ wide - band spectrometer ( wbs ; yoshimori et al .
1992 ) may have made a marginal detection in energy channels up to 50 kev ( h. hudson , private communication ) but given uncertainties in the calibration the implications for the hxr photon spectrum are highly uncertain .
we therefore do not consider wbs data here .
each of the maps shown in fig . 3 is compared with a magnetogram .
the flare is associated with a magnetic bipole just south of the dominant sunspot group in ar 9672 .
the length of the source is @xmath17 in sxr and its width is @xmath18 .
the sxr and hxr sources are coincident ( cf .
veronig & brown 2004 ) and are consistent with a magnetic loop or unresolved loops , the latter suggested by the discrete kernels seen in the trace 171 emissions , presumed here to be footpoint emission .
radio imaging observations of the flare were obtained by the norh at 17 and 34 ghz .
the norh observations were made early in the day , local time , and the angular resolution was therefore reduced in one dimension , yielding an elliptical beam of approximately @xmath19 at 17 ghz , and @xmath20 at 34 ghz , both with a position angle @xmath21 .
4a and 4c show the 17 and 34 ghz maps in total intensity ( stokes i ) overlaid on the magnetogram .
the source morphology evolves very little at radio wavelengths as a function of time ; we therefore show maps near the time of maximum intensity .
the peak brightness temperature of the 17 ghz source is @xmath22 k whereas that of the 34 ghz source is @xmath23 k. fig .
4b shows the 17 ghz map of the circularly polarized flux ( stokes v parameter ) superposed on the magnetogram , demonstrating that it is polarized in the sense of the x - mode with right - circularly polarized emission associated with positive magnetic polarity .
the degree of circular polarization
( i / v ) ranges from 20% left - circularly polarized ( lcp ; negative values of stokes
v ) to 10% right - circularly polarized ( rcp ; positive values of stokes v ) . the observed change in the sense of circular polarization along the length of the loop from rcp to lcp reflects the reversal of magnetic polarity while the asymmetry in the degree of rcp / lcp polarization indicates an asymmetry in the magnitude of the positive and negative magnetic polarity .
the sources observed in the sxr , hxr , and radio bands are coincident and quite similar in morphology .
all imaging data are consistent with the illumination of a simple magnetic loop , or unresolved bundle of loops , straddling a magnetic neutral line .
we note that the loop is not visible in the norh 17 and 34 ghz maps as a discrete feature in the active region prior to the flare .
we estimate the 17 ghz brightness temperature prior to the flare is @xmath24 k. the norp measures the total solar flux in stokes i and v at frequencies @xmath25 , 2 , 3.75 , 9.4 , 17 , 35 , and 80 ghz .
referring to the norp observations summarized in fig . 1b , we note several features .
first , the onset of the 35 ghz emission is delayed by @xmath26 s relative to that at 17 ghz which is , in turn , delayed by @xmath26 s relative to the onset of the hxt l band emission .
second , the time of the 17 ghz flux maximum is delayed by @xmath26 s relative to that at 35 ghz and the maximum of the 9.4 ghz light curve is delayed by 12 s relative to that at 17 ghz .
note , however , that a shoulder " in the 9.4 ghz emission roughly coincides with the 17 and 35 ghz maxima , a point discussed further in 4 .
third , the decay profiles of the 9.4 , 17 , and 35 ghz light curves are nearly identical and are exponential after approximately 23:11:20 ut with a decay time of @xmath27 s. fourth , there is no detectable radio emission at 3.75 ghz .
finally , while the flare was detected at 80 ghz the signal is very noisy ( 4 - 5@xmath28 ) .
we note that both the sensitivity and the calibration of the 80 ghz flux on this date are rather poor , the absolute flux calibration being no better than @xmath29 ( h. nakajima , private communication ) .
these features are confirmed and extended by the ovsa observations . on 2001
october 24 , ovsa employed a high cadence observing mode to produce 24 point spectra between 1.2 - 14.8 ghz every 2 s in total intensity .
a detail of the impulsive onset of the flare is shown in fig .
5a which plots the hxt l band data , three ovsa frequencies ( 11.2 , 12.4 , and 14.8 ghz , labeled with a prefix o " ) , and three norp frequencies ( 9.4 , 17 , and 35 ghz , labeled with a prefix n " ) for a period of 20 s. the frequencies 12.4 , 14.8 , 17 , and 35 ghz show a progressive delay in their flux onset relative to the hxt l band emission .
the onset of emission at 9.4 and 11.2 ghz show the opposite pattern , however .
these trends are also clearly seen in the dynamic spectrum of the ovsa data ( fig .
6 ) . for frequencies
@xmath30 ghz the flux onset is increasingly delayed with frequency whereas the opposite trend is seen for @xmath31 ghz .
5b shows a detail of the decay phase for the same selection of frequencies shown in fig .
each frequency has been scaled so that for times later than 23:11:20 ut the light curves are overlaid , showing that the exponential decay from 9.4 to 35 ghz is essentially identical after 23:11:20 ut . fig .
5b also illustrates the progressive delay of the flux maximum with decreasing frequency .
this feature is more clearly illustrated in fig .
6 , in which it is readily seen ( dashed line ) that the time of the flux maximum increases with decreasing frequency .
there is no detectable emission below @xmath32 ghz .
the radio spectral evolution of the flare is summarized in fig . 7 , which shows composite spectra using ovsa data ( plus symbols ) and norp data ( diamonds ) during the impulsive rise and exponential decay .
the spectral maximum is near 17 ghz during the rise phase but declines to @xmath33 ghz during the decay phase .
given the rather extensive data set it is useful to consolidate and summarize our findings regarding the flare on 2001 october 24 : * the flare is one of the most x - ray poor events of those jointly observed by the norp and _ yohkoh _ hxt . * the norh 17 and 34 ghz images , and the _ yohkoh _ sxt and hxt
l band images are all coincident and similar in morphology .
the source is consistent with a simple magnetic loop or unresolved bundle of loops .
the trace 171 difference image reveals small kernels consistent with loop foot points .
* due to the paucity of x - rays , the photon spectrum was not measured and the electron spectrum is therefore unconstrained by hxr emission .
* the observed radio emission is relatively strong and the spectral coverage is excellent , allowing the following observations to be made : * * radio emission is detected up to 80 ghz .
the spectral maximum is @xmath34 ghz during the impulsive rise , declining to @xmath33 ghz during its decay .
radio emission is undetectable below @xmath32 ghz . * * the impulsive onset of emission shows a progressive delay with _ increasing _
frequency for @xmath35 ghz ; the trend reverses below @xmath36 ghz . * * in contrast , the time of absolute flux maximum shows a progressive delay with _ decreasing _
frequency for all frequencies measured . * * for times later than approximately 23:11:20 ut the decay profile of the radio emission is independent of frequency over two octaves and is characterized by a decay time of @xmath37 s. despite the relatively simple source morphology in the x - ray and radio bands , and the single impulsive spike shown in the hxr and radio light curves , this event displays considerable spectral richness at radio wavelengths . in the next section we develop a source model and perform parametric fits to the time series of radio spectra .
the parametric trends to emerge from these fits then guide our physical interpretation of the event in 4 .
we now direct our attention to the well - observed radio source and fit a simple model to the radio data as a function of time .
a fully self - consistent , three dimensional , time - dependent , inhomogeneous model of the source is beyond the scope of this work .
we instead account for the essential features of the source in terms of a magnetized volume containing thermal plasma and energetic electrons .
similar models have been developed to account for the spectral ( e.g. , klein 1987 ; benka & holman 1992 ; bruggmann et al .
1994 ) , temporal ( e.g. , bruggmann et al . 1994 ) , or spatial ( bastian et al . 1998 ; nindos et al . 2000 ; kundu et al .
2004 ) properties of radio observations . typically , these efforts have been strongly under - constrained in the temporal , spatial , and/or spectral domains . moreover
, model fitting has typically been limited to adjusting model parameters until a reasonable fit by eye is obtained . in the present case , we have comparatively good spectral and temporal coverage of the event at radio wavelengths . while imaging data are limited , the source is apparently rather simple .
it is therefore possible to make simplifying assumptions regarding the source volume and to exploit nonlinear optimization techniques to fit the observed radio spectrum .
moreover , such fits can be performed on the time series of spectra , from which the time variation of fitted parameters can be deduced . as such , the fitting represents a means of time - resolved radio - spectral inversion .
while the model is admittedly crude , it nevertheless allows us to identify trends in the time variation of critical parameters in an unbiased fashion . in other words , rather than postulating a time dependence for physical parameters of interest , the time dependence emerges from the model fits . in this section we describe the model and the underlying assumptions , explore the parametric dependencies of the model , and then present the fitting scheme and the results derived from the fits in detail .
we assume the observed radio emission is the result of gyrosynchrotron radiation from energetic electrons in a magnetic loop .
a striking feature of the radio spectra is the relatively steep roll - off in the emission below @xmath38 ghz , with no emission detectable at all below @xmath39 ghz , a feature that persists for the duration of the event .
the spectrum of the radio emission is clearly nonthermal , yet the brightness temperature of the 17 and 34 ghz sources is unexceptional , characteristic of optically thin emission .
we therefore suggest that the razin effect ( razin 1960 , ginzburg & syrovatskii 1969 ) plays a role in this flare .
the razin effect suppresses gyrosynchrotron emission in the presence of an ambient plasma below a cutoff frequency @xmath40 , where @xmath41 is the thermal electron density and @xmath42 is the source magnetic field . for @xmath43 ghz and and an ambient density @xmath44 @xmath4 ,
@xmath45 g. for a given cutoff frequency @xmath46 , higher magnetic field strengths require higher thermal electron densities .
the thermal free - free absorption coefficient at radio wavelengths is @xmath47 for coronal conditions ( dulk 1985 ) .
the optical depth @xmath48 through a plasma layer @xmath49 is @xmath50 . substituting plausible values for the relevant parameters
@xmath51 k , @xmath52 cm , and @xmath53 @xmath4 we find that @xmath54 , where @xmath55 is the frequency in ghz .
the optical depth is even larger for cooler plasma .
it is clear that if razin suppression is important , free - free absorption is , too .
consequently , we assume that razin suppression and free - free absorption each play a role in determining the shape of the radio spectrum and include both in our model calculations ( cf .
ramaty & petrosian 1972 , klein 1987 ) .
as previously noted , the source morphology changes little during the event .
we therefore consider a homogeneous source with an area @xmath56 @xmath57 ( @xmath58 ) and a depth @xmath59 cm ( 12 " ) , consistent with the x - ray and radio imaging ( see 2.4 ) .
the source volume is assumed to contain thermal plasma with a density @xmath41 and a temperature @xmath60 , permeated by a coronal magnetic field @xmath42 with an angle @xmath61 relative to the line of sight .
finally , it is assumed that an isotropic , power - law distribution of energetic electrons @xmath62 is uniformly mixed with the thermal plasma , with a normalization energy @xmath63 and a high - energy cutoff energy @xmath64 .
the total number density of nonthermal electrons between @xmath63 and @xmath64 is @xmath65 .
no electron transport effects are explicitly included in the model calculation .
the computation of radio emission from even as simple a source model as this depends on many parameters . to make further progress , additional assumptions are necessary .
first , we assume the magnetic field remains constant for the duration of the event .
however , as a means of estimating the dependence of the fitted parameters on the magnetic field , different values of the magnetic field were assumed for different fitting runs .
second we assume the number density of the background thermal plasma also remains constant . in fact , as discussed further in 3.3 , preliminary fits allowed @xmath41 to vary but the fitted values showed little variation with time .
hence , it was typically held fixed .
an additional justification for this second assumption is that , given the weak euv and hxr emission , there was little deposition of energy near the loop footpoints and hence , little evaporation of plasma into the coronal loop .
the incremental increase in density due to evaporation is therefore presumed to be small .
we do allow the temperature of the thermal background plasma to vary with time , however .
finally , we note that the radio emission is insensitive to @xmath63 and therefore set it equal to 100 kev .
to summarize , we model the radio source as a simple homogenous volume with an area @xmath66 and a depth @xmath49 .
the volume is filled with a thermal plasma with a number density @xmath41 and is permeated by a magnetic field @xmath42 oriented at an angle @xmath61 to the line of sight ; @xmath41 , @xmath42 , and @xmath61 are assumed to be constant during the flare but the plasma temperature is allowed to vary . a power - law distribution of @xmath65 energetic electrons per cubic centimeter between energies @xmath67 and @xmath64 is uniformly mixed with the background plasma .
the electron pitch angle distribution is assumed to be isotropic .
typically two or three out of four parameters
@xmath65 , @xmath60 , @xmath64 , and @xmath68 were allowed to vary in fitting the data to the model .
it is useful to make a brief aside at this point to outline the computation of the radio emission spectrum and to elucidate the dependencies of the computed spectrum on various parameters of interest . for the homogeneous source described above
, the source flux at a given frequency is computed as ( ramaty & petrosian 1972 ; benka & holman 1992 ) @xmath69\ ] ] where @xmath70 is the index of refraction , @xmath71 is the total emissivity due to the free - free and gyrosynchrotron mechanisms , @xmath72 is the total absorption coefficient , @xmath73 is the solid angle subtended by the source , and @xmath74 is the distance from the observer to the source ( 1 au ) .
the thermal free - free emission and absorption coefficients used for the computation of the source function are readily calculated ( e.g. , lang 1980 ) ; those for gyrosynchrotron emission and absorption are more problematic .
the expressions for the emissivity and absorption coefficient given by ramaty ( 1969 ) , or variants thereof ( e.g. , benka & holman 1992 ) , are cumbersome and their calculation is computationally intensive . for the iterative @xmath0 minimization scheme we employ , it is far more efficient to use the approximate expressions derived by klein ( 1987 ) which include the effects of razin suppression and are quite accurate for radio frequencies greater than the first few harmonics of the electron gyrofrequency ( @xmath75 ) and for values of @xmath61 that are not too small ( @xmath76 ) .
8 shows model spectra calculated for various sets of parameters .
all four panels show a reference spectrum ( solid line ) due to gyrosynchrotron emission from electrons _ in vacuo _
( @xmath77 ) .
the parameters used for the reference spectrum are : @xmath78 g , @xmath79 , @xmath80 @xmath4 , @xmath81 kev , @xmath82 mev , @xmath83 , @xmath56 @xmath84 , and @xmath85 cm . in fig .
8a the dashed lines indicate spectra where the source parameters are identical to those used to compute the reference spectrum except that an ambient thermal plasma is present in the source .
the ambient plasma has a temperature of @xmath86 k and its density varies from @xmath87 @xmath4 to @xmath88 @xmath4 .
the combined action of free - free absorption and razin suppression strongly reduces the radio emission below 10 - 20 ghz for values of @xmath89 @xmath4 .
moreover , as the density increases , the high - frequency emission is enhanced .
this enhancement is entirely due to the thermal free - free contribution . in fig .
8b , the temperature is again @xmath51 k and the ambient density is held fixed at @xmath53 @xmath4 . here , the magnetic field strength is allowed to vary from @xmath78 to 350 g. all other source parameters are identical to those used to compute the reference spectrum .
note that , in contrast to gyrosynchrotron emission in which razin suppression plays no role , the spectral maximum is insensitive to @xmath42 although the peak flux increases with @xmath42 . in fig .
8c , the density is again held fixed at @xmath53 @xmath4 , @xmath78 g , but the plasma temperature now varies from @xmath90 to @xmath91 k. all other parameters are again the same as those used to compute the reference spectrum . for cool plasma , the combination of razin suppression and thermal absorption strongly reduces the microwave emission . as the temperature of the ambient plasma increases , all other parameters held fixed , the microwave emission increases at lower frequencies because the free - free absorption drops as the temperature increases , but decreases at the highest frequencies owing to the corresponding decrease of the optically thin free - free opacity , and the spectral maximum moves toward lower frequencies . at high enough temperatures
the plasma becomes optically thin to free - free absorption ( @xmath92 k in this example ) and the spectrum becomes insensitive to further increases in temperature except the extreme low - frequency end of the of the spectrum where the emission is due to optically thick thermal radiation of the plasma and is proportional to its temperature . finally , in fig .
8d , the thermal plasma density is held fixed ( @xmath53 @xmath4 ) , @xmath51 k , @xmath78 g , and the number of energetic electrons @xmath65 between @xmath63 and @xmath64 , varies from @xmath93 to @xmath91 @xmath4 .
all other parameters are the same as those used to compute the reference spectrum .
the microwave flux varies with @xmath65 but the spectral maximum is insensitive to variations in @xmath65 .
this point has also been made by belkora et al .
( 1997 ) , fleishman & melnikov ( 2003 ) , and melnikov et al .
( 2007 ) . the optically thin slope steepens with increasing @xmath65 as the influence of the dense thermal plasma becomes relatively less important . at low values of @xmath65
the contribution of free - free emission is significant and acts to flatten the spectrum .
while the source model is relatively simple , it yields a wide variety of spectral shapes which , in turn , yield insight to the source parameters .
we now turn to fitting this simple source model to the observed data .
the excellent radio spectroscopic and temporal coverage of this event allow us to fit the observed radio spectra as a function of time .
lee & gary ( 2000 ) used a similar approach to fit an electron transport model to ovsa data although a different `` goodness of fit '' parameter was optimized . here , we have developed a nonlinear code that adjusts model parameters to minimize the @xmath0 statistic using the downhill simplex method ( press et al .
the @xmath0 statistic is computed as @xmath94 where @xmath95 is the flux density at frequency @xmath96 observed by either ovsa or the norp , @xmath97 is the corresponding model value , @xmath98 is the estimated variance of the observed value , @xmath99 is the number of data points , and @xmath100 is the number of free parameters .
the signal variance @xmath98 at each frequency is empirically known prior to the flare .
it is the result of receiver noise , the sky ( 35 and 80 ghz ) , and the sun itself .
radio emission from the flare makes an additional contribution that can be estimated from the known intensity of the signals and the properties of the instrument .
we have fit the model to composite radio spectra observed by ovsa ( 5 - 14.8 ghz ) and the norp ( 17 , 35 , and 80 ghz ) .
data points below 5 ghz are ignored because they contain no signal .
inspection of the spectra indicates a probable systematic calibration error between the two instruments with a magnitude of @xmath101 ( fig .
the time resolution of the norp data points were averaged to 2 s to match the time resolution of the ovsa data .
a total of 25 spectra were fit to data sampled between 23:10:54 and 23:11:44 ut , as indicated in fig .
the spectra at earlier and later times had insufficient signal - to - noise ratios at too many points to yield reliable fits .
while the spectral coverage of this event at radio frequencies is excellent , it is nevertheless insufficient to allow unique sets of parameters to be established from fits to the data .
excluding frequencies below 5 ghz , we fit to only sixteen data points for each time average .
indeed , as explained in 3.1 , many of the source parameters are held constant and only two or three out of four parameters were varied .
clearly , the models resulting from this procedure , while providing good fits to the data , are non - unique .
we set those parameters that remain constant for all fits to values we judge to be reasonable , but are fully aware that other , similar , values may in fact yield fits that are consistent with the data .
we therefore consider the fits and the trends that emerge from the time series of fitted parameters to be qualitative guides in our physical interpretation of the flare in 4 .
9 shows the results of two- and three - parameter fits to the 25 spectra observed from 23:10:54 to 23:11:44 ut . in the two - parameter fit ( solid line ) , only the number of fast electrons , @xmath65 , and the plasma temperature , @xmath60 , were allowed to vary and the spectral index and cutoff energy were held fixed , with @xmath102 and @xmath82 mev , respectively , as were all other relevant parameters : @xmath103 g , @xmath79 , @xmath81 kev , @xmath56 @xmath84 , and @xmath85 cm .
the value for the spectral index is entirely consistent with previous findings ( e.g. , ramaty et al .
1994 , trottet et al .
1998 , 2000 ; lthi et al . 2004 ) .
three parameter fits are also shown , the third parameter being either @xmath68 ( dashed line ) or @xmath64 ( dotted line ) .
as discussed in 3.1 , little improvement was seen in the ( three - parameter ) fits when the thermal plasma density , @xmath41 was free to vary .
the fitted value of @xmath41 remained nearly constant ( @xmath104 @xmath4 ) in all cases and was therefore held fixed at this value for the fits shown in fig .
when @xmath68 was allowed to vary it , too , remained relatively constant or in some cases showed a tendency to smaller values ; i.e. , spectral hardening . when @xmath64 was allowed to vary , best fit values during the rise phase of the flare yielded @xmath105 mev whereas for times later than approximately 23:11:12 ut , larger values of @xmath64 resulted ( @xmath106 ) mev .
we note that the fits become insensitive to the precise value of @xmath64 when it exceeds @xmath33 mev because the high frequency radio data most sensitive to @xmath64 are sparse and , in the case of 80 ghz , poorly calibrated .
10 shows a comparison between the fitted values for @xmath65 and @xmath60 resulting from the two - parameter fits and the radio and hxr emission as a function of time .
we note that the three - parameter fits yield similar results for the time variation of @xmath65 and @xmath60 .
the shaded area in each panel indicates the range of fitted values that result from varying the magnetic field strength .
the time variation of @xmath65 and @xmath60 are shown for three fixed values of the magnetic field ( 150 , 165 , and 180 g ) .
two points can be made about the apparent trends : first , the total number of energetic electrons tracks the variation of the high frequency ( 35 ghz ) radio emission for the interval that was fit quite well , as expected for optically thin emission .
second , the temperature of the ambient plasma increases with time . with @xmath103 g , @xmath60 increases from @xmath107 k to @xmath108 k during the time range considered .
both @xmath60 and @xmath65 depend on @xmath42 : the fitted value of @xmath65 decreases and that of @xmath60 increases with increasing @xmath42 . referring to fig .
8c , we note that the fits are relatively insensitive to plasma temperature once @xmath60 approaches @xmath109 k ( fig .
10b ) because the optical depth to free - free absorption becomes negligible .
we find no evidence that the observed spectral evolution requires multiple discrete injections of electrons characterized by different energy distribution functions , as suggested by , e.g. , trottet et al .
( 2002 ) , for certain events .
we checked the fitted temperature and the emission measure implied by the assumed density ( i.e. , @xmath110 ) against the goes sxr observations for consistency ( white et al .
a peak temperature near @xmath109 k is consistent with the goes 0.5 - 4 and 1 - 8 data but the density inferred from the goes data is perhaps a factor of two smaller than that assumed here . given that the goes event is very small and the uncertainties in the sxr background are relatively large , and given that the parametric fits depend on a number of assumptions , we conclude that the consistency between the fits and the sxr data is rather good .
the standard model " of flares involves energy deposition in the chromosphere by a flux of nonthermal electrons precipitating from one or more magnetic loops , resulting in heating and evaporation " of chromospheric plasma into the corona where it emits sxr radiation .
we suggest that a difference between this flare and the standard model " is that significant chromospheric evaporation did not occur .
as noted in the previous section , the spectral fits do not require an increase in the thermal plasma density during the course of the flare .
furthermore , there is a paucity of euv and sxr emission from hot thermal plasma , suggesting little increase in emission measure of coronal plasma and hence , little increase in the thermal plasma density .
why is there so little chromospheric evaporation ?
one reason may be that the flaring loop has a large ( albeit asymmetric ) magnetic mirror ratio , defined as @xmath111 , the ratio of the magnetic field strength near the magnetic footpoint to that at the loop top where the magnetic field is weakest , and @xmath112 is the loss - cone angle .
a large value of @xmath113 implies a small value for @xmath112 , little precipitation of energetic electrons from the loop and hence , little euv and hxr footpoint emission .
for example , a mirror ratio @xmath114 implies only @xmath115 of an isotropic particle distribution would precipitate from the magnetic loop if magnetic trapping and weak scattering were the only relevant transport processes ( e.g. , melrose & brown 1976 ) .
a second factor is that the coronal loop , with @xmath53 @xmath4 , is collisionally thick to electrons with energies @xmath116 kev ( veronig & brown 2004 ) .
such electrons would deposit the bulk of their energy in the ambient plasma rather than at the loop footpoints .
one might suppose that if the precipitation of energetic electrons into the footpoints is limited , thermal conduction might nevertheless drive evaporation . however , if the mirror ratio is large the footpoint area is small and evaporation due to conduction can also be strongly reduced .
these considerations are consistent with the observed morphology of the hxt l band source shown in fig .
the idea of a large mirror ratio stands in contradiction to our simple model , for which a magnetically homogeneous source was assumed .
this may not be a concern , however , if the magnetic loop is strongly convergent near the chromosphere ( dowdy et al . 1985 ) .
the time interval during which plasma heating is observed via spectral fitting is from 23:10:50 to 23:11:20 ut , so that @xmath117 s. during this time , the data suggest the temperature of the ambient thermal plasma in the flaring source increased from 2 - 3 mk to 8 - 10 mk .
we note that the source was presumably already heated from @xmath118 mk to 2 - 3 mk prior to the time range for which fits were computed . whether @xmath60 continued to increase significantly beyond 23:11:20 ut is not established because the model fitting becomes insensitive to temperature when @xmath119 k although the goes sxr emission suggest that it did not . since both the radiative and conductive loss times
are long compared to the duration of the flare , the net change of thermal energy density during the observed temperature increase is @xmath120 ergs @xmath4 , yielding a mean energy deposition rate of @xmath121 ergs s@xmath11 @xmath4 . the magnetic energy density @xmath122 erg @xmath4 .
hence , the energy deposition rate required to heat the plasma amounts to @xmath123 of @xmath124 each second and a total energy deposition of @xmath125 ergs . what heats the plasma ?
one possibility , to which we have already alluded , is energy loss via coulomb collisions . by virtue of its high density
the coronal loop is collisionally thick to electrons with energies @xmath126 kev .
unfortunately , no hxr observations of photon energies greater than the hxt l band are available because the flare did not trigger the _ yohkoh _ flare mode .
the spectral index and low - energy cutoff of the distribution of hxr - emitting electrons are unknown and hence , the number of electrons with energies of 10s of kev , can not be inferred from the hxr observations . it may even be possible that thermal bremsstrahlung contributes to the hxt l band counts .
however , if we suppose the mean energy of the hxr - emitting electrons is @xmath127 kev , then @xmath128 such electrons are needed each second to account for the plasma heating via collisions .
we now consider the radio - emitting electrons .
the peak number density of energetic electrons with energies @xmath129 kev was found to be @xmath130 @xmath4 .
note that while @xmath65 is normalized to @xmath81 kev , for @xmath103 g the bulk of the emission at radio frequencies between @xmath131 ghz is in fact contributed by electrons with energies of @xmath132 mev , a result that has been previously noted ( ramaty et al .
1994 , trottet et al .
1998 , 2000 ; lthi et al 2004 ) .
the energy contained in these electrons is only @xmath133 ergs @xmath4 , an insignificant fraction of that required to heat the ambient plasma .
we conclude that the radio - emitting electrons themselves play no role in heating the plasma ; but what if the power - law distribution of radio - emitting electrons extends to lower energies ?
the collisional energy deposition rate is @xmath134 with @xmath135 where the approximate expression for the lifetime @xmath136 applies to mildly relativistic electrons ( melrose & brown 1976 ) . here
, @xmath137 is the lorentz factor , @xmath138 , @xmath139 is the electron energy in units of 100 kev , and @xmath140 is the thermal plasma density in units of @xmath141 @xmath4 . taking @xmath102 ( 3 )
we find @xmath142 kev yields an energy deposition rate of 10 ergs @xmath4 s@xmath11 , in agreement with the mean energy deposition rate , and the total number of electrons @xmath143 kev is found to be @xmath144 @xmath4 s@xmath11 .
the inferred electron energy and number density are similar to those needed to account for plasma heating by coulomb collisions .
we conclude that the population of electrons responsible for the hxt l band emission and plasma heating can be regarded as part of the same power - law distribution of electrons responsible for the radio emission .
while this picture is appealing , it is incomplete .
thus far we have based our discussion on the results of the two - parameter ( @xmath145 ) fits to the time series of radio spectroscopic data for which the electron energy distribution is assumed to be a power - law described by fixed values of @xmath67 , @xmath64 , and @xmath68 . while there are hints in the three - parameter fits of possible evolution of @xmath68 and/or @xmath64 , such variations did not yield significantly better fits during the time interval considered . in
2.3 , however , we considered the impulsive onset of the event and found that onset time progressively increased with frequency for @xmath146 ghz ( fig .
5a ) . the mean energy of the emitting electrons @xmath147 ( dulk 1985 ) ;
hence , higher radio frequencies are emitted by higher energy electrons for a fixed magnetic field .
the progressive delay in the onset of impulsive emission for @xmath35 ghz suggests that the more energetic the electrons , the later they appeared in the source .
the onset times of emission for @xmath148 ghz may at first seem inconsistent with the trend seen at higher frequencies .
however , they can be understood in terms of the overall pattern of strong free - free absorption diminishing with time as the plasma temperature increases .
consider 9.4 ghz : at the time of flux onset the 9.4 ghz flux is strongly reduced by free - free absorption and razin suppression . as the plasma heats the free - free opacity declines and the gyrosynchrotron emission increases .
the triangle at 23:11:02 ut in fig .
1b , and the dotted line in fig . 7 , indicates the time of the 35 ghz flux maximum . in fig .
1b it clearly coincides with the `` shoulder '' seen in the 9.4 ghz light curve , a feature that roughly coincides with the maximum of @xmath65 .
it is not until the plasma heats still further that the free - free opacity becomes negligible at 9.4 ghz and the flux maximum is achieved , well after the maximum of @xmath65 .
therefore , despite the complicating factor of free - free absorption , we conclude that the timing delays seen in the onset of the radio emission relative to the hxt l band , and from lower to higher frequencies , are consistent with progressive acceleration of electrons with time during the impulsive rise . the net time elapsed from the acceleration of electrons with energies of 10s of kev to those with energies of 100s of kev is @xmath149 s , and from 100s of kev to mev energies is @xmath150 s ( fig .
after the initial energization to mev energies , there does not appear to be any further significant evolution of the form electron energy spectrum , as determined by spectral fitting .
we return to this point below . finally , we consider the observed frequency independence of the decay of the radio emission of this event , a feature shared with the flare described by white et al .
this feature is clearly at odds with flare transport models in which the weak diffusion regime and magnetic trapping play a significant role ( e.g. , bruggmann et al .
1994 ; bastian et al . 1998
; lee et al .
2000 ; kundu et al .
if weak diffusion and trapping were relevant , higher energy electrons would be lost from the distribution more slowly than lower energy electrons because the collision frequency @xmath151 and the energy loss rate decrease with energy according to equation ( [ edot ] ) .
consequently , higher frequency radio emission would decay more slowly than lower frequency emission ( see kundu et al .
2001 for a vivid example ) .
yet , the radio decay profiles for the oct 24 event are essentially identical ( fig .
5b ) ; they are exponential with a decay time of 48 s for all frequencies with detectable emission for times later than approximately 23:11:20 ut .
we conclude that while coulomb collisions play a dominant role in the energy loss of hxr - emitting electrons ( 10s of kev ) , they do not appear to play a role in the transport of the mev radio - emitting electrons . instead
, wave - particle interactions mediate electron transport .
to summarize our findings , electron acceleration occurs during the impulsive rise phase of the flare , progressing from energies of 10s of kev to more than an mev in @xmath152 s. there is no compelling evidence for further significant evolution of the electron spectrum during the remainder of the flare based on model fitting .
the flare plasma is collisionally thick to electrons with energies of some 10s of kev ; these electrons heat the plasma .
the increase in plasma temperature leads to a frequency - dependent decrease in free - free opacity , thereby explaining the increasing delay of the radio flux maximum with decreasing frequency .
the decay phase of the flare at radio frequencies is incompatible with coulomb collisions ; we conclude that wave - particle interactions mediate electron transport . given that wave - particle interactions dominate electron transport at mev energies , we briefly consider the role of stochastic processes for electron acceleration .
possibilities for the relevant wave modes include whistler waves ( hamilton & petrosian 1992 , pryadko & petrosian 1997 , petrosian & liu 2004 ) , fast - mode mhd waves ( miller et al .
1996 ) , or others ( see toptygin 1985 , miller et al .
1997 , for reviews ) . to be more concrete ,
consider a turbulent spectrum of whistler waves with a total energy density @xmath153 .
hamilton & petrosian ( 1992 ) have considered stochastic heating and acceleration by whistler waves , including the effects of coulomb collisions . assuming the spectral energy density of whistler wave turbulence is isotropic and can be described by a power law dependence @xmath154 for wave numbers @xmath155 , the average pitch angle and momentum diffusion coefficients , @xmath156 and @xmath157 , are well known ( e.g. , melrose 1980 , eqns . 13 in hamilton & petrosian 1992 ) .
the electron mean free path is given by hamilton & petrosian as @xmath158 .
for a source of size @xmath49 , taken here to be the observed loop length of @xmath159 cm , the lifetime in the source @xmath160 is given by @xmath161 .
expressing the energy density in the turbulent spectrum relative to that in the magnetic field , @xmath162 , we find that for the model source parameters fit in 3 , @xmath163 and @xmath160 depend on values assumed for the turbulence index @xmath164 and @xmath153 . for stochastic acceleration by whistler waves to be relevant we require @xmath165 and @xmath166 s.
these requirements can be met when @xmath167 and @xmath168 . with @xmath169 and @xmath167 ,
@xmath163 is essentially independent of energy and @xmath160 depends only weakly on energy . we therefore conclude that if stochastic processes in a turbulent spectrum of whistler waves are relevant , then the spectrum must be relatively flat ( e.g. , a kolmogorov - like spectrum with @xmath170 ; cf . petrosian & liu 2004 ) . with the estimated turbulence level and spectrum ,
the characteristic acceleration time for mev electrons is very short , a fraction of a second .
this seems to contradict the progression of delays up to @xmath171 s observed between the onset of the hxr l band emission and the radio emission .
the apparent contradiction can perhaps be understood in terms of the evolution time of the turbulence spectrum .
but until the specific turbulent wave mode is firmly established , and the source of the turbulence identified , it is not possible to draw any firm conclusions in this regard .
the flare on 2001 oct 24 was well - observed at radio and x - ray wavelengths .
it was one of the most x - ray poor events jointly observed by the hxt and the norp .
fortuitously , comprehensive x - ray and radio observations of the event .
the spatial morphology of the source is consistent with a simple coronal magnetic loop .
the hxr and radio light curves revealed a progressive delay between the onset of hxt l band emission ( 13.9 - 22.7 kev ) and the onset of norp / ovsa emission , the delay increasing with frequency for @xmath35 ghz .
in contrast , the time of the maximum flux density was progressively delayed with decreasing frequency .
the radio emission decay was independent of frequency .
we have modeled the flare source in terms of an admixture of nonthermal electrons and a dense thermal plasma .
we suggest that the paucity of euv , sxr , and hxr emission as a result of a relatively large mirror ratio in the magnetic loop in which the nonthermal electrons are injected / accelerated and the high ambient plasma density , which rendered the source collisionally thick to electrons @xmath116 kev . the progressive delay in the impulsive onset of radio emission with increasing frequency ( @xmath35 ghz ) is consistent with a progressive energization of electrons from lower to higher energies during the impulsive rise ( @xmath152 s ) .
electrons with energies of 10s of kev deposit their energy in the coronal loop rather than at the foot points , heating the ambient plasma .
the systematic plasma heating with time causes the thermal free - free absorption to decrease with time , thereby explaining the progressive delay of the radio flux maximum with decreasing frequency .
the radio decay time is inconsistent with a trapping scenario .
we suggest that the transport of energetic electrons is instead determined by wave - particle interactions .
we suggest that the data are consistent with electron acceleration and transport mediated by a relatively flat ( kolmogorov ) spectrum of whistler waves although the specifics remain an outstanding question .
|
we present radio and x - ray observations of an impulsive solar flare that was moderately intense in microwaves , yet showed very meager euv and x - ray emission .
the flare occurred on 2001 oct 24 and was well - observed at radio wavelengths by the nobeyama radioheliograph ( norh ) , the nobeyama radio polarimeters ( norp ) , and by the owens valley solar array ( ovsa ) .
it was also observed in euv and x - ray wavelength bands by the trace , goes , and _ yohkoh _ satellites .
we find that the impulsive onset of the radio emission is progressively delayed with _ increasing _ frequency relative to the onset of hard x - ray emission .
in contrast , the time of flux density maximum is progressively delayed with _ decreasing _ frequency . the decay phase is independent of radio frequency .
the simple source morphology and the excellent spectral coverage at radio wavelengths allowed us to employ a nonlinear @xmath0-minimization scheme to fit the time series of radio spectra to a source model that accounts for the observed radio emission in terms of gyrosynchrotron radiation from mev - energy electrons in a relatively dense thermal plasma .
we discuss plasma heating and electron acceleration in view of the parametric trends implied by the model fitting .
we suggest that stochastic acceleration likely plays a role in accelerating the radio - emitting electrons .
|
everyday , billions of individuals generate a large volume of geolocated data by using their mobile phone , gps , public transport cards or credit cards . such a vast amount of data is bringing new opportunities for the research in socio - technical systems @xcite . indeed
, geolocated data allow the identification of when and where people interact with or through ict tools .
each time someone makes a phone call or pays with a credit card the event gets registered contributing to massive databases with potential to provide useful insights on human behavior and mobility @xcite .
for example , the authors of refs .
@xcite used credit card and mobile phone datasets to study statistical characteristics of mobility patterns and showed that the distribution of displacement of all users can be approximated by a levy law . recently
, geolocated data has been also employed to study the spatial structure of cities by detecting hotspots @xcite or to characterize land use patterns in urban areas @xcite with mobile phone records , twitter data @xcite or both together @xcite . on a larger scale ,
comparisons and relations between different cities @xcite or even between countries @xcite have also been also investigated . beyond mere location ,
some datasets offer the opportunity to gather extra information about the type and duration of the interaction or the operation through ict tools .
for instance , it is possible to know from mobile phone records where and when an individual makes a call , but sometimes information such as the i d of the callee and the call duration are also available .
this information enables researchers to move further on the study of human behavior by analyzing the structure , intensity and spatial properties of social interactions .
some examples include the analysis of the structure of social networks @xcite , the correlation between mobility and social network @xcite , information diffusion @xcite and the role played by social groups @xcite
. however , many previous studies lack sociodemographic resolution on the characteristics of the individuals . except for some features such as language or place of work and/or residence identified in @xcite , information about gender , age or occupation are typically missing from studies based on ict data .
this information is of great relevance to characerize the city structure , to estimate population needs in urban planning , transport demand and also for public health .
for example , regarding age , knowing the areas of concentration of younger and older population helps to optimize infrastructure such as location of schools , care facilities , etc .
another aspect for which this information is relevant is the modeling of infectious diseases spreading .
the models rely on the interplay among hosts , which is related to their location and mobility .
recent epidemic modeling has incorporated mobility information as a way to get closer to real disease spreading @xcite .
additionally , demographic factors such as age or gender can also play an important role in disease transmission and , therefore , must be taken into account when modeling certain infections @xcite . furthermore , in a sort of feedback loop , these sociodemographic factors influence mobility as well .
some works based on smaller - scale surveys point out towards a number of significant differences between men and women in terms of their travel purposes and the activities they pursue @xcite .
more recently , quantitative studies of social networks dynamics have also shown that people behave differently according to the gender and age @xcite . in this paper
, we go beyond by analyzing a credit card use database containing over @xmath0 million card transactions in order to explore consumption and mobility patterns of bank customers in the two most populated provinces of spain according to three sociodemographic characteristics : gender , age and occupation .
our dataset comes from an extraction of the banco bilbao vizcaya argentaria ( bbva ) database on credit card transactions .
different extractions of this data have been used in open data challenges @xcite and other scientific works @xcite .
the data contains information about @xmath0 million bank card transactions made in the provinces of madrid and barcelona in @xmath1 .
each transaction is characterized by its amount ( in euro currency ) and the time when the transaction has occurred .
each transaction is also linked to a customer and a business using anonymized customer and business ids .
customers are identified with an anonymized customer i d connected with sociodemographic characteristics ( gender , age and occupation ) and the postcode of his / her place of residence . for convenience sake , we consider five age groups ( @xmath215,30]$ ] , @xmath230,45]$ ] , @xmath245,60]$ ] , @xmath260,75]$ ] , @xmath3 ) and five types of occupations ( student , unemployed , employed , homemaker , and retired ) . in the same way ,
businesses are identified with an anonymized business i d , a business category ( accommodation , automotive industry , bars and restaurants , etc . ) and the geographical coordinates of the credit card terminal .
the geographical extent of our data is restricted to the provinces of barcelona and madrid . for both case studies ,
we only consider the credit card payments made in the province by individuals living in the province ( figure [ fig1 ] ) .
table [ tab1 ] presents some basic statistics on the data collected .
both provinces have similar features in terms of population size , area and number of businesses , but the number of users and transactions are higher in madrid than in barcelona .
the number of users represents about @xmath4 of the total census population in madrid and @xmath5 of that of barcelona .
.summary statistics of the two provinces [ cols="<,^,^",options="header " , ]
the statistical features of the data for barcelona and madrid are very similar .
therefore , the data is aggregated for analyzing general properties in the next two sections and segregated later in the third one to study mobility patterns .
the aggregation provides higher statistical power , while the disaggregation is needed due to the different geographical shapes of both provinces . due to the optimization of space ,
only figures obtained for madrid are displayed in the third section on mobility .
still equivalent results for barcelona are found and can be seen in appendix ( figures s9 - s15 ) . in order to have a first look at the data , we plot in figure [ fig2 ] some descriptive statistics about individuals according to their sociodemographic characteristics .
figure [ fig2 ] shows the proportion of individuals according to gender , age and occupation in the dataset and the corresponding fractions as observed in the census @xcite .
we note an over - representation of men and middle - aged individuals ( 30 - 60 ) in the dataset compared to census data .
moreover , employed people represent about @xmath6 of the individuals , which is two times higher than the proportion of employed people in spain .
therefore , since the data are not representative of the population , in the rest of the manuscript only indicators and measures normalized by the total number of individuals in each groups will be considered .
it is also important to note that the three distributions are not independent , for example , the proportion of individuals according to the age is not the same for student and retired individuals . in the same way
, the proportion of individuals according to the occupation is different for men and women .
for example , there are more female homemakers than male homemakers . for more details , histograms of the three joint distributions
are available in appendix ( figure s1 , s2 , and s3 ) . to highlight differences between individuals having different sociodemographic characteristics , we also plot on figure [ fig2 ] the median number of transactions per user , the median amount of money spent per user and the median average amount of money spent per transaction per user .
we used the median instead of the average because the distributions exhibits a large number of outliers ( see figure s4 , s5 and s6 in appendix for more details ) .
it can be observed that individuals do not spend their money in the same way according to whether they are men or women , young or old and active or inactive .
for instance , the number of transactions and the amount of money spent is higher for women than for men and decreases with age
. furthermore , they are also higher for employed persons and homemakers than for unemployed individuals , students and retired people ( which is probably related to the age ) .
inversely , the average amount of money spent per transaction is higher for men than women and increases with age . to investigate the influence of sociodemographics on the way people spend their money , we plot on figure [ fig3 ] the average fraction of money spent by an individual according to the business category and his / her sociodemographic characteristics .
since the total amount of money spent in 2011 is different from one individual to another , the distribution has been normalized for each user by the total amount of money he / she spent during the year .
note that the distribution is very different for men and women .
indeed , women spend more money than men in fashion , food / hypermarkets , health and wellness / beauty whereas men spend more money than women in automotive industry , bar / restaurants , technology and transport .
we also find that the proportion of money spent in fashion , food / hypermarkets , sports / toys , technology and transport globally decreases with age .
inversely , the amount of money spent in automotive industry , health , travel agencies and wellness / beauty increases with age . finally , the differences between people having different occupation are explored .
for instance , students spend more money in bar / restaurant , fashion , sports / toys and technology than others types of occupation .
since the proportion of individuals according to the occupation is different for men and women , and in order to take away potential bias , we have studied the average fraction of money spent by an individual according to the business category and his / her sociodemographic characteristics but only for employed individuals .
we reach the same conclusions as for the overall sample , see figure s7 in appendix .
to study how the amount of money spent by bbva customers changes over time during an average week , the days of the week have been divided into four groups : one , from monday to thursday representing a normal working day ( hereafter called @xmath7 ) and three more for friday , saturday and sunday ( hereafter called @xmath8 , @xmath9 and @xmath10 ) .
the average amount of money spent per day as a function of the hour of the day is displayed in figure [ fig4]a ( gray curve ) .
globally , the amount of money spent is significantly higher during the week days , friday and saturday than on sunday .
this can be explained by the fact that most of the business were closed on sunday in spain in the time that the data was collected .
the activity on sunday takes place between @xmath11 and @xmath12 with a small peak around @xmath13 . during the week days , friday and saturday money
is spent between @xmath14 and @xmath15 . for these days
the curves show two peaks , one around noon and another one around @xmath12 .
it is interesting to note that for the week days and friday the second peak is higher than the first one whereas the opposite behavior is observed on saturday .
a small peak around @xmath16 corresponding to the nightlife activity is also observed for the three first days . to go further in the analysis , a k - means clustering algorithm with euclidean distance @xcite
is applied in order to identify clusters naturally present in the data .
the purpose is to cluster together individuals exhibiting temporal distribution of money spent .
the total amount of money spent in 2011 is different from one individual to another so we have normalized the temporal distribution of money spent for each user by the total amount of money he / she spent in 2011 . to choose the number of clusters , we use the pseudo - f statistics which describes the ratio of between - cluster variance to within cluster variance @xcite .
the optimal number of clusters is the one for which the highest pseudo - f value is obtained , in our case we found two opposite clusters ( see figure s8 in appendix for more details ) .
figure [ fig4]a displays the results of the clustering analysis , we observe an opposition between active and inactive individuals .
the first cluster represents one third of the individuals and is characterized by a higher activity during the morning and during weekdays in opposition with the second cluster in which individuals tend to spend more money after @xmath17 and during week end days .
it is interesting to note that the first cluster is over - represented by women , old people and homemaker and retired individuals compared to the whole population ( figure [ fig4]b ) . in order to characterize mobility patterns of each user ,
we have considered three variables : @xmath18 , the time elapsed between two consecutive transactions , @xmath19 , the distance traveled between two consecutive transactions , and @xmath20 , the radius of gyration @xcite .
the radius of gyration is defined as @xmath21 where @xmath22 represents the @xmath23 position of the user displacements in 2011 and @xmath24 is the center of mass of his / her motions .
it is important to note that @xmath20 is defined per user whereas @xmath18 and @xmath19 are computed for each displacement .
although @xmath19 and @xmath20 are related , @xmath19 informs us on the distance traveled by users , which might depend on the frequency at which each person uses its credit card , whereas @xmath20 gives us a more holistic view of how people moves around their centers of mass . to avoid the introduction of bias in the mobility patterns analysis , all the consecutive user s positions geo - located in the province and the distances between them are considered whatever the elapsed time between consecutive transactions .
figures [ fig5]a , [ fig6]a and [ fig7]a display the probability density function of the three variables .
the distribution of @xmath18 is a decreasing density function exhibiting circadian rhythms .
the average and median time between two transaction are , respectively , around @xmath25 days and @xmath26 days .
the distribution of @xmath19 show two different regimes .
first the distribution exhibits a slow decay , and then , beyond @xmath0 kilometers the distribution is characterized by a rapid decay .
this cutoff is introduced by the limited geographical scale of the provinces . the probability density function @xmath27 increases very slowly until reaching a maximum around @xmath28 kilometers and
then the distribution is characterized by a rapid decay . in this work
we have also assessed the influence of sociodemographic characteristics on the individual mobility patterns .
the results obtained are plotted on the figure [ fig5 ] , [ fig6 ] and [ fig7 ] .
for each sociodemographic characteristic and each variable , we performed two non - parametric tests to assess the statistical significance of the differences between the different type of users mobility using the mann whitney u test @xcite to compare the distributions and the mood s median test @xcite to compare the medians . for both case
studies the differences between distributions and medians are always significant ( _ p - values _ lower than @xmath29 ) except for the difference between radius of gyration of individuals of age between 15 and 30 and those between 30 and 45 in barcelona .
figure [ fig5 ] displays the inter - event time distribution according to the gender ( figure [ fig5]b ) , the age ( figure [ fig5]c ) and the occupation ( figure [ fig5]d ) .
the average and median inter - event time are higher for men than women and increases with age .
they are also higher for unemployed individuals , students and retired people than for employed persons and homemakers .
we observe an negative correlation between the time elapsed between two consecutive transactions and the number of transactions per individual described in the first section .
the results obtained for @xmath19 and @xmath20 are plotted in figure [ fig6 ] and [ fig7 ] , respectively .
based on these results , one can understand that , depending on his / her sociodemographic characteristics , an individual can travel short or long distances and stays more or less close to his / her center of mass .
three main differences are observed .
first , women travel shorter distances than men and their trajectory stays closer to their center of mass .
second , the average distance traveled between two consecutive positions and the radius of gyration decrease with age .
finally , an opposition between active and inactive individual is highlighted .
indeed , retired , homemaker and , to a lesser extent , unemployed individuals travel shorter distances and stay closer to the center of mass than other people .
as previously mentioned , the distance traveled by an individual between two consecutive transactions might depend on the frequency at which an individual uses his / her credit card , and therefore , the differences between people observed for @xmath19 could be a consequence of the differences observed for @xmath18 .
although the same conclusion are reached for the radius of gyration , which does not depend on the frequency at which someone uses his / her credit card , it could be interesting to study how the average value of @xmath19 evolves as a function of @xmath18 according to the individual s sociodemographic characteristics .
we can observe in figure [ fig8 ] that the differences between the different types of individuals in terms of distances traveled always exist whatever the time elapsed between two consecutive transactions .
it is also worth noting that the value of @xmath30 is not completely independent of @xmath18 . obviously , for small values of @xmath18 ( @xmath31 ) the value of @xmath30 increases with the value of @xmath18 due to physical constraints
but we can also note a valley for @xmath3218,30]$ ] followed by a peak for @xmath3230,42]$ ] .
this phenomenon seems to be more pronounced for active people than for inactive people , possibly reflecting the home - to - work / school commuting . among all these comparisons ,
discrepancy in mobility between men and women is the most challenging . in order to verify that this difference is significant and it is not related to other sociodemographic variables , the kolmogorov - smirnov ( ks ) distance between men and women s @xmath18 , @xmath19 and @xmath20 distributions
are computed ( figure [ fig9 ] ) .
the kolmogorov - smirnov ( ks ) distance between two empirical probability distributions @xmath33 and @xmath34 is defined as @xmath35 where @xmath36 and @xmath37 are the empirical cumulative distribution function of @xmath33 and @xmath34 respectively . since
, the sample size of both distributions may vary from one sociodemographic variable to another we need to normalize @xmath38 according to the sample sizes , @xmath39 where @xmath40 and @xmath41 represent the sample sizes of @xmath33 and @xmath34 , respectively .
this allows for a direct comparisons of the kolmogorov - smirnov distances .
moreover , using this normalization , the null hypothesis that the two data samples come from the same distribution is rejected at level @xmath42 if @xmath43 .
first , we observe that a significant difference between men and women appears whatever the sociodemographic characteristic of the population is filtered out ( i.e. @xmath44 is always higher than 1.95 ) , which means that on average , women have an inter - event time lower than men and men do longer journeys than women .
second , one can observe that this gendered difference is more important for middle age individual than for young and old people , but also that is more pronounced for employed people . to go further ,
we have studied too the influence of the individual s sociodemographic characteristics and the business category on the distance traveled between home and business .
to do so , we computed for each transaction the distance between the individual s place of residence and the business .
as residence location , we use the centroid of the individual s postcode of residence .
finally , these distances were averaged according to individual and business type .
these average distances can be observed in figure [ fig10 ] .
first , we observe that the same differences between type of individuals as the ones highlighted previously are obtained whatever the business category .
for each business category , the distance between home and business is globally higher for men than women , it decreases with age and it is higher for employed and student than for the other occupation categories . although , the average distance between home and business changes according to the category of business . indeed , distances between home and businesses belonging to the categories food / hypermarkets , health , wellness / beauty and book / cd / stationery are lower than for the other categories .
it is interesting to note that these business category are also the type of business in which the number of transactions is higher for women than for men ( figure [ fig11 ] ) .
this partially explains why women travel shorter distance than men to go shopping .
in summary , we have shown in this study that it is possible to use information provided by credit card data to assess the influence of sociodemographic characteristics on the way people move and spend their money .
we highlighted differences in consumption habits and mobility patterns of bank customers according to their gender , age and occupation .
first , we shown that according to the business type the fraction of money spent can be very different from one individual to another . in particular , women tend to spend more money in fashion , food / hypermarkets , health and wellness / beauty than men whereas men spend more money than women in automotive industry , bar / restaurants , technology and transport .
we have also studied the time evolution of the amount of money spent along the week according to the individual s sociodemographic characteristics .
an opposition between two types of individuals has been identified .
the temporal distribution of money spent by the first type of individuals which is over - represented by inactive people is characterized by a higher activity during the morning and during weekdays in opposition with the second type of individuals more active after working hours and during week end days .
then , we investigated the properties of people mobility patterns using three variables : the time elapsed between two consecutive transactions , the distance traveled by an individual between two consecutive transactions and the radius of gyration .
three main differences between groups of people were identified : differences between men and women , young and old people and active and inactive individuals . in the three cases , people of the first group ( men , young people and active people ) travel shorter distances and
their trajectory stays closer to their center of mass than individuals of the second groups ( women , old individual and inactive people ) . among all the differences emphasized in this paper the one between men and women is the most difficult to explain .
in all the comparisons we have carefully checked that this difference was not related to other sociodemographic variables and it was not the case .
it could be interesting to verify whether this difference is related to other social characteristics such as the number of children for example .
indeed , the fact that the difference in terms of mobility patterns between men and women is less pronounced for old people and students may reflect that women with children move differently than women without children . while further data is required to assess whether these differences between individuals are universal , i.e. , to which extend they are specific or not to urban areas or the cities of the country analyzed , our results point toward the possibility that mobility may display significant differences for different types of individuals .
partial financial support has been received from the spanish ministry of economy ( mineco ) and feder ( eu ) under projects modass ( fis2011 - 24785 ) and intense@cosyp ( fis2012 - 30634 ) , and from the eu commission through projects eunoia , lasagne and insight .
the work of ml has been funded under the pd/004/2013 project , from the conselleria de educacin , cultura y universidades of the government of the balearic islands and from the european social fund through the balearic islands esf operational program for 2013 - 2017 .
jjr acknowledges funding from the ramn y cajal program of mineco .
99 watts , d. j. .
_ nature _ * 445 * , 489 ( 2007 ) .
lazer , d. _ et al .
_ computational social science
. _ science _ * 323 * , 721723 ( 2009 ) .
vespignani , a. .
_ science _ * 325 * , 425428 ( 2009 ) .
chowell , g. , hyman , j. , eubank , s. , castillo - chavez , c. . _ phys rev e _ * 68 * , 066102 ( 2003 ) .
barrat , a. , barthlemy , m. , pastor - satorras , r. , vespignani , a. .
* 101 * , 37473752 ( 2004 ) .
brockmann , d. , hufnagel , l. & geisel , t. the scaling laws of human travel .
_ nature _ * 439 * , 462465 ( 2006 ) .
gonzalez , m. c. , hidalgo , c. a. & barabasi , a .-
_ nature _ * 453 * , 779782 ( 2008 ) .
song , c. , qu , z. , blumm , n. & barabsi , a .-
_ science _ * 327 * , 10181021 ( 2010 ) .
bagrow , j. p. & lin , y .- r .
mesoscopic structure and social aspects of human mobility .
_ plos one _ * 7 * , e37676 ( 2012 ) .
louail , t. _ et al . _ from mobile phone data to the spatial structure of cities . _ sci .
rep . _ * 4 * , 5276 ( 2014 ) .
ratti , c. , pulselli , r. m. , williams , s. & frenchman , d. mobile landscapes : using location data from cell phones for urban analysis .
_ environment and planning b : planning and design _ * 33 * , 727748 ( 2006 ) .
reades , j. , calabrese , f. , sevtsuk , a. & ratti , c. cellular census : explorations in urban data collection .
_ pervasive computing , ieee _ * 6 * , 3038 ( 2007 )
. soto , v. & fras - martnez , e. automated land use identification using cell - phone records . in _ proceedings of the 3rd acm international workshop on mobiarch _ , hotplanet 11 , 1722 ( acm , new york , ny , usa , 2011 ) .
pei , t. , sobolevsky , s. , ratti , c. , shaw , s. l. & zhou , c. a new insight into land use classification based on aggregated mobile phone data .
_ arxiv e - print _ * arxiv:1310.6129 * ( 2013 ) .
toole , j. , ulm , m. , gonzlez , m. & bauer , d. inferring land use from mobile phone activity .
paper presented at : _ the acm sigkdd international workshop on urban computing , beijing_. place of publication : _ proceedings of the acm sigkdd international workshop on urban computing , doi:10.1145/2346496.2346498 _ ( 2012 ) .
fras - martnez , v. , soto , v. , hohwald , h. & fras - martnez , e. characterizing urban landscapes using geolocated tweets .
paper presented at : _
socialcom - passat 12 , privacy , security , 2012 international conference on risk and trust ( passat ) and 2012 international confernece on social computing ( socialcom ) , amsterdam_. place of publication : _ proceedings of the privacy , security , 2012 international conference on risk and trust ( passat ) and 2012 international confernece on social computing ( socialcom ) , doi:10.1109/socialcom - passat.2012.19 _ ( 2012 ) .
lenormand , m. _ et al .
_ cross - checking different source of mobility information .
_ plos one _ * 9 * , e105184 ( 2014 ) .
noulas , a. , scellato , s. , lambiotte , r. , pontil , m. & mascolo , c. .
_ plos one _ * 7 * , e37027 ( 2012 ) .
hawelka , b. _ et al .
_ geo - located twitter as a proxy for global mobility patterns . _ cartography and geographic information science _ * 41 * , 260271 ( 2013 ) .
lenormand , m. , tugores , a. , colet , p. & ramasco , j. j. tweets on the road . _
plos one _ * 9 * , e105407 ( 2014 ) .
liben - nowell , d. , novak , j. , kumar , r. , raghavan , p. & tomkins , a. .
usa _ * 102 * , 1162311628 ( 2005 ) .
onnela , j. _ et al . _ structure and tie strengths in mobile communication networks .
usa _ * 104 * , 73327336 ( 2007 ) .
java , a. , song , x. , finin , t. & tseng , b. why we twitter : understanding microblogging usage and communities . in
_ proceedings of the 9th webkdd and 1st sna - kdd 2007 workshop on web mining and social network analysis _ , 5665 ( acm , 2007 ) .
huberman , b. a. , romero , d. m. & wu , f. social networks that matter : twitter under the microscope .
_ first monday _ * 14 * , 12 ( 2008 ) .
eagle , n. , pentland , a. s. & lazer , d. .
usa _ * 106 * , 1527415278 ( 2009 ) .
ferrara , e. .
_ epj data science _ * 1 * , 9 ( 2012 ) . grabowicz , p. a. , ramasco , j. j. , goncalves , b. & eguiluz , v. m. . _
plos one _ * 9 * , e92196 ( 2014 ) .
backstrom , l. , sun , e. & marlow , c. find me if you can : improving geographical prediction with social and spatial proximity . in _ proceedings of the 19th international conference on world wide web _
, 6170 ( acm , 2010 ) .
calabrese , f. , smoreda , z. , blondel , v. d. & ratti , c. .
_ plos one _ * 6 * , e20814 ( 2011 ) .
phithakkitnukoon , s. , smoreda , z. & olivier , p. socio - geography of human mobility : a study using longitudinal mobile phone data . _
plos one _ * 7 * , e39253 ( 2012 ) .
ferrara , e. , varol , o. , menczer , f. & flammini , a. in _ proc .
1st acm conf . on online social networks ( cosn ) _ , 213222 ( 2013 ) .
grabowicz , p. a. , ramasco , j. j. , moro , e. , pujol , j. m. & eguiluz , v. m. .
_ plos one _ * 7 * , e29358 ( 2012 ) .
mocanu , d. _ et al . _ the twitter of babel : mapping world languages through microblogging platforms . _ plos one _ * 8 * , e61981 ( 2013 ) .
rvachev , l.a . &
longini , i.m . a mathematical model for the global spread of influenza .
, 322 ( 1985 ) .
grais , r.f . ,
hugh ellis , j. , & glass , g.e . assessing the impact of airline travel on the geographic spread of pandemic influenza . , 10651072 ( 2003 ) .
eubank , s. _ et al .
_ modelling disease outbreaks in realistic urban social networks . , 180184 ( 2004 ) .
hufnagel , l. , brockmann , d. & geisel , t. forecast and control of epidemics in a globalized world .
, 1512415129 ( 2004 ) .
longini , i.m .
_ et al .
_ containing pandemic influenza at the source , , 10831087 ( 2005 ) .
ferguson , n.m .
_ strategies for containing an emerging influenza pandemic in southeast asia . , 209214 ( 2005 ) .
riley , s. large - scale spatial - transmission models of infectious disease , , 12981301 ( 2007 ) .
colizza , v. , barrat , a. , barthlemy .
m. , valleron , a.j .
& vespignani , a. modeling the worldwide spread of pandemic influenza : baseline case and containment interventions .
, e13 ( 2007 ) .
ciofi degli atti , m.l .
_ et al . _ mitigation measures for pandemic influenza in italy : an individual based model considering different scenarios , , e1790 ( 2008 ) .
balcan , d. _ et al .
usa _ * 106 * , 2148421489 ( 2009 ) .
bajardi , p. _ et al . _ . _
plos one _ * 6 * , e16591 ( 2011 ) .
meloni , s._et al . _ . _ sci .
_ * 1 * , 62 ( 2011 )
. tizzoni , m. _ et al . _
real - time numerical forecast of global epidemic spreading : case study of 2009 a / h1n1pdm .
_ bmc medicine _ * 10 * , 165 ( 2012 ) .
poletto , c. , tizzoni , m. & colizza , v. heterogeneous length of stay of hosts movements and spatial epidemic spread . _ sci .
rep . _ * 2 * , 476 ( 2012 ) .
wallinga , j. , teunis , p. & kretzschmar , m. .
j. epidemiol . _
* 164 * , 936944 ( 2006 ) .
brauer , f. .
_ * 70 * , 18691885 ( 2008 ) .
nishiura , h. .
_ j. trav .
_ * 17 * , 269270 ( 2010 ) .
nishiura , h. , cook , a.r . & cowling , b.j . .
_ interdisciplinary perspectives on infectious diseases _ * 2011 * , 194507 ( 2011 ) .
rocha , l. , liljeros , f. & holme , p. .
_ plos comput .
biol . _ * 7 * , 1001109 ( 2011 ) .
apolloni , a. , poletto , c. & colizza , v. . _ bmc infectious diseases _ * 13 * , 118 ( 2013 ) apolloni , a. , poletto , c. , ramasco , j.j . , jensen , p. & colizza , v. .
_ theoretical biology and medical modelling _ * 11 * , 3 ( 2014 ) .
golob , t. f. & mcnally , m. g. a model of activity participation and travel interactions between household heads .
transportation research part b : methodological _ * 31 * , 177 194 ( 1997 ) . hamed , m. m. & mannering , f. l. modeling travelers postwork activity involvement : toward a new methodology .
_ transportation science _ * 27 * , 381394 ( 1993 ) .
bianco , m. & lawson , c. trip chaining , childcare and personal safety : critical issues in women s travel behavior . in _
proceedings from the second national conference on women s travel issues .
washington dc : us department of transportation , federal highway administration _ ( 1996 ) .
mcpherson , m. , lovin , l. s. & cook , j. m. .
_ annual review of sociology _ * 27 * , 415444 ( 2001 ) . stehle , j. , charbonnier , f. , picard , t. , cattuto , c. & barrat , a. gender homophily from spatial behavior in a primary school : a sociometric study .
_ social networks _ * 35 * , 604 613 ( 2013 ) .
innova challenge http://www.centrodeinnovacionbbva.com/en/innovachallenge/what-innova-challenge .
date of access 03/12/2014 .
sobolevsky , s. _ et al .
_ mining urban performance : scale - independent classification of cities based on individual economic transactions . in _ proceedings of ase
bigdatascience 2014 conference _ ( 2014 ) .
spanish census 2011 ( instituto nacional de estadstica ) : http://www.ine.es/censos2011/tablas/inicio.do .
date of access 12/03/2015 .
hartigan , j. a. & wong , m. a. a k - means clustering algorithm .
_ applied statistics _ * 28 * , 100108 ( 1979 ) .
caliski , t. & harabasz , j. .
_ communications in statistics - simulation and computation _ * 3 * , 127 ( 1974 ) .
mann , h. b. & whitney , d. r. . _ the annals of mathematical statistics _ * 18 * , 5060 ( 1947 ) .
brown , g. w. & mood , a. m. on median tests for linear hypotheses . in _ proceedings of the second berkeley symposium on mathematical statistics and probability _ , 159166 ( university of california press , berkeley , calif . , 1951 ) .
|
human mobility has been traditionally studied using surveys that deliver snapshots of population displacement patterns .
the growing accessibility to ict information from portable digital media has recently opened the possibility of exploring human behavior at high spatio - temporal resolutions .
mobile phone records , geolocated tweets , check - ins from foursquare or geotagged photos , have contributed to this purpose at different scales , from cities to countries , in different world areas .
many previous works lacked , however , details on the individuals attributes such as age or gender . in this work ,
we analyze credit - card records from barcelona and madrid and by examining the geolocated credit - card transactions of individuals living in the two provinces , we find that the mobility patterns vary according to gender , age and occupation .
differences in distance traveled and travel purpose are observed between younger and older people , but , curiously , either between males and females of similar age . while mobility displays some generic features , here we show that sociodemographic characteristics play a relevant role and must be taken into account for mobility and epidemiological modelization .
|
in the minimal standard model of electroweak interactions , the lepton number conservation is assumed and neutrinos are exactly massless weyl particles . however , today s super - kamiokande @xcite , sno @xcite , kamland @xcite and k2k @xcite neutrino oscillation experiments have provided us with very strong evidence that neutrinos are actually massive and lepton flavor mixing does exist .
the most economical modification of the minimal standard model , which can both accommodate neutrino masses and allow lepton number violation to explain the cosmological baryon asymmetry via leptogenesis @xcite , is to introduce two heavy right - handed neutrinos @xmath2 and keep the lagrangian of electroweak interactions invariant under the @xmath3 gauge transformation @xcite .
after the spontaneous electroweak symmetry breaking , this simple but interesting model yields the following neutrino mass term : @xmath4 where @xmath5 with @xmath6 being the charge - conjugation operator ; and @xmath7 denote the left - handed neutrinos . the dirac neutrino mass matrix @xmath8 is a @xmath9 rectangular matrix , and the majorana neutrino mass matrix @xmath10 is a @xmath11 symmetric matrix .
the scale of @xmath8 is characterized by the electroweak scale @xmath12 gev . in contrast
, the scale of @xmath10 can be much higher than @xmath13 , because @xmath14 and @xmath15 are @xmath16 singlets and their corresponding mass term is not subject to the scale of gauge symmetry breaking .
then one may obtain the effective ( light and left - handed ) neutrino mass matrix @xmath17 via the well - known seesaw mechanism @xcite : @xmath18 without loss of generality , both @xmath10 and the charged lepton mass matrix @xmath19 can be taken to be diagonal , real and positive ; i.e. , @xmath20 and @xmath21 , where @xmath22 denote the masses of two heavy majorana neutrinos .
in such a specific flavor basis , the low - energy neutrino phenomenology is governed by @xmath17 , while the cosmological baryon number asymmetry is associated with @xmath8 via the leptogenesis mechanism .
unfortunately , the model itself has no restriction on the structure of @xmath8 . in ref .
@xcite , frampton , glashow and yanagida ( fgy ) have conjectured that @xmath8 may take the form @xmath23 or @xmath24 the texture zeros in @xmath8 could stem from an underlying horizontal flavor symmetry . with the help of eq .
( 2 ) , one may straightforwardly arrive at @xmath25 from eq .
( 3 ) ; or @xmath26 from eq .
note that @xmath27 holds in either case .
is independent of the specific texture zeros taken in eq .
( 3 ) or ( 4 ) . in other words
, the minimal seesaw model itself guarantees that @xmath27 holds automatically . ] note also that @xmath28 holds in the chosen flavor basis , where @xmath29 ( for @xmath30 ) denote the masses of three light neutrinos .
thus one of three neutrino masses must vanish .
as the solar neutrino oscillation data have set @xmath31 @xcite , we are left with either @xmath32 ( normal hierarchy ) or @xmath33 ( inverted hierarchy ) . in ref .
@xcite , the majorana neutrino mass matrix with one texture zero and one vanishing eigenvalue has been classified and discussed in some detail . the main purpose of this paper is to reveal a very striking feature of the minimal seesaw model in which the fgy ansatz is incorporated : all nontrivial cp - violating phases can be calculated in terms of the lepton flavor mixing angles @xmath34 and the ratio of @xmath0 to @xmath1 , where @xmath0 and @xmath1 stand respectively for the typical mass - sqaured differences of solar and atmospheric neutrino oscillations .
this important point , which was not observed in the previous analyses of the minimal seesaw model @xcite , implies that a stringent test of the fgy ansatz can simply be realized once the smallest mixing angle @xmath35 is measured or constrained to a better degree of accuracy . considering both normal and inverted mass hierarchies of three light neutrinos
, we obtain very specific predictions for the cosmological baryon number asymmetry , the effective mass of neutrinoless double beta decay and cp violation in neutrino oscillations .
without loss of generality , one may always redefine the phases of charged lepton fields to make @xmath36 , @xmath37 and @xmath38 of @xmath8 real and positive @xcite .
in other words , only @xmath39 is complex and its phase @xmath40 is the sole source of cp violation in the model under discussion . because both @xmath19 and @xmath10 have been taken to be diagonal , real and positive , @xmath17 can in general be parametrized as follows : @xmath41 where @xmath42 and @xmath43 are two phase matrices , and @xmath44 is given by @xmath45 with @xmath46 , @xmath47 and so on .
the phase parameter of @xmath44 ( dirac phase ) governs the strength of cp violation in neutrino oscillations , while two independent phase parameters of @xmath48 ( majorana phases ) are relevant to the neutrinoless double beta decay @xcite .
the phases of @xmath49 can not be neglected in the parametrization of @xmath17 their essential role is to fulfil a complete match between the phases of @xmath17 in eq .
( 5 ) or ( 6 ) and those defined in eqs . ( 7 ) and
( 8) . it is then obvious that the nontrivial phases of @xmath49 , @xmath44 and @xmath48 should have definite relations with @xmath50 .
note that three mixing angles of @xmath44 can directly be given in terms of the mixing angles of solar , atmospheric and reactor @xcite neutrino oscillations .
namely , @xmath51 , @xmath52 and @xmath53 hold as a good approximation . in view of current experimental data , we have @xmath54 and @xmath55 ( best - fit values
@xcite ) as well as @xmath56 .
the mass - squared differences of solar and atmospheric neutrino oscillations are defined respectively as @xmath57 and @xmath58 .
their best - fit values read @xmath59 and @xmath60 @xcite .
these typical numbers will be used in our numerical calculations . to be more specific
, we shall concentrate on the fgy ansatz for @xmath8 in eq .
( 3 ) or equivalently @xmath17 in eq . ( 5 ) .
some brief comments will be given on the consequences of @xmath8 in eq .
( 4 ) or equivalently @xmath17 in eq . ( 6 ) . indeed
, both possibilities lead to very similar phenomenological results .
if @xmath61 holds , we obtain @xmath62 ev and @xmath63 ev . in this case , only a single majorana phase of cp violation is physically nontrivial .
hence the phase matrix @xmath48 can be simplied to @xmath64 and six independent matrix elements of @xmath17 can be written as @xmath65 \ ; , \nonumber \\ ( m_\nu)_{22 } & = & e^{2i\beta } \left [ m_2 \left ( -s_x s_y s_z + c_x c_y e^{-i\delta } \right ) ^2 e^{2i\sigma } + m_3 s^2_y c^2_z \right ] \ ; , \nonumber \\ ( m_\nu)_{33 } & = & e^{2i\gamma } \left [ m_2 \left ( s_x c_y s_z + c_x s_y e^{-i\delta } \right ) ^2 e^{2i\sigma } + m_3 c^2_y c^2_z \right ] \ ; ; % ( 9)\end{aligned}\ ] ] and @xmath66 \ ; , \nonumber \\ ( m_\nu)_{13 } & = & e^{i(\alpha+\gamma ) } \left [ -m_2 s_x c_z \left ( s_x c_y s_z + c_x s_y e^{-i\delta } \right ) e^{2i\sigma } + m_3 c_y s_z c_z \right ] \ ; , \nonumber \\ ( m_\nu)_{23 } & = & e^{i(\beta+\gamma ) } \left [ -m_2 \left ( s_x c_y s_z + c_x s_y e^{-i\delta } \right ) \left ( -s_x s_y s_z + c_x c_y e^{-i\delta } \right ) e^{2i\sigma } + m_3 s_y c_y c^2_z \right ] \ ; . % ( 10)\end{aligned}\ ] ] because of @xmath67 as shown in eq . ( 5 ) , we straightforwardly obtain @xmath68 \ ; , \nonumber \\ \sigma & = & \frac{1}{2 } \arctan \left [ \frac{c_x s_y \sin\delta } { s_x c_y s_z + c_x s_y \cos\delta } \right ] \ ; , % ( 11)\end{aligned}\ ] ] where @xmath69 .
this result implies that both @xmath70 and @xmath71 can definitely be determined , if and only if the smallest mixing angle @xmath35 is measured . to establish the relationship between @xmath50 and @xmath70
, we need to figure out @xmath72 , @xmath73 and @xmath74 .
because @xmath36 , @xmath37 and @xmath38 are real and positive , @xmath75 , @xmath76 and @xmath77 must be real and positive .
then @xmath72 , @xmath73 and @xmath74 can be derived from eqs .
( 10 ) and ( 11 ) : @xmath78 \ ; , \nonumber \\ \beta & = & -\gamma - \arctan \left [ \frac{c_x c_y s_z \sin\delta } { s_x s_y - c_x c_y s_z \cos\delta } \right ] \ ; , \nonumber \\ \gamma & = & + \frac{1}{2 } \arctan \left [ \frac{s^2_z \sin 2\sigma}{\xi s^2_x c^2_z + s^2_z \cos 2\sigma } \right ] \ ; .
% ( 12)\end{aligned}\ ] ] then the overall phase of @xmath79 , which is equal to the phase of @xmath39 , is given by @xmath80 \ ; .
% ( 13)\ ] ] now that all six phase parameters ( @xmath70 , @xmath71 , @xmath50 , @xmath72 , @xmath73 and @xmath74 ) have been determined in terms of @xmath81 , @xmath82 , @xmath83 and @xmath35 , a measurement of the unknown angle @xmath35 becomes crucial to test the model .
the phase parameters @xmath72 , @xmath73 and @xmath74 can be fixed by taking account of the positiveness of @xmath75 , @xmath76 and @xmath77 . with the help of eqs .
( 14 ) and ( 15 ) , we obtain @xmath91 \ ; , \nonumber \\ \beta & = & \gamma + \pi \ ; , \nonumber \\ \gamma & = & \frac{\delta}{2 } + \frac{1}{2 } \arctan \left [ \frac{s_x s_y \sin \delta}{s_x s_y \cos \delta - c_x c_y s_z } \right ] - \pi \ ; .
% ( 17)\end{aligned}\ ] ] then the overall phase of @xmath79 , which is equal to the phase of @xmath39 , is given by @xmath92 - \pi \ ; .
% ( 18)\ ] ] again , a measurement of the unknown mixing angle @xmath35 will allow us to determine all six phase parameters ( @xmath70 , @xmath71 , @xmath50 , @xmath72 , @xmath73 and @xmath74 ) . using the best - fit values of @xmath0 ,
@xmath1 , @xmath82 and @xmath83 , we illustrate the numerical dependence of six phase parameters @xmath94 on the smallest mixing angle @xmath35 in fig .
1(a ) and fig .
1(b ) for the @xmath61 case and in fig . 2(a ) and fig .
2(b ) for the @xmath33 case .
some discussions are in order .
\(1 ) in the @xmath32 case , @xmath35 is restricted to a very narrow range @xmath95 ( namely , @xmath96 ) .
this result implies that the fgy ansatz with @xmath32 is highly sensitive to @xmath35 and can easily be ruled out if the experimental value of @xmath35 does not really lie in the predicted region . to a good degree of accuracy ,
we obtain @xmath97 , @xmath98 , @xmath99 , and @xmath100 .
these instructive relations can essentially be observed from eqs . ( 11 ) , ( 12 ) and ( 13 ) , because of @xmath101 .
note that we have only shown the dependence of @xmath70 on @xmath35 in the range @xmath102 .
the reason is simply that only this range can lead to @xmath103 ( i.e. , the positive cosmological baryon number asymmetry ) , as one can see later on .
\(2 ) in the @xmath104 case , there is no strong constraint on @xmath35 except that @xmath105 ( or equivalently @xmath106 ) must hold .
we see that @xmath107 , @xmath108 , @xmath109 and @xmath110 hold to a good degree of accuracy .
these results can also be observed from eqs .
( 16 ) , ( 17 ) and ( 18 ) by taking account of @xmath101 .
again , we have used the positive sign of @xmath111 to constrain the allowed range of @xmath70 .
the other phase parameters are required to take possible values between @xmath112 and @xmath113 . as useful by - products , the jarlskog parameter of cp violation ( @xmath114 @xcite ) and the effective mass of neutrinoless double beta decay ( @xmath115 @xcite ) can be calculated .
we show the numerical results of @xmath114 versus @xmath115 in fig .
1(c ) for @xmath32 and in fig .
2(c ) for @xmath33 , respectively .
it is clear that @xmath116 and @xmath117 mev in the @xmath32 case , while @xmath118 and @xmath119 ev in the @xmath104 case .
the present experimental upper bound of @xmath120 is @xmath121 ev at the @xmath122 confidence level @xcite .
because of lepton number violation , two heavy majorana neutrinos @xmath123 ( for @xmath124 and 2 ) may decay into @xmath125 and its cp - conjugate state , where @xmath126 denotes the left - handed lepton doublet and @xmath127 stands for the higgs - boson weak isodoublet .
the decay occurs at both the tree level and the one - loop level ( via self - energy and vertex corrections ) , and their interference leads to a cp - violating asymmetry @xmath128 between the @xmath129-conjugated @xmath130 and @xmath131 processes @xcite .
if the masses of @xmath14 and @xmath15 are hierarchical ( i.e. , @xmath132 ) , the interactions of @xmath14 can be in thermal equilibrium when @xmath15 decays .
the asymmetry @xmath133 is therefore erased before @xmath14 decays , and only the asymmetry @xmath134 produced by the out - of - equilibrium decay of @xmath14 survives . in the flavor basis chosen above , we have @xmath135 ^ 2 } { ( m^\dagger_{\rm d } m_{\rm d})_{11 } } \nonumber \\ & = & \frac{3}{16\pi v^2 } \cdot \frac{\displaystyle m_1 |(m_\nu)_{12}|^2 |(m_\nu)_{23}|^2 \sin 2\phi } { \displaystyle \left [ |(m_\nu)_{11}|^2 + |(m_\nu)_{12}|^2 \right ] it is clear that @xmath134 and @xmath111 only involve two free parameters : @xmath155 and @xmath50 . because @xmath50 is associated with the unknown flavor mixing angle @xmath35 , one may analyze the dependence of @xmath111 on @xmath35 for given values of @xmath155 . for @xmath32 and @xmath104 cases ,
we plot the numerical results of @xmath111 in fig . 1(d ) and fig .
2(d ) respectively . some comments are in order .
\(1 ) in the @xmath32 case , current observational data of @xmath111 require @xmath156 gev for the allowed ranges of @xmath157 . once @xmath157 is precisely measured , it is possible to fix the value of @xmath155 in most cases ( e.g. , @xmath158 gev will be ruled out , if @xmath159 holds ) .
\(2 ) in the @xmath33 case , @xmath160 gev is required by current observational data of @xmath111 .
although @xmath35 is less restricted in this scenario , it remains possible to determine the value of @xmath155 once @xmath35 is measured ( e.g. , @xmath161 gev is expected , if @xmath162 holds ) .
\(3 ) one can carry out a similar analysis of @xmath111 in the framework of supersymmetric seesaw and leptogenesis models .
however , the fgy ansatz does not favor @xmath163 gev , which crucially affects the maximum reheating temperature of the universe after inflation in the generic supergravity models @xcite .
we remark that the cosmological baryon number asymmetry is closely correlated with the jarlskog parameter of cp violation in the fgy ansatz . for illustration
, we plot the numerical correlation between @xmath111 and @xmath114 in fig .
3 , where @xmath164 gev for the @xmath61 case and @xmath165 gev for the @xmath33 case have typically been taken . one can see that the observationally - allowed range of @xmath111 corresponds to @xmath166 in the @xmath61 case and @xmath167 in the @xmath33 case .
the correlation between @xmath111 and @xmath114 is so strong that a measurement of the latter in the long - baseline neutrino oscillation experiments could shed some light on the ball - park magnitude of @xmath155 .
it is worth mentioning that we have neglected possible renormalization - group running effects of neutrino masses and lepton flavor mixing parameters between the scales of @xmath13 and @xmath155 @xcite .
such an approximation is expected to be safe in the @xmath32 case , in which three light neutrinos have a clear mass hierarchy . in the @xmath104 case ,
in which @xmath168 holds , a careful analysis of possible running effects on the fgy ansatz is needed @xcite but it is beyond the scope of this paper .
finally let us comment on possible phenomenological consequences of the fgy ansatz for @xmath8 in eq .
( 4 ) or equivalently @xmath17 in eq . ( 6 ) . in the @xmath61 case
, one can make use of eqs .
( 9 ) and ( 10 ) to calculate relevant phase parameters by setting @xmath169 .
a similar analysis can be done for the @xmath33 case by using eqs .
( 14 ) and ( 15 ) and taking @xmath169 .
we find that the simple replacements @xmath170 and @xmath171 may allow us to write out the expressions of @xmath71 , @xmath50 , @xmath72 , @xmath73 and @xmath74 in the @xmath169 case directly from eqs .
( 11)(13 ) and ( 16)(18 ) .
it turns out that the numerical results of @xmath71 , @xmath50 and @xmath72 are essentially unchanged , but those of @xmath73 , @xmath74 and @xmath114 require the replacements @xmath172 and @xmath173 .
the results for @xmath111 are essentially identical in @xmath169 and @xmath174 cases .
we have analyzed the minimal seesaw model for leptogenesis and neutrino mixing , in which the fgy ansatz is incorporated .
we point out a very striking feature of this model : all nontrivial cp - violating phases can be determined in terms of the lepton flavor mixing angles and the ratio of @xmath0 to @xmath1 .
this important observation allows us to make very specific and testable predictions for the cosmological baryon number asymmetry , the effective mass of neutrinoless double beta decay and cp violation in neutrino oscillations . a precise measurement of the smallest
mixing angle in reactor- and accelerator - based neutrino oscillation experiments will be extremely helpful to examine the fgy ansatz and other presently viable ans@xmath175tze of lepton mass matrices .
t. yanagida , in _ proceedings of the workshop on unified theory and the baryon number of the universe _ , edited by o. sawada and a. sugamoto ( kek , 1979 ) ; m. gell - mann , p. ramond , and r. slansky , in _
supergravity _ , edited by f. van nieuwenhuizen and d. freedman ( north holland , amsterdam , 1979 ) . for recent reviews of leptogenesis with extensive references ,
see : w. buchm@xmath177ller and m. pl@xmath177macher , int
a * 15 * , 5047 ( 2000 ) ; g.f .
giudice , a. notari , m. raidal , a. riotto , and a. strumia , hep - ph/0310123 .
e.w . kolb and m.s .
turner , _ the early universe _ , addison - wesley ( 1990 ) ; h.b .
nielsen and y. takanishi , phys .
b * 507 * , 241 ( 2001 ) ; e.kh .
akhmedov , m. frigerio , and a.yu .
smirnov , jhep * 0309 * , 021 ( 2003 ) .
|
we show that all nontrivial cp - violating phases can be determined in terms of three lepton flavor mixing angles and the ratio of @xmath0 to @xmath1 in the minimal seesaw model in which the frampton - glashow - yanagida ( fgy ) ansatz is incorporated .
this important point allows us to make very specific predictions for the cosmological baryon number asymmetry and cp violation in neutrino oscillations .
a measurement of the smallest neutrino mixing angle will sensitively test the fgy ansatz , in particular in the case that three light neutrinos have a normal mass hierarchy .
|
the recent clear observation of neutrino oscillations @xcite indicates the non - zero masses for the neutrinos @xcite , which are generally considered to be extremely small due to the following mechanisms .
the common form of neutrino mass matrix is expressed through three terms : 1 .
the majorana mass @xmath1 for the left - handed neutrino , that appears as the triplet weak isospin contribution , 2 .
the dirac mass term @xmath2 , involving the interaction with the sterile right - handed neutrino @xmath3 , 3 .
the majorana mass @xmath4 for the singlet @xmath3 over the weak interaction .
then the lagrangian part , determining the neutrino mass , is equal to @xmath5 where @xmath6 denotes the charge conjugation .
the following possibilities are generally discussed .
the first is the absence of sterile @xmath3 , that means there is the only contribution due to the majorana term @xmath1 .
if it is caused by the vacuum expectation @xmath7 of standard higgs , being the isodoublet , then the triplet mass is the square of @xmath7 .
hence , @xmath8 , where @xmath9 appears from an extension of sm , and it is of large scale .
second , the singlet contribution is determined by the physics beyond the sm , so that @xmath10 , where @xmath2 is usually taken in the range , corresponding to the mass scale for the charged fermions , @xmath11
. then the see - saw mechanism @xcite leads to two spices of neutrinos with the small and large masses , @xmath12 and @xmath13 , correspondingly .
these scenarios ( just beyond a super - physics ) generally exhaust the natural explanations for the smallness of neutrino masses .
anyway , the experimental data put @xmath14 ev , which means @xmath15 gev .
note , that we do not know how the small dirac masses can be naturally explained with no involvement of large sterile majorana mass . in the present paper
we offer the scheme , wherein the neutrinos have zero masses in the standard model and acquire the smallness after an extension of sm to include the right - handed neutral currents .
we suggest some non - trivial vacuum correlators , which determine the mass scales in connection to the gauge charges of fermions .
thus , it happens that the large scale @xmath9 is related to the mass of @xmath0 .
from ( [ eq:3a ] ) we deduce that the coupling of vacuum expectations , causing the dirac masses , is determined by the charges of fermions , so that , say , for the neutrino the electric charge equal to zero results in the massless , which , thus , looks quite natural in the sm with the suggested mechanism for the mass generation .
sure , we could introduce the local source , i.e. the higgs field , for the vacuum expectation considered in the model above and make the legendre transformation to substitute the field for the condensates .
then , we believe , the action would take the most usual form - action @xmath26 - \int dx \phi(x ) j(x , x)+{\rm h.c.}\ ] ] to the bare order the equation of motion for the bi - local field results in the straightforward substitution of local field @xmath27 , as it stands in the above consideration for the correlators , developing the vacuum expectation values . after the analysis of divergences in the @xmath28-dependent green functions
, the corresponding counter terms must be added to the action .
then the @xmath28-source can be integrated out , that believes to result in the @xmath27-higgs action , containing the couplings to fermions as well as the suitable potential to develop the spontaneous breaking of symmetry . ] .
suppose that there is the additional @xmath29 local symmetry spontaneously broken at a scale @xmath30 ( for the sake of explicitness we put @xmath31 and @xmath32 ) .
to minimize possible virtual corrections at low energies precisely studied up to the lep measurements and to reproduce the mass relations between the standard gauge bosons , we have to introduce the additional higgs field , being the doublet over @xmath29 , which possesses zero charges over @xmath33 of the sm .
the important challenge is that the standard higgs has to be extended to the field , belonging to the ( @xmath34 ) representation of @xmath35 .
the reason is the desirable renormalizability of the field theory with the spontaneously broken symmetry .
indeed , the fermion mass term @xmath36 certainly is of ( @xmath34 ) , so that the corresponding higgs developing the vev must have the same quantum numbers over the local group .
it can be , say , @xmath37 . in the current discussion
we can suppose that the right - handed doublets are @xmath38 and @xmath39 .
then the charged @xmath29-like gauge bosons are the vector leptoquarks , that can be the reason for the very different physics in the left- and right - handed sectors . ] .
next , the non - zero vevs of neutral higgs fields result in the massless of the photon and lead to the massive neutral currents .
the mass matrix for the local gauge fields @xmath40 , @xmath41 and @xmath42 is determined by the form @xmath43 where @xmath44 , @xmath45 and @xmath46 denote the gauge fields couplings for the @xmath47 group .
then the eigenvalues of the mass matrix are equal to @xmath48 which in the limit of infinitely large @xmath49 tend to the following relations : @xmath50 \approx \frac{1}{4 } v_r^2 g_r^2 . \nonumber\end{aligned}\ ] ] furthermore , it is quite evident to derive that the masses of charged gauged bosons are given by @xmath51 .
\nonumber\end{aligned}\ ] ] thus , we see that up to small corrections the @xmath52 boson mass reproduces the value of standard boson as it is connected to the @xmath53 mass .
the matrix @xmath54 , transforming the gauge fields to the mass eigen - states , has the form @xmath55 with the accuracy up to @xmath56 , where @xmath57 , and @xmath58 is the standard angle by weinberg . in the approximation under consideration
we can see that @xmath54 has the orthogonal form , and the transposition results in the inverse matrix , @xmath59 so that the admixture of @xmath0 in @xmath41 is determined by the ratio of @xmath60 .
the vertices of massive eigen - states are determined by the following relations : 1 .
the photon couples to the electric charge @xmath61 2 .
the left - handed fermions have the standard couplings to @xmath52 @xmath62 3 .
the right - handed fermions acquire the correction to the charge with @xmath52 @xmath63 4 .
the @xmath0 vertex to the left - handed fermions is proportional to that of @xmath52 one , so that @xmath64 and the suppression is due to the smallness of @xmath60 . 5 .
@xmath0 has the charge @xmath65 to the right - handed fermions .
so , some anomalous couplings are introduced due to the @xmath0 physics .
note , that , first , other correlations result in less contributions to the masses , as those are suppressed by powers of @xmath73 .
second , the corrections to the masses of electrically charged fermions seem to be suppressed in the same manner . thus , we see that due to the extension of model to the right - handed local group the neutrinos have the dirac mass of the order of @xmath74 , where - from we extract @xmath75 gev .
if @xmath30 is so large , the current experimental bounds on the anomalous couplings of gauge bosons and the appearance of @xmath0 @xcite are far away from what is expected from the small neutrino masses .
the other possibility is to assume the existence of additional sterile neutrino mass @xmath4 , which activates the see - saw mechanism , but in the model , where the dirac mass can be essentially reduced from @xmath76 gev by several orders of magnitude due to the suppression @xmath73 .
the arrangement of vacuum expectation values for the spontaneous breaking of the local gauge symmetries can be reasonably related to the following qualitative peculiarities , belonging to the corresponding invariant actions .
so , we observe : * the abelian @xmath77-field , coupled in the vector - like way to the fermions of both chiralities , does not appear in the spontaneously breaking phase . * the non - abelian @xmath78 field , possessing the asymptotic freedom ,
is coupled , again , in the vector - like way to the fermions , and it does not acquire the spontaneous breaking of symmetry , too .
however , the back - wise face of asymptotic freedom is the confinement . * the non - abelian @xmath79-field , coupled to the chiral fermions , exposes the spontaneous breaking .
the presence of asymptotic freedom for the latter symmetry depends on the set of matter fields . in what follows
, we exploit the situation , when @xmath79 is asymptotically free .
let us offer the following picture . the structure of vector - like gauge symmetries , i.e. the form of effective potential , preserves them from the developing of non - trivial vacuum correlations , determining the spontaneous breaking of invariances .
the non - abelian theory with chiral fermions does possess the effective action , where the vacuum correlations appear , if the coupling constant is greater than a critical value @xmath80 .
then in gut @xcite with @xmath81 , the chiral @xmath79 will develop the symmetry breaking vev at a low scale @xmath82 , where its coupling @xmath83 will reach the critical value . that can be the reason for the very different values of @xmath84 and @xmath82 . note , that in this approach we know the value of critical constant @xmath80 , since it is well measured in the weak interactions , so that @xmath85 .
as for the neutrino mass generation described above , we are ready to conclude , that the difference between the renormalization group properties , determining the coupling running , for the left- and right - handed symmetries will result in the hierarchy of scales for their characteristic vevs . to be more concrete , consider the gauge symmetry for the right - handed fermions , embedded to the following @xmath29-like doublets : @xmath86 the essential difference from the usual @xmath79 is that the charged vector bosons , possessing the fractional electric charge @xmath87 , are the color triplet @xmath88 and anti - triplet @xmath89 , appropriately .
the corresponding generalized derivative , acting on the fermions , has the form @xmath90 where @xmath91 are the pauli matrices , and the superscript @xmath92 runs over the color anti - triplet and the subscript @xmath92 does the color triplet .
it is quite clear , that , including the quark colors , the `` doublets '' in ( [ eq:10 ] ) could represent the @xmath93 fundamental multiplet , if the couplings of @xmath46 for the right - handed fermions and @xmath94 in @xmath78 would be equal each to other .
if we `` switch off '' the color indexes from the group transformations , the corresponding invariance @xmath95 is @xmath96 on the right - handed fermions .
as we have mentioned , the @xmath95 symmetry is very similar to the famous @xmath79 .
the straightforward consideration leads to that we can reproduce the one - loop calculations for the running of @xmath46 from the corresponding evaluation for @xmath79 , if we substitute for @xmath97 at @xmath98 by @xmath99 , where @xmath100 is the number of colors , and for @xmath101 by @xmath102 .
the running of coupling constants in the @xmath103 field theory is given by the expression @xmath104 where @xmath105 depends on the set of fields .
so , @xmath106 where @xmath107 is the number of chiral fermions , @xmath108 is the number of fundamental scalar multiplets .
therefore , for the @xmath49-coefficients of @xmath109 we get @xmath110 where @xmath111 is the number of fermion generations , @xmath112 are the numbers of corresponding scalars .
next , the @xmath77 coupling constant , normalized as @xmath113 , where @xmath114 is the weak hyper - charge , has the coefficient @xmath115 equal to @xmath116 where we have taken into account the hyper - charge of additional weak doublet due to the extension of standard higgs by @xmath95 : @xmath117 where the fractionally charged higgses are the color anti - triplets with the hyper - charge @xmath118 .
so , @xmath119 denotes the number of standard higgses .
( 130,80 ) ( 0,0 ) ( 0,80)@xmath120 ( 120,0)@xmath121 $ ] ( 10,65)@xmath122 ( 10,40)@xmath123 ( 10,22)@xmath124 ( 58,17)@xmath125 let us discuss the scalar field set , suitable for the problem . first , the number of left - symmetric higgses includes @xmath126 doublets . in realistic models for the generation replication and their mixing , the number of standard higgses usually repeats the number of generations @xcite .
so , we put @xmath127 second , the right - symmetric scalars are those of extensions for the standard higgs and that of @xmath128 singlet to separately break the @xmath95-symmetry .
we put @xmath129 further , we can look at the evolution of couplings to large scales as it is shown in fig.[gut ] , and draw the conclusion on the plausible unification of symmetries at @xmath130 gev .
we present this picture for the illustration of other feature : the @xmath131 coupling reaches the critical region , @xmath132 , at the scales , which are only one or two orders of magnitude less than the gut energy because of the appropriate properties in the renormalization group .
of course , the numerical estimate qualitatively depends on @xmath133 , which can vary over the structure of right - handed symmetry .
another note concerns the extra - higgses , which , according to the evolution performed , are much lighter than @xmath84 . as for the model under discussion
, we could add only that , obviously , there is the right - handed symmetry at @xmath84 : @xmath93 with the violation in the way @xmath134 . to complete , we have to emphasize that the correlators in ( [ eq:7])-([eq:8 ] ) belong to the @xmath135 representation over @xmath136 , and , hence , contribute , a little bit , to vev of the extended higgs , and not to the @xmath95-doublet , developing the @xmath30 scale itself .
we have shown how the model of mass generation can be constructed on the basis of higgs mechanism , wherein the scalar isodoublet field is related to the vacuum correlations and fermion charges , so that * the neutrino is massless in the standard model because of its zero electric charge , * the smallness of neutrino masses can be caused by the hierarchy of the correlation scales for the spontaneous breaking of the standard local symmetry and the right - handed extension , @xmath137 , which leads to the dirac kind of mass , * the unification of coupling constants makes @xmath30 to be only one or two orders of magnitude less than @xmath84 , * we could reduce the scale of sterile majorana mass by involving both the see - saw mechanism and suppressed dirac terms due to the above approach .
finally , the author would like to express the gratitude to prof .
a.wagner and members of desy theory group for their kind hospitality during my visit to desy , where this paper was written , as well as to prof . a.k.likhoded for discussions .
99 chooz collaboration , phys .
b420 * , 397 , ( 1998 ) ; + super - kamiokande collaboration , phys .
lett . * 81 * , 1158 ( 1998 ) ; + homestake collaboration , b.t.cleveland _ et al .
( proc . suppl . )
* b38 * , 47 ( 1995 ) ; + kamiokande collaboration , y.suzuki , nucl .
( proc . suppl . )
* b38 * , 54 ( 1995 ) ; + gallex collaboration , p.anselmann _ et al .
_ , phys . lett . *
b357 * , 237 ( 1995 ) ; + sage collaboration , j.n.abdurashitov _ et al .
_ , phys . lett . *
b328 * , 234 ( 1994 ) ; + super - kamiokande collaboration , phys . lett . *
b433 * , 9 ( 1998 ) , phys . rev
* 81 * , 1562 ( 1998 ) ; + kamiokande collaboration , k.s.hirata _ et al . _ ,
b205 * , 416 ( 1988 ) , _ ibid . _ * b280 * , 146 ( 1992 ) , y.fukuda _ et al .
b335 * , 237 ( 1994 ) ; + imb collaboration , d.casper _ et al .
lett . * 66 * , 2561 ( 1991 ) , r.becker-szendy _ et al .
_ , phys . rev . * d46
* , 3720 ( 1992 ) ; + soudan2 collaboration , t.kafka , nucl . phys .
( proc.suppl . ) * b35 * , 427 ( 1994 ) , m.c.goodman , _ ibid . _
* 38 * ( 1995 ) 337 , w.w.m.allison _ et al .
b391 * , 491 ( 1997 ) .
n. hata and p. langacker , phys
* d56 * , 6107 ( 1997 ) ; + j. n. bahcall , pi .
krastev , and a. yu .
smirnov , phys .
* d58 * ( 1998 ) , hep - ph/9807216 ; + v. barger , s. pakvasa , t. j. weiler , and k. whisnant , phys . lett . *
b437 * , 107 ( 1998 ) ; + m. jezabek and y. sumino , hep - ph/9807310 ; + r. n. mohapatra and s. nussinov , hep - ph/9809415 ; + g. c. branco , m. n. rebelo , and j. i. silva - marcos , hep - ph/9810328 ; + a. j. baltz , a. s. goldhaber , and m. goldhaber , hep - ph/9806540 ; + h.fritzsch and x.xing , phys . lett.*b372 * , 265 ( 1996 ) ; + m.fukugita , m.tanimoto and t.yanagida , hep - ph/9709388 ; + y.koide , phys . rev .
* d39 * , 1391 ( 1989 ) ; + g.altarelli and f.feruglio , hep - ph/9807353 ; + l.wolfenstein , phys . rev .
* d18 * , 958 ( 1978 ) ; + r.n.mohapatra and s.nussinov , phys . lett . *
b346 * , 75 ( 1995 ) ; + p harrison , d.perkins and w.scott , phys . lett .
* b 349 * , 137 ( 1995 ) ; + h. fritzsch and z .- z .
xing , hep - ph/9808272 and 9807234 ; + d. o. caldwell and r. n. mohapatra , phys . rev .
* d48 * , 3259 ( 1993 ) , * d50 * , 3477 ( 1994 ) ; + j. t. peltoniemi and j. w. f. valle , nucl . phys .
* b406 * , 409 ( 1993 ) ; + d. caldwell , hep - ph/9804367 ; + h. georgi and s. l. glashow , hep - ph/9808293 .
m. gell - mann , p. ramond , and r. slansky , in _ supergravity _
f. van nieuwenhuizen and d. freedman , ( north holland , amsterdam , 1979 ) p. 315 ; t. yanagida , _ proc . of the workshop on unified theory and the baryon number of the universe _ , kek , japan , 1979 ; s. weinberg , phys .
43 * , 1566 ( 1979 ) .
p.langacker , preprint upr-0806 t , hep - ph/9805486 ; + m.cvetic , d.a.demir , j.r.espinosa , l.everett , p.langacker , phys . rev . *
d56 * , 2861 ( 1997 ) ; + g.cleaver , m.cvetic , j.r.espinosa , l.everett , p.langacker , phys . rev . * d57 * , 2701 ( 1998 ) ; + a a.likhoded , g.valencia , o.p.yushchenko , phys .
* d57 * , 2974 ( 1998 ) ; + a.a.likhoded , t.han , g.valencia , phys . rev .
* d53 * , 4811 ( 1996 ) ; + m.beccaria , f.m.renard , s.spagnolo , c.verzegnassi , hep - ph/9810323 ; + j.ellison , j.wudka , hep - ph/9804322 .
h.georgi , s.l.glashow , phys .
lett . * 32 * , 438 ( 1974 ) ; + h.georgi , h.quinn , s.weinberg , phys .
* 33 * , 451 ( 1974 ) .
z.kakushadze , g.shiu , s.h.h.tye , y.vtorov-karevsky , int
a13 * , 2551 ( 1998 ) ; + s.adler , phys . rev .
* d58 * , 015012 ( 1998 ) ; + v.v.kiselev , ihep 98 - 21 ( 1998 ) , hep - ph/9806523 .
|
the characteristic scale of neutral current , provided by an extension of standard model with a local group over the right - handed fermions , determines the smallness of neutrino masses of dirac kind .
the experimental observation of neutrino oscillations imposes the stringent limit on the @xmath0 physics appearance at low energies .
= -1.7 cm = -1 cm * neutrino masses and @xmath0 physics * v.v.kiselev + institute for high energy physics + 142284 , protvino , moscow region , russia . + e - mail : [email protected]
|
control - based continuation is a systematic method for performing bifurcation studies on physical experiments .
based on modern feedback control schemes it enables dynamical phenomena to be detected and tracked as system parameters are varied in a similar manner to how nonlinear numerical models can be investigated using numerical continuation .
control - based continuation was originally developed as an extension to pyragas time - delayed feedback control @xcite to make it more robust and suitable for parameter studies @xcite , though its current implementation contains no elements of time - delayed feedback .
the use of feedback control for the investigation of nonlinear systems is not new ; in addition to time - delayed feedback , methods such as ogy control @xcite have previously been employed to provide non - invasive control to stabilise unstable orbits and investigate dynamical phenomena .
indeed , previous authors have gone as far as to implement such control schemes within parameter continuation studies to numerically simulate particular experiments such as atomic force microscopes @xcite or reaction kinetics @xcite .
control - based continuation goes beyond these particular methods to allow the use of almost any feedback control scheme and , as such , it is a general purpose tool applicable to a wide range of physical experiments .
control - based continuation has been successfully applied to a range of experiments including a parametrically excited pendulum @xcite , nonlinear energy harvesters @xcite and a bilinear oscillator @xcite . in each case
, periodic orbits have been tracked through instabilities such as saddle - node bifurcations ( folds ) thus revealing a great deal of dynamical information about the system in question , including the location of a codimension-2 cusp bifurcation in one case @xcite . as well as bifurcations , other dynamic features such as backbone curves
can also be tracked with control - based continuation @xcite . although the basic scheme for control - based continuation is well established ( an overview is provided in sec . [
sec : cbc ] ) , it lacks many of the features of standard numerical continuation schemes such as bifurcation detection
. only saddle - node bifurcations ( folds ) can be detected readily and that is because they are geometric features in the solution surface .
bifurcations such as period - doubling bifurcations are not geometric features and can go undetected due to the stabilising affect of the feedback controller . in this paper
we consider only periodically forced systems and hence study periodic orbits , though there is no reason that the methods developed should not be applicable to autonomous systems as well . in sec .
[ sec : linear ] , we propose a method for calculating the stability ( the floquet multipliers and associated stable and unstable eigendirections ) based on the estimation of a local linearisation around a stabilised periodic orbit .
we demonstrate the effectiveness of this method in sec .
[ sec : apparatus ] and sec .
[ sec : results ] by applying it to a ( physical ) nonlinear mass - spring - damper system where the nonlinearity is geometric in nature the springs are mounted perpendicular to the direction of motion .
numerical continuation is a path following method used to track solution branches as parameters of the system in question are varied . in a nonlinear system
, these solution branches can encounter bifurcations at particular parameter values which results in a qualitative change in the dynamics of the system .
numerical continuation enables these bifurcations to be detected and tracked in turn .
it is typically applied to differential equation models but it can be used more widely , for example on finite element models . at a basic level , numerical continuation tracks the solutions of an arbitrary nonlinear function , a _ zero problem _ given by @xmath0 where @xmath1 is the system state and @xmath2 is the system parameter(s ) .
numerical continuation works in a predictor - corrector fashion ; at each step a new solution @xmath3 is predicted from previously determined solutions and then the solution is corrected using a nonlinear root finder applied to the function @xmath4 ( typically a newton iteration ) .
the use of a nonlinear root finder means that the stability or instability of solutions is unimportant . in certain circumstances ( for example , near a fold or saddle - node bifurcation )
the function @xmath4 must be augmented with an additional equation the pseudo - arclength equation which enables the numerical continuation scheme to track solution curves that double back on themselves . in these circumstances , without the pseudo - arclength equation the correction step for a fixed set of parameter values @xmath2 will fail since no solution exists . for extensive information and guidance on numerical continuation see the textbooks @xcite .
numerical software is also readily available in the form of coco @xcite and auto @xcite amongst others .
control - based continuation is a means for defining a zero - problem based on the outputs of a physical experiment , thus enabling numerical continuation to be applied directly without the need for a mathematical model . to do this
there are two key challenges to overcome .
1 ) in general , it is not possible to set the state @xmath1 of the physical system and so it is not possible to evaluate @xmath4 at arbitrary points .
2 ) the physical system must remain around a stable operating point while the experiment is running .
while a numerical model going unstable might prove to be a mild annoyance , a physical system going unstable can prove dangerous . in order to overcome these challenges ,
a feedback controller is used to stabilise the system and the control target ( or reference signal ) acts as a proxy for the system state .
non - invasive , that is , the controller does not affect the locations of any invariant sets in the experiment such as equilibria or period orbits .
this requirement for non - invasiveness defines the zero problem ; a control target must be chosen such that the control action @xmath5 in this paper , we consider the case of a periodically forced experiment and , as such , only consider periodic motions . in this case
it is appropriate to consider a fourier discretisation of eq .
( [ eqn : control_zero ] ) and so find the coefficients of the fourier series of the control target .
( in other circumstances different discretisations may be appropriate . )
@xmath6 to solve eq .
( [ eqn : cbc_zero ] ) standard root finding algorithms can be used , however , any required derivatives must be estimated using finite differences from experimental data .
consequently , gradient - based methods can be slow despite their good convergence rates .
in previous publications a newton - broyden method , which avoids recomputing derivative information for successive iterates , has proven effective @xcite . in this paper , since the control acts through the same mechanism as the forcing , we are able to use a quicker method which exploits the fact that we are performing a parameter study in the forcing amplitude @xcite .
consider the case where the total input to the system is given by @xmath7 where @xmath8 is the forcing signal the case of sinusoidal forcing where @xmath9 for an arbitrary control target , the fourier coefficients of eq .
( [ eqn : u_fourier ] ) will be non - zero .
however , the contribution in the fundamental mode ( coefficients @xmath10 and @xmath11 ) can be lumped together with the coefficients @xmath12 and @xmath13 of the forcing term giving a new effective forcing amplitude of @xmath14 in essence , instead of setting the forcing amplitude and trying to calculate the correct corresponding control target , we set the control target and measure the corresponding forcing amplitude .
though this procedure leaves the @xmath15 and the higher harmonics untouched , the corresponding control target coefficients required to set the control action to zero can be quickly determined using a fixed - point iteration . in an iterative manner
the remaining coefficients of the control target @xmath16 are simply set equal to the measured coefficients of the system response @xmath17 , the iteration finishes when the response and the control target remain equal for a certain period of time . for the system described below
, this takes a single iteration .
a fixed - point iteration can not be applied to the coefficients of the fundamental mode since , generically , instabilities in the system will manifest in the fundamental mode .
for a typical ordinary differential equation model , the floquet multipliers and hence stability of a periodic orbit are determined by integrating over one period the first variational equation @xmath18 where @xmath19 represents the deviation from in the context of control - based continuation , even determining whether an orbit is stable or unstable is problematic due to the presence of stabilising feedback control . in @xcite a number of measures are suggested to overcome this problem but all require turning off the feedback control for a period of time ; in many situations this is not desirable as damage could be caused to the experiment or even the experimenter . as such , here
we attempt to fit a time - varying linearisation using techniques from the system identification community . from the fitted time - varying linearisation we are able to calculate the corresponding floquet multipliers and hence the stability properties of the periodic orbit . rather than try to fit a continuous time model of the response to perturbations such as eq .
( [ eqn : variational ] ) , which requires the estimation of derivatives from experimental data , we instead fit the coefficients of a discrete - time multiple - input multiple - output ( mimo ) auto - regressive model with exogenous inputs ( arx ) of the form @xmath20 where @xmath21_{i=0\ldots m-1}$ ] , @xmath22_{i=0\ldots m-1}$ ] and @xmath23_{i=0\ldots m-1}$ ] are vectors of data points of the perturbed system response , the perturbed system input ( due to the control action and an additional random perturbation ) and the model error respectively , sampled across a single period of oscillation and @xmath24 is the backward shift operator ( * ? ? ?
thus @xmath25 corresponds to a discretisation of the perturbed system response using @xmath26 points over the period ; this discretisation is illustrated in fig .
[ fig : sysid ] .
@xmath27 and @xmath28 are square @xmath29 matrices of polynomials in @xmath24 ; here we restrict the polynomials to being first order ( at most ) .
thus eq .
( [ eqn : arx ] ) acts as a period map with all the dynamics of interest encoded in @xmath27 .
( the matrix @xmath28 though required for system identification are not used to infer stability . ) and @xmath30 ; the period of the orbit is normalised to @xmath31 .
here @xmath32 $ ] .
since @xmath30 , @xmath33 is purely a function of @xmath34 $ ] and @xmath35 $ ] .
similarly , @xmath36 is purely a function of @xmath37 $ ] and @xmath38 $ ] , where @xmath24 is the backward shift operator . thus the linearised model eq .
( [ eqn : arx ] ) allows the construction of a linear period map from @xmath39 to @xmath40 . ] at each point we assume linear observability to state that the next point in the time evolution of the linearisation is determined entirely by the previous @xmath41 points alone .
we assume that @xmath42 ( increase the value of @xmath26 as appropriate ) to obtain a banded matrix structure for @xmath27 of @xmath43 and for @xmath28 @xmath44 thus the system identification procedure must identify the @xmath45 coefficients to fully identify the linear model .
the optimal values for @xmath26 and @xmath41 can be estimated using the akaike information criterion ( aic ) @xcite or similar .
note that while increasing @xmath41 increases the data requirements for successful system identification , increasing @xmath26 does not since more information is taken from the existing time series .
since the experiment is operating in closed loop , the number of available system identification methods is somewhat limited . in this paper
we make use of the direct method for closed - loop identification due to its simplicity however , other methods such as joint input - output identification can be used if required @xcite . in order to provide sufficiently informative results for system identification purposes
, the random perturbation should have a sufficiently broadband spectrum .
however , to minimise extraneous noise ( that is , random perturbations which are not captured by the discretisation ) the bandwidth of the random perturbation should be less than the nyquist frequency corresponding to the discretisation in eq .
( [ eqn : arx ] ) . must be estimated using a method such as the prediction error method ( pem ) since the straightforward use of linear least - squares will result in bias ( * ? ? ?
* chapter 10 ) .
however , linear least - squares provides a quick and effective way of starting the iterative pem optimisation .
once a linearised model has been identified the floquet multipliers of the periodic orbit can be determined from the matrix @xmath27 .
corresponds to the first @xmath41 rows and @xmath41 columns of the matrix given by @xmath46 and the floquet multipliers of the periodic orbit are the eigenvalues of @xmath47 .
in addition to the floquet multipliers , the eigenvectors of @xmath47 correspond to the stable and unstable directions of the periodic orbit at a particular point in the oscillation .
to test the effectiveness of this methodology outlined in this paper we apply control - based continuation to a nonlinear tuned mass damper ( ntmd ) similar to the one described in @xcite .
the ntmd consists of a mass able to move horizontally on a low friction bearing system while restrained by two springs that are mounted perpendicularly to the direction of motion , thus providing a geometric nonlinearity .
the ntmd is then excited at the base .
this configuration results in a hardening spring - type characteristic . a photograph of the experiment is shown in fig . [ fig : exp](a ) along with a schematic of the experiment in fig . [ fig : exp](b ) .
the details of the actuation and measurement equipment are as follows .
the ntmd is excited using an aps-113 long - stroke electrodynamic shaker in current control mode using a maxon ads-50/10 - 4qdc motor controller .
typical base displacements are sinusoidal with a frequency ranging from @xmath48@xmath49 and an amplitude ranging from @xmath50@xmath51 .
the peak response amplitude is limited to @xmath52 ; at resonance , this limitation restricts the amplitude of the base motion to approximately @xmath53 .
the motion of the base and the moving mass are measured using laser displacement sensors ( an omron zx2-ld100 and an omron zx1-ld300 respectively ) .
in addition to the displacement measurements , the force provided by the shaker is measured using an mcl - type load cell . for the real - time control , a linear proportional - derivative ( pd ) controller
is used with manually tuned gains .
the methodology is relatively insensitive to the control gains used provided they are sufficient to stabilise any unstable orbits that are encountered .
the controller is implemented on a beaglebone black fitted with a custom data acquisition board ( hardware schematics and associated software are open source and freely available @xcite ) .
all measurements are made at @xmath54 with no filtering .
a random perturbation signal is generated , when necessary , on the real - time control board using the box - muller transformation to generate gaussian pseudo - random numbers which are then filtered using a sixth - order butterworth filter with a cut - off arbitrarily set at @xmath55 ( below the nyquist frequency of the discretisation used for eq .
( [ eqn : arx ] ) ) .
estimations of the fourier coefficients of the response and the control action are also calculated in real - time on the control board . however ,
this was for convenience rather than a necessity .
the basic control - based continuation algorithm described in sec .
[ sec : cbc ] was used to do repeated continuations in the forcing amplitude ( the amplitude of displacement of the shaking table ) for fixed values of the forcing frequency . the forcing frequency , while fixed for individual continuation runs ,
was varied between @xmath56 and @xmath49 in steps of @xmath57 . at each data point , full time series measurements were made .
these are shown as black dots in fig .
[ fig : solution_surface ] where the forcing frequency and forcing amplitude ( in mm ) are plotted against the response amplitude , which we define as the magnitude of the first component in the fourier series .
continuation runs that vary the forcing amplitude for different fixed values of the forcing frequency .
the forcing frequency is changed in steps between @xmath56 and @xmath49 .
the shaded surface is calculated using gaussian process regression on the measured data points .
the location of the unstable periodic orbits in this figure can be inferred from the geometry of the solution surface ; the plotted curve ( solid line ) represents a 1d fold curve inside which are the unstable periodic orbits .
this fold curve is calculated using the gaussian process regressor . ] to aid visualisation , a continuous surface constructed from the individual data points is also plotted in fig .
[ fig : solution_surface ] .
this continuous surface is created using gaussian progress regression on the collected data points where the hyper - parameters for the gaussian process are calculated by maximising the marginal likelihood of the hyper - parameters ( * ? ? ?
the use of gaussian process regression ( or any other similar scheme for interpolating the multi - dimensional data ) also allows for geometric features of the solution surface to be easily extracted .
one pertinent feature is the fold in the solution surface which , from dynamical systems theory , indicates a change in stability of the periodic orbits . as such , the fold curve shown in fig .
[ fig : solution_surface ] ( blue curve ) was extracted using numerical continuation with coco @xcite directly on the regression surface defined by the gaussian process . using gaussian process regression .
the forcing amplitudes of the base motion are @xmath58 , @xmath59 , @xmath60 , @xmath61 and @xmath62 respectively .
fold points ( limit points ) which determine the hysteresis region are marked with black dots . ]
other features of interest are frequency response curves which can also be obtained through numerical continuation on the regression surface by fixing the forcing amplitude to a prescribed value .
figure [ fig : freq_sweep ] shows such frequency response curves , including unstable periodic orbits , for fixed forcing amplitudes of @xmath63 , @xmath59 , @xmath60 , @xmath61 and @xmath62 . for high - amplitude forcing , there is little to distinguish the results to those obtained from a duffing equation with hardening nonlinearity .
however , for low - amplitude forcing there seems to be a significant influence from frictional nonlinearities in the bearing system suspending the mass . in order to verify that the unstable orbits found in the experiment are true unstable periodic orbits and not artefacts of the control scheme
, we repeatedly drive the system to a particular unstable periodic orbit and then turn off the stabilising controller . as can be seen from the time series shown in fig .
[ fig : unstable_escape_timeseries ] , starting from the unstable periodic orbit both the stable low - amplitude and stable high - amplitude periodic orbits can be reached the stable manifold of the unstable orbit acts as a separatrix between the two stable orbits as we would expect from dynamical systems theory . out of @xmath64 separate time series
recorded starting from the same unstable periodic orbit , @xmath65 end at the high - amplitude periodic orbit and the remaining end at the low - amplitude periodic orbit . to recreate the state space .
the time coordinate in panel ( b ) is normalised such that a period of forcing takes one time unit . ] in order to apply the method outlined in sec .
[ sec : linear ] , once an orbit has been obtained using control - based continuation it must be perturbed with a random input signal .
one such example with a perturbation size of @xmath66 is shown in fig .
[ fig : orbit_delay_embedding ] . to avoid estimating the velocity of motion , a time - delay coordinate @xmath67 is used to reconstruct the state - space of the experiment . here
the value of @xmath68 used is @xmath69 where @xmath70 is the period of the forcing .
confidence interval of the absolute values of the floquet multipliers estimated against the size of the perturbation applied .
each point represents @xmath71 separate measurements .
the points marked ( a ) ( red ) are calculated for a stable periodic orbit in open - loop conditions ; the points marked ( b ) ( blue ) are calculated for the same stable periodic orbit in closed - loop conditions ; and the points marked ( c ) ( green ) are calculated for an unstable periodic orbit in closed - loop conditions .
] there is a single parameter in sec .
[ sec : linear ] for which there is no algorithmic way to determine an appropriate value , that is the amplitude of the perturbation applied to the periodic orbit .
thus in order to determine an appropriate amplitude we select two periodic orbits , one stable and one unstable and calculate the floquet multipliers using the fitted linear time - varying model eq .
( [ eqn : arx ] ) for a variety of perturbation sizes and repeat the experiments @xmath71 times . throughout this paper
we set @xmath72 and @xmath73 in eq .
( [ eqn : arx ] ) .
the results of the floquet multiplier estimations are shown in fig .
[ fig : control_effect ] .
the points marked ( a ) ( in red ) are the absolute values of the floquet multipliers estimated for a stable periodic orbit while the experiment is running in open - loop ; similarly , the points marked ( b ) ( in blue ) are the floquet multipliers estimated for the same stable periodic orbit while the experiment is running in closed - loop .
finally , the points marked ( c ) ( in green ) are the floquet multipliers estimated for an unstable periodic orbit while the experiment is running in closed - loop .
each floquet multiplier is marked with the @xmath74 confidence range .
for both the stable and the unstable periodic orbits it can be seen that the @xmath74 confidence range narrows significantly for larger perturbation sizes ; the larger perturbations allow the floquet multipliers to be estimated more consistently
. however , the consistency of the results does not imply accuracy as with estimating derivatives from finite differences , taking a large step introduces errors caused by the nonlinearities in the system . though , since the mean magnitude of the floquet multipliers does not change significantly beyond a perturbation size of @xmath75 , we can have reasonable confidence in the results and so a perturbation size of @xmath66 is used throughout the remainder of this paper . furthermore , fig .
[ fig : control_effect ] shows that running the system in closed - loop rather than open - loop does not have a significant affect on the estimation of the floquet multipliers the error bars of the points ( a ) and ( b ) overlap considerably .
unfortunately no accurate independent estimations of the floquet multipliers are available to check the results of the system identification .
convergence tests on the stable periodic orbit were performed , however the transient dynamics of the electrodynamic shaker as the experiment equilibrated rendered the results meaningless . furthermore , escape tests from the unstable periodic orbit such as those seen in fig .
[ fig : unstable_escape_timeseries ] produced estimates ranging from @xmath76 to @xmath77 depending on the measurement cut - offs used . as such , we consider the best way to judge the accuracy of the estimations is via comparison with geometric phenomena such as fold points where a floquet multiplier should pass through @xmath78 in the complex plane . .
stable periodic orbits are shown by solid blue dots and unstable orbits are shown by red circles .
the maximum displacement of the experiment is limited to @xmath52 .
panel ( b ) shows the for three separate continuation runs ( each shown with a different symbol ) .
it can be seen that most of the stable periodic orbits have complex conjugate floquet multipliers which then become real close to the saddle - node bifurcations ; between the two saddle - node bifurcations , one of these floquet multipliers lies outside the unit circle while the other lies inside the unit circle . ]
subsequently , a continuation in the forcing amplitude was performed with floquet multipliers estimated for each obtained solution ; the results are shown in fig .
[ fig : branch_stability](a ) with stable periodic orbits marked as solid blue dots and unstable periodic orbits marked as red circles .
the results agree very well with what is expected from dynamical systems theory ; the orbits between the two fold points are all unstable .
to ensure repeatability , the same continuation run was carried out three times and the absolute values of the corresponding floquet multipliers are plotted in fig .
[ fig : branch_stability](b ) with different symbols for the different runs . as can be seen from fig .
[ fig : branch_stability](b ) the results are very consistent with all the runs showing good quantitative as well as qualitative agreement .
as expected , a complex conjugate pair of floquet multipliers becomes real close to the fold point after which a single real multiplier passes through the unit circle .
this process reverses close to the second fold point . using time - delay coordinates .
the red curve represents the unstable periodic orbit and the two green curves represent the stable high- and low - amplitude periodic orbits .
the blue curves show the transient dynamics between the unstable orbits and the stable orbits .
the two different stable orbits are reached with approximately equal probability .
the time coordinate in panel ( b ) is normalised such that a period of forcing takes one time unit . ]
as well as providing information about the floquet multipliers of the periodic orbits , the monodromy matrix obtained from eq .
( [ eqn : arx ] ) also provides the stable and unstable eigendirections of the periodic orbit .
figure [ fig : unstable_escape ] shows the time series from fig .
[ fig : unstable_escape_timeseries ] plotted in state - space , with the unstable periodic orbit ( marked in red ) lying between the two stable periodic orbits ( marked in green ) . from fig .
[ fig : unstable_escape](b ) it can be seen that the escape from the unstable periodic orbit occurs along a two - dimensional unstable manifold .
separate time series of the escape from the unstable periodic orbit shown in fig .
[ fig : unstable_escape_timeseries ] and fig .
[ fig : unstable_escape ] .
each dot ( blue ) corresponds to an intersection with the poincar section defined by @xmath79 , where @xmath70 is the forcing period .
the dots trace out the 1d unstable manifold of the unstable orbit .
the positions of the stable high- and low - amplitude periodic orbits are marked with green crosses .
the stable and unstable eigenspaces estimated using the methodology outlined in this paper are marked by red arrows .
the unstable eigenspace shows good agreement with the dots from the time series measurements . ]
figure [ fig : unstable_escape_poincare ] shows a poincar section at a fixed time in the forcing cycle that was created with the data from fig . [ fig : unstable_escape ] combined with data from a further @xmath80 independent time series measurements ; the blue dots denote intersections of the trajectories with the poincar section .
the intersection of the two - dimensional unstable manifold with the poincar section results in a well defined one - dimensional curve showing how trajectories leave the unstable periodic orbit ( marked with a red dot ) and approach the two stable periodic orbits ( marked with green crosses ) . superimposed on fig .
[ fig : unstable_escape_poincare ] is a set of red arrows which mark the unstable and stable eigendirections calculated from the monodromy matrix of the periodic orbit using only local perturbations ( as previously described ) .
the eigendirections represent the linearisation of the manifolds at the unstable periodic orbit and , as such , show remarkably good agreement with the measured unstable manifold which was calculated in open - loop conditions .
this opens up the possibility for using eigendirection information in other calculations on the physical system , for example the estimation of basins of attraction .
this paper has proposed a new method for estimating floquet multipliers and their associated eigendirections for periodic orbits that encountered when using control - based continuation on a physical experiment . a linear time - varying model in the form of an auto - regressive model with exogenous inputs ( arx )
is fitted to each periodic orbit using small perturbations to the orbit to explore the nearby state - space .
the method was demonstrated on a nonlinear mass - spring - damper - type experiment that has a hardening spring characteristic .
the nonlinearity is provided by placing springs perpendicular to the direction of motion , thus creating a geometric nonlinearity .
the floquet multiplier estimations were shown to agree with what is expected from dynamical systems theory and the associated eigendirections match well with open - loop measurements taken .
all the experimental data used in this paper has been deposited into the university of bristol research data repository and is publicly available for download @xcite .
d.a.w.b . is supported by epsrc first grant ep / k032739/1 .
d.a.w.b also gratefully acknowledges discussions with and input from ludovic renson , alicia gonzalez - buelga and simon neild .
10 url # 1`#1`urlprefixhref # 1#2#2 # 1#1 s. misra , h. dankowicz , m. r. paul , event - driven feedback tracking and control of tapping - mode atomic force microscopy , proceedings of the royal society a 464 ( 2096 ) ( 2008 ) 21132133 .
http://dx.doi.org/10.1098/rspa.2007.0016 [ ] .
c. i. siettos , i. g. kevrekidis , d. maroudas , coarse bifurcation diagrams via microscopic simulators : a state - feedback control - based approach , international journal of bifurcation and chaos 14 ( 1 ) ( 2004 ) 207220 . http://dx.doi.org/10.1142/s0218127404009193 [ ] .
j. sieber , a. gonzalez - buelga , s. a. neild , d. j. wagg , b. krauskopf , experimental continuation of periodic orbits through a fold , physical review letters 100 ( 24 ) ( 2008 ) 244101 . http://dx.doi.org/10.1103/physrevlett.100.244101 [ ] .
d. a. w. barton , s. g. burrow , numerical continuation in a physical experiment : investigation of a nonlinear energy harvester , asme journal of computational and nonlinear dynamics 6 ( 1 ) ( 2011 ) 011010 .
http://dx.doi.org/10.1115/1.4002380 [ ] .
d. a. w. barton , b. p. mann , s. g. burrow , control - based continuation for investigating nonlinear experiments , journal of vibration and control 18 ( 4 ) ( 2012 ) 509520 .
http://dx.doi.org/10.1177/1077546310384004 [ ] .
e. bureau , f. schilder , i. ferreira santos , j. juel thomsen , j. starke , experimental bifurcation analysis of an impact oscillator tuning a non - invasive control scheme , journal of sound and vibration 332 ( 22 ) ( 2013 ) 58835897 . http://dx.doi.org/10.1016/j.jsv.2013.05.033 [ ] .
e. bureau , f. schilder , m. elmegrd , i. f. santos , j. j. thomsen , j. starke , experimental bifurcation analysis of an impact oscillator determining stability , journal of sound and vibration 333 ( 21 ) ( 2014 ) 54645474 . http://dx.doi.org/10.1016/j.jsv.2014.05.032 [ ] .
k. d. murphy , p. v. bayly , l. n. virgin , j. a. gottwald , measuring the stability of periodic attractors using perturbation - induced transients : applications to two non - linear oscillators , journal of sound and vibration 172 ( 1 ) ( 1994 ) 85102 .
http://dx.doi.org/10.1006/jsvi.1994.1160 [ ] .
l. n. virgin , m. d. todd , c. j. begley , s. t. trickey , e. h. dowell , basins of attraction in experimental nonlinear oscillators , international journal of bifurcation and chaos 8 ( 3 ) ( 1998 ) 521533 . http://dx.doi.org/10.1142/s0218127498000334 [ ] .
i. kozinsky , h. w. c. postma , o. kogan , a. husain , m. l. roukes , basins of attraction of a nonlinear nanomechanical resonator , physical review letters 99 ( 20 ) ( 2007 ) 207201 .
http://dx.doi.org/10.1103/physrevlett.99.207201 [ ] .
e. j. doedel , b. e. oldeman , http://indy.cs.concordia.ca/auto/[auto-07p : continuation and bifurcation software for ordinary differential equations ] , tech .
, concordia university , montreal ( canada 2012 ) .
d. a. w. barton , experimental data for `` control - based continuation : bifurcation and stability analysis for physical experiments '' , available online ( 2015 ) .
|
control - based continuation is technique for tracking the solutions and bifurcations of nonlinear experiments .
the basic idea is to apply the method of numerical continuation to a feedback - controlled physical experiment . since in an experiment it is not ( generally ) possible to set the state of the system directly , the control target is used as a proxy for the state .
the challenge then becomes to determine a control target such that the control is non - invasive , that is , it stabilises a steady - state ( or periodic orbit ) of the original open - loop experiment without altering it otherwise .
once implemented , control - based continuation enables the systematic investigation of the bifurcation structure of a physical system , much like if it was numerical model
. however , stability information ( and hence bifurcation detection and classification ) is not readily available due to the presence of stabilising feedback control .
this paper uses methods from the system identification community to extract stability information in the form of floquet multipliers from the closed - loop experiment , thus enabling the direct detection of bifurcations .
in particular , it is shown that a periodic auto - regressive model with exogenous inputs ( arx ) can be constructed that approximates the time - varying linearisation of the experiment around a particular periodic orbit .
this method is demonstrated using a physical nonlinear tuned mass damper . numerical continuation , bifurcation theory , system identification , feedback control
|
the cooling and thermal balance of the primordial , zero - metallicity gas by molecular hydrogen ( h@xmath11 ) and deuterated hydrogen ( hd ) are key ingredients in the formation process of the first baryonic objects in the universe , such as dwarf - sized galaxies and pop - iii stars ( see for recent reviews on this topic barkana & loeb 2001 ; ciardi & ferrara 2004 ; bromm & larson 2004 ) . it is well known that h@xmath11 and hd molecules form in the universe after recombination ( e.g. , lepp & shull 1984 ; puy et al .
1993 ; palla , galli , & silk 1995 ; galli & palla 1998 ; stancil , lepp , & dalgarno 1998 ) . the formation of h@xmath11 and hd molecules in primordial gas is also possible in a postshock flow ( mac low & shull 1986 ; shapiro & kang 1987 ; uehara & inutsuka 2000 ) , where the gas density and temperature , as well as the molecular fractions , are different from those of the expanding homogeneous gas . the h@xmath11 molecule is considered typically as the main coolant in the primordial medium ( e.g. , tegmark et al . 1997 ; abel , bryan & norman 2000 ; bromm , coppi & larson 2002 ) .
thus , rather detailed calculations were carried out for it ( see for the most recent results and references therein le bourlot , pineau des forts & flower 1999 ; flower et al .
2000 ; shaw et al . 2005 ) .
nevertheless , in the case of dense primordial gas , the role of the hd molecule in the thermal and cooling process may become comparable to or more important than the one of h@xmath11 .
therefore , detailed calculations of the hd cooling function ( the rate of cooling per hd molecule , hereafter cf ) are needed .
the hd molecule is an efficient coolant in the primordial medium because it has a permanent electric dipole moment that allows high probabilities for the radiative rotational transitions from the rotational levels pumped by collisions with h and he . besides the hd molecule was efficiently produced in the reaction of chemical fractionation @xmath12 , which leads to a strong enhancement of the initial abundance of hd , up to roughly @xmath13/[h_{2}]=10^{-2}-10^{-3}$ ] ( e.g. puy et al .
1993 ; galli & palla 1998 ; flower 2000 ) .
the hd molecule can cool the primordial gas to temperatures below @xmath14 k because it has allowed dipole rotational transitions characterized by energies two times smaller than those of the quadrupole transitions of h@xmath15 . in that regard higher temperatures ( @xmath16k ) , cooling by hd
can be again as important as that by h@xmath15 .
this question will be discussed in the present paper .
the hd cf was calculated for the first time by dalgarno & mccray ( 1972 ) for low temperatures and boltzmann distributed rotational level populations .
varshalovich & khersonskii ( 1976 ) improved the calculations by taking into account the departure from local thermodynamic equilibrium .
more recently , flower et al . ( 2000 , and see also more references therein ) have calculated the hd cf for low and intermedium kinetic temperatures and a large range of densities .
these authors considered the rotational transitions within the vibrational ground state @xmath17 of hd .
note that in the case of higher kinetic temperatures ( @xmath18 ) , the populations of the vibrational levels become significant , and cooling by ro - vibrational transitions of hd may be important due to ( i ) the relatively high energy of vibrational quanta , and ( ii ) the relatively high probabilities of the ro - vibrational radiative transitions .
therefore , the inclusion of the the vibrational structure of the hd molecule in the calculation of its cf might be important .
the aim of this paper is to recalculate the hd cf for a wide range of kinetic temperatures and densities by taking into account ro - vibrational transitions within ground @xmath17 and the first three excited @xmath19 vibrational levels .
further exploration of the influence of the hd cf calculated here on the thermal evolution of the primordial medium will be presented elsewhere . in
2 we give details of our calculations .
the revised hd cf is given and compared with that from flower et al .
( 2000 ) in 3 .
finally , in 4 we discuss our results and speculate about possible implications .
it is well known that in the low density limit ( @xmath20@xmath7 ) , the relative populations of the levels of hd molecule depart significantly from the boltzmann distribution .
therefore , to calculate the hd cf in the general case , we need the correct values of the ro - vibrational level populations outside the thermodynamical equilibrium . this can be calculated by using the equations of detailed population level balance : @xmath21where @xmath22 is the population of the ro - vibrational level @xmath23 , @xmath24 and @xmath25 are the probabilities of the radiative and collisional transitions , respectively . we assume a steady state for the population levels . this assumption has been used and discussed to be valid in our context by many other authors ( e.g. , flower et al .
2000 ; flower & pineau des forts 2001 ; le petit , roueff & le bourlot 2002 ) .
the population of the ro - vibrational levels of the @xmath26 molecule was calculated in the wide range of number densities @xmath27 from @xmath28 to @xmath29@xmath7 , and for the kinetic temperature @xmath30 from @xmath31 to @xmath4 k. the ro - vibrational radiative transition probabilities for the hd molecule were taken from abgrall et al .
( 1982 ) . regarding the collisional transition probabilities , it was shown in flower et al .
( 2000 ) that the hd cf is insensitive to the h / h@xmath15 density ratio .
therefore , one may take into account the excitation and de - excitation of hd only by collisions with the h atoms .
the probabilities of the collisionally excited pure rotational transitions of hd ( up to the third vibratioanally excited state , @xmath32 , and up to the 8-th rotational level @xmath33 ) were taken from roueff & flower ( 1999 ) and roueff & zeippen ( 1999 ) .
we consider only the dipolar transitions ( @xmath34 ) .
owing to the permanent dipolar moment of the hd molecule , these transitions dominate collisional population transfer ( flower & roueff 1999 ) .
flower & roueff ( 1999 ) computed probability coefficients for several ro - vibrational transitions in hd induced by collisions with atomic and molecular hydrogen in a range of temperatures from 100 to 2000 k. in that paper , some values of the transition probabilities for the vibrational relaxation @xmath35 are presented and compared with the corresponding coefficients for h@xmath15 .
the rate coefficients for vibrational relaxation of hd and h@xmath11 in collisions with h are similar in magnitude for a given temperature .
besides , the probabilities for both molecules change approximately in the same way with temperature . on the basis of these results and in the absence of extensive calculations for the hd+h collisional ro - vibrational transition probabilities in the range of temperatures we want to explore , we will use the corresponding probabilities of the h@xmath15+h ( tin , lepp & dalgarno 1998 ) given in electronic form on the website http://www.physics.unlv.edu/astrophysics/h2h2rates/index.html ( lepp , tin & dalgarno 1997 , in preparation ) .
the total cf per unit of volume is defined , for example for the hd molecule , as : @xmath36where @xmath37 is the hd number density , and @xmath38 is the hd cf in unities of erg s@xmath39 per hd molecule : @xmath40here , @xmath41 are the probabilities of the radiative transitions with the emission of the photon @xmath42 , and @xmath43 are the probabilities of radiative transition with the absorption of the corresponding field photon . the symbols @xmath44 and @xmath45 are for the corresponding einstein coefficients and @xmath46 is the radiation field .
the populations of the ro - vibrational levels @xmath22 are those calculated from the balance equation ( 1 ) .
this definition of @xmath38 is more general than the one used in flower et al . (
2000 ) , because it takes into account the potential effects of a radiation field in the cooling process . in the case of @xmath47 , eq .
( 3 ) tends to the case considered in flower et al .
( 2000 ) . at high redshifts , the cosmic microwave background radiation ( cmbr ) temperature , @xmath48 , may become comparable to the kinetic temperature of the gas @xmath30 ; this will significantly affect the hd cf at that @xmath30 ( see below for more details ) .
1 we present the variation of @xmath38 with @xmath30 and @xmath27 .
the dotted - line curves correspond to the hd cf , neglecting the collisional ro - vibrational transitions as in flower et al .
our results agree very well with those of these authors .
note that we did not consider collisions of hd with he and h@xmath15 because , as it was mentioned above , flower et al .
( 2000 ) showed that @xmath38 is essentially a function of @xmath30 and @xmath27 only .
the solid - line curves in fig .
1 are the cfs including ro - vibrational collisional transitions for a 4-level vibrational structure of the hd molecule ( @xmath49 ) .
one sees that the latter cfs depart from the former , as @xmath30 and @xmath27 are larger . for @xmath50@xmath7 and @xmath51k ,
the calculated value of @xmath38 including the vibrational transitions is more than a factor of fifty larger than the @xmath38 calculated neglecting collisional ro - vibrational transitions .
note that in fig . 1 , @xmath38 has been calculated for @xmath52 k ( the cmbr temperature at present ) , i.e. , well within the limit @xmath47 , when the radiation field does not participate in the thermal balance of the gas and eq .
( 3 ) is reduced to the equation for @xmath38 used in flower et al .
this way were able to compare our results with those of these authors .
the exact hd cf data presented here are available at http://www.cifus.uson.mx/personal_pages/anton/ data / hd_cooling / hd_cool.html . to facilitate the use of the hd cf in computational programs
we give also a polynomial approximation for the hd cf .
the approximation is very accurate in the ranges of @xmath27 and @xmath30 studied here , from @xmath28 to @xmath6 @xmath7 , and from @xmath31 to @xmath4 k , respectively .
the approximation is writen in the form : @xmath53where the coefficients @xmath54 are tabulated in table 1 .
.polynomial coefficients @xmath54 [ cols="<,^,^,^,^,^",options="header " , ] from fig .
1 one clearly sees how @xmath38 increases at high densities after including the collisional ro - vibrational level transitions . in the case of low densities ,
the contribution of these transitions to the cooling of hd is less important .
nevertheless , at high temperatures @xmath38 is still larger than in the case when these transitions are omitted ( pure rotational transitions shown by dashed line at fig .
therefore , the analytical approximation commonly used for the low density limit ( see e.g. , galli & palla 2002 ) should be slightly modified .
the approximation given in galli & palla ( 2002 ) is based on the calculations for only two low rotational transitions @xmath55 and @xmath56 , because the collisional probabilities of these transitions are rather large as compared with those of higher rotational levels .
however , it should be stressed that when the vibrational structure is taken into account , then there are other ro - vibrational transitions with comparable probabilities of collisional excitation . based on our results ,
we suggest the following modified approximation for the low - density limit hd cf : @xmath57 this aproximation can be applied for the gas density @xmath27 up to @xmath58 @xmath7 .
the effects of the radiation field @xmath46 in @xmath38 become important when @xmath59 ( flower 2000 ; flower & pineau des forts 2001 ) . in this case , the radiative absorption by the hd molecule dominates and the second term in eq .
( 3 ) , as well as the @xmath60 term in the probability @xmath61 , should be considered . the second term in eq .
( 2 ) can be interpreted as a heating function . at the redshfits at which the first baryon objects are expected to form ( @xmath62 ) ,
the cmb temperature can be similar to the typical kinetic temperatures of the small primordial gas clouds . in fig .
2 we present the complete hd cfs ( with the collisional ro - vibrational transitions included ) considering a radiative field with @xmath63 for redshifts @xmath64 , @xmath65 and @xmath66 ( dotted , dashed and long - dashed lines , respectively ) and for two densities , @xmath67 and @xmath6 @xmath7 .
one sees that when @xmath59 , @xmath38 falls dramatically . the second term in eq .
( 3 ) actually comes to dominate and the hd molecular lines act as heating sources .
this will inhibit the collapse of a gas cloud .
therefore the minimum temperature of the primordial collapsing clouds is limited by the cmbr temperature ( or by the heating due to other radiation fields ) and not by hd ( and h@xmath15 ) cooling .
we note that the steady state assumption for the hd population levels ( see 2 ) remains even at the high redshifts considered here .
the typical time scales for population level changes are determined mainly by the inverse of the transition einstein coefficients . for the hd molecule transitions these coefficientes imply typical time scales much less than the time scales related to variations in @xmath68 and/or to the gas clouds collapse in the epochs considered here ( @xmath69 ) .
we have calculated the hd cf , @xmath38 , for a wide range of kinetic temperatures and gas densities by taking into account the ro - vibrational structure of the molecule for both the radiative and collisional transitions .
we have found that , when including the collisional ro - vibrational transitions , the cooling efficiency of hd molecule is higher than previously reported .
the main increasing of @xmath38 as compared to previous works is at the high - temperature side and for high densities ( see fig .
1 ) ; for @xmath70 @xmath7 , the differences in @xmath38 reach an order of magnitude and more at @xmath71 . our results in support of recent claims ( flower 2000 ; uehara & inutsuka 2000 ; flower & pineau des forts 2001 ; nakamura & umemura 2002 ; galli & palla 2002 ) , suggest that the hd molecule may play an important role in the thermal balance and cooling of primordial gas at high densities . in fig .
3 we attempt to compare the relative contributions to the cf of the gas by the h@xmath11 and hd molecules . because the ratio of abundances of hd to h@xmath11 is significantly smaller than 1 , the @xmath38 curves should be reduced by a factor equal to this ratio in order to compare the contributions to the cooling of gas by both molecules . as it was mentioned above
, this ratio for the primordial gas , after chemical fractionation , is approximately @xmath72 . in fig .
3 , we reduce @xmath38 by a factor of @xmath13/[h_{2}]=10^{-2.4}$ ] according to galli & palla ( 2002 ) for @xmath64 . as one can see in fig .
3 , the role of hd in the cooling of low - density gas is still unimportant with respect to the one of h@xmath11 , while for high densities , hd becomes as efficient a coolant as h@xmath11 , being even more efficient at low temperatures .
the h@xmath11 cf has been calculated according to flower et al .
( 2000 ) .
the contribution to the gas cooling of hd found here for high gas densities and temperatures larger than @xmath10 k is close to the corresponding contribution of h@xmath11 .
this will surely affect calculations of galaxy formation in the regime when dense gas cools down from temperatures above @xmath10 k. this regime may arise in photoionization - heated gas or in shock - heated dense material in dwarf galaxy - size dark matter halos with masses @xmath73 m@xmath74 ( bromm & larson 2004 ) .
the presence of an efficient coolant , besides that h@xmath11 , able to operate at @xmath75 k and at high densities ( as is the case of hd ) is probably also relevant to understand the formation of the first globular clusters .
bromm & clarke ( 2002 ) have shown that , under the action of an efficient coolant at @xmath76 k , stellar clusters with masses @xmath77 may form into small dark matter subhalos that later on are tidally destroyed during the violent relaxation of dwarf galaxy - sized halos at @xmath78 ( see also weil & pudritz 2001 ) .
the cooling efficiency of h@xmath15 or hd molecules increases with density until the population levels reachs the local thermodynamical equilibrium at a critical density @xmath79 , in such a way that at @xmath79 the probability of collisional de - exitation becomes equal to the spontaneous radiative probability . beyond this density
, the cooling is saturated . for molecular hydrogen ,
the critical density is rather low , @xmath80 @xmath7 , due to the relatively small values of the h@xmath15 einstein coefficients . in the case of the hd molecule ,
the einstein coefficients are larger as compared to those of h@xmath15 by @xmath81 orders of magnitude .
one may estimate the ratio of critical densities @xmath82 as @xmath83 , which is @xmath84 for radiative vibrational transitions .
the value of @xmath85 is indeed around two orders of magnitude higher than @xmath86 . from fig . 1 ( see also flower et al .
2000 ) , one sees that @xmath87 @xmath7 . to conclude
, we remark the importance of taking into consideration the hd cf calculated here in the simulations and models of the first baryonic objects in the universe . in particular , those processes related to dense gas that cool down from temperatures above @xmath10 k will be affected by our results .
we are thankful to the anonymous referee for his / her remarks and comments that helped to improve the paper .
this work was supported by grant promep /103.5/03/1147 to a. l. in addition , r. n - l .
gratefully acknowledges a phd fellowship by conacyt .
we are grateful to j. benda for grammar corrections to the manuscript , and c. guzmn for computer assistance .
|
we report new calculations of the cooling rate of primordial gas by the hd molecule , taking into account its ro - vibrational structure .
the hd cooling function is calculated including radiative and collisional transitions for @xmath0 rotational levels , and for the vibrational levels @xmath1 and 3 .
the ro - vibrational level population is calculated from the balance equation assuming steady state .
the cooling function is evaluated in the ranges of the kinetic temperatures , @xmath2 , from @xmath3 to @xmath4 k and the number densities , @xmath5 , from 1 to @xmath6 @xmath7 .
we find that the inclusion of collisional ro - vibrational transitions increases significantly the hd cooling efficiency , in particular for high densities and temperatures . for @xmath8 and @xmath9
k the cooling function becomes more than an order of magnitude higher than previously reported .
we give also the hd cooling rate in the presence of the cosmic microwave radiation field for radiation temperatures of 30 , 85 and 276 k ( redshifts of 10 , 30 and 100 ) .
the tabulated cooling functions are available at http://www.cifus.uson.mx/personal_pages/anton/data/hd_cooling/hd_cool.html .
we discuss the relevance to explore the effects of including our results into models and simulations of galaxy formation , especially in the regime when gas cools down from temperatures above @xmath10k .
[ firstpage ] cosmology : first stars galaxies : formation molecular processes
|
high energy @xmath0-rays are one of the most direct ways of studying the non - thermal universe .
sources of very high energy radiation enable us to explore the highest energy accelerators in the cosmos , in situations with extreme gravitational and magnetic fields .
although high energy @xmath0-rays have been studied for some time , rapid development in the field came only in the 1990s .
the most significant thrust to high energy astronomy above 30 mev came with the launch of the _ compton gamma ray observatory _ ( cgro ) in 1991 , which carried onboard the energetic gamma - ray experiment telescope ( egret ) , along with three other instruments . until june 2000 ,
when cgro was de - orbitted , egret produced a wealth of astrophysical results , and was responsible for the detection of more than 270 point sources of @xmath0-rays @xcite .
the space - based results were complemented by ground - based atmospheric cherenkov telescopes ( acts ) , which detected @xmath0-rays with energies above 250 gev from a few of the same sources .
particularly , the detection of the crab nebula above 250 gev by more than eight separate experiments has firmly established ground - based @xmath0-ray astronomy on a solid foundation @xcite .
it is impossible to do justice to the exciting developments in high energy @xmath0-ray astronomy in the few allotted pages .
this article , therefore , selectively reviews a few highlights of the field , and refers the reader for more details to some of the more comprehensive reviews that have appeared in the literature recently @xcite .
the review is organized as follows : sections 2 and 3 of this article summarize the techniques and results from egret and the ground - based experiments , respectively .
while the experimental techniques are distinct , there is considerable overlap in the scientific issues addressed by the space- and ground - based experiments .
the three other @xmath0-ray experiments , batse , osse , and comptel onboard cgro are not included in this review , but are described elsewhere @xcite .
some recent developments in ground - based experiments , and future directions are given in section 4 .
egret was primarily a spark chamber instrument that detected @xmath0-rays in the energy range @xmath1 30 mev to 30 gev using the pair production technique .
the instrument had a lifetime of approximately 9 years , from 1991 may to 2000 june , and proved to be immensely successful .
it had the standard components of a high - energy @xmath0-ray instrument : an anticoincidence dome to discriminate against charged particles , a spark chamber particle track detector with interspersed high-@xmath2 material to convert the @xmath0-rays into electron - positron pairs , a triggering telescope to detect the presence of the pair with the correct direction of motion , and an energy measurement system , which in the case of egret was constructed of nai(tl ) crystals .
the anticoincidence dome was successful in rejecting the charged particle background , which outnumber the @xmath0-rays by a factor of @xmath3 .
egret had an effective area of 1500 @xmath4 in the energy range 0.2 gev to 1 gev , decreasing to about one - half the on - axis value at @xmath5 off - axis .
the threshold sensitivity of egret ( @xmath6 mev ) for a single 2-week observation was @xmath7 photons @xmath8 s@xmath9 .
details of the egret instrument are given elsewhere @xcite .
prior to egret , two other successful satellite experiments , sas-2 and cos - b , pioneered the field of high energy @xmath0-rays .
these experiments were responsible for making some of the first maps of the @xmath0-ray sky and producing the first @xmath0-ray source catalogs @xcite .
the @xmath0-ray sky in the 30 mev to 30 gev range comprises both diffuse radiation as well as point sources .
the diffuse radiation , as described in the following section , has a galactic as well as an extragalactic component , and needs to be modeled in order to do point source analysis of the egret data .
the point sources detected by egret above 100 mev are shown in fig .
egret has seen several different kinds of sources , as shown in table 1 .
in addition to the sources listed in the table , egret has also detected 5 @xmath0-ray bursts @xcite , the x - ray binary cen x-3 @xcite , and 1 solar flare @xcite .
it is interesting to note that more than 60% of the egret sources are unidentified , with no firmly established counterparts at other wavelengths .
some of these sources have remained unidentified since the first surveys of the @xmath0-ray sky with the cos - b satellite @xcite , and are one of the outstanding mysteries of the egret mission .
some researchers have found correlations between unidentified egret sources in the galactic plane and supernova remnants @xcite , while others report correlations with ob associations and massive stars @xcite .
a review of the various efforts to understand the nature of these sources is given in @xcite .
recently , it has been noted that the mid - latitude unidentified egret sources form a population distinctly different from the sources along the galactic plane @xcite .
further , it has been suggested that these mid - latitude sources are probably associated with the gould belt of massive stars and gas clouds , about 600 light years away @xcite .
counterparts to individual unidentified sources may only be found after the launch of future , more sensitive @xmath0-ray missions such as glast ( see section 4 ) .
.sources in the third egret ( 3eg ) catalog [ cols="<,<",options="header " , ] + * 27 agn have been marginally identified .
the majority of the sources away from the galactic plane are extragalactic and have been identified with active galactic nuclei ( agn ) .
other than the unidentified sources , agn constitute the largest class of egret sources .
egret tends to detect the `` blazar '' class of agn , which are characterized by emissions that include high radio and optical polarization , and rapid flux variability at all wavelengths .
a large fraction of these sources exhibit apparent superluminal motion , as evidenced from vlbi radio observations .
egret has observed variability in several blazars on timescales of days to months .
a review on the blazars observed by egret may be found in @xcite .
the @xmath0-ray spectrum of blazars in the 30 mev to 30 gev range is well - described by a single power - law of spectral index @xmath10 .
the broad - band spectral energy distribution ( sed ) of one egret flat - spectrum radio quasar ( fsrq ) , pks 0528 + 134 , is shown in fig .
2 . the figure demonstrates the typical characteristics of the sed of a quasar observed by egret , showing that the @xmath0-ray luminosity dominates the spectrum .
in fact , in most fsrqs the @xmath0-ray power dominates or equals the power at other wavebands .
the apparent luminosity , assuming isotropic emission , is @xmath11 ergs s@xmath9 @xcite .
2 includes broadband observations at several different epochs and demonstrates another important characteristic of egret blazars spectral variability between different states of the source .
egret observations of blazars help to constrain models of emission for these objects .
rapid variability , high compactness , and superluminal motion suggest that the emission is likely to be beamed and doppler - boosted into the line of sight @xcite . simplifying many details
, it is generally believed that agn are powered by accretion of matter on to a supermassive black hole . for the case of blazars ,
collimated jets composed of shocks or plasma blobs move with relativistic speeds outward along the axes of the accretion disk , making a small angle with the axis .
gamma - ray emission models for blazars are generally divided into two broad classes : leptonic and hadronic .
details of these models are not discussed here but may be found in @xcite .
2 shows an example of a leptonic jet model applied to the case of pks 0528 + 134 @xcite .
it has been suggested that hadronic models may be distinguished from leptonic models by the observation of high energy neutrinos that may be produced as a result of photoproduction @xcite .
there are several outstanding questions regarding the details of the different models that may only be resolved in the future .
in addition to blazars , egret has detected @xmath0-rays from 6 pulsars .
these are the crab , vela , psr 1706 - 44 , b1951 + 31 , b1055 - 58 , and geminga .
in addition , soft @xmath0-rays have been seen from the pulsar b1509 - 58 , although no pulse was detected in the egret energy range .
a few other pulsars have been marginally identified .
all strongly - detected @xmath0-ray pulsars are double - pulsed , although the light curves exhibit a wide variety of patterns . unlike blazars ,
the flux history of pulsars show relatively constant emission over time .
the @xmath0-ray spectra of these sources have power - law indices in the range -1.39 to -2.07 .
geminga is an interesting pulsar , as it is radio - quiet , and its period was originally derived from x - ray measurements .
an up - to - date review on egret pulsars and how they impact models of pulsar @xmath0-ray emission may be found in @xcite .
one of the most exciting results in high energy astrophysics is undoubtedly the detection of @xmath0-ray bursts ( grb ) .
more than 2500 bursts have been recorded by batse , but the detection of 5 of these bursts by egret has demonstrated that grb are not necessarily a low - energy @xmath0-ray phenomena . a review on the high energy emission from grb is presented elsewhere in these proceedings @xcite .
egret - detected bursts are described in @xcite .
egret observations of the @xmath0-ray sky above 30 mev have provided a unique view of the diffuse @xmath0-ray radiation , which constitutes the bulk of the emission detected by egret .
the diffuse emission is found to have a galactic component arising from cosmic - ray interactions with the local interstellar gas and radiation @xcite , as well as an almost uniformly distributed component believed to originate outside the galaxy @xcite .
the average spectrum of the extragalactic diffuse emission is well - described by a single power - law with an index @xmath12 in the 30 mev to 100 gev range .
the precise origin of the extragalactic diffuse emission is not well - known , and possibly includes both diffuse origin as well as contributions from unresolved point sources .
these are discussed in @xcite .
recent theories include contributions by the upscattering of cosmic microwave background radiation by highly relativistic electrons , formed as a result of gravity - induced shock waves during the formation of large scale structure in the intergalactic medium @xcite .
models based on discrete source contributions have considered an array of different source classes . of these
the most popular have been unresolved blazars .
recent estimates of blazar contribution based on a @xmath0-ray evolution function derived from @xmath0-ray data is roughly 25 to 30 percent @xcite , although earlier estimations have found a larger contribution @xcite .
at very high energies satellite experiments are made impractical due to the rapidly falling photon fluxes . however , at these energies @xmath0-ray astronomy can be done using ground - based instruments by detecting the secondary particles or radiation produced as a result of the interaction of high energy @xmath0-rays in the upper atmosphere . at energies
@xmath14 gev , imaging atmospheric cherenkov telescopes ( acts ) have been used successfully to image the flashes of cherenkov light produced as a result of the electromagnetic cascades initiated by high energy @xmath0-rays .
the field was pioneered by the whipple 10-m telescope , which was the first to detect tev @xmath0-rays from an astrophysical source @xcite .
imaging acts detect the cherenkov flashes using fast photomultipliers ( pmts ) located at the focal plane of large optical reflectors .
the pmt signals are subsequently read out using a fast electronics chain . in comparison to satellite experiments ,
ground - based acts have much larger collection areas , and better angular resolution , although much smaller fields of view .
acts can only operate on cloudless , moonless nights , and thus have a lower duty cycle .
the whipple observatory was followed by several successful high resolution acts , e.g. , cat , cangaroo , durham , gt-48 , hegra , pachmari , shalon , tactic , telescope array , among others .
a complete list of current telescopes and their properties may be found in @xcite .
air shower arrays , which operate above 30 tev , and directly detect the secondary particles in the air showers are not included in the review ( see @xcite for a discussion ) .
figure 3 shows the skymap of point sources detected by ground - based experiments at energies @xmath15 gev .
in addition to those shown in the figure , a few other sources have been detected at a lower significance @xcite .
the source list consists of both galactic , as well extragalactic objects .
the crab nebula was the first source to be detected unambiguously in tev @xmath0-rays .
since then it has been detected several times with various different instruments and is clearly the standard candle in the 300 gev to 3 tev range @xcite .
no pulsed emission has been observed from the crab at tev energies .
in addition to the crab , two other plerionic supernova remnants ( snrs ) , vela and psr 1706 - 44 , and a couple of shell - type snrs have been detected by acts ( see @xcite for a review ) .
mrk 421 was the first extragalactic source to be detected at @xmath15 gev @xcite , and has been observed by several different ground - based instruments @xcite . at a redshift of 0.031 , it is the closest bl lac object seen by egret .
the source exhibits extremely rapid variability at tev energies with doubling times of the order of 1 hour or less @xcite .
its tev flux can vary from a quiescent level of 0.3 to @xmath16 crab flux during a @xmath0-ray flare .
mrk 421 exhibits the fastest time scale @xmath0-ray variability seen in any blazar to date .
figure 4 shows the spectacular flare from the source in 1996 .
cross - correlations of various data sets of mrk 421 indicate a significant correlation of the x - ray and tev @xmath0-ray flux variability @xcite . at a redshift of 0.033 ,
mrk 501 is the second closest x - ray - selected bl lac ( xbl ) known and was the second blazar discovered at tev energies by the whipple group @xcite .
it has since been observed by hegra , cat , tactic , and the telescope array project @xcite .
the detection of mrk 501 in the egret data came only after the tev detection @xcite , making it the first object to be discovered as a @xmath0-ray source from the ground .
the broadband spectrum of mrk 501 during the outburst of 1996 may is shown in fig .
egret reported a gev outburst from the source when mrk 501 was detected at 5.3 @xmath17 above 500 gev @xcite .
mrk 501 also shows significant variability at x - ray , low - energy @xmath0-ray and tev energies .
this source was detected for the first time in the 100 kev to 1 mev range by osse , the resulting sed showing this emission to be most likely of synchrotron origin @xcite .
this represents the highest synchrotron cutoff seen in any blazar at present . at tev energies ,
the source exhibited an extraordinary outburst in 1997 @xcite , when its intensity was observed to be several times that of the crab .
a comprehensive review of the other sources detected at tev energies is given in @xcite .
a comparison of the gev and tev skymaps show that only a handful of the egret sources have been detected at tev energies , although a simple extrapolation of the egret spectra predict flux levels well above the sensitivity of current atmospheric cherenkov telescopes .
the cutoffs in the spectra could be either due to intergalactic absorption or intrinsic absorption at the source .
@xmath0-rays traversing intergalactic distances may be absorbed by photon - photon pair production ( @xmath18 ) , with a tev photon preferentially interacting with 0.5 ev ( ir ) background photon .
the presence of this extragalactic background light ( ebl ) is probably the main explanation for the lack of detection of the majority of egret sources at tev energies .
the spectral cutoffs of the sources is expected to be in the 50 to 250 gev range , and is one of the primary motivations for exploring this regime .
the study of the ebl is important for several reasons .
direct measurements of the ebl are sparsely sampled and fraught with uncertainties due to the presence of foreground ir radiation @xcite .
the tev spectra of mrk 421 and mrk 501 have been used to indirectly derive limits on the ebl @xcite .
clearly , more spectral measurements of agn at different redshifts , particularly in the 50 to 250 gev range , are required in order to set restrictive limits on the ebl .
the future prospects of high energy astronomy are very promising .
recent results from space- and ground - based experiments have raised numerous important questions about the emission mechanisms in astrophysical sources .
there is also the tantalizing possibility of probing novel astrophysical phenomena which could arise as a result of new physics beyond the standard model @xcite . in the future , it is clear that one needs more sensitive instruments , both satellite - borne as well as ground - based .
it is particularly important to explore the energy region between 20 and 250 gev .
recently , two experiments stacee and celeste have demonstrated that lower energy thresholds can be achieved by using existing large arrays of solar heliostat mirrors to collect cherenkov light .
both these experiments have reported the detection of the crab nebula @xcite and are in the final stages of construction . a third experiment , solar-2 is currently being built @xcite . for the first time , the energy gap between space- and ground - based experiments is starting to be bridged .
experiments that will be built in the future include magic in spain , hess in namibia , veritas in arizona , and super - cangaroo in australia ( see @xcite and references therein ) .
the project grace in india will involve four independent experimental systems that will span nearly ten decades of photon energy ( @xmath19 kev - 100 tev ) and do coordinated studies of @xmath0-ray sources @xcite .
these experiments will complement each other and together span the energy range from 20 gev to 10 tev .
in addition to the acts , milagro , a new kind of @xmath0-ray detector with a large field - of - view and continuous operation has recently come on line .
a prototype of this detector , milagrito , reported the possible detection of the correlated tev @xmath0-ray burst grb 970417a @xcite , as well as mrk 501 @xcite .
the ground - based experiments will be complemented by two next generation space experiments , glast and agile .
glast , which is expected to be launched in 2005 , is projected to be a state - of - the art detector which will use a si - strip tracker and a csi calorimeter @xcite .
glast s sensitivity will be a factor of @xmath20 better than egret , and is expected to detect two orders of magnitude more sources than egret .
glast will have some overlap with ground - based instruments in the 30 - 300 gev regime .
a smaller egret - sized experiment , the italian agile @xcite , sensitive in 30 mev to 50 gev energy range , is expected to be launched in 2002 and bridge the gap before glast .
experiments in high energy @xmath0-rays will be complemented by neutrino and cosmic ray experiments @xcite , thus promising significant scientific progress in the future .
i wish to thank m. catanese , j. kataoka , r. ong , and p. sreekumar for providing some of the data and figures presented here , as well as at the neutrino 2000 conference , and d. hanna and b. dingus for reading the manuscript .
this research is supported in part by the national science foundation and by an award from research corporation .
r. c. hartman , et al . , apj , 123 ( 1999 ) 79 .
m. catanese & t. c. weekes , astro - ph 9906501 ( 1999 ) .
r. a. ong , phys .
, 305 ( 1998 ) 95 .
r. a. ong , astro - ph 0003014 ( 2000 ) . c. m. hoffman , et al . ,
phys . , 71 ( 1999 ) 897 .
n. gehrels & c. shrader , aip conf . proc . 410
( 1997 ) 3 .
d. j. thompson , et al .
, apjs , 102 ( 1995 ) 259 . b. n. swanenburg , et al . , apj , 243 ( 1981 ) l69 .
b. l. dingus , j. r. catelli , e. j. schneid , aip conf .
( 1998 ) 349 .
w. t. vestrand , p. sreekumar , & m. mori , apj , 483 ( 1997 ) l49 .
g. kanbach , et al . , a&as , 97 ( 1993 ) 349 .
yadigaroglu & r. w. romani , apj , 476 ( 1997 ) 356 . s. j. sturner & c. d. dermer , a&a , 281 ( 1995 ) l17 .
j. a. esposito , et al .
, apj , 461 ( 1996 ) 820 .
p. kaaret & j. cottam , apj , 462 ( 1998 ) l65 .
g. e. romero , et al .
, a&a , 348 ( 1999 ) 868 .
r. mukherjee , i. a. grenier , d. j. thompson , aip conf .
( 1997 ) 384 .
n. gehrels , et al . , nature , 404 ( 2000 ) 363 .
i. a. grenier , nature , 404 ( 2000 ) 344 .
r. c. hartman , et al .
, aip conf .
( 1997 ) 307 .
r. mukherjee , et al .
, apj , 527 ( 1999 ) 132 .
r. mukherjee , et al .
, apj , 490 ( 1997 ) 116 .
r. schlickeiser , sp .
, 75 ( 1996 ) 299 .
a. muecke , astro - ph 0004052 d. j. thompson , @xmath0 2000 proc .
, heidelberg .
e. waxman , neutrino 2000 proc .
, sudbury .
s. d. hunter , et al .
, apj , 481 ( 1997 ) 205 .
p. sreekumar , et al . ,
apj , 494 ( 1998 ) 523 .
a. loeb & e. waxman , nature , 405 ( 2000 ) 156 .
j. chiang & r. mukherjee , apj , 496 ( 1998 ) 752 .
f. w. stecker & m. h. salamon , apj , 464 ( 1996 ) 600 .
t. c. weekes , et al .
, apj , 342 ( 1989 ) 379 . t. c. weekes , et al .
, astro - ph/9910394 .
m. punch , et al . ,
nature , 358 ( 1992 ) 477 .
j. gaidos , et al . ,
nature , 383 ( 1996 ) 319 .
j. h. buckley , et al .
, apj , 472 ( 1996 ) l9 . j. quinn , et al . , apj , 456 ( 1996 ) l83 .
s. m. bradbury , et al .
, a&a , 320 ( 1997 ) l5 . m. punch , et al . , 25th icrc , 3 ( 1997 ) 253 . c. l. bhat , iauc ( 1997 ) 6709 .
n. hayashida , et al .
, apj , 504 ( 1998 ) l71 . j. kataoka , et al . ,
apj , 514 ( 1999 ) 138 .
p. sreekumar , et al .
, aip conf .
( 1999 ) 318 .
m. catanese , et al . , apj , 487 ( 1997 ) l143 .
f. aharonian , et al . , a&a , 349 ( 1999 ) 11 . m. g. hauser , et al .
, apj , 508 ( 1998 ) 25 .
puget , et al . , a&a , 308 ( 1996 ) l5 .
s. d. biller , et al .
, 80 ( 1998 ) 2992 .
t. stanev & a. franceschini , apj , 494 ( 1998 ) , l59 .
g. jungman , m. kamionkowski , & k. griest , phys . rep .
, 267 ( 1996 ) 195 .
s. oser , et al . , 26th icrc , 3 ( 1999 ) 464 . m. de naurois , 26th icrc , 5 ( 1999 ) 211 .
t. tumer , et al . ,
aph , 11 ( 1999 ) 271 . c. l. bhat , `` towards a major atmospheric cerenkov detector - v , '' kruger park , south africa ( 1997 ) .
g. sinnis , et al .
b ( proc . suppl . ) , 43 ( 1995 ) 141 .
r. atkins , et al . , apj , 533 ( 2000 ) l119 .
r. atkins , et al .
, apj , 525 ( 1999 ) l25 .
n. gehrels & p. michelson , aph , 526 ( 1999 ) 297 .
m. tavani , et al . ,
a&as , 138 ( 1999 ) 569 .
|
this article reviews the present status of high energy @xmath0-ray astronomy at energies above 30 mev .
observations in the past decade using both space- and ground - based experiments have been primarily responsible for giving a tremendous boost to our knowledge of the high energy universe .
high energy @xmath0-rays have been detected from a wide range of galactic and extragalactic astrophysical sources , such as @xmath0-ray bursters , pulsars , and active galaxies .
these observations have established high energy @xmath0-ray astronomy as a vital and exciting field , that has a bright future .
this review summarizes the experimental techniques , observations and results obtained with recent experiments , and concludes with a short description of future prospects .
|
we would like to thank w. vogelsang for valuable discussions .
the work of j.k . was supported by the grant - in - aid for scientific research no .
the work of k.t . was supported by the grant - in - aid for scientific research no .
|
we calculate qcd corrections to transversely polarized drell - yan process at a measured @xmath0 of the produced lepton pair in the dimensional regularization scheme . the @xmath0 distribution is discussed resumming soft gluon effects relevant for small @xmath0 .
hard processes with polarized nucleon beams enable us to study spin - dependent dynamics of qcd and the spin structure of nucleon .
the helicity distribution @xmath1 of quarks within nucleon has been measured in polarized dis experiments , and @xmath2 of gluons has also been estimated from the scaling violations of them .
on the other hand , the transversity distribution @xmath3 , i.e. the distribution of transversely polarized quarks inside transversely polarized nucleon , can not be measured in inclusive dis due to its chiral - odd nature,@xcite and remains as the last unknown distribution at the leading twist .
transversely polarized drell - yan ( tdy ) process is one of the processes where the transversity distribution can be measured , and has been undertaken at rhic - spin experiment .
we compute the 1-loop qcd corrections to tdy at a measured @xmath0 and azimuthal angle @xmath4 of the produced lepton in the dimensional regularization scheme . for this purpose
, the phase space integration in @xmath5-dimension , separating out the relevant transverse degrees of freedom , is required to extract the @xmath6 part of the cross section characteristic of the spin asymmetry of tdy.@xcite the calculation is rather cumbersome compared with the corresponding calculation in unpolarized and longitudinally polarized cases , and has not been performed so far .
we obtain the nlo @xmath7 corrections to the tdy cross section in the @xmath8 scheme .
we also include soft gluon effects by all - order resummation of logarithmically enhanced contributions at small @xmath0 ( `` edge regions of the phase space '' ) up to next - to - leading logarithmic ( nll ) accuracy , and obtain the first complete result of the @xmath0 distribution for all regions of @xmath0 at nll level .
we first consider the nlo @xmath7 corrections to tdy : @xmath9 , where @xmath10 denote nucleons with momentum @xmath11 and transverse spin @xmath12 , and @xmath13 is the 4-momentum of dy pair . the spin dependent cross section @xmath14 is given as a convolution @xmath15 where @xmath16 is the factorization scale , and @xmath17\ ] ] is the product of transversity distributions of the two nucleons , and @xmath18 is the corresponding partonic cross section .
note that , at the leading twist level , the gluon does not contribute to the transversely polarized process due to its chiral odd nature .
we compute the one - loop corrections to @xmath18 , which involve the virtual gluon corrections and the real gluon emission contributions , e.g. , @xmath19 , with @xmath20 .
we regularize the infrared divergence in @xmath21 dimension , and employ naive anticommuting @xmath22 which is a usual prescription in the transverse spin channel.@xcite in the @xmath8 scheme , we eventually get,@xcite to nlo accuracy , @xmath23 , \label{cross section}\end{aligned}\ ] ] where @xmath24 with @xmath25 , @xmath26 is the rapidity of virtual photon , and @xmath4 is the azimuthal angle of one of the leptons with respect to the initial spin axis . for later convenience
, we have decomposed the cross section into the two parts : the function @xmath27 contains all terms that are singular as @xmath28 , while @xmath29 is of @xmath7 and finite at @xmath30 .
writing @xmath31 as the sum of the lo and nlo contributions , we have@xcite @xmath32 , and @xmath33 \nonumber\\ \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!+ & & \!\!\!\!\!\!\!\!\left ( \frac{1}{(q_t^2)_+ } + \delta ( q_t^2 ) \ln \frac{q^2}{\mu_f^2 } \right)\!\!\ ! \left [ \int^1_{x_1 ^ 0 } \frac{d z}{z } \delta p_{qq}^{(0 ) } ( z)\ \delta h \left ( \frac{x_1 ^ 0}{z } , x_2 ^ 0 ; \
\mu_f \right ) + ( x_1 ^ 0 \leftrightarrow x_2 ^ 0 ) \right ] \biggr\ } , \label{eq : x}\end{aligned}\ ] ] where @xmath34 are the relevant scaling variables with @xmath35 , and @xmath36 is the lo transverse splitting function.@xcite in ( [ eq : x ] ) , the terms involving @xmath37 come from the virtual gluon corrections , while the other terms represent the recoil effects due to the real gluon emissions . for the analytic expression of @xmath29 , see ref.@xcite .
eq .
( [ cross section ] ) gives the first nlo result in the @xmath8 scheme .
we note that there has been a similar nlo calculation of tdy cross section in massive gluon scheme.@xcite we also note that , integrating ( [ cross section ] ) over @xmath0 , our result coincides with the corresponding @xmath0-integrated cross sections obtained in previous works employing massive gluon scheme@xcite and dimensional reduction scheme,@xcite via the scheme transformation relation.@xcite the cross section ( [ cross section ] ) becomes very large when @xmath38 , due to the terms behaving @xmath39 and @xmath40 in the singular part @xmath27 .
it is well - known that , in unpolarized and longitudinally polarized dy , large `` recoil logs '' of similar nature appear in each order of perturbation theory as @xmath41 , @xmath42 , and so on , corresponding to ll , nll , and higher contributions , respectively , and that the resummation of those `` double logarithms '' to all orders is necessary to obtain a well - defined , finite prediction of the cross section.@xcite because the ll and nll contributions are universal,@xcite we can work out the all - order resummation of the corresponding logarithmically enhanced contributions in ( [ cross section ] ) up to the nll accuracy , based on the general formulation@xcite of the @xmath0 resummation .
this can be conveniently carried out in the impact parameter @xmath43 space , conjugate to the @xmath0 space . as a result
, the singular part @xmath27 of ( [ cross section ] ) is modified into the corresponding resummed part , which is expressed as the fourier transform , @xmath44 here @xmath45 , and the large logarithmic corrections are resummed into the sudakov factor @xmath46 with @xmath47 .
the functions @xmath48 , @xmath49 as well as the coefficient functions @xmath50 are calculable in perturbation theory , and at the present accuracy of nll , we get:@xcite @xmath51 , @xmath52 , @xmath53 . we have utilized a relation@xcite between @xmath48 and the dglap kernels in order to obtain the two - loop term of @xmath48 .
the other contributions have been determined so that the expansion of the above formula ( [ resum ] ) in powers of @xmath54 reproduces @xmath27 of ( [ cross section ] ) , ( [ eq : x ] ) to the nlo accuracy .
eq .
( [ cross section ] ) with ( [ resum ] ) presents the first result of the nll @xmath0 resummation formula for tdy .
the nlo parton distributions in the @xmath8 scheme have to be used .
one more step is necessary to make the qcd prediction of tdy . similarly to other all - order resuumation formula
, our result ( [ resum ] ) is suffered from the ir renormalons due to the landau pole at @xmath55 in the sudakov factor , and it is necessary to specify a prescription to avoid this singularity . here
we deform the integration contour in ( [ resum ] ) in the complex @xmath43 space , following the method introduced in the joint resummation.@xcite obviously prescription to define the @xmath43 integration is not unique reflecting ir renormalon ambiguity , e.g. , `` @xmath56 prescription '' to `` freeze '' effectively the @xmath43 integration along the real axis is frequently used.@xcite the renormalon ambiguity should be eventually compensated in the physical quantity by the power corrections @xmath57 ( @xmath58 ) due to non - perturbative effects .
correspondingly , we make the replacement @xmath59 in ( [ resum ] ) with the `` minimal '' ansatz for non - perturbative effects , @xcite @xmath60 with a non - perturbative parameter @xmath61 .
fig.1 shows the @xmath0 distribution of tdy at @xmath62 gev , @xmath63 gev , @xmath64 , and with a model for the transversity @xmath3 that saturates the soffer s inequality at a low scale.@xcite solid line shows the nlo result using ( [ cross section ] ) , and the dashed and dot - dashed lines show the nll result using ( [ cross section ] ) , ( [ resum ] ) , @xmath60 , with @xmath65 gev@xmath66 and @xmath67 , respectively .
|
it is of great importance to understand semileptonic decays of kaons ( @xmath0 ) , since it plays a significant role in determining the ckm matrix element @xmath4 precisely @xcite .
furthermore , it provides a simple phenomenological basis for testing the breaking of flavor su(3 ) symmetry : in exact flavor su(3 ) symmetry , the kaon semileptonic form factor @xmath5 becomes unity .
the ademollo - gatto @xcite theorem asserts that of second order are the corrections of flavor su(3 ) symmetry breaking to the form factors of vector currents at zero momentum transfer ( @xmath6 ) . however ,
when goldstone bosons are involved , the ademollo - gatto theorem must be modified : while the singlet part of flavor su(3 ) symmetry breaking preserves flavor su(3 ) symmetry , it breaks chiral su(3)@xmath7su(3 ) symmetry .
langacker and pagels showed that the corrections of flavor su(3 ) symmetry breaking appear to first order due to the presence of that singlet part @xcite .
the effect of flavor su(3 ) symmetry breaking on the kaon semileptonic decay form factor is known to be around @xmath8 , which is rather small . the well - known soft - pion callan - treiman @xcite theorem connects the ratio of the pion and kaon decay constants to the semileptonic form factors of the kaon at @xmath9 ( callan - treiman point ) .
any chiral quark model should satisfy the callan - treiman theorem with flavor su(3 ) symmetry breaking . experimentally , there are a certain amount of data to judge theoretical calculations @xcite .
thus , the kaon semileptonic decay form factor provides a basis to examine the validity and reliability of any theoretical theory and model for hadrons .
there has been a great number of theoretical work : chiral perturbation theory ( @xmath1pt ) @xcite , lattice qcd ( lqcd ) @xcite , a dyson - schwinger method @xcite , constituent quark models @xcite , and so on . in the present work
, we will investigate the @xmath0 form factor within the framework of the nonlocal chiral quark model ( @xmath1qm ) derived from the instanton vacuum .
we will consider the leading order in the large @xmath10 expansion and flavor su(3 ) symmetry breaking explicitly .
the meson - loop corrections , which are of @xmath11 order , are neglected .
the model has several virtues : all relevant qcd symmetries are satisfied within the model , and there are only two parameters : the average size of instantons ( @xmath2 fm ) and average inter - instanton distance ( @xmath3 fm ) , which can be determined by the internal constraint such as the saddle - point equation @xcite .
these values for @xmath12 and @xmath13 have been supported in various lqcd simulations recently @xcite .
there is no further adjustable parameter in the model . as being discussed previously , since the effects of flavor su(3 ) symmetry
breaking are essential in the present work , we employ the modified low - energy effective partition function with flavor su(3 ) symmetry breaking @xcite .
this partition function extends the former one derived in the chiral limit @xcite .
it has been proven that the partition function with flavor su(3 ) symmetry breaking is very successful in describing the low - energy hadronic properties such as various qcd condensates , magnetic susceptibilities , meson distribution amplitudes , and so on @xcite .
however , the presence of the nonlocal interaction between quarks and pseudo - goldstone bosons breaks the ward - takahashi identity for nther currents .
since the kaon semileptonic decay form factors involve the vector current , we need to deal with this problem . while ref .
@xcite proposed a systematic way as to how the conservation of the nther current is restored , one has to handle the integral equation .
@xcite derived the light - quark partition function in the presence of the external gauge fields . with this gauged partition function
, it was shown that the low - energy theorem for the transition from two - photon state to the vacuum via the axial anomaly was satisfied @xcite . moreover , the magnetic susceptibility of the qcd vacuum and the meson distribution amplitudes were obtained sucessfully @xcite .
thus , in the present work , we will investigate the kaon semileptonic decay ( @xmath0 ) form factors , using the gauged low - energy effective partition function from the instanton vacuum with flavor su(3 ) symmetry breaking explicitly taken into account .
we sketch the present work as follows : in section ii , we briefly explain the general formalism relevant for studying the @xmath0 form factor . in section iii , we introduce the nonlocal chiral quark model from the instanton vacuum . in section iv ,
the numerical results are discussed , and are compared with those of other works .
the final section is devoted to summarize the present work and to draw conclusions .
in the present work , we are interested in the following kaon semileptonic decays ( @xmath0 ) in two different isospin channels : @xmath14 where @xmath15 and @xmath16 stand for the leptons ( either the electron or the muon ) and neutrinos . the relevant diagrams for the @xmath0 form factor are depicted in figure [ fig0 ] in which we define the momenta for the particles involved . ,
@xmath17 and @xmath18 , where @xmath19 , @xmath20 and @xmath21 stand for the loop - integral variable , initial kaon and vector - field momenta , respectively.,width=377 ] the nonlocal contributions in figure [ fig0](b ) and [ fig0](c ) arise from the gauged effective chiral action that will be discussed in section iii .
the decay amplitude ( @xmath22 ) can be expressed as follows @xcite : @xmath23,\ ] ] where @xmath24 is the well - known fermi constant : @xmath25 .
@xmath26 denotes the cabibbo angle .
we define , respectively , the weak leptonic and hadronic matrix elements ( @xmath27 and @xmath28 ) with the @xmath29 vector current ( @xmath30 ) as : @xmath31 where @xmath32 is the isospin factor , and set to be unity and @xmath33 for @xmath34 and @xmath35 , respectively .
the matrix @xmath36 denotes the combination of the two gell - mann matrices , @xmath37 , for the relevant flavor in the present problem .
the @xmath38 denotes the quark field .
the momentum transfer is defined as @xmath39 .
@xmath40 represent the vector form factors with the corresponding lepton @xmath15 ( @xmath41-wave projection ) .
alternatively , the form factor @xmath42 can be expressed in terms of the scalar form factor ( @xmath43 , @xmath44-wave projection ) and the vector form factor @xmath45 defined as follows : @xmath46.\ ] ] hence , the @xmath43 can be written as the linear combination of @xmath45 and @xmath47 : @xmath48f_{l-}(t).\ ] ] since the isospin breaking effects are almost negligible , we will consider only the @xmath49 decay channel .
input values for the numerical calculations are given as follows : @xmath50 mev and @xmath51 mev , respectively .
the up- and down - quark masses are taken as their average value : @xmath52 mev , while the strange - quark mass @xmath53 as around 150 mev .
it has been well - known that the experimental data for @xmath54 can be reproduced qualitatively well by the linear and quadratic fits @xcite : @xmath55 , \nonumber\\ { \rm quadratic}&:&\,f_{l+,0}(t ) = f_{l+,0}(0)\left[1+\frac{\lambda'_{l+,0}}{m^2_{\pi}}(t - m^2_l ) + \frac{\lambda''_{l+,0}}{2m^4_{\pi}}(t - m^2_l)^2\right ] , \end{aligned}\ ] ] where @xmath56 is the lepton mass .
the slope parameter @xmath57 is deeply related to the @xmath58 decay radius ( @xmath59 ) as follows @xcite : @xmath60 moreover , this radius can be expressed in terms of the gasser - leutwyler low - energy constant @xmath61 in the large @xmath10 limit @xcite : @xmath62 to obtain the decay rate @xmath63 , we use the convention defined in ref .
@xcite : @xmath64 this expression can be further simplified as a function of @xmath65 @xcite : @xmath66 \nonumber\\ & \simeq&\frac{g^2_f|v_{us}|^2}{24\pi^3}|\vec{p}_{\pi}|^3f^2_{e+}(t ) \,\,\,\,{\rm for}\,\,m_l = m_e\simeq0 , \end{aligned}\ ] ] where the three momentum of the pion @xmath67 is defined by : @xmath68^{\frac{1}{2}}. \end{aligned}\ ] ] this three momentum of the pion constrains the physically accessible region for the decay , i.e. : @xmath69
in this section , we show how to derive the hadronic matrix element given in eq .
( [ eq : def3 ] ) within the framework of the nonlocal @xmath1qm from the instanton vacuum .
we begin by the low - energy effective qcd partition function derived from the instanton vacuum @xcite : @xmath70 .
\label{effectiveaction } \end{aligned}\ ] ] @xmath71 is the dynamically generated quark mass being momentum - dependent , whereas @xmath72 stands for the current - quark mass with flavor @xmath73 . the nonlinear background pseudo - goldstone field @xmath74 is given by @xmath75 with the meson decay constants @xmath76 mev and @xmath77 mev fixed to the experimental data .
the meson field @xmath78 is defined as @xmath79 $ ] .
the octet pseudoscalar meson field @xmath80 is defined as follows : @xmath81 refs .
@xcite showed how to improve the low - energy effective qcd partition function in eq .
( [ effectiveaction ] ) by taking into account effects of flavor su(3 ) symmetry breaking , so that the dynamical quark mass acquires the contribution of the @xmath72 corrections : @xmath82 , \label{eq : dqm}\ ] ] where @xmath83 is the dynamical quark mass with zero momentum transfer in the chiral limit .
its value is determined by the saddle - point equation : @xmath84 mev .
@xmath85 is the momentum - dependent part which arises from the fourier transform of the fermionic zero - mode solutions in the instantons .
however , we will employ the simple - pole type parameterization for @xmath85 : @xmath86 which shows a very similar behavior to the original expression of @xmath85 . the value of @xmath87 in the square parenthesis of eq .
( [ eq : dqm ] ) can be computed within the model @xcite : @xmath88 as mentioned previously , the momentum - dependent dynamical quark mass @xmath71 breaks the conservation of the nther ( vector ) currents .
@xcite derived the light - quark partition function in the presence of the external vector field : @xmath89 .
\label{gpf } \end{aligned}\ ] ] the effective chiral action then becomes : @xmath90 , \end{aligned}\ ] ] where @xmath91 denotes the trace over space - time , flavor , and spin spaces . calculating the functional derivative of @xmath92 with the external vector field @xmath93
, we obtain the relevant operator expression for the @xmath94 semileptonic decay form factors : @xmath95_{v=0 } \nonumber\\ & = & -n_c{\rm tr}\left [ \frac{1}{i\rlap{/}{\partial}+im_f+i\sqrt{m_f(i\partial_{\mu } ) } u^{\gamma_5}_{fg}\sqrt{m_g(i\partial_{\mu } ) } } \gamma_\mu\lambda^{4-i5 } \right]_{v=0 } \cr & & + n_c{\rm tr}\left [ \frac{\left[\frac{\partial}{\partial v_\mu } \left(\sqrt{m_f(i\partial_{\mu})}\right ) u^{\gamma_5}_{fg}\sqrt{m_g(i\partial_{\mu } ) } -i\sqrt{m_g(i\partial_{\mu})}u^{\gamma_5}_{fg } \frac{\partial}{\partial v_\mu}\left ( \sqrt{m_g(i\partial_{\mu})}\right)\right]\lambda^{4-i5 } } { i\rlap{/}{\partial}+im_f+i\sqrt{m_f(i\partial_{\mu } ) } u^{\gamma_5}_{fg}\sqrt{m_g(i\partial_{\mu } ) } } \right]_{v=0}. \end{aligned}\ ] ] the first term in eq .
( [ eq : functional ] ) is usually called a _ local _ contribution , and the other two terms the _ nonlocal _ ones . since we are interested in the decay process with two on - mass shell pseudoscalar mesons , that is , the pion and kaon as shown in eq .
( [ eq : def3 ] ) , the local contribution can be written as follows : @xmath96^{k\pi}_{{\rm local},v=0 } \nonumber\\ & = & \frac{2n_c}{f_{\pi}f_k}{\rm tr } \left[\frac{\sqrt{m_f(i\partial_{\mu } ) } \gamma_5{\cal m}^a\lambda^a\sqrt{m_g(i\partial_{\mu } ) } } { { \cal d}(i\partial_{\mu } ) } \frac{\gamma^{\mu}\lambda^{4-i5}}{{\cal d}(i\partial_{\mu } ) } \frac{\sqrt{m_f(i\partial_{\mu } ) } \gamma_5{\cal m}^b\lambda^b \sqrt{m_g(i\partial_{\mu})}}{{\cal d}(i\partial_{\mu } ) } \right ] .
\end{aligned}\ ] ] the pseudoscalar meson field @xmath97 can be either the kaon or the pion , depending on flavor .
@xmath98 denotes the abbreviation for the quark - propagator : @xmath99.\ ] ] then , the corresponding matrix element can be obtained as follows : @xmath100^{k\pi}_{{\rm local},v=0}\bigg|\pi(p_{\pi } ) \bigg\rangle.\ ] ] the matrix element for the local contribution can be immediately expressed as @xmath101 \left[k^2_b+\overline{m}^2_s(k_b)\right ] \left[k^2_c+\overline{m}^2_q(k_c)\right ] } \nonumber\\ & & \hspace{2cm}\times \bigg[\left[k_a\cdot{k_b}+\overline{m}_q(k_a)\overline{m}_s(k_b)\right ] k_{c\mu}-\left[k_b\cdot{k_c}+\overline{m}_s(k_b)\overline{m}_q(k_c)\right]k_{a\mu } \nonumber\\ & & \hspace{3cm}+ \left[k_a\cdot{k_c}+\overline{m}_q(k_a)\overline{m}_q(k_c)\right]k_{b\mu}\bigg ] , \end{aligned}\ ] ] where @xmath102 .
the relevant momenta are defined as @xmath103 , @xmath17 and @xmath18 , in which @xmath19 , @xmath20 and @xmath21 denote the internal quark , initial kaon , and transfered momenta , respectively , as depicted in figure [ fig0 ] .
the trace @xmath104 runs over dirac spin space .
similarly , we can evaluate the nonlocal contributions as follows @xcite : @xmath105 \left[k^2_b+\overline{m}^2_s(k_b)\right ] \left[k^2_c+\overline{m}^2_q(k_c)\right ] } \nonumber\\ & & \times \left[\overline{m}_q(k_c)k_a\cdot k_b+\overline{m}_s(k_b)k_a\cdot k_c-\overline{m}_q(k_a)k_b\cdot k_c + \overline{m}_q(k_a)\overline{m}_s(k_b)\overline{m}_q(k_c ) \right ] -\left(b\leftrightarrow c\right ) , \nonumber\\ & & f^{\rm nonlocal(c)}_{\mu } = -\frac{4n_c}{f_{\pi}f_k}\int\frac{d^4k}{(2\pi)^4 } \frac { \sqrt{m_q(k_a)}\sqrt{m_s(k_b)}\sqrt{m_q(k_c)}_{\mu}\sqrt{m_q(k_a ) } \left[k_a\cdot k_b+\overline{m}_q(k_a)\overline{m}_s(k_b)\right ] } { \left[k^2_a+\overline{m}^2_q(k_a)\right]\left[k^2_b+\overline{m}^2_s(k_b)\right ] } \nonumber\\ & & + \frac{4n_c}{f_{\pi}f_k}\int\frac{d^4k}{(2\pi)^4 } \frac{\sqrt{m_q(k_a)}\sqrt{m_s(k_b)}\sqrt{m_q(k_c)}\sqrt{m_q(k_a)}_{\mu } \left[k_a\cdot k_b+\overline{m}_q(k_a)\overline{m}_s(k_b)\right ] } { \left[k^2_a+\overline{m}^2_q(k_a)\right]\left[k^2_b+\overline{m}^2_s(k_b)\right ] } \nonumber\\ & & \hspace{4cm}+\left(b\leftrightarrow c\right ) , \end{aligned}\ ] ] where @xmath106 .
the local ( a ) , nonlocal ( b ) and nonlocal ( c ) contributions correspond to the diagrams ( a ) , ( b ) and ( c ) in figure [ fig0 ] , respectively .
we now discuss various numerical results for the kaon semileptonic decay ( @xmath0 ) form factors in the present work .
we facilitate the breit - momentum framework for the calculation , since we are free to choose an arbitrary momentum framework because of the lorentz invariance .
the relevant momenta for the calculation are defined as follows ( @xmath107 ) : @xmath108 we first consider the case of @xmath109 . since the electron mass is negligible in comparison to those of the pion and the kaon , it can be set to be zero . in the left panel of figure [ fig1 ] , we draw the numerical results for @xmath110 ( solid ) , @xmath111 ( dotted ) and @xmath112 ( dashed ) within the physically accessible regions constrained by eq .
( [ eq : ar ] ) .
note that the scalar form factor @xmath112 is derived by using eq . ( [ eq : swave ] ) . [
cols="^,^ " , ]
in the present work , we have investigated the kaon semileptonic decay ( @xmath0 ) form factors within the framework of the gauged nonlocal chiral quark model from the instanton vacuum .
the effect of flavor su(3 ) symmetry breaking were taken into account .
we calculated the vector form factors ( @xmath113 ) , scalar form factor ( @xmath114 ) , slope parameters ( @xmath115 ) , decay width ( @xmath116 ) , etc as demonstrated in table [ table0 ] .
we found that the present results of the kaon semileptonic decay form factors are in a qualitatively good agreement with experiments .
we emphasize that there were no adjustable free parameters in the present investigation .
all results were obtained with only two parameters from the instanton vacuum , i.e. the average instanton size ( @xmath117 fm ) and inter - instanton distance ( @xmath3 fm ) . in the present investigation ,
we have considered only the leading - order contributions in the large @xmath10 limit .
while these contributions reproduce the observables relevant for kaon semileptonic decay in general , it seems necessary to take into account the meson - loop @xmath11 corrections in order to reproduce quantitatively the kaon decay constant @xmath118 and the low - energy constant @xmath119 . as noticed already in refs .
@xcite , these meson - loop corrections can play an important role in producing the kaon properties .
the related works are under progress .
the present work is supported by the korea research foundation grant funded by the korean government(moehrd ) ( krf-2006 - 312-c00507 ) .
the work of s.i.n . is supported by the brain korea 21 ( bk21 ) project in center of excellency for developing physics researchers of pusan national university , korea .
s.i.n . would like to appreciate the fruitful comments from m. khlopov .
99 n. cabibbo , phys .
* 10 * , 531 ( 1963 ) . h. leutwyler and m. roos , z. phys .
c * 25 * , 91 ( 1984 ) .
m. ademollo and r. gatto , phys .
* 13 * , 264 ( 1964 ) .
p. langacker and h. pagels , phys .
* 30 * , 630 ( 1973 ) . j. gasser and h. leutwyler , nucl .
b * 250 * , 517 ( 1985 ) .
m. bae and d.t .
cornwell , phys .
d * 10 * , 2694 ( 1974 ) . c. g. callan and s. b. treiman , phys .
* 16 * , 153 ( 1966 ) .
w. m. yao _ et al .
_ [ particle data group ] , j. phys .
g * 33 * , 1 ( 2006 ) .
l. m. chounet , j. m. gaillard and m. k. gillard , phys .
rept . * 4 * , 199 ( 1972 ) .
j. bijnens , g. colangelo , g. ecker and j. gasser , arxiv : hep - ph/9411311 .
n. tsutsui _ et al .
_ [ jlqcd collaboration ] , pos * lat2005 * , 357 ( 2006 ) .
d. becirevic _ et al .
_ , nucl .
b * 705 * , 339 ( 2005 ) . c. r. ji and p. maris , phys .
d * 64 * , 014032 ( 2001 ) .
kalinovsky , k. l. mitchell and c. d. roberts , phys . lett .
b * 399 * , 22 ( 1997 ) .
n. isgur , phys .
d * 12 * , 3666 ( 1975 ) .
h. m. choi and c. r. ji , phys .
d * 59 * , 034001 ( 1999 ) .
h. m. choi and c. r. ji , phys .
b * 460 * , 461 ( 1999 ) .
e. v. shuryak , nucl .
b * 203 * , 93 ( 1982 ) .
d. diakonov and v. y. petrov , nucl .
b * 245 * , 259 ( 1984 ) .
d. diakonov and v. y. petrov , nucl .
b * 272 * , 457 ( 1986 ) .
m. c. chu , j. m. grandy , s. huang and j. w. negele , phys .
d * 49 * , 6039 ( 1994 ) .
j. w. negele , nucl .
proc . suppl .
* 73 * , 92 ( 1999 ) .
t. degrand , phys .
d * 64 * , 094508 ( 2001 ) .
m. musakhanov , eur .
j. c * 9 * , 235 ( 1999 ) .
m. musakhanov , arxiv : hep - ph/0104163 .
m. musakhanov , nucl . phys .
a * 699 * , 340 ( 2002 ) .
s. i. nam and h. -ch .
kim , arxiv : hep - ph/0605041 , accepted for publication in plb .
h. y. ryu , s. i. nam and h. -ch .
kim , arxiv : hep - ph/0610348 , submitted to prd .
m. m. musakhanov and h. -ch .
kim , phys .
b * 572 * , 181 ( 2003 ) .
s. i. nam and h. -ch .
kim , phys .
d * 74 * , 076005 ( 2006 ) .
kim , m. musakhanov and m. siddikov , phys .
b * 608 * , 95 ( 2005 ) .
p. v. pobylitsa , phys .
b * 226 * , 387 ( 1989 ) .
h. j. munczek and d. w. mckay , phys .
d * 39 * , 888 ( 1989 ) [ erratum - ibid .
d * 46 * , 5209 ( 1992 ) ] .
h. j. munczek and a. m. nemirovsky , phys .
d * 28 * , 181 ( 1983 ) .
d. w. mckay and h. j. munczek , phys .
d * 28 * , 187 ( 1983 ) .
d. w. mckay and h. j. munczek , phys .
d * 32 * , 266 ( 1985 ) .
d. atkinson , h. de groot and p. w. johnson , phys . rev .
d * 43 * , 218 ( 1991 ) .
r. d. bowler and m. c. birse , nucl .
a * 582 * , 655 ( 1995 ) .
a. apostolakis _ et al .
_ [ cplear collaboration ] , phys .
b * 473 * , 186 ( 2000 ) .
e. p. shabalin , sov .
j. nucl .
phys . * 51 * , 296 ( 1990 ) [ yad .
51 * , 464 ( 1990 ) ] . c. r. ji and h. m. choi , phys .
b * 513 * , 330 ( 2001 ) .
a. afanasev and w. w. buck , phys .
d * 55 * , 4380 ( 1997 ) .
g. ecker , prog .
phys . * 35 * , 1 ( 1995 ) .
b. w. lee , phys .
* 20 * , 617 ( 1968 ) .
h. -ch .
kim , m. m. musakhanov and m. siddikov , phys .
b * 633 * , 701 ( 2006 ) .
g. calderon and g. lopez castro , phys .
d * 65 * , 073032 ( 2002 ) .
s. i. nam , h. -ch .
kim , a. hosaka and m. m. musakhanov , phys .
d * 74 * , 014019 ( 2006 ) .
j. gasser and h. leutwyler , nucl .
b * 250 * , 465 ( 1985 ) .
d. scora and n. isgur , phys .
d * 52 * , 2783 ( 1995 ) .
v. cirigliano , m. knecht , h. neufeld , h. rupertsberger and p. talavera , eur .
j. c * 23 * , 121 ( 2002 ) .
v. cirigliano , h. neufeld and h. pichl , eur .
j. c * 35 * , 53 ( 2004 ) .
d. becirevic _ et al .
_ , eur .
phys . j. a * 24s1 * , 69 ( 2005 ) .
a. o. barut and k. c. tripathy , phys .
rev . * 178 * , 2278 ( 1969 ) .
|
we investigate the kaon semileptonic decay ( @xmath0 ) form factors within the framework of the nonlocal chiral quark model ( @xmath1qm ) from the instanton vacuum , taking into account the effects of flavor su(3 ) symmetry breaking .
we also consider the problem of gauge invariance arising from the momentum - dependent quark mass in the present work .
all theoretical calculations are carried out without any adjustable parameter , the average instanton size ( @xmath2 fm ) and the inter - instanton distance ( @xmath3 fm ) having been fixed .
we also show that the present results satisfy the callan - treiman low - energy theorem as well as the ademollo - gatto theorem . using the @xmath0 form factors , we evaluate relevant physical quantities .
it turns out that the effects of flavor su(3 ) symmetry breaking are essential in reproducing the kaon semileptonic form factors .
the present results are in a good agreement with experiments , and are compatible with other model calculations .
|
the correspondence between geometric spaces and commutative algebras is a familiar and basic idea of algebraic geometry .
noncommutative topology started with the famous gelfand - naimark theorems : every commutative c*-algebra is the algebra of continuous functions vanishing at infinity on a locally compact space and vice versa . in particular , a noncommutative c*-algebra may be viewed as `` the algebra of continuous functions vanishing at infinity '' on a `` quantum space '' .
the aim of noncommutative geometry is to develop the basic concepts of topology , measure theory and differential geometry in algebraic terms and then to generalize the corresponding classical results to the setting of noncommutative algebras .
the question whether there is a way to translate the geometric concept of a fibre bundle to noncommutative geometry is quite interesting in this context . in the case of vector bundles a refined version of the theorem of serre and swan @xcite gives the essential clue : the category of vector bundles over a manifold @xmath4 is equivalent to the category of finitely generated projective modules over @xmath5 .
it is therefore reasonable to consider finitely generated projective modules over an arbitrary algebra as noncommutative vector bundles " .
the case of principal bundles is so far not treated in the same satisfactory way . from a geometrical point of view
it is not sufficiently well understood what a noncommutative principal bundle " should be .
still , there are several approaches towards the noncommutative geometry of principal bundles : for example , there is a well - developed abstract algebraic approach known as hopf galois extensions which uses the theory of hopf algebras ( cf . @xcite or ( * ? ? ?
* chapter vii ) ) .
another topologically oriented approach can be found in @xcite ; here the authors use @xmath6-algebraic methods to develop a theory of principal noncommutative torus bundles based on theorem ( cf .
* corollary 15 ) ) .
furthermore , the authors of @xcite introduce @xmath6-algebraic analogs of freeness and properness , since by a classical result ( of differential geometry ) having a free and proper action of a lie group @xmath7 on a manifold @xmath8 is equivalent saying that @xmath8 carries the structure of a principal bundle with structure group @xmath7 . in @xcite
we have developed a geometrically oriented approach to the noncommutative geometry of principal bundles based on dynamical systems and the representation theory of the corresponding transformation groups .
+ as is well - known from classical differential geometry , the relation between locally and globally defined objects is important for many constructions and applications . for example
, a principal bundle @xmath9 can be considered as a geometric object that is glued together from local pieces which are trivial , i.e. , which are of the form @xmath10 for some open subset @xmath11 of @xmath4 .
thus , a natural step towards a geometrically oriented theory of noncommutative principal torus bundles " is to describe the _ trivial _ objects first , i.e. , to determine and to classify the trivial noncommutative principal torus bundles .
this was done in @xcite : [ trivial ncp t^n - bundles ] ( trivial noncommutative principal torus bundles ) .
a ( smooth ) dynamical system @xmath1 is called a ( _ smooth _ ) _ trivial noncommutative principal @xmath0-bundle _ , if each isotypic component @xmath12 , @xmath13 , contains an invertible element .
this definition is inspired by the following observation : a principal @xmath0-bundle @xmath14 is trivial if and only if it admits a trivialization map .
such a trivialization map consists basically of @xmath3 smooth functions @xmath15 satisfying @xmath16 for all @xmath17 and @xmath18 . from an algebraical point of view
this condition means that each isotypic component of the ( naturally ) induced dynamical system @xmath19 contains an invertible element .
conversely , each trivial noncommutative principal @xmath0-bundle of the form @xmath19 induces a trivial principal @xmath0-bundle of the form @xmath20 .
the crucial point here is to verify the freeness of the induced action of @xmath0 on @xmath8 .
an important class of examples , which will also show up in this paper , is provided by the so - called noncommutative tori : [ nc n - tori as ncptb ] ( a ) ( noncommutative @xmath3-tori ) let @xmath21 be a real skew - symmetric @xmath22 matrix .
noncommutative @xmath3-torus _
@xmath23 is the universal unital c*-algebra generated by unitaries @xmath24 with @xmath25 moreover , there is a continuous action @xmath26 of @xmath0 on @xmath23 by algebra automorphisms , which is on generators given by @xmath27 where @xmath28 . in particular
, @xmath29 shows that the triple @xmath30 is a trivial noncommutative principal @xmath0-bundle .
\(b ) the _ smooth noncommutative n - torus _
@xmath31 is the unital subalgebra of smooth vectors for the previous action .
its elements are given by ( norm - convergent ) sums @xmath32further , a deeper analysis shows that the induced action of @xmath0 on @xmath31 is smooth .
thus , the triple @xmath33 is a smooth trivial noncommutative principal @xmath0-bundle .
another hint for the quality of this definition for trivial noncommutative principal torus bundles is the observation that they have a natural counterpart in the theory of hopf galois extensions : in fact , up to a suitable completion , they correspond to the so - called cleft @xmath34$]-comodule algebras ; loosely speaking , being cleft is a triviality condition in the theory of hopf galois extensions : ( realtion to cleft hopf galois extensions ) .
an algebra @xmath2 is a @xmath34$]-comodule algebra if and only if @xmath2 is a @xmath35-graded algebra ( cf .
* lemma 4.8 ) ) .
moreover , we conclude from ( * ? ? ?
* example 2.1.4 ) that a @xmath35-graded algebra @xmath36 is a hopf
galois extension ( of @xmath37 ) if and only if @xmath2 is strongly graded , i.e. , @xmath38 for all @xmath39 . now , a short calculation shows that a @xmath34$]-comodule algebra @xmath2 is cleft if and only if each grading space @xmath12 contains an invertible element . for more background on hopf
galois extensions , in particular for the definition of cleft extensions , we refer to ( * ? ?
* section 2.2 ) or ( * ? ? ?
* appendix a ) . in view of the previous discussion ,
it is the next natural step is to work out a convenient localization method for non - commutative algebras or , more generally , for dynamical systems . for this
we first note that the idea of localization comes from algebraic geometry : given a point @xmath40 in some affine variety @xmath41 , one likes to investigate the nature of @xmath41 in an arbitrarily small neighbourhood of @xmath40 in the zariski topology .
now , small neighbourhoods of @xmath40 in @xmath41 correspond to large algebraic subsets @xmath42 .
for example , let @xmath42 be the zero set of some algebraic function @xmath43 on @xmath41 , which does not vanish at @xmath40 .
then the affine ring @xmath44 $ ] is obtained from @xmath45 $ ] by adjoining a multiplicative inverse for @xmath43 , i.e. , taking the coproduct of @xmath45 $ ] with the free ring in one generator and dividing out the ideal @xmath46 ; this is called _ inverting _ @xmath43 .
[ k(x)_f = o(x_f ) ] _ _ ( _ _ cf .
* chapter ii , proposition 2.2 ( b ) ) _ _ ) _ _ let @xmath41 be an affine variety and @xmath43 an element in the coordinate ring @xmath45 $ ] . if @xmath47 , then the map @xmath48_f\rightarrow \mathcal{o}_x(x_f),\,\,\,\frac{g}{f^n}\mapsto\left(x\mapsto\frac{g(x)}{f^n(x)}\right)\]]is an isomorphism of rings .
this construction is not valid for the algebra of smooth functions on some manifold @xmath4 . indeed ,
if @xmath49 and @xmath50 , then not every smooth function @xmath51 is of the form @xmath52 for @xmath53 and some @xmath54 .
this is due to the fact that there are too many " smooth functions
. another important remark in this context is that the natural restriction map @xmath55 , @xmath56 is in general neither injective nor surjective . + in section [ slns ] we present a construction which will fix this problem . to be more precise
, we present an appropriate method of localizing ( possibly noncommutative ) algebras in a smooth way , while in section 3 we are concerned with the problem of calculating the spectrum of such a localized algebra .
+ section 4 is devoted to discussing algebra bundles @xmath57 with a possibly infinite - dimensional fibre @xmath2 over a finite - dimensional manifold @xmath4 .
in particular , we show how to endow the corresponding space @xmath58 of sections with a topology that turns it into a ( unital ) locally convex algebra .
the general philosophy here is use the existing results for mapping spaces and reduce the occurring questions of continuity to mapping spaces .
+ in section 5 we finally prove a smooth analogue of proposition [ k(x)_f = o(x_f ) ] for the algebra @xmath5 of smooth functions on some manifold @xmath4 .
in fact , we even show that if @xmath2 is a unital frchet algebra , @xmath59 an algebra bundle and @xmath60 , then the smooth localization of @xmath58 with respect to @xmath43 is isomorphic ( as a unital frchet algebra ) to @xmath61 . here , .
this result means that it is possible to reconstruct @xmath62 by localization out of data from @xmath5 .
the crucial idea is to embed @xmath11 in an appropriated way as a closed submanifold into @xmath63 and , of course , the methods we use come from differential geometry , like the regular value theorem ( also known as the submersion theorem ) .
thus , it is not straightforward if similar results also hold in the context of @xmath6-algebras .
+ section 6 is dedicated to describing the spectrum of the algebra of sections of an algebra bundle with finite - dimensional commutative fibre .
we obtain a beautiful result which connects algebra with geometry : _ let @xmath2 be a finite - dimensional unital commutative algebra .
further , let @xmath59 be an algebra bundle .
if @xmath64 , then the spectrum @xmath65 is an @xmath3-fold covering of @xmath4 . _ in section [ acncpt^nb ] we use the ideas of sections [ slns ] to introduce a method of localizing dynamical systems @xmath66 with respect to elements of @xmath67 , i.e. , with respect to elements of the commutative fixed point algebra of the induced action of @xmath7 on the center @xmath68 of @xmath2 .
in fact , this construction turns out to be the starting point for our approach to noncommutative principal torus bundles : given a principal bundle @xmath9 and an open subset @xmath11 of @xmath4 such that @xmath69 is trivial , we show that the localization of the induced smooth dynamical system @xmath70 around `` @xmath11 '' leads to the smooth dynamical system @xmath71 which is in turn nautrally isomorphic to the smooth dynamical system @xmath72 coming from the trivial principal bundle @xmath73 .
+ the main goal of section [ a geometric approach to ncpb ] is to present a geometrically oriented approach to the noncommutative geometry of principal torus bundles . as already mentioned a trivial noncommutative principal torus bundle
is a dynamical system @xmath1 with the additional property that each isotypic component contains an invertible element . in view of the previous discussion ,
our main idea is inspired by the classical setting : loosely speaking , a dynamical system @xmath1 is called a noncommutative principal @xmath0-bundle , if it is locally " a trivial noncommutative principal @xmath0-bundle , i.e. , section [ acncpt^nb ] enters the picture .
we prove that this approach extends the classical theory of principal torus bundles and a present some noncommutative examples .
indeed , we first show that each trivial noncommutative principal torus bundle carries the structure of a noncommutative principal torus bundle in its own right .
we further show that examples are provided by sections of algebra bundles with trivial noncommutative principal torus bundle as fibre , sections of algebra bundles which are pull - backs of principal torus bundles and sections of trivial equivariant algebra bundles . at the end of this section we present a very concrete example . +
another advantage of our approach is that it seems to have a nice classification theory . indeed , in ( * ? ? ?
* section 4 ) we gave a complete classification of trivial noncommutative principal torus bundles ( up to a suitable completion ) .
thus it is , of course , a natural ambition to work out a classification theory for ( non - trivial ) noncommutative principal torus bundles .
the final section is therefore devoted to hints and ideas for an appropriate classification theory for ( non - trivial ) noncommutative principal torus bundles .
+ appendix a is devoted to some results concerning the projective tensor product of locally convex spaces . in appendix
b we discuss some useful results on the smooth exponential law which will be used several times within this paper .
all manifolds appearing in this paper are assumed to be finite - dimensional , paracompact , second countable and smooth .
for the necessary background on ( principal ) bundles and vector bundles we refer to @xcite .
all algebras are assumed to be complex if not mentioned otherwise . given an algebra @xmath2 , we write @xmath74 ( with the topology of pointwise convergence on @xmath2 ) for the spectrum of @xmath2 and @xmath75 for the corresponding group of automorphisms in the category of @xmath2
. moreover , a dynamical system is a triple @xmath66 , consisting of a unital locally convex algebra @xmath2 , a topological group @xmath7 and a group homomorphism @xmath76 , which induces a continuous action of @xmath7 on @xmath2 .
we thank k .- h . neeb and helge glckner for fruitful and inspiring discussions .
moreover , we thank christoph zellner for proofreading of parts of this paper .
from the viewpoint of noncommutative differential geometry , one might ask whether proposition [ k(x)_f = o(x_f ) ] is still true for the algebra of smooth function on a manifold @xmath4 . unfortunately
, if @xmath43 is an element in @xmath5 and @xmath50 , then not every smooth function @xmath77 is of the form @xmath52 for @xmath53 and some @xmath54 . in this section
we provide a construction which will fix this problem . for the following propositions we recall the smooth compact open topology for smooth vector - valued function spaces of definition [ smooth compact open topology ] and the projective tensor product topology of definition [ proj .
tensor product iii ] .
[ c(m , a ) loc .
con . algebra ] if @xmath4 is a manifold and @xmath2 a locally convex algebra , then @xmath78 is a locally convex algebra . to prove the claim we just have to verify that the multiplication in @xmath78 is continuous : \(i ) for this we first note that the tangent space @xmath79 of @xmath2 carries a natural locally convex algebra structure given by the tangent functor @xmath80 , i.e. , defined by @xmath81if @xmath82 is the multiplication of @xmath2 , then @xmath83 is the multiplication of @xmath79 .
iterating this process , we obtain a locally convex algebra structure on @xmath84 for each @xmath54 .
\(ii ) for elements @xmath85 the functoriality of @xmath80 implies that @xmath86 in particular , we conclude that the embedding @xmath87is a morphism of algebras . by (
* proposition d.5.7 ) , the multiplication on the product algebra on the right is continuous , which implies the continuity of the multiplication on @xmath78 .
. tensor product iv ] if @xmath4 is a manifold and @xmath88 a complete locally convex space , then the locally convex space @xmath89 is complete ( cf .
* proposition 4.2.15 ( a ) ) ) .
moreover , a short observation shows that the subspace generated by the image of the bilinear map @xmath90 consists of functions whose image is contained in a finite dimensional subspace of @xmath88 .
writing @xmath91 for this subspace , we conclude from an example in [ gro55 ] , chapter 2 , 3.3 ,
theorem 13 that the map @xmath92is an isomorphism of locally convex spaces and can therefore be extended to an isomorphism @xmath93 of complete locally convex spaces .
the crucial point here is to use the fact that @xmath5 is a nuclear space .
note that this isomorphism means that the smooth compact open topology and the projective tensor product topology coincides on @xmath94 .
[ proj . tensor product for algebras ] if @xmath4 is a manifold and @xmath2 is a complete locally convex algebra , then there is a unique locally convex algebra structure on the locally convex space @xmath95 for which @xmath95 and @xmath78 are isomorphic as complete locally convex algebras .
according to proposition [ aob as l.c .
algebra ] , @xmath96 is a locally convex algebra .
thus , ( * ? ? ?
* corollary d.1.7 ) applied to the multiplication map of @xmath96 implies that @xmath95 is also a locally convex algebra . if @xmath97 denotes the multiplication map of @xmath95 and @xmath98 the multiplication map of the locally convex algebra @xmath78 ( cf .
proposition [ c(m , a ) loc .
con . algebra ] ) , then it remains to verify the identity @xmath99 where @xmath100 denotes the isomorphism of locally convex spaces stated in remark [ proj .
tensor product iv ] : in fact , an easy observation shows that ( [ verify identity ] ) holds on @xmath96 and thus the claim follows from the principle of extension of identities .
the next definition is crucial for the aim of this paper since it is the beginning of a smooth localization method : [ sm .
( smooth localization of algebras ) . for @xmath54 and
a unital locally convex algebra @xmath2 we write @xmath101for the unital locally convex algebra of smooth @xmath2-valued functions on @xmath102 ( cf .
proposition [ c(m , a ) loc .
con . algebra ] ) .
we further define for each @xmath103 and @xmath104 a smooth @xmath2-valued function on @xmath102 by @xmath105if @xmath106 , then we write @xmath107 for the closure of the two - sided ideal generated by these functions .
finally , we write @xmath108for the corresponding locally convex quotient algebra and @xmath109for the corresponding continuous quotient homomorphism .
the algebra @xmath110 is called the ( _ smooth _ ) _ localization _ of @xmath2 with respect to @xmath111 .
we continue with a bunch of remarks on the smooth localization of an algebra : for @xmath54 and a complete locally convex algebra @xmath2 , proposition [ proj .
tensor product for algebras ] , applied to @xmath112 , implies that @xmath113as complete locally convex algebras .
this result corresponds to the classical picture in commutative algebra of adjoining @xmath3 indeterminate elements to a ring @xmath114 by taking the coproduct of @xmath114 and the free @xmath115-algebra in @xmath3 generators : @xmath116\otimes r\cong r[t_1,\ldots , t_n].\]]in particular , the algebra @xmath117 may be thought of the outcome of adjoining @xmath3 indeterminates to the algebra @xmath2 in a smooth way . [
rem to sm .
] unlike in ordinary commutative algebra , where every element of the `` classical localization '' @xmath118 of @xmath2 with respect to @xmath104 can be written in the form @xmath119 for some @xmath120 and @xmath54 , we do , not even in the case @xmath121 , have an explicit description of the elements of @xmath110 . moreover , the outcome is , in general , quite different from the classical localization procedure ; for example , if one localizes an algebra @xmath2 with respect to a non - zero quasi - nilpotent element @xmath122 , i.e. , an element with @xmath123 which is not nilpotent , then @xmath124 since the function @xmath125 is invertible in @xmath126 .
on the other hand we have @xmath127 .
we will point out another difference in corollary [ c(m)_f = c(m_f ) ] .
nevertheless , by the universal property of the classical localization procedure , we have a canonical homomorphism from @xmath118 to @xmath128 .
[ inverting 1 and 0 ] let @xmath2 be an arbitrary ( possibly noncommutative ) unital algebra .
then localizing @xmath2 with respect to 0 leads to the zero - algebra , i.e. , @xmath129 and @xmath130 .
on the other hand , if @xmath122 in @xmath2 is an invertible element , then localization of @xmath2 with respect to @xmath122 changes nothing , i.e. , there is a canonical isomorphism between @xmath2 and @xmath118 .
this is not clear at all for the smooth localization @xmath128 .
[ a_z frechet again ] ( frchet algebras ) .
if @xmath2 is a unital frchet algebra , then the same holds for @xmath110 .
indeed , this assertion follows from the fact that the quotient of a frchet space by a closed subspace is again frchet ( cf .
@xcite ) .
[ inverting 1 ] let @xmath2 be a complete unital locally convex algebra .
then the map @xmath131is a surjective morphism of _ _ ( _ _ complete _ ) _ unital locally convex algebras with kernel @xmath132 \(i ) according to ( * ? ? ?
* proposition i.2 ) , the evaluation map @xmath133is smooth . in particular , the map @xmath134 is continuous .
further , a short observation shows that @xmath134 is surjective and a homomorphism of algebras .
therefore , @xmath134 is a surjective morphism of ( complete ) unital locally convex algebras .
\(ii ) it remains to determine the kernel @xmath135 : clearly , @xmath136 .
therefore , let @xmath137 with @xmath138 .
we define a smooth @xmath2-valued function on @xmath139 by @xmath140now , an easy calculation leads to @xmath141i.e . , to @xmath142 and
thus @xmath143 .
[ inverting 1 iso ] in the situation of proposition [ inverting 1 ] , the map @xmath144is an isomorphism of _ _ ( _ _ complete _ ) _ locally convex algebras . in view of proposition [ inverting 1 ] and the definition of the quotient topology , the map @xmath145 is a bijective morphism of locally convex algebras .
further , a short observation shows that the map @xmath146where @xmath147 denotes the canonical inclusion , is a continuous inverse of the map @xmath145 . in the forthcoming chapter we will need the following property of @xmath128 : [ a_a
... is a ( left ) a - modul ] for all @xmath106 there is a continuous @xmath2-bimodule structure on @xmath110 , given for all @xmath148 and @xmath149 by @xmath150.a':=[afa']=afa'+i_{a_1,\ldots , a_n}.\ ] ] the claim follows from the continuity of the @xmath2-bimodule structure on @xmath151 , given for all @xmath148 and @xmath149 by @xmath152and the definition of the quotient topology .
( another localization method ) .
another interesting localization method can be found in ( * ? ? ?
* chapter 3 ) . indeed , given a commutative @xmath139-algebra @xmath2 and an open subset @xmath11 of the _ real spectrum _
@xmath153 of @xmath2 ( with the topology of pointwise convergence on @xmath2 ) , the authors of @xcite define @xmath154 to be the ring of fractions of @xmath2 with respect to the multiplicative subset of all elements in @xmath2 without zeroes in @xmath11 , i.e. , elements in @xmath154 are ( equivalence classes of ) fractions @xmath155 , where @xmath156 and @xmath157 for all @xmath158 .
in particular , they show that if @xmath159 , then @xmath160 for each open subset @xmath11 of @xmath102 ( cf .
* chapter 3 , example 3.3 ) ) .
of course , an important restriction of this approach is the commutativity of the algebra @xmath2 .
before coming up with concrete examples of smoothly localized algebras , we will first be concerned with the problem of calculating the spectrum of @xmath110 for some unital locally convex algebra @xmath2 and elements @xmath106 . in the following we write @xmath161 for the set of continuous characters of @xmath2 .
we start with discussing the spectrum of the algebra of smooth functions on a manifold : [ spec of c(m , k ) set ] if @xmath4 is a manifold , then each character @xmath162 is an evaluation in some point @xmath163 .
a proof of this statement can be found in ( * ? ? ?
* corollary 4.3.2 ) .
the next proposition shows that the correspondence between @xmath4 and @xmath164 is actually a topological isomorphism : [ spec of c(m ) top ] let @xmath4 be a manifold .
then the map @xmath165 is a homeomorphism .
\(i ) the surjectivity of @xmath166 follows from lemma [ spec of c(m , k ) set ] . to show that @xmath166 is injective , choose elements @xmath167 of @xmath4 .
since @xmath4 is manifold , there exists a function @xmath43 in @xmath5 with @xmath168 .
then @xmath169implies that @xmath170 , i.e. , @xmath166 is injective .
\(ii ) next , we show that @xmath166 is continuous : let @xmath171 be a convergent sequence in @xmath4
. then we have @xmath172i.e .
, @xmath173 in the topology of pointwise convergence .
hence , @xmath166 is continuous .
\(iii ) we complete the proof by showing that @xmath166 is an open map : for this let @xmath11 be an open subset of @xmath4 , @xmath174 in @xmath11 and @xmath175 a smooth real - valued function with @xmath176 and @xmath177 .
since the map @xmath178is continuous , a short calculations shows that @xmath179 is a neighbourhood of @xmath174 containing the open subset @xmath180 .
hence , @xmath166 is open .
the following observation is well - known , but by a lack of a reference , we give the proof : [ spectrum of tensor products ] let @xmath2 and @xmath181 be two unital locally convex algebras .
then the map @xmath182is a homeomorphism .
\(i ) we first note that lemma [ continuity of maps eof to eof ] implies that the map @xmath166 is well - defined . to show that the map @xmath166 is bijective , we recall that each simple tensor @xmath183 can be written as @xmath184 .
therefore , every continuous character @xmath185 is uniquely determined by the elements of the form @xmath186 and @xmath187 for @xmath104 and @xmath188 .
in particular , the restriction of @xmath189 to the subalgebra @xmath190 , resp . , @xmath191 corresponds to a continuous character of @xmath2 , resp .
, @xmath181 , i.e. , @xmath192hence , @xmath166 is surjective .
a similar argument shows that the map @xmath166 is injective .
\(ii ) finally , we leave it as an easy exercise to the reader to verify the continuity of @xmath166 and its inverse @xmath193 .
[ cia stuff ] ( continuous inverse algebras ) .
( a ) a unital locally convex algebra @xmath2 is called continuous inverse algebra , or cia for short , if its group of units @xmath194 is an open subset in @xmath2 and the inversion map @xmath195 is continuous at @xmath196 .
they are encountered in k - theory and noncommutative geometry , usually as dense unital subalgebras of @xmath6-algebras .
\(b ) for a compact manifold @xmath4 , the frchet algebra of smooth functions @xmath5 is the prototype of such a continuous inverse algebra . more generally , if @xmath4 is a compact manifold and @xmath2 a continuous inverse algebra , then @xmath78 , equipped with the smooth compact open topology , is a continuous inverse algebra ( cf .
* proposition 7.1 ) ) .
the example @xmath197 shows that , if @xmath2 is a continuous inverse algebra , @xmath78 need not have an open unit group .
\(c ) if @xmath2 is a commutative cia , then each character @xmath198 is continuous . moreover , given a maximal proper ideal @xmath199 in @xmath2 , then @xmath199 is the kernel of some character @xmath198 .
a reference for the last two statements is ( * ? ? ?
* chapter 2 ) .
we are now ready to prove the following theorem on continuous characters of smooth vector - valued function spaces : [ spectrum of c(m , a ) ] let @xmath4 be a manifold and @xmath2 be a unital locally convex algebra . further let @xmath200 . then the following assertions hold : * if @xmath4 is compact and @xmath2 is a cia , then each maximal ideal of @xmath181 is closed . *
if @xmath2 is complete , then each continuous character @xmath201 is an evaluation homomorphism @xmath202for some @xmath163 and @xmath203 .
\(a ) we first note that remark [ cia stuff ] ( ii ) implies that @xmath181 is a cia .
hence , the unit group @xmath204 is an open subset of @xmath181 .
if @xmath205 is a maximal ideal , then @xmath199 intersects @xmath206 trivially , and since @xmath206 is open , the same holds for the closure @xmath207 .
hence , @xmath207 also is a proper ideal , so that the maximality of @xmath199 implies that @xmath199 is closed .
\(b ) according to proposition [ proj .
tensor product for algebras ] , we know that @xmath95 and @xmath78 are isomorphic as complete unital locally convex algebras . in particular , @xmath96 is a dense unital subalgebra of @xmath78 , and thus each ( continuous ) character @xmath201 restricts to a character on @xmath96 , which is , by lemma [ spectrum of tensor products ] , an evaluation in some point @xmath208 .
this character is continuous with respect to the projective tensor product topology and can therefore be uniquely extended to a continuous character on @xmath181 .
[ spectrum of c(m , a ) a cia ] let @xmath4 be a manifold and @xmath2 be a complete cia .
if @xmath200 , then each continuous character @xmath209 is an evaluation homomorphism @xmath202for some @xmath163 and @xmath210 . in particular ,
if @xmath4 is compact , then each character of @xmath181 is an evaluation homomorphism .
the first part of the corollary immediately follows from theorem [ spectrum of c(m , a ) ] ( b ) and remark [ cia stuff ] ( c ) , which states that every character of @xmath2 is continuous . if @xmath4 is additionally compact , then remark [ cia stuff ] ( b ) implies that @xmath181 is a ( complete ) cia .
therefore , the second assertion ( again ) follows from remark [ cia stuff ] ( c )
. it would be nice to find a purely algebraic proof of corollary [ spectrum of c(m , a ) a cia ] , which shows that , even in the non - compact case , every ( ! ) character of @xmath181 is an evaluation homomorphism . by lemma [ spec of c(m , k ) set ] this is , for example , true for @xmath211 and should also work for function algebras on compact spaces , i.e. , for commutative c*-algebras .
for further investigations on this question , we refer to the paper @xcite .
if @xmath4 is compact and @xmath2 a complete commutative cia , then each maximal proper ideal of @xmath200 is the kernel of an evaluation homomorphism @xmath202for some @xmath163 and @xmath210 . according to remark [ cia stuff ] ( b )
, @xmath181 is a commutative cia .
hence , remark [ cia stuff ] ( c ) implies that each maximal proper ideal of @xmath181 is the kernel of a character of @xmath181 .
therefore , the claim follows from corollary [ spectrum of c(m , a ) a cia ] .
the following theorem shows that , under a certain condition on the continuous spectrum of @xmath2 , the bijection of theorem [ spectrum of c(m , a ) ] ( b ) becomes an isomorphism of topological spaces : [ spec of c(m , a ) top ] let @xmath4 be a manifold and @xmath2 be a complete unital locally convex algebra .
further , let @xmath200 and assume that @xmath212 is locally equicontinuous . then the map @xmath213is a homeomorphism .
\(i ) since @xmath2 is unital , @xmath5 is a central subalgebra of @xmath181 .
moreover , we note that @xmath2 is embedded in @xmath181 as the constant - valued functions .
hence , @xmath214 implies that @xmath215 and @xmath216 , i.e. , @xmath166 is injective .
further , theorem [ spectrum of c(m , a ) ] ( b ) implies that the map @xmath166 is surjective .
\(ii ) next , we prove the continuity of the inverse map @xmath193 : for this we choose a convergent net @xmath217 in @xmath218 . because @xmath5 is a central subalgebra of @xmath181 , we conclude that @xmath219 in @xmath164 . by proposition [ spec of c(m )
top ] , the last condition is equivalent to @xmath220 in @xmath4 . since @xmath2 is embedded in @xmath181 as the constant - valued functions , we get @xmath221 in @xmath2 .
\(iii ) to prove continuity of @xmath166 , let @xmath222 , @xmath223 and @xmath224 .
we first choose an equicontinuous neighbourhood @xmath225 of @xmath226 in @xmath161 such that @xmath227next , we choose a neighbourhood @xmath228 of @xmath229 in @xmath2 such that @xmath230for all @xmath231 and @xmath232 . finally , we choose a neighbourhood @xmath11 of @xmath174 in @xmath4 such that @xmath233 for all @xmath234
. then @xmath235 implies @xmath233 and @xmath232 and therefore @xmath236it follows that @xmath237 for all @xmath235 .
[ equicont ] ( sources of algebras with equicontinuous spectrum ) .
( a ) the spectrum @xmath238 of each cia @xmath2 is equicontinuous .
in fact , let @xmath11 be a balanced @xmath239-neighbourhood such that @xmath240
. then @xmath241 ( cf .
the proof of ( * ? ? ?
* lemma 2.2.4 ) ) .
\(b ) moreover , if @xmath2 is a _
@xmath242-seminormed _ algebra , then ( * ? ?
* corollary 7.3.9 ) implies that @xmath161 is equicontinuous .
[ d(a ) ] for a unital algebra @xmath2 and @xmath104 we write @xmath243 for the set of characters which do not vanish on @xmath122 . moreover ,
if @xmath106 , then we define @xmath244 [ d(a ) is open ] each set of the form @xmath245 is an open subset of @xmath238 . by the definition of @xmath245 , it suffices to show that each set of the form @xmath246 is open .
therefore , let @xmath104 be arbitrary and note that the function @xmath247 , @xmath248 is continuous .
the claim now follows from @xmath249 .
next we want to describe the spectrum of @xmath128 for a complete locally convex algebra @xmath2 and an element @xmath104 .
we first need the following proposition : [ spectrum of quotient algebras ] let @xmath2 be a unital locally convex algebra and let @xmath199 be a closed two - sided ideal of @xmath2 .
further , let @xmath250 denote the quotient homomorphism .
then the map @xmath251is a homeomorphism with inverse given by @xmath252 since the quotient homomorphism @xmath250 becomes continuous with respect to the quotient topology , the map @xmath166 is well - defined .
now , a short calculation shows that both maps @xmath166 and @xmath253 are continuous and inverse to each other .
[ spec a_a ] let @xmath2 be a complete unital locally convex algebra such that @xmath212 is locally equicontinuous .
if @xmath106 , then the map @xmath254):=\chi\left(f\left(\frac{1}{\chi(a_1)},\ldots,\frac{1}{\chi(a_n)}\right)\right)\]]is a homeomorphism .
\(i ) a short observation shows that the map @xmath255is clearly a homeomorphism onto its image .
\(ii ) to proceed we show that the set @xmath256 is homeomorphic to @xmath257 : for this we first put @xmath258 . then proposition [ spectrum of quotient algebras ] implies that the space @xmath257 is homeomorphic to @xmath259 .
furthermore , theorem [ spec of c(m , a ) top ] implies that this last set is homeomorphic to @xmath260next , we observe that the condition @xmath261 is equivalent to @xmath262 for all @xmath103 , i.e. , we have @xmath263the corresponding homeomorphism @xmath264 is given by @xmath265):=\chi\left(f\left(\frac{1}{\chi(a_1)},\ldots,\frac{1}{\chi(a_n)}\right)\right).\ ] ] \(iii ) the claim now follows from @xmath266 .
in this section we are dealing with algebra bundles @xmath57 with a possibly infinite - dimensional fibre @xmath2 over a finite - dimensional manifold @xmath4 and show how to endow the corresponding space @xmath58 of sections with a topology that turns it into a ( unital ) locally convex algebra .
[ algebra bundle ] ( algebra bundles ) .
let @xmath2 be a locally convex algebra .
an _ algebra bundle _
( with fibre @xmath2 ) is a quadruple @xmath59 , consisting of an infinite - dimensional manifold @xmath267 , a manifold @xmath4 and a smooth map @xmath57 , with the following property : all fibres @xmath268 , @xmath163 , carry algebra structures , and each point @xmath163 has an open neighbourhood @xmath11 for which there exists a diffeomorphism @xmath269satisfying @xmath270 and all maps @xmath271are algebra isomorphisms .
[ sections of algebra bundles ] ( sections of algebra bundles ) .
let @xmath59 be an algebra bundle .
then the corresponding space @xmath272of _ smooth sections _ carries the structure of an algebra . indeed , given two sections @xmath273 and @xmath274 , the vector space structure on @xmath58
is defined by @xmath275moreover , the product @xmath276 defines a multiplication map on @xmath58 .
if @xmath2 has a unit , then the unit of @xmath58 is given by the section @xmath277 .
in this case @xmath5 is a unital subalgebra of @xmath58 .
our next goal is to endow @xmath58 with a topology .
[ top on space of sections ] ( a topology on @xmath58 ) .
let @xmath59 be an algebra bundle and @xmath278 a bundle atlas for @xmath59 .
we endow @xmath58 with the initial topology @xmath279 generated by the maps @xmath280for @xmath281 . here
, the right - hand side carries the smooth compact open topology from definition [ smooth compact open topology ] .
this topology turns @xmath58 into a locally convex algebra ( cf .
proposition [ c(m , a ) loc .
con . algebra ] ) .
we will see soon that the topology on @xmath58 does not depend on the particular choice of the bundle atlas @xmath278 .
[ locally defines sections ] let @xmath59 be an algebra bundle and @xmath278 a bundle atlas for @xmath59 .
moreover , let @xmath282 _ _ ( _ _ on @xmath283 _ _ ) _ _ and @xmath284 for @xmath285 and @xmath286 _ _ ( _ _ cf .
construction [ top on space of sections]_)_. then the following assertions hold : * if @xmath286 and @xmath287 for @xmath281 and @xmath288 , then @xmath289 for each @xmath285 . *
conversely , if @xmath290 satisfies @xmath289 for each @xmath285 , then the map @xmath291defines a section of the bundle @xmath59 with @xmath292 for each @xmath281 .
\(a ) this is just a simple calculation involving the bundle charts @xmath278 .
\(b ) let @xmath293 .
then @xmath294shows that the map @xmath295 is well - defined .
it obviously defines a section of the bundle @xmath59 with @xmath292 for each @xmath281 .
[ imphi ] suppose we are in the situation of lemma [ locally defines sections ] .
further , consider the algebra homomorphism @xmath296then the following assertions hold : * we have @xmath297 * the algebra homomorphism @xmath298 is a topological embedding with closed image .
\(a ) the first assertion is a direct consequence of lemma [ locally defines sections ] .
\(b ) clearly , the map @xmath298 is injective .
moreover , the topology @xmath299 just defined in construction [ top on space of sections ] turns it into a topological embedding . to see that @xmath300 is closed ,
let @xmath301 be a net in @xmath58 such that @xmath302 converges to some @xmath290 . then @xmath303 for each @xmath281 and @xmath288 . from this
we conclude that @xmath304for all @xmath285 and @xmath293 and thus that @xmath305 by part ( a ) .
[ sco topology = initial topology ] ( smooth compact open topology@xmath306initial topology ) . given a manifold @xmath4 and a locally convex algebra @xmath2 ,
the topology @xmath279 on @xmath78 induced by an atlas @xmath278 of @xmath4 ( consider @xmath78 as the sections of the trivial bundle @xmath307 and use construction [ top on space of sections ] ) coincides with the smooth compact opnen topology traditionally considered on @xmath78 ( cf .
definition [ smooth compact open topology ] ) . a very nice proof of this statement ( and more background on topologies on function spaces ) can be found in ( * ? ? ?
* proposition 4.19 ( d ) ) .
[ topology unique ] let @xmath59 be an algebra bundle and @xmath278 a bundle atlas for @xmath59 .
further , let @xmath308 be another bundle atlas for @xmath59 .
then the identity map @xmath309is an isomorphism of locally convex algebras . in particular , the topology on @xmath58 does not depend on the particular choice of the bundle atlas .
the universal property of the initial topology @xmath310 implies that the identity map @xmath311 is continuous if and only if the maps @xmath312are continuous for each @xmath313 .
therefore , we fix @xmath313 and note that the continuity of @xmath314 follows from proposition [ imphi ] ( b ) , remark [ sco topology = initial topology ] and the continuity of the map @xmath315where @xmath316 denotes the smooth map defined by the transition function @xmath317 .
a similar argument shows that the inverse " map @xmath318 is continuous .
thus , @xmath311 is an isomorphism of locally convex algebras .
[ a frechet gammaa frechet ] let @xmath2 be a frchet algebra and @xmath59 an algebra bundle .
then @xmath58 carries a unique structure of a frchet algebra , when endowed with the topology of construction [ top on space of sections ] .
if @xmath278 is a countable bundle atlas for @xmath59 , then proposition [ imphi ] ( b ) implies that @xmath58 carries the structure of a frchet algebra , since the right - hand side is a frchet algebra and the image of @xmath298 is closed . that this structure is unique is now a consequence of theorem [ topology unique ] .
the following remark can be used to verify that certain maps to @xmath58 are smooth .
we recall that the space @xmath319 carries the structure of an infinite - dimensional manifold : [ smooth structure on sections ] ( verifying smoothness on @xmath58 ) .
if @xmath59 is an algebra bundle and @xmath278 a bundle atlas for @xmath59 , then the map @xmath320is a smooth embedding . indeed ,
proposition [ imphi ] ( b ) implies that the map @xmath298 is continuous and linear .
hence , @xmath298 is smooth .
since @xmath298 has closed image , we conclude from ( * ? ? ?
* lemma 2.2.7 ) , that a map @xmath321 from a manifold @xmath322 to @xmath58 is smooth if and only if the composition @xmath323 is smooth .
in this section we finally want to prove a smooth analogue of proposition [ k(x)_f = o(x_f ) ] for the algebra @xmath5 of smooth functions on some manifold @xmath4 .
in fact , we show that if @xmath2 is a unital frchet algebra , @xmath59 an algebra bundle and @xmath60 , then the smooth localization of @xmath58 with respect to @xmath43 ( cf .
definition [ sm .
loc . ] ) is isomorphic ( as a unital frchet algebra ) to @xmath61 , where @xmath50 . to be more precise ,
we show that the map @xmath324\mapsto f\circ\left(\frac{1}{f}\times\operatorname{id}_{m_f}\right)\]]is an isomorphism of unital frchet algebras . we start with a very useful lemma of hadamard : [ hadamards lemma ] _ _ ( _ _ hadamard s lemma__)__. let @xmath88 be a complete locally convex space .
further , let @xmath11 be an open convex subset of @xmath325 containing @xmath239 and @xmath326 be a smooth function that vanishes on @xmath327 .
if @xmath328 denotes the projection to the j - th factor , then @xmath329holds for suitable smooth functions @xmath330 . for @xmath331 in @xmath11
we define @xmath332 and @xmath333 .
we further define a smooth @xmath88-valued curve @xmath175 on @xmath334 $ ] by @xmath335\rightarrow e,\,\,\ , h(t):=f(z+ty).\]]then @xmath336 , and for @xmath337 we obtain @xmath338i.e . , @xmath339 as desired .
[ manifold mod sub = sub ] _ _ ( _ _ a factorization theorem for algebra bundles__)__. let @xmath2 be a complete locally convex algebra and @xmath59 an algebra bundle .
further , let @xmath340 be a smooth function with @xmath341 as a regular value .
if @xmath342 is the corresponding closed submanifold of @xmath4 , then the restriction map @xmath343 , @xmath344 is a surjective morphism of locally convex algebras .
its kernel @xmath345 is equal to @xmath346 in particular , @xmath347 is a closed two - sided ideal of @xmath58 .
the proof of this theorem is devided into three parts : \(i ) to show that the map @xmath348 is a surjective morphism of locally convex algebras , we first note that @xmath348 is obviously linear , multiplicative and continuous as a restriction map .
thus , it remains to prove its surjectivity : for this we have to show that any section @xmath349 can be extended to a global section on @xmath4 .
locally this can be done , since @xmath350 looks like @xmath351 in @xmath102 ( for @xmath352 ) and @xmath295 like a smooth @xmath2-valued function on @xmath351 . to get a global extension we have to use a partition of unity .
\(ii ) next , we show the equality of the kernel of @xmath348 and the ideal @xmath347 . since @xmath348 is continuous , this will in particular imply that @xmath347 is a closed ideal of @xmath58 : we immediately verify that @xmath353 . for the other inclusion ,
let @xmath286 with @xmath354 , i.e. , @xmath355 on @xmath350 .
we claim that each @xmath163 has an open @xmath98-neighbourhood @xmath11 such that @xmath356holds for suitable sections @xmath357 .
if this is the case , we choose an open cover @xmath358 of @xmath4 such that @xmath359holds for suitable sections @xmath360 and a partition of unity @xmath361 subordinated to this cover .
since each section @xmath362 can be extended to a section on @xmath4 ( by defining it to be zero outside @xmath363 ) , we conclude that @xmath364 \(iii ) we now prove the claim of part ( ii ) . for this , let first be @xmath365 .
then there exists @xmath366 and an open @xmath98-neighbourhood @xmath11 such that @xmath367 on @xmath11 .
therefore , we can write @xmath368now , let @xmath369 . by the implicit function theorem we may choose a chart @xmath370 around @xmath98 such that @xmath371 looks like the projection onto the first @xmath372-coordinates and @xmath373 like a smooth @xmath2-valued function on @xmath102 .
if we write @xmath374 and use lemma [ hadamards lemma ] , then we conclude that @xmath375for suitable sections @xmath376 . in particular
, from this we immediately obtain the desired representation @xmath377 [ algebra section 0 ] if @xmath59 is an algebra bundle and @xmath322 a manifold , then the following assertions hold : * if @xmath278 is a bundle atlas for @xmath59 , then the algebra map @xmath378is a topological embedding with closed image given by @xmath379 _ _
lemma [ locally defines sections ] and lemma [ smooth exp law ] _ ) _ . * if @xmath380 denotes the canonical projection onto @xmath4 and @xmath381 the corresponding pull - back bundle , then the map @xmath382is an isomorphism of unital locally convex algebras .
\(a ) the first assertion immediately follows from [ gl05 ] , lemma 4.2.6 applied to proposition [ imphi ] .
\(b ) for the second assertion we first note that both spaces , @xmath383 and @xmath384 , carry the structure of a locally convex algebra and that the map @xmath253 is an algebras homomorphism . to see that @xmath253 is bijective it is enough to note that each section @xmath295 of the pull - back bundle @xmath381 has the form @xmath385for some smooth function @xmath386 satisfying @xmath387 for each @xmath388 , i.e. , for some @xmath389 . it therefore remains to show that the map @xmath253 is bicontinuous : for this let @xmath278 be a bundle atlas for @xmath59 .
then the bicontinuity of @xmath253 follows from the definition of the topology on @xmath384 ( which is induced from the bundle atlas @xmath278 ) and part ( a ) of the lemma .
[ algebra section 1 ] let @xmath2 be a unital locally convex algebra and @xmath59 an algebra bundle . if @xmath390 are smooth functions and @xmath391 , then the map @xmath392is a surjective morphism of unital locally convex algebras .
its kernel @xmath393 is equal to @xmath394 in particular , @xmath395 is a closed two - sided ideal of @xmath396 .
the proof of this proposition is divided into three parts : \(i ) we first note that @xmath341 is a regular value for the function @xmath397 in particular , @xmath398is a closed submanifold of @xmath399 .
further , we note that @xmath350 is diffeomorphic to @xmath400 .
\(ii ) next , lemma [ algebra section 0 ] ( b ) applied to the algebra bundle @xmath59 and @xmath401 implies that the map @xmath402is an isomorphism of unital locally convex algebras .
a similar argument shows that the space @xmath403 of sections of the pull - back bundle @xmath381 restricted to the submanifold @xmath350 is isomorphic ( as a unital locally convex algebra ) to @xmath404 .
\(iii ) the assertions of the proposition now follow from theorem [ manifold mod sub = sub ] applied to the restriction map @xmath405 [ algebra section 2 ] _ _ ( _ _ smooth localization of sections of algebra bundles__)__. suppose we are in the situation of proposition [ algebra section 1 ] . if @xmath2 is a unital frchet algebra
, then the map @xmath406 is open .
moreover , the induced map @xmath407\mapsto f\circ\left(\frac{1}{f_1}\times\cdots\times\frac{1}{f_k}\times\operatorname{id}_{m_{f_1,\ldots , f_k}}\right)\]]is an isomorphism of unital frchet algebras _ _ ( _ _ cf . definition [ sm .
loc . ] for the definition of @xmath408__)__. if @xmath2 is a unital frchet algebra , then so are @xmath58 , @xmath409 ( cf .
corollary [ a frechet gammaa frechet ] ) and @xmath396 .
in particular , the open mapping theorem ( cf .
* chapter iii , section 2.2 ) ) implies that the map @xmath406 is open . since the quotient of a frchet space by a closed subspace is again frchet ( cf .
@xcite ) , we can use the open mapping theorem again to see that the map @xmath410 is an isomorphism of unital frchet algebras .
although @xmath411 is not the ring of fractions with some power of @xmath43 as denominator , theorem [ algebra section 2 ] yields the desired smooth analogue of proposition [ k(x)_f = o(x_f ) ] : [ c(m)_f = c(m_f ) ] let @xmath4 be a manifold and @xmath2 a unital frchet algebra .
further , let @xmath390 be smooth functions and @xmath412
. then the map @xmath413\mapsto f\circ\left(\frac{1}{f_1}\times\cdots\times\frac{1}{f_k}\times\operatorname{id}_{m_{f_1,\ldots , f_k}}\right)\]]is an isomorphism of unital frchet algebras .
the assertion directly follows from theorem [ algebra section 2 ] applied to the trivial algebra bundle @xmath414 .
we now present a very useful and well - known theorem of analysis : [ whitneys theorem ] _ _ ( _ _ whitneys theorem__)__. any open subset @xmath11 of a manifold @xmath4 is the complement of the zeroset of a smooth function @xmath415 , i.e. , @xmath416such a function @xmath43 is called a @xmath11-defining function . a proof for the case @xmath112 can be found in [ gupo74 ] , exercise 1.5.11 .
a combination with a partition - of - unity argument proves the general case .
[ algebra section 3 ] if @xmath2 is a unital frchet algebra , @xmath59 an algebra bundle and @xmath417 a @xmath11-defining function corresponding to an open subset @xmath11 of @xmath4 , then the map @xmath418\mapsto f\circ\left(\frac{1}{f_u}\times\operatorname{id}_u\right)\]]is an isomorphism of unital frchet algebras .
this assertion is a direct consequence of theorem [ algebra section 2 ] .
in the last two sections we were concerned with sections of algebra bundles .
the goal of this section is to describe the spectrum of the algebra of sections of an algebra bundle with finite - dimensional commutative fibre .
we show that if @xmath59 is such an algebra bundle , then the spectrum of the corresponding algebra @xmath58 of sections is an @xmath3-fold covering of @xmath4 ( here @xmath419 ) .
[ wedderburn - artin ] _ _ ( _ _ wedderburn - artin_)_. each finite - dimensional semisimple unital algebra is isomorphic to a direct product of matrix algebras , i.e. , to @xmath420conversely , each finite direct product of matrix algebras is a semisimple unital algebra . for the proof of this statement
we refer , for example , to ( * ? ? ?
* theorem 2.4.3 ) .
[ tsssabfdcf1 ] each finite - dimensional semisimple unital algebra which is commutative is isomorphic to a direct product of copies of @xmath421 .
this assertion is a direct consequence of theorem [ wedderburn - artin ] .
if a unital algebra @xmath2 is not semisimple , it is quite natural to measure the defect for @xmath2 not being semisimple .
this can be done with the so - called jacobson radical : [ the jacobson radical ] ( the jacobson radical ) .
jacobson radical _ @xmath422 of a finite - dimensional unital algebra @xmath2 can be defined as * the largest nilpotent ideal of @xmath2 . *
the intersection of all maximal ideals of @xmath2 .
* the smallest ideal of @xmath2 such that the quotient algebra @xmath423 is semisimple .
in fact , for the equivalence of these definition we refer to ( * ? ? ?
* section 3.1 ) .
[ tsssabfdcf2 ] let @xmath2 be a finite - dimensional unital algebra .
further , let @xmath424 be the jacobson radical of @xmath2 and @xmath425 the corresponding semisimple unital quotient algebra .
then the following assertions hold : * each algebra automorphism @xmath426 leaves @xmath424 invariant , i.e. , restricts to an algebra automorphism @xmath427 of @xmath424 .
we write @xmath428for the corresponding group homomorphism . *
in particular , each algebra automorphism @xmath426 descends to an algebra automorphism @xmath429 of @xmath430 .
we write @xmath431for the corresponding group homomorphism .
\(a ) the first assertion follows from the definition of @xmath422 ( cf .
definition [ the jacobson radical ] ( a ) or ( b ) ) .
\(b ) the second assertion is a consequence of part ( a ) .
[ tsssabfdcf3 ] suppose we are in the situation of lemma [ tsssabfdcf2 ] .
further , let @xmath59 be an algebra bundle and @xmath432 be a bundle atlas for @xmath59 .
if @xmath433 for @xmath285 and @xmath434 are the corresponding transition functions , then the following assertions hold : * there exists an algebra _ _ ( _ _ sub- _ ) _ bundle @xmath435 of @xmath59 with bundle charts @xmath436 such that @xmath437 * there exists an algebra bundle @xmath438 with bundle charts such that @xmath439 in particular , the algebra bundle @xmath438 can be realized as the quotient algebra bundle of @xmath59 with respect to @xmath435 .
the corresponding quotient map is given by @xmath440 * we obtain a short exact sequence @xmath441 of _ _ ( _ _ frchet- _ ) _ algebras . here , the second arrow denotes the natural inclusion and @xmath442 ( a),(b ) the first two assertions are a consequence of lemma [ tsssabfdcf2 ] and ( * ? ? ?
* part i , section 3 , theorem 3.2 ) .
\(c ) the third assertion follows from an easy calculation and part ( a ) and ( b ) .
[ finite open cover ] ( finite open cover property ) .
if @xmath59 is an algebra bundle over a compact manifold @xmath4 , then it is obviously true that there exists a _
finite _ open cover @xmath443 of @xmath4 such that each @xmath444 is trivial .
note that this property is also true for _ arbitrary _ manifolds : indeed ( * ? ? ?
* theorem 7.5.16 ) states that if @xmath59 is an algebra bundle over a manifold @xmath4 of dimension @xmath3 , then there is a cover @xmath445 of @xmath4 , such that @xmath444 is trivial .
[ tsssabfdcf4 ] let @xmath435 be the algebra _ _
( _ _ sub- _ ) _ bundle of proposition [ tsssabfdcf3 ] _ _ ( _ _ a_)_. then @xmath446 consists of nilpotent elements , i.e. , for each @xmath447 there exists a natural number @xmath54 such that @xmath448 . in view of remark [ finite open cover ] and proposition [ imphi ] ,
@xmath449 can be embedded into a finite product of the form @xmath319 . since each function
@xmath450 appearing in the image @xmath451 of an element @xmath452 takes values in @xmath422 , there exists a natural number @xmath54 such that @xmath453 .
thus , we obtain @xmath448 as desired .
[ tsssabfdcf5 ] let @xmath2 be a finite - dimensional unital algebra and @xmath59 be an algebra bundle .
further , let @xmath438 be the algebra bundle of proposition [ tsssabfdcf3 ] _ _ ( _ _ b_)_. then the map @xmath454where @xmath286 with @xmath455 , is a well - defined homeomorphism with inverse given by @xmath456for @xmath457 .
\(i ) we first show that @xmath166 is well - defined . for this
we have to verify that each character @xmath458 vanishes on @xmath449 ( cf .
proposition [ tsssabfdcf3 ] ( c ) ) : indeed , if @xmath458 and @xmath452 , then proposition [ tsssabfdcf4 ] implies that there exists a natural number @xmath54 such that @xmath448 .
hence , @xmath459 follows from @xmath460 .
\(ii ) now , a short calculation shows that both maps @xmath166 and @xmath253 are continuous and inverse to each other . in the following
we describe the spectrum of the algebra of sections of an algebra bundle with finite - dimensional semisimple commutative fibre : [ tsssabfdcf6 ] let @xmath4 be a manifold and @xmath461 for some @xmath54 .
then the map @xmath462where @xmath463 , is a homeomorphism .
the assertion immediately follows from theorem [ spec of c(m , a ) top ] and the observation that .
[ tsssabfdcf7 ] for @xmath54 let @xmath464 and @xmath461 .
then the following assertions hold : * we have @xmath465 , where @xmath466 denotes the symmetric group of order @xmath3 . * the map @xmath467is an isomorphism of groups . *
let @xmath4 be a manifold . if we identify @xmath2 with @xmath468 , then the map @xmath469where @xmath470is an isomorphism of unital frchet algebras .
( a),(b ) the proof of the first two statements is left as an easy exercise to the reader .
\(c ) the last assertion follows from lemma [ smooth exp law ] applied to @xmath471 .
[ tsssabfdcf8 ] let @xmath2 be a finite - dimensional commutative unital algebra and .
further , let @xmath59 be an algebra bundle and @xmath432 be a bundle atlas for @xmath59 . if @xmath433 for @xmath285 and @xmath472 are the corresponding transition functions of @xmath438 _ _ ( _ _ cf .
proposition [ tsssabfdcf3 ] _ ) _ , then there exists a fibre bundle @xmath473 , i.e. , an @xmath3-fold covering @xmath474 with bundle charts @xmath475 such that @xmath476__(__cf . lemma [
tsssabfdcf7]_)_. the assertion of this proposition follows from lemma [ tsssabfdcf7 ] ( b ) and ( * ? ? ?
* part i , section 3 , theorem 3.2 ) .
[ tsssabfdcf9 ] suppose we are in the situation of proposition [ tsssabfdcf8 ] .
further , consider the algebra homomorphism @xmath477with @xmath478 .
then the following assertions hold : * if @xmath479 _ _ ( _ _ on @xmath480 _ _ ) _ _ , then @xmath481 * the algebra homomorphism @xmath482 is a topological embedding with closed image .
the proof of this proposition is similar to the proof of proposition [ imphi ] .
[ tsssabfdcf10 ] suppose we are in the situation of proposition [ tsssabfdcf8 ] .
further , let @xmath483be the embedding of proposition [ imphi ] applied to the algebra bundle @xmath438 and @xmath482 the embedding of proposition [ tsssabfdcf9 ] .
then the map @xmath484is an isomorphism of unital frchet algebras . here
, @xmath485 denotes the map of lemma [ tsssabfdcf7 ] _ _ ( _ _ c _ ) _ applied to @xmath486 and @xmath487 . in order to prove the assertion ,
it is enough to verify that the image of the map @xmath488 is equal to @xmath489 .
in fact , this immediately follows from lemma [ tsssabfdcf7 ] ( a ) and ( b ) .
[ tsssabfdcf11 ] _ _ ( _ _ the spectrum as an @xmath3-fold covering__)__. let @xmath2 be a finite - dimensional commutative unital algebra .
further , let @xmath59 be an algebra bundle and @xmath432 be a bundle atlas for @xmath59 .
if @xmath490 is as in proposition [ tsssabfdcf8 ] and @xmath491 denotes the usual evaluation map in some @xmath492 , then the map @xmath493is a homeomorphism .
here , @xmath494 is the map of proposition [ tsssabfdcf10 ] and @xmath457 _ _ ( _ _ cf .
proposition [ tsssabfdcf3]_)_. in particular , if @xmath495 , then the spectrum @xmath65 is an @xmath3-fold covering of @xmath4 .
we first note that the map @xmath496 is well - defined .
next , let @xmath497be the homeomorphism of proposition [ spec of c(m ) top ] and @xmath498the homeomorphism induced by the map @xmath494 of proposition [ tsssabfdcf10 ] .
if @xmath253 is the homeomorphsim of theorem [ tsssabfdcf5 ] , then a short observation shows that @xmath499 . from this
we conclude that @xmath500 is a homeomorphism as it is a composition of homeomorphisms .
[ tsssabfdcf12 ] suppose we are in the situation of theorem [ tsssabfdcf11 ] with @xmath461 .
then the map @xmath501for @xmath281 with @xmath502 , @xmath503 with @xmath504 and @xmath505 , is a homeomorphism .
this assertion is a direct consequence of theorem [ tsssabfdcf11 ] .
in this section we present a method of localizing dynamical systems @xmath66 with respect to elements of the commutative fixed point algebra of the induced action of @xmath7 on the center @xmath68 of @xmath2 . in fact , this construction turns out to be the starting point for our approach to noncommutative principal torus bundles .
let @xmath66 be a dynamical system and @xmath68 be the center of @xmath2 .
if @xmath506 and @xmath104 , then the equation @xmath507 shows that the map @xmath26 restricts to a continuous action of @xmath7 on @xmath68 by algebra automorphisms . in the following we write @xmath508 for the fixed point algebra of the induced action of @xmath7 on @xmath68 , i.e. , @xmath509
we further write @xmath510 for the smooth localization of @xmath2 with respect to @xmath511 and @xmath512for the corresponding continuous quotient homomorphism ( cf .
definition [ sm .
loc . ] ) .
[ a_z is locally convex space again ] with the notations from above , the following assertion holds : the map @xmath513,z)\mapsto [ z\cdot f]\]]defines on @xmath510 the structure of a right locally convex @xmath508-module .
since @xmath508 is commutative , the assertion directly follows from lemma [ a_a ... is a ( left ) a - modul ] .
[ a_z is algebra ] the locally convex space @xmath510 becomes a locally convex algebra , when equipped with the multiplication @xmath514,[f']):=[f\cdot f'].\ ] ] this assertion is an easy consequence of the construction of @xmath510 .
indeed , if @xmath98 denotes the ( continuous ) multiplication map of the algebra @xmath515 , then the continuity of @xmath516 follows from @xmath517 and the fact that the map @xmath518 is continuous , surjective and open .
[ remark on smooth dynamical systems ] ( smooth dynamical systems ) .
we call a dynamical system @xmath66 _ smooth _ if @xmath7 is a lie group and the group homomorphism @xmath76 induces a smooth action of @xmath7 on @xmath2 .
a classical example of such a smooth dynamical system is induced by a smooth action @xmath519 of a lie group @xmath7 on a manifold @xmath4 . in particular ,
each principal bundle @xmath9 induces a smooth dynamical system @xmath70 , consisting of the frchet algebra of smooth functions on the total space @xmath8 , the structure group @xmath7 and a group homomorphism @xmath520 , induced by the smooth action @xmath521 of @xmath7 on @xmath8 . for this and further ( non - classical ) examples of smooth dynamical systems we refer the interested reader to the examples of section [ a geometric approach to ncpb ] or @xcite .
[ dynamical system and trivial algebra bundles ] let @xmath66 be a _ _ ( _ _ smooth _ ) _ dynamical system and @xmath4 a manifold .
then the map @xmath522defines a _ _ ( _ _ smooth _ ) _ continuous action of @xmath7 on @xmath78 by algebra automorphisms . in particular ,
the triple @xmath523 is a _ _ ( _ _ smooth _ ) _ dynamical system .
we first note that the map @xmath524is smooth ( cf .
* proposition i.2 ) ) .
then lemma [ smooth exp law ] implies that the map @xmath525 is smooth if and only if the map @xmath526is smooth . since @xmath527 , we conclude that @xmath528 is smooth as a composition of smooth maps . clearly , @xmath525 defines an action of @xmath7 on @xmath78 by algebra automorphisms .
[ t^n action on spec again ] if @xmath66 is a dynamical system and @xmath511 , then the map @xmath529\underline{})\mapsto[\alpha(g)\circ f]\]]defines a continuous action of @xmath7 on @xmath510 by algebra automorphisms . in particular , the triple @xmath530 is a dynamical system .
if @xmath531 , then @xmath532 implies that the map @xmath533 is well - defined .
further , a simple calculation shows that the map @xmath533 defines an action of @xmath7 on @xmath510 by algebra automorphisms . that this action is continuous is a consequence of the continuity of the map @xmath534(cf .
lemma [ dynamical system and trivial algebra bundles ] applied to @xmath535 ) , @xmath536 and the fact that @xmath537 is continuous , surjective and open . [ smooth localized dynamical system ] ( smooth localization of dynamical systems ) .
let @xmath66 be a dynamical system and @xmath511 .
we call the dynamical system @xmath530 of proposition [ t^n action on spec again ] the _ smooth localization of @xmath66 _ associated to the element @xmath511 .
[ example on smooth localization ] let @xmath9 be a principal bundle and @xmath70 be the induced smooth dynamical system ( cf . remark [ remark on smooth dynamical systems ] ) .
further , let @xmath11 be an open subset of @xmath4 such that @xmath538 is a trivial principal @xmath7-bundle over @xmath11 , i.e. , such that @xmath539 holds as @xmath7-manifolds .
if @xmath43 is a @xmath11-defining function ( cf .
theorem [ whitneys theorem ] ) , then @xmath540 is a @xmath538-defining function lying in the subalgebra @xmath541 .
thus , we conclude from corollary [ c(m)_f = c(m_f ) ] that the map @xmath542\mapsto\left(p\mapsto f\left(\frac{1}{h(p)},p\right)\right)\]]is an isomorphism of unital frchet algebras .
moreover , the map @xmath543 is @xmath7-equivariant .
in fact , we have @xmath544)=\alpha(g,\phi_u([f]))\]]for all @xmath531 and @xmath545\in c^{\infty}(p)_{\{h\}}$ ] . in particular , the corresponding localized dynamical system @xmath546 of proposition [ t^n action on spec again ] carries the structure of a smooth dynamical system , since it is isomorphic to the smooth dynamical system @xmath547 corresponding to the trivial principal @xmath7-bundle @xmath538 over @xmath11 . in the remaining part of this section
we will be concerned with the spectrum of @xmath510 : [ t^n action on spec ii again ] if @xmath66 is a dynamical system and @xmath511 , then the map @xmath548defines an action of @xmath7 on @xmath549 .
this claim follows from a simple calculation and is therefore left to the reader .
[ spec a_z ] ( @xmath550 ) .
let @xmath2 be a complete unital locally convex algebra such that @xmath161 is locally equicontinuous .
then theorem [ spec a_a ] implies that the map @xmath551):=\chi\left(f\left(\frac{1}{\chi(z)}\right)\right)\]]is a homeomorphism .
the main goal of this section is to present a geometrically oriented approach to the noncommutative geometry of principal torus bundles .
as already mentioned in the introduction , a trivial noncommutative principal torus bundles is a dynamical system @xmath1 with the additional property that each isotypic component contains an invertible element . since the freeness property of a group action is a local condition ( cf .
remark [ free = locally free ] ) , our main idea is inspired by the classical setting : loosely speaking , a dynamical system @xmath1 is called a noncommutative principal @xmath0-bundle , if it is locally " a trivial noncommutative principal @xmath0-bundle , i.e. , section [ acncpt^nb ] enters the picture .
we prove that this approach extends the classical theory of principal torus bundles and present some noncommutative examples .
indeed , we first show that each trivial noncommutative principal torus bundle carries the structure of a noncommutative principal torus bundle in its own right .
we further show that examples are provided by sections of algebra bundles with trivial noncommutative principal torus bundle as fibre , sections of algebra bundles which are pull - backs of principal torus bundles and sections of trivial equivariant algebra bundles . at the end of this section we present a very concrete example . given a dynamical system @xmath66 , we ( again )
write @xmath508 for the fixed point algebra of the induced action of @xmath7 on the center @xmath68 of @xmath2 , i.e. , @xmath552 .
in particular , if @xmath8 is a manifold , @xmath7 a lie group group and @xmath70 a ( smooth ) dynamical system , then @xmath553 .
[ ncpt^nb again ] ( noncommutative principal torus bundles ) .
we call a ( smooth ) dynamical system @xmath1 a _ _ _ ( _ _ smooth _ ) _ noncommutative principal @xmath0-bundle _ if for each @xmath554 there exists an element @xmath511 with @xmath555 such that the corresponding localized dynamical system @xmath556 of proposition [ t^n action on spec again ] is a ( smooth ) trivial noncommutative principal @xmath0-bundle ( cf
. definition [ trivial ncp t^n - bundles ] ) .
the previous definition of noncommutative principal torus bundles is inspired by the classical setting .
in fact , we will see in theorem [ nct^nb for manifold again ] that it actually agrees with the classical definition of principal torus bundles if we identify open subsets @xmath11 of the base manifold with the corresponding @xmath11-defining functions and use corollary [ c(m)_f = c(m_f ) ] .
[ remark on ncpt^nb again ] ( an equivalent point of view ) . given @xmath511
, we recall that @xmath557then a short observation shows that a ( smooth ) dynamical system @xmath1 is a ( smooth ) noncommutative principal @xmath0-bundle if and only if there exists a family of elements @xmath558 satisfying the following two conditions : * the family @xmath559 is an open covering of @xmath560 .
* the localized dynamical systems @xmath561 are ( smooth ) trivial noncommutative principal @xmath0-bundles .
[ t^n action on spec iii again ] let @xmath2 be a commutative unital locally convex algebra and @xmath1 be a noncommutative principal @xmath0-bundle .
further , let @xmath554 and @xmath511 with @xmath555 such that the corresponding _ _ ( _ _ smooth _ ) _ localized dynamical system @xmath556 is a trivial noncommutative principal @xmath0-bundle .
then the @xmath0-action map @xmath562of lemma [ t^n action on spec ii again ] is free .
the assertion follows from fact that each trivial noncommutative principal torus bundle defines a free action of @xmath0 on its spectrum ( cf .
* proposition 1.4 ) ) .
in this part of the section we show how to recover the classical definition of principal torus bundles .
we start with some general result relating smooth group actions and the corresponding smooth dynamical systems . [ extending lemma for dynamical systems ] let @xmath8 be a compact manifold , @xmath7 a lie group and @xmath70 a smooth dynamical system .
then each character @xmath563 is the restriction of an evaluation homomorphism @xmath564 let @xmath563 be a character . if all functions of @xmath565 vanish in the point @xmath566 in @xmath8 , then the maximality of @xmath199 implies that @xmath567 , i.e. , that @xmath568 .
so we have to show that such a point exists .
let us assume that this is not the case .
from that we shall derive the contradiction @xmath569 : \(i ) if such a point does not exist , then for each @xmath17 there exists a function @xmath570 with @xmath571 .
the family @xmath572 is an open cover of @xmath8 , so that there exist @xmath573 and a smooth @xmath7-invariant function @xmath574 on @xmath8 . here
, the last statement follows from the fact that each algebra automorphism of @xmath575 is automatically a * -automorphism ( cf .
* remark 1.8 ) ) .
\(ii ) in view of part ( i ) we obtain a function @xmath576 which is nowhere zero .
thus , @xmath577 is invertible in @xmath575 .
we claim that its inverse @xmath578 is also @xmath7-invariant : for this we choose an arbitrary element @xmath531 and note that @xmath579 implies @xmath580 in particular , @xmath581 .
since @xmath531 was arbitrary , we conclude that @xmath582 .
\(iii ) finally , part ( ii ) leads to the contradiction @xmath583 , i.e. , @xmath569 as desired .
[ extension of char on fixed point algebras vi ] let @xmath8 be a compact manifold , @xmath7 a compact lie group and @xmath70 a smooth dynamical system . then the map @xmath584is a homeomorphism .
we divide the proof of this proposition into four parts : \(i ) we first note that the quotient @xmath585 is a compact hausdorff space .
further , the map @xmath166 is well - defined since each function @xmath586 is @xmath7-invariant .
\(ii ) proposition [ spec of c(m ) top ] and the definition of the quotient topology on @xmath585 implies that @xmath166 is continuous .
\(iii ) the surjectivity of @xmath166 is a consequence of lemma [ extending lemma for dynamical systems ] ( alternatively we can use ( * ? ? ?
* corollary 2.5 ) ) . to show that @xmath166 is injective , we choose elements @xmath587 with @xmath588 . since @xmath8 is a manifold , there exist disjoint @xmath7-invariant open subsets @xmath11 and @xmath225 of @xmath8 containing the compact subsets @xmath589 and @xmath590 , respectively .
next , we choose a positive smooth function @xmath43 on @xmath8 which is equal to @xmath591 on @xmath589 and vanishes outside @xmath11
. then the function @xmath592where @xmath593 denotes the normalized haar measure on @xmath7 , defines a smooth @xmath7-invariant function satisfying @xmath594here , we have used that the smooth action map @xmath26 induces a smooth action of @xmath7 on @xmath8 ( cf .
* proposition 2.7 ) ) .
hence , @xmath595implies that @xmath596 , i.e. , @xmath166 is injective .
\(iv ) so far we have seen that @xmath166 is a bijective continuous map from a compact hausdorff space onto a hausdorff space .
thus , the claim follows from a well - known theorem from topology guaranteeing that @xmath166 is a homeomorphism . for a compact manifold @xmath8 , a compact lie group @xmath7 and a dynamical system @xmath70 we write @xmath597 , @xmath598 for the canonical quotient map .
moreover , we identify @xmath599 with an open subset of @xmath585 via the homeomorphism of proposition [ extension of char on fixed point algebras vi ] and define @xmath600for the following lemma we recall that @xmath601 : [ c(p ) localized ii again ] let @xmath8 be a compact manifold , @xmath7 a compact lie group and @xmath70 a dynamical system . if @xmath602 and @xmath511 , then the map @xmath603is a homeomorphism .
this lemma is an easy consequence of proposition [ spec of c(m ) top ] .
[ c(p ) localized iii again ] suppose we are in the situation of lemma [ c(p ) localized ii again ] .
then the map @xmath604):=f\left(\frac{1}{z(p)},p\right)\]]is a homeomorphism .
in particular , @xmath605 becomes a smooth manifold when endowed with the smooth structure turning the map @xmath606 into a diffeomorphism .
the claim is a direct consequence of remark [ spec a_z ] ) and lemma [ c(p ) localized ii again ] .
indeed , we just have to note that @xmath607 .. [ free = locally free ] ( freeness is a local condition ) .
let @xmath608 be an action of a topological group @xmath7 on a topological space @xmath41 and let @xmath609 be the corresponding quotient map .
then a short observation shows that the action @xmath610 is free if and only if there exists an open cover @xmath358 of the quotient space @xmath611 ( with respect to the quotient topology ) such that each restriction map @xmath612where @xmath613 , is free .
[ nct^nb for manifold again ] _ _ ( _ _ reconstruction theorem__)__. for a manifold @xmath8 , the following assertions hold : * if @xmath8 is compact and @xmath19 is a smooth noncommutative principal @xmath0-bundle , then we obtain a principal @xmath0-bundle @xmath20 , where @xmath614 here , we have identified @xmath8 with the set of characters via the map @xmath166 from proposition [ spec of c(m ) top ] . *
conversely , if @xmath14 is a principal @xmath0-bundle , then the corresponding smooth dynamical system @xmath19 is a smooth noncommutative principal @xmath0-bundle .
\(a ) we first note that the induced action map @xmath610 is actually smooth ( cf .
* proposition 2.7 ) ) . since @xmath0 is compact
, @xmath610 is proper and in view of the quotient theorem ( cf .
* kapitel viii , abschnitt 21 ) ) , it remains to verify the freeness of @xmath610 : for this , we first note that the map @xmath610 is free if and only if there exists an open cover of @xmath560 of the form @xmath559 such that each restriction map @xmath615is free ( cf .
remark [ free = locally free ] ) .
moreover , proposition [ c(p ) localized iii again ] implies that @xmath616holds for each @xmath511 , @xmath17 and @xmath617 .
therefore , the freeness of @xmath610 is a consequence of remark [ remark on ncpt^nb again ] ( i ) and lemma [ t^n action on spec iii again ] .
\(b ) conversely , we have to show that the condition of definition [ ncpt^nb again ] is satisfied : this will be done in the following three steps : \(i ) we first note that @xmath618 ( cf .
* proposition 2.4 ) ) .
hence , @xmath560 is homeomorphic to @xmath4 by proposition [ spec of c(m ) top ] .
\(ii ) next , we choose an open cover @xmath358 of @xmath4 such that each @xmath619 is a trivial principal @xmath7-bundle over @xmath620 , i.e. , such that @xmath621 holds for each @xmath281 .
further , we choose for all @xmath281 a @xmath620-defining function @xmath622 and note that each function @xmath623 is a ( smooth ) @xmath619-defining function .
\(iii ) for @xmath17 we choose @xmath281 with @xmath624 . then @xmath625 is an element in @xmath508 with @xmath626 and we conclude from corollary [ c(m)_f = c(m_f ) ] that the map @xmath627\mapsto\left(p\mapsto f\left(\frac{1}{h_i(p)},p\right)\right)\]]is an isomorphism of unital frchet algebras .
moreover , the map @xmath628 is @xmath0-equivariant . in fact
, we have @xmath629)=\alpha(g,\phi_i([f]))\]]for all @xmath617 and @xmath545\in c^{\infty}(p)_{\{h_i\}}$ ] .
in particular , the corresponding localized dynamical system @xmath630 carries the structure of a smooth trivial noncommutative principal @xmath0-bundle , since it is isomorphic to the smooth trivial noncommutative principal @xmath0-bundle @xmath631 corresponding to the trivial principal @xmath0-bundle @xmath632 over @xmath620 .
we show that each trivial noncommutative principal @xmath0-bundle carries the structure of a noncommutative principal @xmath0-bundle : [ tnctb are nctb ] _ _ ( _ _ trivial noncommutative principal torus bundles__)__. each trivial noncommutative principal @xmath0-bundle @xmath1 carries the structure of a noncommutative principal @xmath0-bundle .
if @xmath1 is a trivial noncommutative principal @xmath0-bundle , then @xmath633 and @xmath634 holds for each @xmath554 .
in particular , corollary [ inverting 1 iso ] implies that @xmath635 holds as unital locally convex algebras .
therefore , @xmath1 is a noncommutative principal @xmath0-bundle in its own right .
we show that if @xmath2 is a unital frchet algebra and @xmath1 a smooth trivial noncommutative principal @xmath0-bundle such that @xmath636 is isomorphic to @xmath421 , then the algebra of sections of each algebra bundle with fibre " @xmath1 is a noncommutative principal @xmath0-bundle .
we start with the following proposition : [ trivial ncp t^n - bundles from trivial algebra bundles ] if @xmath1 is a _ _ ( _ _ smooth _ ) _ trivial noncommutative principal @xmath0-bundle and @xmath4 a manifold , then the triple @xmath637 of lemma [ dynamical system and trivial algebra bundles ] is also a _ _ ( _ _ smooth _ ) _ trivial noncommutative principal @xmath0-bundle .
the claim directly follows from the fact that the algebra @xmath2 is naturally embedded in @xmath78 through the constant maps . in fact , if @xmath13 , then the corresponding isotypic component @xmath638 is equal to @xmath639 and thus contains invertible elements .
[ automorphism of dynamical systems ] ( the automorphism group of a dynamical system ) .
let @xmath66 be a dynamical system .
the group @xmath640is called the _ automorphism group of _ the dynamical system @xmath66 .
if @xmath641 , then [ ne08b ] , proposition 1.4.8 implies that @xmath642where @xmath643 , @xmath644 . in particular , if @xmath645 , then @xmath646 [ g - automorphism group of quantumtori ] if @xmath31 is the smooth noncommutative @xmath3-torus and @xmath33 the corresponding smooth dynamical system of example [ nc n - tori as ncptb ] ( b ) , then @xmath647indeed , we first recall that @xmath31 is generated by unitaries @xmath648 .
if now @xmath649 , @xmath18 and @xmath650 ( @xmath651 ) is such a unitary , then @xmath652leads to @xmath653 for some @xmath654 .
since @xmath650 is a unitary and @xmath145 is a @xmath655-automorphism ( @xmath31 is a @xmath655-algebra ) , we conclude that @xmath656 . in particular , each element in @xmath657 corresponds to an element in @xmath0 and vice versa .
[ cocycle description of bundles ] _ _ ( _ _ cocycle description of bundles__)__. let @xmath66 be a smooth dynamical system and @xmath4 a manifold .
further , let @xmath658 be an open cover of @xmath4 and @xmath433 for @xmath285 .
if @xmath659 is a collection of functions @xmath660 satisfying @xmath661then the following assertions hold : * there exists an algebra bundle @xmath59 and bundle charts @xmath662 such that @xmath663 * the map @xmath664for @xmath281 with @xmath665 and @xmath666 with @xmath667 , defines a smooth action of @xmath7 on @xmath267 by fibrewise algebra automorphisms .
\(a ) a proof of the first statement can be found in ( * ? ? ?
* part i , section 3 , theorem 3.2 ) .
\(b ) the crucial point is to show that the map @xmath610 is well - defined : for this let @xmath285 with @xmath668 , @xmath666 with @xmath667 and @xmath669 with @xmath670 .
then @xmath671 leads to @xmath672 further , a short calculation shows that @xmath610 defines an action of @xmath7 on @xmath267 by fibrewise algebra automorphisms .
its smoothness follows from the local description by a smooth function .
[ dynamical system and algebra bundles ] suppose we are in the situation of proposition [ cocycle description of bundles ] .
then the map @xmath673defines a smooth action of @xmath7 on @xmath58 by algebra automorphisms . in particular
, the triple @xmath674 is a smooth dynamical system . that the map @xmath525 defines an action of @xmath7 on @xmath58 by algebra automorphisms is a consequence of proposition [ cocycle description of bundles ] ( b ) . to verify the smoothness of @xmath525 , we first choose a bundle atlas @xmath278 of @xmath59 and use the definition of the smooth structure on @xmath58 : in fact , remark [ smooth structure on sections ]
implies that the map @xmath525 is smooth if and only if each map @xmath675is smooth .
next , we recall that each map @xmath676is smooth by lemma [ dynamical system and trivial algebra bundles ] . since @xmath677 for each @xmath281 , each map @xmath678 is smooth as a composition of smooth maps . for the next theorem we recall that the smooth dynamical systems @xmath33 of example [ nc n - tori as ncptb ] ( b ) provide a class of examples of trivial noncommutative principal @xmath0-bundles for which @xmath679 is isomorphic to @xmath421 .
[ ncp g_bundle with fibre trivial ncp g bundle ] let @xmath2 be a unital frchet algebra and @xmath1 a smooth trivial noncommutative principal @xmath0-bundle such that @xmath680 is isomorphic to @xmath421 .
further , let @xmath4 be a manifold , @xmath658 an open cover of @xmath4 and @xmath433 for @xmath285 .
if @xmath659 is a collection of functions @xmath681 satisfying @xmath661then the smooth dynamical system @xmath682 of proposition [ dynamical system and algebra bundles ] is a noncommutative principal @xmath0-bundle .
\(i ) we first note that @xmath683 .
in particular , the spectrum @xmath560 is homeomorphic to @xmath4 .
\(ii ) next , we choose for each @xmath281 a @xmath620-defining function @xmath622 .
further , for @xmath163 , we choose @xmath281 with @xmath684 .
then @xmath622 is an element in @xmath5 with @xmath685 and we conclude from corollary [ algebra section 3 ] that the map @xmath686\mapsto f\circ\left(\frac{1}{f_i}\times\operatorname{id}_{u_i}\right)\]]is an isomorphism of unital frchet algebras .
moreover , the map @xmath687 is @xmath0-equivariant .
in fact , we have @xmath688)=\beta(t,\phi_{u_i}([f]))\]]for all @xmath617 and @xmath545\in\gamma\mathbb{a}_{\{f_i\}}$ ] . since the natural isomorphism between the space @xmath689 and @xmath690 is also a @xmath0-equivariant isomorphism of unital frchet algebras ( cf .
proposition [ cocycle description of bundles ] ( a ) ) , it follows from proposition [ trivial ncp t^n - bundles from trivial algebra bundles ] that the corresponding dynamical system @xmath691 of proposition [ t^n action on spec again ] carries the structure of a smooth trivial noncommutative principal @xmath0-bundle .
[ non - triviality of the previous construction ] ( non - triviality of the previous construction ) . in this example
we show that the previous construction actually leads to non - trivial examples .
for this we apply theorem [ ncp g_bundle with fibre trivial ncp g bundle ] to the trivial noncommutative principal @xmath0-bundle @xmath33 . in view of example
[ g - automorphism group of quantumtori ] we have @xmath692 in particular , a similar argument as in ( * ? ? ?
* proposition 2.1.14 ) implies that there is a one - to - one correspondence between the algebra bundles arising from proposition [ cocycle description of bundles ] and principal @xmath0-bundles .
thus , if @xmath693 is such an algebra bundle which corresponds to a non - trivial principal @xmath0-bundle , then also @xmath693 is non - trivial as algebra bundle .
we claim that the associated smooth dynamical system @xmath682 of proposition [ dynamical system and algebra bundles ] is a non - trivial noncommutative principal @xmath0-bundle .
to prove this claim we assume the converse , i.e. , that @xmath682 is a trivial noncommutative principal @xmath0-bundle .
for this , we rename the open subsets @xmath620 of @xmath4 of theorem [ ncp g_bundle with fibre trivial ncp g bundle ] to @xmath694 in order to avoid an abuse of notation and proceed as follows : \(i ) we first recall that @xmath31 is generated by unitaries @xmath24 and that its elements are given by ( norm - convergent ) sums @xmath32here , @xmath695
\(ii ) for each @xmath651 let @xmath696 be the isotypic component corresponding to the canonical basis element @xmath697 of @xmath35 .
then the definition of the action @xmath525 implies that @xmath698 where @xmath699 denotes the isotypic component of the fibre @xmath700 corresponding to @xmath701 . since @xmath682 is assumed to be a trivial noncommutative principal @xmath0-bundle , we may choose in each space @xmath696 an invertible element @xmath702 .
\(iii ) if @xmath703 ( cf . construction [ top on space of sections ] ) , then @xmath704 for some smooth function @xmath705 .
moreover , the compatibility property for @xmath706 and ( [ referenz g - automorphism group of quantumtori ] ) imply that @xmath707 for all @xmath285 . hence
, we may choose a section @xmath702 which is locally given by @xmath708 for some smooth function @xmath709 .
\(iv ) we now show that the map @xmath710is an equivalence of algebra bundles over @xmath4 .
indeed , @xmath145 is bijective and fibrewise an algebra automorphism .
moreover , the map @xmath145 is smooth if and only if the map @xmath711is smooth for each @xmath281 . since @xmath712we conclude that @xmath713 , i.e. , that @xmath714 is smooth as a composition of smooth maps .
a similar argument shows the smoothness of the inverse map .
\(v ) we finally achieve the desired contradiction : in view of part ( iv ) , @xmath267 is a trivial algebra bundle contradicting the construction of @xmath267 , i.e. , that @xmath267 is non - trivial as algebra bundle .
this proves the claim .
[ nontriviality of the bundle corresponding to a^2 ] in this remark we want to point out that there exist non - trivial algebra bundles which are trivial as noncommutative principal torus bundles : in fact , ( * ? ? ?
* corollary 12.7 ) implies that the @xmath715-tori @xmath716 are mutually non - isomorphic for @xmath717 .
further , if @xmath21 is rational , @xmath718 , @xmath719 , @xmath720 relatively prime , then @xmath716 is isomorphic to the algebra of continuous sections of an algebra bundle over @xmath721 with fibre @xmath722 ( cf .
* proposition 12.2 ) or @xcite , proposition e.2.5 for the smooth case ) .
therefore , the ( continuous ) bundle corresponding to such a rational quantum torus @xmath716 , @xmath21 rational with @xmath723 , is non - trivial as algebra bundle .
since the smooth noncommutative @xmath715-torus @xmath724 is a dense subalgebra of @xmath716 , the same conclusion holds for the ( smooth ) bundle corresponding to @xmath724
. nevertheless , the associated dynamical systems @xmath725 and @xmath726 of example [ nc n - tori as ncptb ] are trivial noncommutative principal @xmath721-bundles .
we show that if @xmath2 is a unital frchet algebra with trivial center , @xmath59 an algebra bundle and @xmath727 a principal @xmath0-bundle , then the algebra of sections of the pull - back bundle @xmath728is a noncommutative principal @xmath0-bundle .
we start with the following lemma : [ pull - back 0 ] if @xmath2 is a unital locally convex algebra and @xmath4 a manifold , then the map @xmath729defines a smooth action of @xmath0 on @xmath730 by algebra automorphisms .
in particular , the triple @xmath731 is a smooth trivial noncommutative principal @xmath0-bundle . for the proof we just have to note that the algebra @xmath732 is naturally embedded in @xmath730 through the unit element of @xmath2 .
the smoothness of the map @xmath26 can be proved similarly to lemma [ dynamical system and trivial algebra bundles ] .
[ pull - back i ] if @xmath59 is an algebra bundle , @xmath733 a principal bundle and @xmath734 the pull - back bundle over @xmath8 , then the map @xmath735defines a smooth action of @xmath7 on @xmath734 .
we first note that the map @xmath610 is well - defined .
its smoothness follows from local considerations and we leave the details to the reader . [ pull - back ii ] suppose we are in the situation of lemma [ pull - back i ] .
if @xmath736 , then the map @xmath737defines a smooth action of @xmath7 on @xmath738 by algebra automorphisms . in particular , the triple @xmath739 is a smooth dynamical system .
an easy calculation shows that the map @xmath26 is a well - defined action of @xmath7 on @xmath738 by algebra automorphisms .
next , we choose a bundle atlas @xmath278 of @xmath59 with the additional property that each @xmath740 is trivial .
then @xmath741 is a bundle atlas of the pull - back bundle @xmath734 over @xmath8 and we can use the definition of the smooth structure on @xmath738 to verify the smoothness of the map @xmath26 : in fact , remark [ smooth structure on sections ] implies that @xmath26 is smooth if and only if each map @xmath742is smooth . for this
we note that each map @xmath743is smooth and further that @xmath744 holds for each @xmath281 .
thus , each map @xmath678 is smooth as a composition of smooth maps .
[ pull - back iv ] let @xmath2 be a unital frchet algebra with trivial center , @xmath59 an algebra bundle and @xmath727 a principal bundle . if @xmath734 is the pull - back bundle over @xmath8 and @xmath736 , then the smooth dynamical system @xmath745 of proposition [ pull - back ii ] is a smooth noncommutative principal @xmath0-bundle .
\(i ) we first note that @xmath746 and therefore that @xmath747 ( cf .
* proposition 2.4 ) ) .
in particular , the spectrum @xmath560 is homeomorphic to @xmath4 .
\(ii ) next , we choose an open cover @xmath358 of @xmath4 such that both @xmath748 and @xmath749 are trivial , i.e. , such that @xmath750 and hold for each @xmath281 .
further , we choose for each @xmath281 a @xmath620-defining function @xmath622 and note that each function @xmath623 is a ( smooth ) @xmath751-defining function @xmath622 .
\(iii ) for @xmath17 we choose @xmath281 with @xmath624 . then @xmath625 is an element in @xmath508 with @xmath626 and we conclude from corollary [ algebra section 3 ] that the map @xmath752\mapsto f\circ\left(\frac{1}{h_i}\times\operatorname{id}_{v_i}\right)\]]is an isomorphism of unital frchet algebras .
moreover , the map @xmath753 is @xmath0-equivariant .
in fact , we have @xmath754)=\alpha(t,\phi_{v_i}([f]))\]]for all @xmath617 and @xmath545\in\mathcal{a}_{\{h_i\}}$ ] . since the natural isomorphism between the space @xmath755 and @xmath756 is also a @xmath0-equivariant isomorphism of unital frchet algebras , the dynamical system @xmath757 carries the structure of a smooth trivial noncommutative principal @xmath0-bundle ( cf .
lemma [ pull - back 0 ] ) .
[ non - triviality of the previous construction for example 3 ] ( non - triviality of the previous construction ) . in this example
we show that the previous construction actually leads to non - trivial examples
. therefore let @xmath758 be the special unitary group of rank @xmath759 .
it is a well - known fact that @xmath758 is a 1-connected , i.e. , connected and simply - connceted , lie group ( cf .
* proposition 16.1.3 ) ) .
moreover , @xmath758 contains a maximal torus of dimension 2 . identifying @xmath721 with this torus leads to a non - trivial principal bundle @xmath760 , where denotes the canonical quotient map and @xmath610 is the natural subgroup action .
next , let us consider the trivial algebra bundle over @xmath761 with fibre @xmath762 .
the pull - back along @xmath760 leads to the trivial algebra bundle over @xmath758 with fibre @xmath762 .
we claim that the associated smooth dynamical system @xmath763 of proposition [ pull - back ii ] is a non - trivial noncommutative principal @xmath721-bundle . here
we have , of course , .
to prove this claim we assume the converse , i.e. , that @xmath763 is a trivial noncommutative principal @xmath721-bundle and proceed as follows : \(i ) since @xmath763 is assumed to be a trivial noncommutative principal @xmath721-bundle , there exist invertible elements @xmath764 and @xmath765 .
\(ii ) part ( i ) now implies that @xmath766 .
since @xmath767 is invertible , so is @xmath43 , i.e. , @xmath43 takes values in @xmath768 . in view of (
* proposition 1.9 ) , we may assume that @xmath43 takes values in @xmath769 .
moreover , the function @xmath43 satisfies @xmath770the same construction applied to @xmath771 gives an invertible element @xmath772 which takes values in @xmath769 and satisfies @xmath773here we have used that the action of @xmath721 on @xmath738 restricts to an action on @xmath774 .
\(iii ) let @xmath775 , @xmath776 be the @xmath3-fold covering map of the @xmath591-torus . since @xmath758 is simply connected , we can lift @xmath43 to a smooth function @xmath777 satisfying @xmath778 .
further , a short observation shows that @xmath779 .
in particular , we obtain an invertible element @xmath780 which takes values in @xmath769 .
a similar construction leads to an invertible element @xmath781 which takes values in @xmath769 .
\(iv ) finally , part ( iii ) leads to the desired contradiction : indeed , we conclude that the smooth functions @xmath782 and @xmath783 define an equivalence of principal @xmath721-bundles over @xmath761 given through the map @xmath784this is not possible , since @xmath761 is simply - connected .
\(a ) for a discussion on infinitesimal objects for the noncommutative space @xmath785 we refer to [ mas08 ] , section 3 or ( * ? ? ?
* section 11.3 ) .
\(b ) the same argument as in example [ non - triviality of the previous construction for example 3 ] works for every principal bundle @xmath14 which has a simply connected total space @xmath8 .
we now consider again a principal bundle @xmath9 and , in addition , a unital locally convex algebra @xmath2 .
if @xmath786 defines a smooth action of @xmath7 on @xmath2 by algebra automorphisms , then @xmath787defines a ( free ) action of @xmath7 on @xmath788 and one easily verifies that the trivial algebra bundle @xmath789 is @xmath7-equivariant .
moreover , a short observation shows that the map @xmath790defines a smooth action of @xmath7 on @xmath791 by algebra automorphisms .
we recall that the corresponding fixed point algebra is isomorphic ( as @xmath5-module ) to the space of sections of the associated algebra bundle @xmath792over @xmath4 .
in fact , we refer to ( * ? ? ?
* construction 2.1.13 ) if @xmath2 is finite - dimensional .
if @xmath2 is not finite - dimensional , then one can use that bundle charts for @xmath9 induces bundle charts for the associated algebra bundle .
[ v.1 ] the above situation applied to the trivial principal bundle @xmath793 leads to a smooth trivial noncommutative principal @xmath0-bundle @xmath794 with fixed point algebra @xmath78 . for the proof we again just have to note that the algebra @xmath732 is naturally embedded in @xmath795 through the unit element of @xmath2
the smoothness of the map @xmath26 can be proved similarly to lemma [ dynamical system and trivial algebra bundles ] .
[ v.2 ] if @xmath727 is a principal bundle , @xmath2 a unital frchet algebra with trivial center and @xmath786 a smooth action of @xmath7 on @xmath2 , then the smooth dynamical system @xmath796 is a smooth ncp @xmath7-bundle .
\(i ) we first note that @xmath618 ( cf .
* proposition 2.4 ) ) .
hence , @xmath560 is homeomorphic to @xmath4 by proposition [ spec of c(m ) top ] .
\(ii ) next , we choose an open cover @xmath358 of @xmath4 such that each @xmath797 is a trivial principal @xmath0-bundle over @xmath620 , i.e. , such that @xmath798 holds for each @xmath281 .
further , we choose for all @xmath281 a @xmath620-defining function @xmath622 and note that each function @xmath623 is a ( smooth ) @xmath799-defining function @xmath622 .
\(iii ) for @xmath17 we choose @xmath281 with @xmath624 . then @xmath625 is an element in @xmath508 with @xmath626 and we conclude from corollary [ algebra section 3 ] that the map @xmath800\mapsto f\circ\left(\frac{1}{h_i}\times\operatorname{id}_{p_i}\right)\]]is an isomorphism of unital frchet algebras .
moreover , the map @xmath801 is @xmath0-equivariant .
in fact , we have @xmath802)=\alpha(t,\phi_{p_i}([f]))\]]for all @xmath617 and @xmath545\in c^{\infty}(p , a)_{\{h_i\}}$ ] . since the natural isomorphism between @xmath803 and @xmath756 is also a @xmath0-equivariant isomorphism of unital frchet algebras , the dynamical system @xmath804 carries the structure of a smooth trivial noncommutative principal @xmath0-bundle ( cf .
lemma [ v.1 ] ) .
[ non - triviality of the previous construction for example 4 ] ( non - triviality of the previous construction ) .
non - trivial examples can be constructed similarly as in example [ non - triviality of the previous construction for example 3 ] .
let @xmath805 be the universal covering of @xmath769 , @xmath806 and note that the map @xmath807defines a ( smooth ) action of @xmath808 on @xmath2 .
we thus can form the associated algebra bundle @xmath809\mapsto\pi(r),\label{associated algebra bundle}\end{aligned}\ ] ] with fibre @xmath2 . the space @xmath58 of section of the bundle _ ( _ [ associated algebra bundle ] _ ) _ carries the structure of a non - trivial noncommutative principal bundle @xmath769-bundle . the proof of this claim is divided into the following five steps : \(i ) we first note that the map @xmath810,\]]is an isomorphism of unital frchet algebras . indeed , both @xmath811 and @xmath58 are unital frchet algebras .
an easy calculation shows that the map @xmath166 is an isomorphism of unital algebras , and its continuity is a consequence of the topology on @xmath58 ( cf .
definition [ top on space of sections ] ) . finally , the open mapping theorem ( cf .
* chapter iii , section 2.2 ) ) implies that @xmath253 is open .
\(ii ) from lemma [ smooth exp law ] , we conclude that @xmath812is isomorphic ( as a unital frchet algebra ) to @xmath813 \(iii ) the action of @xmath808 on @xmath814 corresponding to ( [ equation example ] ) is given by @xmath815moreover , this action is free . to verify its properness , we take two compact subsets @xmath816 and @xmath817 of @xmath814
. then there exists @xmath818 such that the projection of @xmath816 onto @xmath139 is contained in the interval @xmath819 $ ] .
this implies that the set @xmath820 is finite and thus that @xmath610 is proper ( cf .
[ ne08b ] , example 1.2.2 ( a ) ) . in view of the quotient theorem ( cf .
* kapitel viii , abschnitt 21 ) ) , we get a @xmath808-principal bundle @xmath821where @xmath822 and @xmath823 denotes the canonical quotient map .
in particular , we conclude from ( i ) , ( ii ) and the previous discussion that @xmath58 is canonically isomorphic to @xmath5 .
\(iv ) next , one easily verifies that the map @xmath824,z)\mapsto [ ( r , z_1,z_2z)]\]]defines a smooth action of @xmath769 on @xmath4 , which is is free and proper .
we thus get a @xmath769-principal bundle @xmath825 , where @xmath826 denotes the canonical quotient map .
in particular , theorem [ nct^nb for manifold again ] ( b ) implies that the triple @xmath827 is a smooth noncommutative principal bundle @xmath769-bundle .
\(v ) it remains to show that @xmath4 is non - trivial as principal @xmath769-bundle over @xmath828 .
for this it is enough to show that @xmath829 . indeed ,
if @xmath830 , then @xmath831 . we proceed as follows : let @xmath832 be the ( left- ) semidirect product of @xmath139 and @xmath833 defined by the homomorphism @xmath834then the assignment @xmath835is the lifting of the action @xmath610 of part ( iii ) to ( the universal covering ) @xmath836 ( @xmath837 as manifolds ) . in particular , the homogeneous space defined by the discrete subgroup @xmath838 of @xmath832
is diffeomorphic to @xmath4 , i.e. , @xmath839we thus conclude from the corresponding exact sequence of homotopy groups ( cf .
[ br93 ] , theorem vii.6.7 or [ ne08b ] , theorem 6.3.17 ) that the map @xmath840):=(n , m , m'),\]]where @xmath841 is such that @xmath842 holds for a continuous lift of the loop @xmath843 in @xmath4 with @xmath844 , is an isomorphism of groups .
this proves the claim .
in section [ a geometric approach to ncpb ] we have introduced the concept of ( non - trivial ) noncommutative principal torus bundles . since we gave a complete classification of trivial noncommutative principal torus bundles in (
* section 4 ) , it is , of course , a natural ambition to work out a classification theory for ( non - trivial ) noncommutative principal torus bundles . to be more precise ,
the problem is the following : [ open problem classification of ncp torus bundles ] ( classification of noncommutative principal torus bundles ) .
let @xmath181 be a unital locally convex algebra .
work out a good classification theory for noncommutative principal @xmath0-bundles @xmath1 for which @xmath845 , i.e. , classify all noncommutative principal @xmath0-bundles @xmath1 for which @xmath845 .
as is well - know from classical differential geometry , the relation between locally and globally defined objects is important for many constructions and applications . for example , a ( non - trivial ) principal bundle @xmath9 can be considered as a geometric object that is glued together from local pieces which are trivial , i.e. , which are of the form @xmath10 for some ( arbitrary small ) open subset @xmath11 of @xmath4 .
this approach immediately leads to the concept of @xmath7-valued cocycles and therefore to a cohomology theory , called the ech cohomology of the pair @xmath846 .
this cohomology theory gives a complete classification of principal bundles with structure group @xmath7 and base manifold @xmath4 ( cf .
* kapitel ix.5 ) ) .
( a noncommutative ech cohomology ) . in view of the previous discussion , a possible approach to the open problem [ open problem classification of ncp torus bundles ] is to work out a _ noncommutative version _ of the ech cohomology . for this
, it is worth to study the paper @xcite .
moreover , ideas of the classification methods for loop algebras of @xcite might be of particular interest .
also , a possible point of view for the relation between locally and globally defined objects in the noncommutative world can be found in @xcite ; their subspaces are described by ideals and one glues algebras along ideals , using a pull - back construction .
( classification of locally trivial hopf - galois extensions ) . a classification procedure for noncommutative principal torus bundles may as well lead to a classification theory of so - called _ locally trivial hopf galois extensions _ ( cf .
p.m. hajac ( editor ) : quantum symmetries in noncommutative geometry , in preparation ) .
this part of the appendix is devoted to some results concerning the projective tensor product of locally convex spaces .
. tensor product iii ] ( the projective tensor product ) .
let @xmath88 and @xmath767 be two locally convex spaces .
further , let @xmath847 be the algebraic tensor product of @xmath88 and @xmath767 . the _ projective tensor product _
@xmath848 of @xmath88 and @xmath767 is the algebraic tensor product @xmath847 equipped with the strongest locally convex topology for which the canonical projection " @xmath849is continuous .
the corresponding completion of @xmath848 with respect to this topology is denoted by @xmath850 .
[ universal property of projective tensor product topology ] ( universal property of the projective tensor product ) .
we recall that the projective tensor product @xmath848 of two locally convex spaces @xmath88 and @xmath767 has the following universal property : for every locally convex space @xmath508 the canonical ( algebraic ) isomorphism of the space of bilinear maps of @xmath851 into @xmath508 onto the space of linear maps of @xmath848 into @xmath508 induces an ( algebraic ) isomorphism of the space of continuous bilinear maps of @xmath851 into @xmath508 onto the space of continuous linear maps of @xmath848 into @xmath508 . in particular , a bilinear map @xmath852 is continuous if and only if there exists a continuous linear map @xmath853 satisfying @xmath854 . a nice reference for these statements can be found in ( * ? ? ? *
part iii ) .
[ continuity of maps eof to eof ] let @xmath855 and @xmath771 be four locally convex spaces .
further , let @xmath856 and @xmath857 be two continuous linear maps .
then the linear map @xmath858is continuous .
if @xmath859 denotes the canonical projection " , then the claim follows from the fact that @xmath860 is continuous as a composition of continuous maps .
[ aob as l.c .
algebra ] let @xmath2 and @xmath181 be two locally convex algebras .
then @xmath861 becomes a locally convex algebra , when equipped with the multiplication @xmath862 \(i ) suppose that the algebra structure on @xmath2 is given by the continuous bilinear map @xmath863and that the algebra structure on @xmath181 is given by the continuous bilinear map @xmath864then remark [ universal property of projective tensor product topology ] implies that the map @xmath865 , resp .
@xmath866 , induces a continuous linear @xmath867resp . @xmath868
\(ii ) to show that the map @xmath98 defines the structure of a locally convex algebra on @xmath861 , we use remark [ universal property of projective tensor product topology ] again : indeed , the bilinear map @xmath98 is continuous if and only if the linear map @xmath869is continuous . here
, we have used the canonical isomorphism @xmath870now , a short observations shows that @xmath871 , and therefore lemma [ continuity of maps eof to eof ] implies that @xmath98 is continuous .
in the last part of the appendix we discuss some useful results on the smooth exponential law which will be used several times in this paper . for arbitrary sets @xmath41 and @xmath42 , let @xmath872 be the set of all mappings from @xmath41 to @xmath42
. then the following exponential law " holds : for any sets @xmath873 and @xmath508 , we have @xmath874as sets . to be more specific , the map @xmath875is a bijection , where @xmath876the inverse map is given by @xmath877where @xmath878next we consider an important topology for spaces of smooth functions : [ smooth compact open topology ] ( smooth compact open topology ) .
if @xmath4 is a locally convex manifold and @xmath88 a locally convex space , then the _ smooth compact open topology _ on @xmath89 is defined by the embedding @xmath879where the spaces @xmath880 carry the compact open topology . since @xmath881 is a locally convex space isomorphic to @xmath882 , the spaces
@xmath880 are locally convex and we thus obtain a locally convex topology on @xmath89 . it is natural to ask if there exists an exponential law " for the space of smooth functions endowed with the smooth compact open topology , i.e. , if we always have @xmath883 as a locally convex space , for all locally convex smooth manifolds @xmath4 , @xmath322 and locally convex spaces @xmath88 . in general
, the answer is _
no_. nevertheless , it can be shown that ( [ exp law ] ) holds if @xmath322 is a finite - dimensional manifold over @xmath139 . to be more precise : [ smooth exp law ] let @xmath4 be a real locally convex smooth manifold , @xmath322 a finite - dimensional smooth manifold and @xmath88 a topological vector space .
then a map @xmath884 is smooth if and only if the map @xmath885is smooth .
moreover , the following map is an isomorphism of locally convex spaces : @xmath886
|
a ( smooth ) dynamical system with transformation group @xmath0 is a triple @xmath1 , consisting of a unital locally convex algebra @xmath2 , the @xmath3-torus @xmath0 and a group homomorphism , which induces a ( smooth ) continuous action of @xmath0 on @xmath2 . in this paper
we present a new , geometrically oriented approach to the noncommutative geometry of principal torus bundles based on such dynamical systems .
our approach is inspired by the classical setting : in fact , after recalling the definition of a trivial noncommutative principal torus bundle , we introduce a convenient ( smooth ) localization method for noncommutative algebras and say that a dynamical system @xmath1 is called a noncommutative principal @xmath0-bundle , if localization leads to a trivial noncommutative principal @xmath0-bundle .
we prove that this approach extends the classical theory of principal torus bundles and present a bunch of ( non - trivial ) noncommutative examples .
|
within the last decade the need has arisen for a method that incorporates quantum electrodynamics ( qed ) radiative corrections into the many - body problem for heavy atoms and ions .
for example , account of qed radiative corrections , that is , the vacuum polarization and self - energy corrections , were critical in the interpretation of the measurement @xcite of the parity violating amplitude in cesium , which restored an apparent deviation from the standard model of particle physics ( see , e.g. , @xcite ) .
there are other areas where full account of qed and many - body effects are required , including in the transition frequencies of many - electron highly - charged ions ( see , e.g. , @xcite ) and in the spectra and chemical properties of superheavy elements @xcite .
ongoing improvements in experimental precision and developments in many - body methods across a range of precision atomic applications necessitate the accurate treatment of combined many - body and qed effects in the generic atomic problem .
the one - loop radiative corrections are made up of the ( local ) vacuum polarization and the ( non - local ) self - energy terms .
the vacuum polarization is dominated by the ( lowest order in @xmath4 ) uehling potential , and incorporation of this potential into the many - body problem is relatively straightforward .
the rigorous , highly - accurate _ ab initio _ procedures for calculating qed self - energy corrections to properties of hydrogen - like and few - electron atoms and ions can not be applied with the same success to more complex many - body systems ( see , for example , the review @xcite ) .
there are exact qed calculations performed in frozen atomic potentials ( see , e.g. , @xcite ) , though here the many - body aspect of the problem may not be readily determined . on the other hand ,
the qed part of the problem may be simplified and the full many - body treatment employed .
several such `` radiative potentials '' have been advocated for use in many - body calculations .
these include the local potentials @xcite and the more sophisticated non - local potential @xcite , all fitted to self - energy shifts in hydrogenlike ions .
there are a number of recent works where qed radiative corrections have been included into the many - body problem using one or other of these radiative potentials ( e.g. , @xcite )
. however , there has as yet been no in - depth study of the interplay between radiative corrections and many - body effects or of the breakdown of different terms .
therefore , it is important to demonstrate the significance of different many - body effects so that they may be considered appropriately in high - precision calculations .
derevianko _ et al . _
@xcite studied the effects of core relaxation on valence @xmath3 and @xmath2 level shifts of neutral cs due to the uehling potential in the relativistic hartree - fock approximation .
they demonstrated how important these relaxation effects are , particularly for orbitals with orbital quantum number @xmath0 . in this paper
we extend the work of derevianko _ et al .
_ @xcite to include the effects on valence @xmath1 levels .
inclusion of relaxation corrections leads to several orders of magnitude enhancement of the uehling correction to the @xmath1-level binding energies .
we study also the effect of second- and higher - order many - body perturbation theory on the uehling correction to valence binding energies through inclusion of the second - order and all - orders correlation potentials .
we have found significant corrections to uehling shifts for all orbitals .
the underlying mechanisms leading to the large many - body effects are studied and calculations are performed for the series of alkali - metal atoms na through to the superheavy element e119 .
this work serves as a preliminary study of the more general qed - many - body interplay , where the full radiative corrections ( including the larger self - energy ) are considered .
since the uehling potential may be determined to very high precision and is a local operator which may readily be included into atomic many - body codes , the study of the many - body effects is not hindered by any complexity arising from the qed terms themselves . while the form of the self - energy terms is more sophisticated than uehling , and the radiative shift is generally larger and of opposite sign , the short range of the interaction means that the same many - body enhancement mechanisms apply .
the structure of the paper is the following . in section
[ sec : finite ] we present the finite - size form of the uehling potential used in this work . in section
[ sec : first ] , the first - order valence shifts are calculated for frozen atomic cores corresponding to core - hartree , kohn - sham , and hartree - fock approximations .
the many - body effects of core relaxation and valence - core correlations are studied in sections [ sec : relax ] and [ sec : correlations ] , respectively . in these sections ,
details of the many - body enhancement mechanisms are presented and calculations performed for na through to e119 , with more detailed results given for cs .
the uehling potential @xcite gives the lowest - order in @xmath5 vacuum polarization correction to the coulomb potential @xmath6 ; @xmath7 is the nuclear charge and @xmath8 is the fine - structure constant .
for the point - nucleus case , where @xmath9 , the uehling potential may be expressed as @xmath10 here and throughout we use atomic units ( @xmath11 , @xmath12 ) . to obtain the finite - nucleus expression for the uehling potential , the expression for the point - like nucleus
is folded with the nuclear density @xmath13 , @xmath14 where we use the normalization @xmath15 .
taking the nuclear density to be spherically symmetric , and after angular integrations , the well - known finite - nucleus expression for the uehling potential is obtained @xmath16 there are publicly available procedures for performing this integral .
for example , the code of hnizdo @xcite or the expansion of fullerton and rinker @xcite .
we will instead consider the case of the step - function density ( homogeneous charge distribution ) and reduce the double integration , eq .
( [ eq : double ] ) , to a single integration .
the uehling potential then reduces to the following form , divided into parts valid inside and outside the nucleus , @xmath17 \ , & \qquad r\leq r_n \\ \frac{\alpha}{\pi}\frac{z}{r}\int_1^{\infty}dt \sqrt{t^2 - 1}\big(\frac{1}{t^2}+\frac{1}{2t^4}\big ) \big(\frac{2}{\varkappa^3}\big ) e^{-2tr/\alpha } \big [ \varkappa \cosh ( \varkappa)-\sinh(\varkappa)\big ] \ , & \qquad r >
r_n \ \ , \end{array}\ ] ] where @xmath18 is the nuclear radius and @xmath19 .
the first term inside the square brackets of the upper line of eq .
( [ eq : step ] ) may be simplified by integrating over @xmath20 analytically , yielding @xmath21 , where all factors have been included .
( [ eq : step ] ) agrees with that given in ref .
@xcite in a different form . throughout the paper
we take @xmath22 and @xmath23 , where the nuclear root - mean - square radii @xmath24 are taken from ref .
we have found that there is agreement to all digits presented for binding - energy shifts calculated using the step - function density expression for the uehling potential , eq .
( [ eq : step ] ) , compared to the full double integral eq .
( [ eq : double ] ) with a two - parameter fermi distribution used for the nuclear density @xmath25 , provided the same root - mean - square nuclear radius is used . in this work the integration of eq .
( [ eq : step ] ) is performed using the gnu scientific library adaptive integration routine qagi @xcite .
the first - order uehling correction to the binding energies of the valence electron is given by @xmath26 where @xmath27 is the valence electron wave function of state @xmath28 .
the zeroth - order valence energies @xmath29 and wave functions @xmath27 are found by solving the relativistic equations @xmath30 where @xmath31 and @xmath32 are the nuclear and electronic potentials , @xmath33 and @xmath34 are dirac matrices , and @xmath35 is the momentum operator .
we take the nuclear density to correspond to the two - parameter fermi distribution , with the nuclear thickness ( 90% to 10% fall - off ) taken to be @xmath36 for all atoms and the half - density radii found from the root - mean - square radii of ref .
@xcite . for e119
, we take @xmath37 , within the range predicted from hartree - fock - bcs theory @xcite . [ cols="<,<,^,^,^,^,^,^,^ " , ] the effect of the correlation potential is to pull the wave functions closer to the nucleus .
this leads to larger uehling shifts for all waves and is more important for the orbitals with @xmath0 .
the difference between the results using @xmath38 or @xmath39 is significant , affecting the second digit in the results .
however , the fitted results shown in columns five and eight of table [ tab : sigma ] , agree very well , to about 2% for @xmath40 and significantly better for the other waves .
that is , _ as long as the correlation potential is fitted to reproduce the experimental energies , the correlation corrections to the uehling shifts are insensitive to the details of @xmath41_. therefore , we have performed calculations for na through to e119 with the second - order correlation potential @xmath38 fitted to reproduce experimental energies ( for e119 , we use fitting factors from fr ) .
these results , which also include the important core - relaxation effects , are presented in the sixth column of table [ tab : all - order ] . in the final column of table [
tab : all - order ] we present , along with the results of ref .
@xcite , the vacuum polarization calculations performed by thierfelder and schwerdtfeger @xcite in the relativistic hartree - fock approach .
their results are in reasonable agreement with our relaxed core results @xmath42 .
we are not aware of other many - body vacuum polarization results for the alkali atoms .
( see inset ) . ]
it is worth emphasizing the effect of the correlation potential on the valence electron wave functions .
the wave functions are modified by the correlation potential in the region @xmath43 ; see fig .
[ fig : cs6sorbitals ] , where we illustrate the relativistic hartree - fock and brueckner upper radial component for the @xmath40 orbital of cs . for @xmath44 ,
the brueckner wave functions are proportional to the hartree - fock wave functions , @xmath45 .
the effect that the correlation corrections have in this region is simply in the normalization of the wave functions . obviously , since the uehling potential acts at a distance close to the nuclear radius , @xmath46 . if the range of the correction to the hartree - fock potential @xmath47 were within @xmath48 , then @xmath49 also , i.e. , @xmath50 .
we do notice some deviation from this relation , which is a result of the long range of the relaxation correction to the hartree - fock potential ; this range indeed extends beyond @xmath51 , as we saw in the previous section .
overall , though , the relation above holds fairly well . from consideration of the level of agreement between the fitted brueckner results of cs from table [ tab : sigma ]
, we estimate the uncertainty of our final results for the alkali atoms na to e119 ( column six of table [ tab : all - order ] ) to be a few percent .
while the aim of this work is to study the effect of many - body corrections on the qed radiative shifts , applied to the uehling potential specifically , we note that there are other contributions to the vacuum polarization shift at the one - loop level that we have not included , for instance the higher - order in @xmath4 corrections and remaining electron screening effects corresponding to a vacuum polarization loop in the photon exchange between electrons .
a more precise study would include these contributions , though we expect them to contribute at a level that is smaller than the estimated uncertainty in this work .
we have studied the mechanisms of core relaxation and valence - core correlations on the uehling shift for the valence levels of the alkali atoms na to e119 .
we observe sizeable corrections to the @xmath3-level shifts , while for the other orbitals @xmath0 the relative size of the corrections is larger . for the @xmath52 orbitals we have seen for several atoms that the relaxation and first - order uehling shifts are roughly the same size and of opposite sign . for @xmath53 and for the @xmath1 levels ,
the effect of core relaxation is enormous , with corrections being orders of magnitude larger than the first - order result and usually of opposite sign .
account of the valence - core correlations is also important for all waves . for rb , cs , and
fr , the @xmath1-level shifts become comparable to the size of the @xmath3-level shifts ( one order of magnitude smaller ) .
the atomic theory uncertainty for calculations of transition frequencies in heavy neutral alkali atoms is at the level of 0.1% @xcite , limited by the incomplete treatment of electron - electron correlations .
this is roughly the same level where the radiative corrections enter .
( the full radiative shift is comprised , at one - loop level , of the vacuum polarization and self - energy shifts , with the self - energy shift typically an order of magnitude larger than the vacuum polarization shift and of opposite sign . ) if the remaining uncertainty in the electronic theory can be controlled , the huge many - body corrections for the radiative shifts for @xmath1 levels could make high - precision studies of transition frequencies involving these levels a particularly sensitive test of combined qed and many - body effects .
the results of this work are relevant for studies of radiative potentials potentials that mimic self - energy qed radiative corrections for use in heavy atoms and ions .
the many - body enhancement mechanisms that operate for the uehling potential apply also to the self - energy due to the short range of the interaction .
we are grateful to a. yelkhovsky for useful discussions .
the all - order correlation potential @xmath39 was calculated using the dzuba - flambaum - sushkov codes maintained by v. dzuba .
this work was supported in part by the australian research council , grant de120100399 .
a. kramida , yu .
ralchenko , j. reader , and nist asd team ( 2014 ) .
nist atomic spectra database ( ver .
5.2 ) , [ online ] .
available : http://physics.nist.gov/asd .
national institute of standards and technology , gaithersburg , md .
|
we calculate vacuum polarization corrections to the binding energies in neutral alkali atoms na through to the superheavy element e119 .
we employ the relativistic hartree - fock method to demonstrate the importance of relaxation of the electronic core and the correlation potential method to study the effects of second and higher orders of perturbation theory .
these many - body effects are sizeable for all orbitals , though particularly important for orbitals with angular momentum quantum number @xmath0 .
the orders of magnitude enhancement for @xmath1 waves produces shifts that , for rb and the heavier elements , are larger than those for @xmath2 waves and only an order of magnitude smaller than the @xmath3-wave shifts .
the many - body enhancement mechanisms that operate for vacuum polarization apply also to the larger self - energy corrections .
|
in @xcite , chuang and tan determine the loewy structures of specht modules in defect 2 over odd characteristic .
the authors prove results for the schur algebra , and then translate to results for the symmetric group by applying the schur functor .
the authors mention that the symmetric group results may be proven entirely in the representation theory of the symmetric group . throughout this paper @xmath3
is the symmetric group on @xmath4 letters and @xmath0 is a field of characteristic @xmath1 at least 5 .
the goal of this paper is to establish some loewy structure results for specht modules which lie in the principal block of @xmath2 . for the remainder of the paper , the principal block of @xmath2
will be denoted by @xmath5 .
weight @xmath6 specht modules are well understood .
when @xmath7 the specht modules are simple and projective , and each @xmath1-block contains exactly one specht module . for @xmath8
the ext@xmath9-quiver is a line segment , and the specht modules have loewy length and composition length at most 2 . over odd characteristic , if @xmath10 then a specht module has loewy length at most 3 and composition length at most 5 ( cf .
proposition 6.2 of @xcite and theorem 5.6 of @xcite ) .
in this paper we prove four structure results for specht modules which lie in the principal block of @xmath2 .
the first result shows that specht modules in the block have loewy lengths bounded above by 4 .
the second result is a necessary and sufficient condition on a partition @xmath11 so that the corresponding specht module ( in the principal block ) attains the maximum possible loewy length in corollary [ cor : nscondition ] .
the third result shows that if @xmath12 is a reducible specht module in the principal block with @xmath11 @xmath1-regular then its second radical layer is determined by the ext@xmath9-quiver of the block .
for such a partition @xmath11 we also prove that if @xmath12 has loewy length 3 then @xmath13 .
finally , the fourth result gives an ( attained ) upper bound of 14 on the composition length for specht modules in the block .
the approach taken to prove the mentioned results relates partitions in the principal block of @xmath2 to partitions in weight 2 blocks of @xmath14 then inducing back up to the principal block .
this approach is motivated by the work of martin and russell in @xcite , where the authors show that a substantial portion of the @xmath15-quiver for the principal block of @xmath2 is built from @xmath15-quivers of the defect 2 blocks of @xmath14 . for the case
@xmath16 , the ext@xmath9-quiver of the principal block , @xmath17 , is constructed in @xcite .
the author gives the loewy structures of the projective indecomposable modules of @xmath18 ( cf .
theorems 2.15 , 2.17 of @xcite ) .
since every projective indecomposable module of the symmetric group is a young module one is able to deduce the following : if @xmath11 is a partition in @xmath18 that is both @xmath19-regular and @xmath19-restricted then @xmath12 has loewy length 4 . now
if @xmath11 is a partition in @xmath18 that is @xmath19-regular and not @xmath19-restricted then one may apply lemma 2.14 of @xcite and the arguments of lemmas [ lem : sinduce ] , [ lem : nicerest ] , [ lem : getit ] , and theorem [ thm : gotit ] of this paper to conclude @xmath12 has loewy length at most 3 .
so the specht modules in @xmath18 have loewy length at most 4 , and a specht module having loewy length 4 necessarily corresponds to a partition that is both 3-regular and 3-restricted .
the representations in the blocks @xmath18 and @xmath5 appear to behave similarly even though @xmath18 has non - abelian defect group and @xmath5 has abelian defect group .
in fact , after analyzing the ext@xmath9-quiver of @xmath18 , we see proposition [ prop : mullextend ] and corollary [ cor : ks ] of this paper are also true for @xmath18 .
in section [ section : prelim ] we recall some background material in the representation theory of the symmetric group . in section [ section : partitions ]
we prove a few results for certain weight 3 partitions of @xmath20 . in section [ section : irreducibles ]
we classify all specht modules lying in certain defect 2 blocks of @xmath21 that are irreducible .
we then determine the specht modules in @xmath22 that have an irreducible restriction to a defect 2 block of @xmath14 .
section [ section : loewy ] contains the two main results of the paper , corollaries [ cor : nscondition ] and [ cor : ks ] .
in addition , we see that the composition factors of the heart of the specht module @xmath12 are determined by the ext@xmath9-quiver of @xmath5 whenever @xmath11 is both @xmath1-regular and @xmath1-restricted ( with help from the mullineux map ) . in section [ section :
bound ] we prove that the largest attainable composition length for any specht module in the principal block of @xmath2 is 14 .
finally , in section [ section : conclude ] we summarize the results of this paper and discuss future research for specht modules in arbitrary blocks of weight 3 .
a partition of @xmath4 is a sequence of non - negative integers @xmath23 such that @xmath24 for all @xmath25 and @xmath26 . for brevity ,
write @xmath27 if @xmath11 is a partition of @xmath4 .
assume @xmath11 and @xmath28 are partitions of @xmath4 .
say @xmath11 _ dominates _
@xmath28 , and write @xmath29 , if @xmath30 for all @xmath31 .
if @xmath29 and @xmath32 , we say @xmath11 _ strictly dominates _
@xmath28 and write @xmath33 .
this gives a partial order on the partitions of @xmath4 called the _
dominance order_. finally , let @xmath34 denote the lexicographic order on partitions .
let @xmath35 be a partition of @xmath4 .
the _ young diagram _ for @xmath11 is the subset of @xmath36 : @xmath37 = \{(i , j ) \in \mathbb{n } \times \mathbb{n } : 1\leq i \leq k , \
1\leq j \leq \lambda_i \}.\ ] ] there is a natural correspondence between partitions of @xmath4 and young diagrams with @xmath4 boxes .
we identify a partition @xmath11 with its young diagram and write @xmath11 to mean @xmath38 $ ] .
let @xmath39 denote the conjugate partition to @xmath11 , that is , the partition with young diagram : @xmath40 = \{(j , i ) \in \mathbb{n } \times \mathbb{n } : 1\leq i \leq k , \
1\leq j \leq \lambda_i \}.\ ] ] in other words , @xmath41 where @xmath42 is the number of boxes in the @xmath43th column of the young diagram for @xmath11
. a node @xmath44 of @xmath11 is called a _ removable _
node for @xmath11 if @xmath45 is a partition . in this case
define @xmath46 .
a node @xmath47 is called an _ addable _
node for @xmath11 if @xmath48 is a partition . in this case
define @xmath49 .
if @xmath50 and @xmath51 are nodes we say @xmath52 is _ above _ @xmath53 or @xmath53 is _ below _
@xmath52 if @xmath54 . fix a prime @xmath1 . define the @xmath1-residue of a node @xmath55 in @xmath36 to be @xmath56 .
we denote the @xmath1-residue of the node @xmath53 by res@xmath53 .
we refer to a node of @xmath1-residue @xmath25 as an @xmath25-node .
a removable @xmath25-node @xmath53 is _ normal _ if there is an injective function from the set of addable @xmath25-nodes above @xmath53 to the set of removable @xmath25-nodes above @xmath53 which maps each addable node @xmath52 to a node @xmath57 below @xmath52 .
finally , a removable @xmath25-node is called _ good _ if it is the lowest normal @xmath25-node .
.1 in * example .
* let @xmath58 and @xmath59 .
the removable nodes of @xmath11 are @xmath60 , @xmath61 , @xmath62 , and @xmath63 , and we fill the box at node @xmath64 in the young diagram with res@xmath64 .
we also mark each addable node of @xmath11 with its 5-residue .
011 ( 5,6,4,2 ^ 2,1 ^ 2 ) now @xmath65 is a normal @xmath66-node since there is no addable @xmath66-node above @xmath65 .
similarly , @xmath67 is a normal @xmath68-node .
the node @xmath69 is not a normal @xmath19-node since there is an addable @xmath19-node @xmath52 above it but no removable @xmath19-node @xmath57 strictly between @xmath53 and @xmath52 .
similarly , @xmath70 is not a normal @xmath66-node .
both @xmath65 and @xmath67 are good nodes .
a partition @xmath71 is _
@xmath1-regular _ if there does not exist @xmath72 such that @xmath73 .
if such @xmath25 exists then call @xmath11 a _ @xmath1-singular _ partition . finally , @xmath11 is _ @xmath1-restricted _ if @xmath39 is @xmath1-regular .
-.065 in suppose @xmath74 is a node of the partition @xmath11 .
we define the @xmath74-_hook length _ , denoted @xmath75 , to be the number of boxes in the young diagram that sit either to the right of @xmath74 in the @xmath25th row or beneath @xmath74 in the @xmath43th column including @xmath74 .
more precisely , @xmath76 .
the _ hook - length diagram _ of @xmath11 is the young tableau of shape @xmath11 where @xmath75 is the @xmath74-entry . for any integer @xmath77 ,
let @xmath78 be the largest power of @xmath1 dividing @xmath77 .
we define the @xmath1-_power diagram _ of @xmath11 to be the young tableau of shape @xmath11 where @xmath79 is the @xmath74-entry .
.1 in * example . *
the hook - length diagram and 5-power diagram of @xmath58 are given below .
( we omit the zeros in the @xmath1-power diagram . )
( < 11><10><5><4><2><1>,8521,52,41,2,1 ) ( ; ; ; 1;;;,;;1;;,1 ; , ; ; , ; , ; ) the @xmath1-power diagram of @xmath11 plays an important role when trying to determine whether an ordinary representation of a symmetric group remains irreducible over a field characteristic @xmath1 .
if @xmath11 is a partition of @xmath4 , the associated specht is denoted by @xmath12 ( cf .
the permutation module @xmath80 has the unique indecomposable summand @xmath81 containing @xmath12 .
the module @xmath81 is called a _ young _ module . over characteristic
zero @xmath82 is a complete list of non - isomorphic simple modules of @xmath83 .
over positive characteristic @xmath1 the list @xmath84 is a complete list of ( non - isomorphic ) simple @xmath83-modules ( cf .
@xcite ) .
for any @xmath83-module @xmath85 , let @xmath86 denote the @xmath83-module dual to @xmath85 .
if @xmath85 has an @xmath87-module filtration @xmath88 with @xmath89 we write @xmath90 to represent the filtration . if the module @xmath85 is filtered by some modules @xmath91 we may write @xmath92 where @xmath93 is the multiplicity of the factor @xmath94 . for
a simple @xmath87-module @xmath95 , let @xmath96 $ ] denote the multiplicity of @xmath95 as a composition factor of @xmath85 .
finally , let @xmath97 denote the one - dimensional representation of @xmath3 defined by : for @xmath98 , @xmath99 if @xmath100 is an even permutation or @xmath101 if @xmath100 is an odd permutation .
we list two well known facts .
[ prop : james ] let @xmath0 be any field . 1 .
( @xcite , theorem 4.9 ) if @xmath95 is a simple @xmath83-module then @xmath102 .
( @xcite , theorem 8.15 ) if @xmath11 is a partition of @xmath4 then @xmath103 .
the next important result is due to gordon james .
[ prop : rad ] assume @xmath11 and @xmath28 are @xmath1-regular partitions of @xmath4 .
then : 1 . @xmath104 = 1 $ ] , and 2 .
@xmath105 \neq 0 $ ] implies @xmath106 .
we say that the specht module @xmath12 is a _ hook representation _ if @xmath107 .
over odd characteristic hook representations are either irreducible or have two composition factors . in the latter case
the representation is no longer irreducible , yet it remains indecomposable .
we summarize the work of peel in the next proposition .
( @xcite ) [ prop : peel ] suppose @xmath1 is an odd prime .
if @xmath1 divides @xmath4 then part of the decomposition matrix of @xmath3 is given below .
[ cols="^,^,^,^,^,^,^ " , ] peel also proved that the hook representations remain irreducible if @xmath1 does not divide @xmath4 ; however , we only need the case of when @xmath1 divides @xmath4 for this paper . in particular , over odd characteristic hook representations have loewy length at most 2 and at most 2 composition factors .
we cite a theorem of kleshchev and sheth .
[ prop : ks ] let @xmath108 and let @xmath11 and @xmath28 be partitions of @xmath4 with at most @xmath109 nonzero parts . if @xmath11 does not strictly dominate @xmath28 then @xmath110 it remains open whether proposition [ prop : ks ] holds without the assumption of at most @xmath109 parts . in section
[ section : loewy ] we see that one may remove the restriction on the number of parts for @xmath1-regular partitions in @xmath5 . in @xcite ,
james and mathas introduced a family of partitions , now called @xmath1-jm partitions , and conjectured that a specht module @xmath12 remains irreducible over a field of odd characteristic @xmath1 if and only if @xmath11 is a member of this family of partitions .
this conjecture was proved in @xcite .
we now define what it means to be a @xmath1-jm partition .
a partition @xmath11 is a _
@xmath1-jm - partition _ if the following property holds : whenever @xmath74 is a node of the young diagram of @xmath11 and @xmath111 either all entries in row @xmath25 of the @xmath1-power diagram are equal or all entries in column @xmath43 of the @xmath1-power diagram are equal . ( @xcite )
[ prop : irredspecht ] let @xmath1 be an odd prime .
then the specht module @xmath12 is irreducible over a field of characteristic @xmath1 if and only if @xmath11 is a @xmath1-jm partition .
.1 in * example . *
consider the partition @xmath58 from section [ subsection:22 ] .
we give the 5-power and 3-power diagrams of @xmath11 below .
we see that @xmath11 is not a 5-jm partition , but it is ( vacuously ) a 3-jm partition . hence @xmath112 is reducible over characteristic five and irreducible over characteristic three .
the _ rim _ of @xmath11 is the collection of all nodes @xmath74 of @xmath11 such that @xmath113 is not a node of @xmath11 .
a _ removable rim @xmath1-hook _ for @xmath11 is a connected sequence of @xmath1-many nodes in the rim of @xmath11 , say @xmath114 , such that @xmath115 is a partition .
we may repeatedly remove rim @xmath1-hooks from @xmath11 to obtain a partition @xmath116 that has no removable rim @xmath1-hook .
the partition @xmath116 is the _
@xmath1-core _ of @xmath11 .
the number of @xmath1-hooks removed , say @xmath117 , until we reach the @xmath1-core is called the _ @xmath1-weight _ ( or weight ) of the partition .
.1 in * example .
* let @xmath59 and @xmath118 . the young diagram for @xmath11
is given below , and the collection of boxes marked @xmath119 represents the rim of @xmath11 .
( ; ; ; ; , ; ; , ) the following sequence demonstrates the repeated removal of rim 5-hooks until the 5-core is reached . the collection of boxes marked @xmath120 represents the rim 5-hook that is to be removed at each stage .
@xmath121 so @xmath118 has 5-core @xmath122 .
obtaining the 5-core of @xmath11 required removing a total of 2 rim 5-hooks , so the weight of @xmath11 is 2 .
the @xmath1-blocks of the symmetric group are classified by a theorem commonly referred to as ` the nakayama conjecture . '
the conjecture of nakayama was first proved simultaneously by brauer and robinson .
we state the result in the next proposition .
[ prop : n ] the specht modules @xmath12 and @xmath123 lie in the same @xmath1-block of @xmath87 if and only if @xmath11 and @xmath28 have the same @xmath1-core .
thus the @xmath1-blocks of the symmetric group are labeled by @xmath1-cores .
the @xmath1-core of @xmath11 corresponds to a @xmath1-block @xmath52 which contains the specht module @xmath12 ( and all of its composition factors ) . in this case , we say @xmath11 is in the block @xmath52 . now suppose @xmath28 is a partition of @xmath4 with the same @xmath1-core as @xmath11 and also assume @xmath11 has weight @xmath117 .
then @xmath28 is in @xmath52 , and @xmath28 is also a weight @xmath117 partition of @xmath4 .
hence all partitions of @xmath4 in @xmath52 necessarily have the same @xmath1-weight @xmath117 .
we say @xmath117 is the weight of the block @xmath52 .
the principal block , @xmath5 , of @xmath2 is the unique @xmath1-block of @xmath2 containing the trivial module @xmath124 . by the nakayama conjecture ,
a specht module @xmath12 is in @xmath5 if and only if @xmath11 is a partition of @xmath125 that has empty @xmath1-core .
equivalently , @xmath11 is a weight 3 partition of @xmath20 .
we give a fact about the decomposition numbers for the symmetric group for weights 2 and 3 in the next proposition .
( @xcite theorem i , @xcite theorem 1.1)[prop : decnumb ] suppose @xmath11 and @xmath28 are weight @xmath117 partitions of @xmath4 with @xmath28 @xmath1-regular and @xmath126 or @xmath19 .
then @xmath105 \leq 1 $ ] .
a convenient way to display a partition is on an abacus with vertical runners labeled @xmath127 ( from left to right ) .
we will refer to this labeling of the runners as the _ left - to - right _ numbering .
we adopt the convention of calling the first position ( upper left - hand corner ) on the the display ` position 1 ' . so the positions on the first column of the abacus are @xmath128 , et cetera , and the positions of the first two rows are @xmath129 .
for example , if @xmath130 then the positions of the abacus are as follows .
@xmath131 now suppose @xmath132 and @xmath133 are positions on an abacus display for @xmath11 .
if @xmath134 , we either say that position @xmath133 comes after position @xmath132 or position @xmath132 comes before position @xmath133 .
suppose the partition @xmath135 has @xmath136 non - zero parts .
for an integer @xmath137 define the _ beta - number _
@xmath138 for @xmath139 with the convention that if @xmath140 then @xmath141 for @xmath142 .
we may use this set of beta - numbers for @xmath11 to form an abacus display for @xmath11 which has a bead ` @xmath120 ' at position @xmath143 for each @xmath144 and ` @xmath145 ' at any position @xmath146 that is not a member of the specified set of beta - numbers .
we call a position of the display having ` @xmath120 ' _ occupied _ and a position with ` @xmath145 ' _ unoccupied _ ( or empty ) .
note that one may see the beta - number defined as : @xmath147 .
our definition ensures the beta - numbers for partitions in @xmath5 are at least one when @xmath148 .
.1 in * example .
* let @xmath130 , @xmath149 , and @xmath150 .
we represent @xmath11 on an abacus with 15 beads .
then a set of beta - numbers of @xmath11 is : @xmath151 , @xmath152 , @xmath153 , @xmath154 , @xmath155,and @xmath156 for @xmath157 . for each @xmath158
we place a bead at position @xmath143 . the abacus display for @xmath11
is given in figure [ fig : fig1 ] .
@xmath159 fix an abacus display for the partition @xmath11 .
suppose the display has a bead at position @xmath160 and no bead at position @xmath161 .
pushing said bead ` to the left ' to position @xmath161 corresponds to removing a ( removable ) node from @xmath11 .
call such a bead _
removable_. now suppose there is a bead at position @xmath162 and no bead at position @xmath163 .
pushing the bead ` to the right ' to position @xmath163 corresponds to adding an addable node to @xmath11 .
call such a bead _
finally , we call a bead on the display at position @xmath164 _ proper _ if there is an unoccupied position before position @xmath164 .
otherwise , we call a bead _ improper_. for example , referring to figure [ fig : fig1 ] , we see that the abacus display has removable beads at positions 12 , 14 , and 21 and addable beads at positions 12 , 15 , and 22 .
also , the beads at positions 1 through 10 are all improper while all other beads are proper .
suppose an abacus display for @xmath11 has a removable bead in row @xmath165 of runner @xmath25 . if @xmath166 we call the removable bead _
normal _ provided that for any @xmath167 the number of beads on runner @xmath25 in rows @xmath168 is at least the number of beads on runner @xmath169 in rows @xmath168 .
if @xmath170 we call the bead _ normal _ provided that for any @xmath167 the number of beads on runner @xmath170 in rows @xmath171 is at least the number of beads on runner @xmath1 in rows @xmath172 .
we call such a bead _ normal _ because it corresponds to a normal node of @xmath11
. now suppose the display has a bead at position @xmath173 and no bead at position @xmath174 ( i.e. , one row above the bead on the same runner ) .
pushing the bead ` up ' to position @xmath175 corresponds to removing a ( removable ) rim @xmath1-hook from @xmath11 .
thus we obtain an abacus display for the @xmath1-core of @xmath11 by pushing the beads up as far as possible .
similarly , if there is a bead at position @xmath162 and no bead at position @xmath176 then pushing the bead at position @xmath162 ` down ' to position @xmath176 corresponds to adding an addable rim @xmath1-hook to @xmath11 .
the beads in figure [ fig : fig1 ] corresponding to removable rim 5-hooks of @xmath149 are at positions 21 and 22 . given an abacus display for the partition @xmath11 we may obtain an abacus display for the @xmath1-core of @xmath11 ( with the same number of beads ) by ` pushing up ' the beads on the display as far as possible .
if the display has a multiple of @xmath1 beads then one records the mentioned process on a @xmath1-quotient which is a sequence of partitions @xmath177 $ ] where @xmath178 is defined as follows . if you can not push a bead up on runner @xmath43
then @xmath179 .
suppose @xmath180 are the beads on runner @xmath43 that can be pushed up .
we define @xmath181 to be the number of unoccupied positions on runner @xmath43 above @xmath182 .
we call @xmath183 the _ weight _ of the bead @xmath182 .
we refer the reader to section 2.7 of @xcite for a detailed discussion on the @xmath1-quotient , but we note here that if @xmath11 has @xmath1-weight @xmath117 then @xmath184 . for example , using figure [ fig : fig1 ] , we see that @xmath185 has 5-weight 3 and 5-quotient @xmath186 $ ]
. the @xmath1-quotient will be useful in determining which specht modules in @xmath5 are irreducible ( see theorem [ thm : jm ] ) .
fayers provides a test involving a @xmath1-quotient of the partition @xmath11 ( with the left - to - right numbering ) in proposition 2.1 of @xcite which determines whether @xmath11 is a @xmath1-jm partition
. it will be helpful to use a different form of the test given by fayers in @xcite . in order to state the test as in @xcite
we need to define a special numbering on the runners of the abacus and also define the ` pyramid ' of a block .
fix an abacus display for @xmath11 that has a multiple of @xmath1 beads with runners @xmath127 . push all beads up as far as possible to obtain an abacus display , @xmath187 , for the @xmath1-core of @xmath11 . for
each runner @xmath25 , let @xmath188 be the position of the first empty space in @xmath187 on runner @xmath25 .
now write the @xmath189 s in ascending order @xmath190 ( reindexing if needed ) and renumber the runners of the display so that position @xmath188 is on runner @xmath25 .
we refer to this renumbering of the runners as the _ non left - to - right numbering_. observe that we have renumbered the runners from left - to - right as @xmath191 for the unique permutation @xmath192 .
for @xmath193 , define @xmath194 the array @xmath195 is called the _ pyramid _ of the block in which @xmath11 lies .
define the @xmath1-quotient of @xmath11 in the non left - to - right numbering to be the sequence @xmath196 $ ] where @xmath197 .
we refer to this @xmath1-quotient of @xmath11 as the _ reordered _ @xmath1-quotient of @xmath11 .
.1 in * example .
* let @xmath149 and @xmath59 .
we use the abacus display for @xmath11 from figure [ fig : fig1 ] . pushing all beads up on the display we obtain an abacus display for the @xmath198-core ( see figure [ fig : reorder](a ) ) , and see @xmath11 has @xmath198-quotient @xmath186 $ ] in the left - to - right numbering . now
we have @xmath200 and all other @xmath201 are zero .
the runners are renumbered from left to right as 2 , 5 , 1 , 3 , and 4 .
the renumbering is given by the permutation @xmath202 .
so the reordered 5-quotient of @xmath11 is @xmath203 = [ \lambda^{(3 ) } , \lambda^{(1 ) } , \lambda^{(4 ) } , \lambda^{(5 ) } , \lambda^{(2 ) } ] = [ \emptyset , ( 2 ) , \emptyset , \emptyset , ( 1)].\ ] ] we are now ready to state the test due to fayers .
[ @xcite , proposition 4.9 ] [ prop : fjm ] let @xmath1 be an odd prime . take an abacus display for the partition @xmath11 with non left - to - right numbering , and suppose the corresponding reordered @xmath1-quotient is @xmath204 $ ] .
then @xmath11 is a @xmath1-jm partition if and only if : 1 .
@xmath205 for @xmath206 .
2 . @xmath207 is a @xmath1-restricted @xmath1-jm partition .
3 . @xmath208 is a @xmath1-regular @xmath1-jm partition .
@xmath209 for all @xmath210 .
next we define abacus notation for partitions in @xmath5 . by the nakayama conjecture , @xmath12 is in the principal block of @xmath2 if and only if @xmath11 has an empty @xmath1-core .
the empty @xmath1-core can be represented by an abacus display with three beads on each runner ( so the display has @xmath148 beads ) .
hence @xmath11 has an abacus display with three beads on each runner ( the @xmath211 display for @xmath11 ) .
we use the @xmath211 abacus display to study partitions in @xmath5 .
[ def:3p ] let @xmath11 be a partition in the principal block of @xmath2 . 1 .
write @xmath212 if the abacus display for @xmath11 has one bead of weight 3 on runner @xmath25 .
2 . write @xmath213 if the abacus display for @xmath11 has one bead of weight 2 on runner @xmath25 and one bead of weight one on ( not necessarily distinct ) runner @xmath43 .
3 . write @xmath214 if the abacus display for @xmath11 has one bead of weight 1 on ( not necessarily distinct ) runner @xmath25,@xmath43 , and @xmath136
. .1 in * example . * the @xmath215 display for the partition @xmath216 is given below .
so @xmath11 is represented in @xmath215 notation by @xmath217 .
@xmath218 for a partition @xmath11 let @xmath219 denote the number of removable nodes for @xmath11 and let @xmath220 denote the number of normal nodes for @xmath11 .
one may easily determine @xmath221 and @xmath222 from an abacus display for @xmath11 with @xmath1-many runners .
.1 in * example .
* take @xmath223 , and consider @xmath224 and @xmath225 in @xmath5 .
the removable beads of each display are marked with a box .
so @xmath226 and @xmath227 . in figure
[ fig : remove](a ) we see that removable beads at positions 20 and 23 correspond to normal nodes while the bead at position 14 does not .
hence @xmath228 . next , in figure [ fig : remove](b )
, the beads at positions 18 and 20 corresponds to normal nodes of @xmath28 .
so @xmath229 .
analysis of the @xmath230 displays gives the next proposition .
[ prop : nremovables ] we have the following .
.1 in ( 1 ) for @xmath231 .1 in @xmath232 and @xmath233 .1 in ( 2 ) for @xmath234 .1 in @xmath235 and @xmath236 .1 in ( 3 ) for @xmath237 .1 in @xmath238 and @xmath239 .1 in ( 4 ) for @xmath240 and @xmath241 .1 in @xmath242 and @xmath243 .1 in ( 5 ) for any @xmath1-regular partition @xmath11 in @xmath5 : 1 . if @xmath223 then @xmath244 .
2 . for @xmath245 , @xmath246 if @xmath247 with @xmath248 and @xmath249 , @xmath250 or @xmath251 otherwise .
let @xmath11 be a @xmath1-regular partition of @xmath4 .
mullineux conjectured there is a simple algorithm to determine the @xmath1-regular partition @xmath252 of @xmath4 such that @xmath253 .
ford and kleshchev proved the conjecture in @xcite .
the map @xmath162 is a bijection from the set of @xmath1-regular partitions of @xmath4 onto itself , and it is called the _
mullineux map_. the next proposition contains the defining property of the mullineux map .
[ prop : mullk ] suppose @xmath53 is a good node for a @xmath1-regular partition @xmath11 of @xmath4 , with @xmath254
. then there exists a good node @xmath52 for @xmath252 with @xmath255 and @xmath256 . by definition ,
a good node for @xmath11 is normal , but a normal node for @xmath11 need not be good
. if the normal nodes for @xmath11 are of distinct @xmath1-residues then the set of good nodes and the set normal nodes for @xmath11 coincide .
if @xmath11 is a @xmath1-regular partition in the principal block of @xmath2 then @xmath257 since each removable node is of a distinct @xmath1-residue . that the removable nodes have distinct residues is easily deduced after analyzing the @xmath211 displays of definition [ def:3p ] .
a few proofs require computing the image of a partition under the mullineux map .
if @xmath258 is the @xmath211 abacus display for the @xmath1-regular partition @xmath11 in @xmath5 , let @xmath259 denote the @xmath211 abacus display for @xmath252 .
we follow the procedure given by ford and kleshchev to compute the effect the mullineux map has on an abacus display ( see @xcite definition 1.3 up to definition 1.6 ) .
we now define the parity of a partition @xmath11 .
suppose we remove rim @xmath1-hooks @xmath260 to obtain the @xmath1-core for @xmath11 .
define the _ leg - length _ of @xmath261 to be @xmath262 .
now define @xmath263 .
parity _ of @xmath11 , denoted @xmath264 , is @xmath265 .
we cite two important results of fayers and tan .
[ prop : mullparity ] for any @xmath1-regular partition @xmath11 of weight 3 , we have @xmath266 .
( @xcite , theorem 1.1 ) [ prop : bipart ] suppose that @xmath0 is a field of characteristic at least 5 and that @xmath52 is a weight 3 block of @xmath83 .
if @xmath11 and @xmath28 are @xmath1-regular partitions in @xmath52 with @xmath267 , then @xmath268 . in particular , the @xmath15-quiver of @xmath52 is bipartite .
suppose @xmath11 is a partition of @xmath4 that is not a @xmath1-core , and let @xmath77 be a removable rim @xmath1-hook for @xmath11 .
it is not hard to see that if @xmath77 has @xmath165 rows then @xmath77 has @xmath269 columns .
so @xmath77 has leg - length @xmath270 , and @xmath77 corresponds to a removable rim @xmath1-hook , say @xmath271 , for @xmath39 with leg - length @xmath272 .
from this we deduce @xmath11 and @xmath39 have the same parity when @xmath1 is odd .
this observation will be applied later in the proof of theorem [ thm : rr ] .
it is well known that the young module @xmath81 is self - dual and indecomposable . moreover ,
if @xmath11 is @xmath1-restricted then @xmath81 has the simple socle isomorphic to @xmath273 and @xmath81 is projective . in @xcite , hemmer and nakano
show that the multiplicities in a specht or dual specht module filtration of an @xmath87-module are well defined .
the authors also show that the @xmath87- module @xmath81 has a specht module filtration of the form : @xmath274 where @xmath275 and @xmath276 for all @xmath277 .
we give two important results about young modules in the next proposition .
[ prop : young ] let @xmath11 be a partition of @xmath4 . 1 . if @xmath11 is @xmath1-restricted then the young module @xmath81 is isomorphic to the projective cover of @xmath278 .
if @xmath279 and @xmath11 is @xmath1-regular then @xmath280 = 1 $ ] . in @xcite , martin and russell construct the @xmath15-quiver for @xmath5 from @xmath15-quivers for defect 2 and defect 1 blocks of @xmath14 . in doing so ,
the authors establish rules for restricting to defects 2 and 1 as well as rules for inducing back up to @xmath5 . for @xmath234 ,
let @xmath281 denote the defect 2 block of @xmath14 corresponding to the @xmath1-core @xmath282 , and let @xmath283 be the defect 1 block of @xmath14 with @xmath1-core @xmath284 .
we now turn our attention to how these blocks relate to the removable beads on the @xmath211 display .
fix the @xmath211 display for a partition @xmath11 .
suppose the display has a removable bead on runner @xmath285 at position @xmath162 .
pushing said bead from position @xmath162 to position @xmath161 produces a non - zero restriction of the specht module @xmath12 to the block @xmath281 .
further , we remove a removable @xmath286-node from @xmath11 producing a partition @xmath287 of @xmath288 . in this case , write @xmath289 .
martin and russell showed the following : if the @xmath211 display for @xmath11 has a removable bead on runner @xmath234 then there is exactly one partition @xmath290 in @xmath5 such that @xmath291 and @xmath292 .
it is easily seen that we also have @xmath293 .
conversely , if @xmath287 is a partition in the block @xmath281 of @xmath14 and @xmath234 then there are exactly two partitions @xmath11 and @xmath28 in @xmath5 such that @xmath294 with @xmath295 .
it is not hard to see that @xmath296
. if the @xmath211 for @xmath11 has a removable bead on runner @xmath170 then there are two partitions @xmath28 and @xmath297 in @xmath5 such that @xmath298 and @xmath299 .
conversely , if the partition @xmath287 is in @xmath283 then there are three partitions @xmath300 in @xmath5 so that @xmath299 for each @xmath301 , define a subset of the set of @xmath1-regular partitions in @xmath5 , @xmath302 by kleshchev s modular branching rules for simple modules ( see theorem @xmath303 of @xcite ) , a @xmath1-regular partition @xmath11 is in @xmath304 if and only if the @xmath211 display for @xmath11 has a normal bead on runner @xmath25 . if @xmath305 then @xmath306 is simple
moreover , it is known that @xmath307 is self - dual with simple head and simple socle both isomorphic to @xmath308 . by theorem 3.9 of @xcite , if @xmath309 and @xmath290 is @xmath1-regular then @xmath310 and @xmath311 = 1 $ ] .
now , for @xmath285 , define a function @xmath312 from @xmath313 to the set of partitions of @xmath288 lying in the block @xmath281 as follows : for @xmath305 , define @xmath314 to be the partition @xmath287 in @xmath281 such that @xmath289 .
martin and russell prove that @xmath305 is @xmath1-singular if and only if @xmath314 is @xmath1-singular .
furthermore , if @xmath11 and @xmath28 are in @xmath313 and @xmath295 then @xmath315 . we gather results of martin , russell , and tan in the following proposition .
[ prop : mt ] let @xmath11 and @xmath28 be @xmath1-regular partitions in @xmath5 . 1 .
suppose @xmath305 ( @xmath316 )
. then @xmath308 extends @xmath317 if and only if either @xmath318 and @xmath319 extends @xmath320 or @xmath321 and @xmath322 \neq 0 $ ] .
2 . suppose @xmath319 is a simple module of @xmath281 for some @xmath25 ( @xmath234 ) .
then @xmath307 has loewy length 3 .
suppose @xmath319 extends @xmath320 in @xmath281 for some @xmath25 ( @xmath285 ) .
then @xmath307 extends @xmath317 , where @xmath323 , and @xmath317 occurs in the second loewy layer of the non - split extension of @xmath307 by @xmath317 .
4 . suppose @xmath319 extends @xmath320 in @xmath281 for some @xmath25 ( @xmath234 ) .
let @xmath324 be a non - split extension of @xmath319 by @xmath320 .
then @xmath325 has loewy length 4 .
[ cor : cpext ] suppose @xmath11 is a @xmath1-regular partition in @xmath313 ( @xmath234 ) . if @xmath290 is @xmath1-regular then @xmath326 .
since @xmath305 , @xmath327 is simple . by proposition [ prop : mt](2 ) ,
@xmath307 has loewy length 3 with simple head and simple socle isomorphic to @xmath308 .
now @xmath328 = 1 $ ] and @xmath293 , so @xmath329 must lie in the second loewy layer of @xmath307 . hence , @xmath330 .
part ( 1 ) of the proposition is proposition 3.1 of @xcite which is a summary of the work of martin and russell in @xcite .
parts ( 2 ) , ( 3 ) , and ( 4 ) are proposition 3.3 , lemma 3.4 , and proposition 3.5 of @xcite , respectively
. an important consequence of proposition [ prop : mt ] ( 2),(3),(4 ) is the next lemma .
[ lem : sinduce ] suppose the specht module @xmath331 lies in the block @xmath281 of @xmath14 where @xmath234 .
we have the following . 1 .
if @xmath331 has loewy length 1 then @xmath332 has loewy length 3 .
if @xmath331 has loewy length 2 then @xmath332 has loewy length at most 4 .
if @xmath331 has loewy length 3 then @xmath332 has loewy length at most 5 .
\(1 ) if @xmath331 is simple then @xmath333 , where @xmath334 is the @xmath1-regularization of the partition @xmath335 . by proposition [ prop : mt](2 ) ,
@xmath336 has loewy length 3 .
now ( 2 ) follows directly from proposition [ prop : mt](4 ) .
finally , ( 3 ) follows from proposition [ prop : mt](3),(4 ) and proposition [ prop : bipart ] .
we now cite a result that places an upper bound on the loewy of the specht module that lie in a defect 2 block of the symmetric group .
[ prop : w2 ] let @xmath52 be a defect 2 block of @xmath83 , where @xmath0 is a field of characteristic @xmath337 .
further , assume @xmath11 is a partition in @xmath52
. then the loewy length of @xmath12 is bounded above by 3 .
moreover , @xmath12 has loewy length 3 if and only if @xmath11 is both @xmath1-regular and @xmath1-restricted . the rules for restricting specht modules to defect 2 blocks of @xmath14 discussed above , the work of carter and payne in @xcite , and the modular branching rules of kleshchev give the following proposition .
[ prop : cp ] assume that the @xmath211 display for the partition @xmath11 has a removable bead on runner @xmath234 . 1 . if @xmath11 and @xmath290 are @xmath1-regular then @xmath338 . furthermore , we have @xmath339 = 1 $ ] .
2 . there is a short exact sequence @xmath340 at times we will need to analyze the ext@xmath9-quiver of the block @xmath281 .
suppose that the @xmath341 display for @xmath11 has a removable bead on runner @xmath234 at position @xmath162 .
pushing the bead from position @xmath162 to position @xmath161 gives an abacus display ( for the partition @xmath287 in @xmath281 ) which has 3 beads on the first @xmath342 runners , 4 beads on runner @xmath169 , 2 beads on runner @xmath25 , and 3 beads on the remaining @xmath343 runners .
so we may represent @xmath287 in @xmath344 notation .
in fact , all partitions in @xmath281 can be represented in this notation
. however , the ext@xmath9-quiver of the block @xmath281 constructed in @xcite has partitions represented in @xmath345 notation .
the authors provide a way to get from the @xmath344 notation to the @xmath345 notation .
lemma 3.10 of @xcite gives the following proposition .
[ prop : images ] the partitions @xmath346 ( @xmath347 ) listed below are expressed in @xmath348 notation .
.1 in ( 1 ) for @xmath349 , 1 .
@xmath350 ; 2 .
@xmath351 ; 3 .
@xmath352 .
\(2 ) for @xmath353 , @xmath249 , 1 . @xmath354 ; 2 . @xmath355
@xmath356 ; 4 .
@xmath357 .
we only represent @xmath346 in @xmath348 notation when we need to analyze the ext@xmath9-quiver of @xmath358 . in all other instances , @xmath346
is assumed to be in @xmath359 notation .
as discussed in the previous section , we may associate each partition in @xmath5 with its @xmath211 display . at times
we refer to a @xmath211 display as the partition it represents and vice versa . after analyzing the @xmath211 displays
, we have proposition [ prop:3p ] .
[ prop:3p ] the following statements are true . 1 .
@xmath360 is both @xmath1-regular and @xmath1-restricted if and only if @xmath361 .
@xmath362 is both @xmath1-regular and @xmath1-restricted if and only if @xmath363 .
3 . suppose @xmath364
. then @xmath365 is both @xmath1-regular and @xmath1-restricted if and only if @xmath366 .
a partition in @xmath5 is both @xmath1-regular and @xmath1-restricted if and only if it has @xmath211 display of the form ( 1 ) , ( 2 ) , or ( 3 ) .
@xmath367 if and only if @xmath11 has @xmath211 representation of the form @xmath368 or @xmath369 for some @xmath370 . 6 .
@xmath11 is neither @xmath1-regular nor @xmath1-restricted if and only if @xmath11 has @xmath211 display of the form @xmath371 for some @xmath285 . in this case ,
@xmath11 is a hook if and only if @xmath11 has @xmath211 display of the form @xmath373 , @xmath371 , or @xmath374 for some @xmath285 .
@xmath11 is @xmath1-regular , not @xmath1-restricted , and not a hook if and only if @xmath11 has @xmath375 display of the form : 1 .
@xmath376 for @xmath231 , 2 .
@xmath213 for @xmath377 , or 3 .
@xmath378 .
when working over a field of characteristic @xmath1 one often wishes to know which of the ordinary irreducible representations of the symmetric group remain irreducible modulo @xmath1 . by proposition [ prop :
irredspecht ] , we know that the @xmath1-jm partitions label such representations . presently we determine which partitions in @xmath5 are @xmath1-jm partitions .
we apply the test of proposition [ prop : fjm ] .
[ rem : irred ] any partition @xmath11 in @xmath5 satisfies conditions ( 2 ) and ( 3 ) of proposition [ prop : fjm ] since @xmath379 and @xmath380 .
suppose @xmath381 is a partition of @xmath382 .
then @xmath381 is both @xmath1-regular and @xmath1-restricted since its young diagram has less than @xmath1 boxes .
furthermore , for any node @xmath74 of @xmath381 we have @xmath383 .
hence @xmath381 is also a @xmath1-jm partition .
we now show the only irreducible specht modules in @xmath5 are @xmath384 and @xmath385 .
[ thm : jm ] let @xmath11 be a partition in @xmath5 . then @xmath11 is a @xmath1-jm partition if and only if @xmath386 . obviously , if @xmath387 then it is a @xmath1-jm partition .
suppose @xmath11 is a @xmath1-jm - partition in @xmath5 .
fix the @xmath211 abacus display for @xmath11 . pushing all beads up as far as possible on the @xmath211 display for @xmath11
gives display @xmath187 ( for the empty partition ) and the left - to - right @xmath1-quotient @xmath388 $ ] .
it is clear that @xmath389 for all @xmath285 .
thus the left - to - right and reordered @xmath1-quotients of @xmath11 are the same and @xmath390 for any @xmath193 . by remark [ rem : irred ]
, we need to only consider conditions ( 1 ) and ( 4 ) proposition [ prop : fjm ] . since @xmath11 is a @xmath1-jm partition , @xmath391 for @xmath237 by proposition [ prop : fjm](1 ) .
so the @xmath1-quotient becomes @xmath392 $ ] , and we have @xmath393 .
now , since @xmath390 for any @xmath210 , condition ( 4 ) of proposition [ prop : fjm ] becomes @xmath394 we claim this implies either @xmath395 and @xmath396 or @xmath397 and @xmath398 [ in which case @xmath399 or @xmath400 .
we may choose @xmath401 and @xmath402 in equation [ eq : lam ] .
then @xmath403 .
since @xmath393 , @xmath404 and @xmath405 can not both be zero .
so either @xmath406 and @xmath407 or @xmath408 and @xmath409 .
the first case implies @xmath395 and @xmath396 and the second case implies @xmath397 and @xmath398 .
we conclude that if @xmath11 in @xmath5 is a @xmath1-jm partition then @xmath386 .
this section is concluded with a lemma that allows one to relate certain specht modules in @xmath5 to specht modules in defect 2 blocks of @xmath14 .
application of the lemma is seen in the section 5 .
[ lem : remove ] let @xmath11 be a partition in the principal block of @xmath2 . if @xmath11 is both @xmath1-regular and @xmath1-restricted then @xmath11 has a removable node @xmath53 such that the following hold : 1 .
@xmath410 has @xmath1-weight 2 , and 2 .
@xmath410 is both @xmath1-regular and @xmath1-restricted .
moreover , if @xmath411 then it has at least one normal node that satisfies both ( i ) and ( ii ) .
let @xmath11 be a partition in the principal block of @xmath2 which is both @xmath1-regular and @xmath1-restricted .
then the @xmath211 display for @xmath11 must be of the form ( 1 ) , ( 2 ) , or ( 3 ) of proposition [ prop:3p ] .
.1 in * case 1
. * @xmath412 has @xmath211 display of form @xmath213 , @xmath413 .
assume @xmath414 .
push the bead at position @xmath415 to position @xmath416 ( see figure [ fig : case1](a ) ) .
this gives an abacus display for a weight 2 partition @xmath410 , which is clearly @xmath1-regular .
notice that the longest string of spaces between beads on the abacus display for @xmath410 has length at most @xmath109 ( which occurs when @xmath417 ) .
hence @xmath410 is @xmath1-restricted . finally , @xmath53 is normal since it is the top removable node of @xmath11 ( the last occupied position on the display is @xmath415 ) .
now consider the display of the form @xmath418 , @xmath237 .
push the bead at position @xmath419 ( see figure [ fig : case1](b ) ) to position @xmath420 .
again , we obtain a partition @xmath410 that satisfies ( i ) and ( ii ) . if @xmath421 then there is no addable bead on runner @xmath422 after position @xmath420 .
hence @xmath53 is normal for @xmath421 . finally , when @xmath423 we see exactly two removable nodes ( corresponding to beads at positions @xmath424 and @xmath425 ) that satisfy conclusions ( i ) and ( ii ) ; however , neither of these nodes is normal .
.1 in * case 2 .
* @xmath412 has @xmath211 display of form @xmath362 , @xmath426 .
first assume @xmath427 .
push the bead at position @xmath428 to position @xmath429 ( see figure [ fig : case2](a ) ) . the resulting abacus display represents a partition that satisfies both ( i ) and ( ii ) , and since position @xmath428 is the last occupied position the corresponding removable node is normal . if @xmath430 then push the bead at position @xmath419 to position @xmath420 ( see figure [ fig : case2](b ) ) .
the resulting abacus display represents a partition that satisfies both ( i ) and ( ii ) .
now since runner @xmath431 has no addable bead at a position after position @xmath420 the corresponding removable node is normal .
.1 in * case 3
. * @xmath412 has @xmath211 display of form @xmath214 , @xmath432 and @xmath433 .
we first assume @xmath434 . if @xmath435 then @xmath436 .
push the bead at position @xmath428 to position @xmath429 ( see figure [ fig : case3a](a ) ) .
this gives a display for a weight 2 partition that is both @xmath1-regular and @xmath1-restricted .
the node that corresponds to the bead at position @xmath428 is normal since runner @xmath169 has no addable bead at a position after position @xmath429 . if @xmath421 we push the bead at position @xmath419 to position @xmath420 .
again we obtain a partition that satisfies ( i ) and ( ii ) of the lemma .
the node that corresponds to the bead at position @xmath419 is normal since runner @xmath431 has no addable bead at a position after position @xmath420 .
next assume @xmath437 .
push the bead at position @xmath438 to position @xmath439 to obtain a partition satisfying ( i ) and ( ii ) ( see figure [ fig : case3b ] ) .
the node that corresponds to the bead at position @xmath438 is normal since runner @xmath440 has no addable bead at a position after position @xmath439 .
@xmath441 it follows from the proofs of cases ( 1)-(3 ) that @xmath442 has at least one normal node that satisfies both ( i ) and ( ii ) .
[ cor : i2 ] assume @xmath11 is a partition in @xmath5 that is both @xmath1-regular and @xmath1-restricted and @xmath442 .
then @xmath305 for some @xmath234 .
in this section we determine the specht modules in the block @xmath281 ( @xmath234 ) that remain irreducible over the field @xmath0 .
we then induce these modules back up to the block @xmath5 to see which specht modules in @xmath5 have an irreducible restriction to the block @xmath281 . for @xmath234 , define @xmath443 . after applying proposition [ prop : fjm ] to each block @xmath281
we have the following lemma .
[ lem : simpledefect2]the following is a complete list of partitions in the block @xmath281 ( @xmath234 ) represented in @xmath344 notation that label irreducible specht modules . 1 .
@xmath444 for @xmath349 ; 2 .
@xmath445 ; 3 .
@xmath446 we prove the result for the block @xmath447 .
the blocks @xmath448 will have a similar calculation .
the @xmath1-core @xmath449 for the block @xmath447 can be represented on an abacus with 4 beads on runner 1 , 2 beads on runner 2 , and 3 beads on runners @xmath450 .
hence @xmath451 , @xmath452 for @xmath453 , and @xmath454 .
so we renumber the runners from left to right as @xmath455 .
notice that in this renumbering @xmath456 and @xmath457 .
the block @xmath447 has pyramid @xmath458 if @xmath459 and @xmath460 if @xmath461 . now suppose @xmath11 is a @xmath1-jm partition in @xmath447 .
since @xmath11 is a @xmath1-jm partition , the reordered @xmath1-quotient of @xmath11 has the form @xmath462 $ ] by proposition [ prop : fjm](1 ) .
note that @xmath11 has @xmath1-weight 2 , so @xmath207 and @xmath208 satisfy conditions ( 2 ) and ( 3 ) of proposition [ prop : fjm ] , respectively .
so we consider condition ( 4 ) , and observe that @xmath463 .
we consider three cases .
.1 in * case 1 . *
assume @xmath464 and @xmath465 .
first suppose @xmath466 .
then for @xmath193 we have the following .
@xmath467 since @xmath468 for all @xmath459 the reordered @xmath1-quotient of @xmath11 does not contradict condition ( 4 ) .
now @xmath11 has left - to - right @xmath1-quotient @xmath469 $ ] .
so @xmath470 . if @xmath471 then one can verify that the reordered @xmath1-quotient satisfies condition ( 4 ) .
now the left - to - right @xmath1-quotient is @xmath472 $ ] .
hence @xmath473 .
.1 in * case 2 . *
assume @xmath474 and @xmath475 .
if @xmath476 then we have the following .
@xmath477 it is easily seen that the reordered @xmath1-quotient satisfies condition ( 4 ) .
the left - to - right @xmath1-quotient is @xmath478 $ ] , so @xmath479 .
now suppose @xmath480 .
observe that @xmath481 and @xmath482 .
so this @xmath1-quotient does not satisfy condition ( 4 ) .
.1 in * case 3 .
* assume @xmath483 .
one may easily verify that this reordered @xmath1-quotient satisfies condition ( 4 ) .
the left - to - right @xmath1-quotient is @xmath484 $ ] .
so @xmath485 .
.1 in therefore @xmath486 .
lemma [ lem : simpledefect2 ] and the modular branching rules give us the next proposition .
[ prop : spechtsimpleinduce ] inducing the irreducible specht modules in the block @xmath281 ( @xmath234 ) up to @xmath5 produces the following specht module filtrations . furthermore ,
if a specht module in @xmath5 has an irreducible restriction to some block @xmath281 ( @xmath234 ) then it must be seen in one of these filtrations . 1 . for @xmath487 @xmath488 2 .
for @xmath349 @xmath489 3 . for @xmath490 @xmath491
in this section we study the loewy length of @xmath12 where @xmath11 is a partition in the principal block of @xmath492 .
we see that the maximum possible loewy length of a specht module in the block @xmath5 is 4 .
compare this with the result of chuang and tan given in proposition [ prop : w2 ] . in @xcite , martin and tan
prove that the projective indecomposable modules of principal blocks of @xmath87 ( @xmath493 ) have common loewy length 7 .
tan later generalizes this result to all weight 3 blocks ( with @xmath380 ) of @xmath87 in @xcite .
this fact aids in finding a common loewy length among the specht modules in a given defect 3 block of @xmath87 that correspond to partitions that are both @xmath1-regular and @xmath1-restricted .
after the upper bound on the loewy lengths is established , we prove some loewy structure results for the specht modules in @xmath5 . we claim that all the specht modules that lie in @xmath5 have loewy length at most 4 .
we begin by showing that any specht module @xmath12 lying in a defect 3 block of @xmath87 , where @xmath11 is a @xmath1-regular and @xmath1-restricted , has loewy length 4 .
[ thm : rr ] suppose the partition @xmath11 lies in a defect 3 block @xmath52 of @xmath87
. if @xmath11 is both @xmath1-regular and @xmath1-restricted then the specht module @xmath12 has loewy length 4 .
since @xmath11 is @xmath1-restricted , @xmath81 is isomorphic to the projective cover of @xmath494 by proposition [ prop : young](1 ) .
so @xmath81 has loewy length 7 .
now , @xmath495 = 1 $ ] ( see proposition [ prop : young](2 ) ) implies that @xmath308 must lie in the fourth loewy layer of @xmath81 since @xmath81 is self - dual and indecomposable .
we conclude that @xmath12 has loewy length 4 .
suppose the partition @xmath11 lies in a defect 3 block of @xmath83 . if @xmath11 is @xmath1-regular and @xmath1-restricted then @xmath496 = 1 \text { and } [ \text{rad}^iy^{\lambda}/\text{rad}^{i+1}y^{\lambda}:d^{\lambda } ] = 0\ ] ] for all @xmath497 we just proved that the specht modules in a defect 3 block which have corresponding partition that is @xmath1-regular and @xmath1-restricted all have common loewy length 4 .
we claim that all the specht modules in @xmath5 have loewy length at most 4 . the idea behind the proof is to restrict to `` nice '' defect 2 blocks of @xmath14 ( i.e.
, we avoid restricting to specht modules having loewy length 3 if possible ) then apply lemma [ lem : sinduce](1 ) and ( 2 ) .
we also note that if @xmath11 is a hook partition of @xmath20 then @xmath12 has loewy length at most 2 . now
the only partitions in @xmath5 that are neither @xmath1-regular nor @xmath1-restricted have @xmath211 display of the form @xmath371 for some @xmath285 .
so the only partitions in @xmath5 that are neither @xmath1-regular nor @xmath1-restricted are the hooks @xmath498 , @xmath316 . therefore it suffices to consider @xmath11 that is @xmath1-regular , not @xmath1-restricted , and not a hook ( by propositions [ prop : james ] , [ prop : peel ] , and theorem [ thm : rr ] ) .
for such a partition @xmath11 , we analyze the restriction of @xmath12 to a particular defect 2 block of @xmath14 and apply proposition [ prop : w2 ] in next lemma .
[ lem : nicerest ] let @xmath11 be a partition in @xmath5 that is @xmath1-regular , not @xmath1-restricted , and not a hook
. then there exists @xmath234 such that @xmath499 has loewy length at most 2 , except for when @xmath500 . any partition in @xmath5 that satisfies the given hypothesis must have @xmath211 display of the form : 1 .
@xmath376 where @xmath501 , 2 .
@xmath502 where @xmath377 , or 3 .
@xmath378 .
\(1 ) first assume @xmath166 .
push the removable bead at position @xmath415 of @xmath376 to position @xmath416 .
this gives a partition @xmath503 in @xmath504 notation .
clearly , @xmath287 is @xmath1-regular . for each @xmath505 ,
the position @xmath506 of @xmath507 is unoccupied , and the last occupied position of @xmath507 is @xmath416 .
so @xmath508 is @xmath1-singular . by proposition
[ prop : w2 ] , the module @xmath509 has loewy length at most 2 .
for the case @xmath170 , observe that @xmath510 .
.1 in ( 2 ) assume @xmath427 .
push the removable bead at position @xmath511 to position @xmath416 .
this gives a partition @xmath512 in @xmath513 notation that is @xmath1-regular but not @xmath1-restricted . by proposition [ prop : w2 ]
, @xmath514 has loewy length at most 2 .
now assume @xmath430 .
first consider @xmath515 . pushing the removable bead at position @xmath419 to position @xmath420
gives the partition @xmath516 in @xmath517 notation which is @xmath1-regular but not @xmath1-restricted . by proposition
[ prop : w2 ] , @xmath518 has loewy length at most 2 .
for the case @xmath519 , observe that @xmath520 has loewy length at most 2 .
.1 in ( 3 ) the partition @xmath11 with display @xmath521 has exactly one removable node , @xmath53 , of @xmath1-residue @xmath522 . by inspection
, @xmath410 is both @xmath1-regular and @xmath1-restricted .
hence @xmath523 has loewy length 3 by proposition [ prop : w2 ] .
the exceptional partition , @xmath500 , of lemma [ lem : nicerest ] corresponds to a specht module that has loewy length 3 . to demonstrate this we show @xmath12 embeds into the specht module @xmath524 which has loewy length 4 and simple socle isomorphic to @xmath525 .
[ lem : help ] the specht module @xmath526 is isomorphic to a submodule of @xmath527 .
in particular , @xmath528 has loewy length 3 and simple socle isomorphic to @xmath529 .
let @xmath530 be a non - zero homomorphism ( see proposition [ prop : cp ] ) .
let @xmath531 and @xmath532 . by parity and theorem [ thm : rr ] ,
we know @xmath308 is in the second loewy layer of @xmath123 since @xmath533 . observe that @xmath11 has exactly one removable node @xmath53 ( which has residue @xmath522 ) .
hence @xmath534 .
furthermore , @xmath535 , @xmath536 , and @xmath295 .
clearly , @xmath410 is both @xmath1-regular and @xmath1-restricted , so @xmath537 has loewy length 3 and simple socle @xmath538 .
regard @xmath539 as an @xmath14-module homomorphism .
then @xmath540 is non - zero , so @xmath539 is one - to - one since @xmath541 = 1 $ ] ( see proposition [ prop : decnumb ] ) .
the following lemmas help us prove that a specht module in @xmath5 corresponding to a partition that is @xmath1-regular , not @xmath1-restricted , and not a hook has loewy length at most 3 . ultimately , these lemmas help to prove the converse of theorem [ thm : rr ] .
[ lem : getit ] let @xmath542 be a @xmath1-regular , not @xmath1-restricted , and non - hook partition in @xmath5 .
suppose there exists a @xmath1-regular partition @xmath381 in @xmath5 such that the following hold : 1 .
@xmath543 ( for some @xmath234 ) with @xmath544 and 2 .
@xmath545 has loewy length at most 2 .
then @xmath12 has loewy length at most 3 .
we have @xmath546 and @xmath547 .
now @xmath548 has loewy length at most 4 ( by lemma [ lem : sinduce ] ) .
observe that @xmath548 has a specht filtration @xmath549 with @xmath550 at the top . by frobenius reciprocity
, @xmath308 is not in the head of @xmath548 since @xmath551 .
hence @xmath552 .
so @xmath12 has loewy length strictly less than that of @xmath553 .
we conclude that @xmath12 has loewy length at most 3 .
we are now ready to prove the converse of theorem [ thm : rr ] .
it suffices to prove the converse is true for @xmath1-regular partitions in @xmath5 that are not @xmath1-restricted and also not hooks by propositions [ prop : james ] and [ prop : peel ] .
[ thm : gotit ] let @xmath11 be a partition in @xmath5 that is @xmath1-regular but not @xmath1-restricted .
then @xmath12 has loewy length at most 3 .
let @xmath11 be a partition in @xmath5 that is @xmath1-regular but not @xmath1-restricted . by lemma [ lem :
help ] , we may assume @xmath554 .
further , we may assume @xmath11 is not a hook ( by proposition [ prop : peel ] ) .
from lemma [ lem : nicerest ] , we can find a weight 2 block @xmath281 ( @xmath234 ) of @xmath14 such that @xmath555 has loewy length at most 2 , so @xmath556 has loewy length at most 4 .
in particular , we have the following pairs of restrictions .
1 . @xmath557 for @xmath558 2 .
@xmath559 for @xmath560 3 .
@xmath561 for @xmath562 , @xmath249 4 .
@xmath563 for @xmath560 define the sets : 1 .
@xmath564 , 2 .
@xmath565 , 3 .
@xmath566 and 4 . @xmath567 if @xmath568 then @xmath12 has loewy length at most 3 by lemma [ lem : getit ] and the pairs of restrictions given above .
we just need to prove the theorem for the partitions @xmath569 and @xmath570 .
we have @xmath571 is irreducible by proposition [ prop : spechtsimpleinduce](3 ) .
hence @xmath572 and @xmath573 both have loewy length at most 3 by lemma [ lem : sinduce](1 ) and since @xmath574 .
[ cor : nscondition ] if @xmath12 is in @xmath5 then @xmath12 has loewy length at most 4 .
moreover , a partition @xmath11 in @xmath5 is @xmath1-regular and @xmath1-restricted if and only if @xmath12 has loewy length 4 .
we conclude this subsection by classifying the specht modules in @xmath5 having loewy length 2 .
this classification will aid in describing the socle of a specht module in @xmath5 that has loewy length 3 .
[ lem : spechtlength2help ] let @xmath11 be a @xmath1-regular partition in @xmath5 and suppose there is @xmath234 so that @xmath289 . 1 . if @xmath575 is in the socle of @xmath576 then @xmath577 .
2 . if @xmath305 , @xmath290 is @xmath1-regular , and @xmath578 is irreducible then @xmath12 and @xmath576 both have loewy length at most 2 .
if @xmath12 has loewy length 2 and @xmath551 then @xmath578 is irreducible .
we have a short exact sequence .
@xmath579 ( 1 ) note that @xmath575 is in the socle of @xmath580 if and only if @xmath581 the result now follows since @xmath582 .
we proceed to prove ( 2 ) .
since @xmath578 is irreducible , @xmath576 and @xmath12 both have loewy length at most 3 by lemma [ lem : sinduce](1 ) .
now , we have @xmath583 since @xmath584 and @xmath585 .
so @xmath576 has loewy length at most 2 since @xmath580 has loewy length 3 .
now suppose @xmath575 is in the socle of @xmath12 . if @xmath12 is irreducible then we are done . if @xmath12 is reducible then @xmath586 since @xmath306 and @xmath587 = 1 $ ] .
( i.e. , the only composition factor of @xmath12 that restricts to @xmath319 is @xmath308 . )
so @xmath575 is not in the socle of @xmath580 by part ( 1 ) .
we conclude that @xmath12 has loewy length at most 2 .
finally , ( 3 ) follows since the specht modules are indecomposable over odd characteristic .
the next lemma follows from the restrictions ( 1 ) , ( 2 ) , ( 3 ) , and ( 4 ) from the proof of theorem [ thm : gotit ] and since @xmath588 . [
lem : killhead ] if @xmath589 is a partition in @xmath5 that is @xmath1-regular , not @xmath1-restricted , and not a hook then there is @xmath234 so that @xmath590 is non - zero and @xmath547 .
consider the following sets .
1 . @xmath591 2 .
@xmath592 3 .
@xmath593 4 .
@xmath594 let @xmath595 .
the next theorem classifies the specht modules in @xmath5 that have loewy length 2 .
[ thm : spechtlength2 ] let @xmath11 be a partition in @xmath5
. then @xmath12 has loewy length 2 if and only if @xmath596 .
let @xmath596 .
if @xmath597 then @xmath11 is a hook partition different from both @xmath598 and @xmath599 .
so @xmath12 has loewy length 2 by proposition [ prop : peel ] .
assume @xmath600 . if @xmath601 then @xmath602 .
now @xmath603 is irreducible by proposition [ prop : spechtsimpleinduce](3 ) .
hence , by lemma [ lem : spechtlength2help](2 ) , @xmath572 and @xmath573 both have loewy length 2 . now observe that @xmath604 and @xmath605 .
so @xmath606 and @xmath607 both have loewy length 2 since @xmath608 .
conversely , suppose the specht module @xmath12 in @xmath5 has loewy length 2 . by proposition [ prop:3p ] , if @xmath11 is neither @xmath1-regular nor @xmath1-restricted then @xmath609 for some @xmath285 .
so @xmath610 .
note that a hook partition in @xmath5 has @xmath375 notation of the form @xmath373 , @xmath371 , or @xmath374 for @xmath285 .
so we may assume that @xmath11 is not a hook .
first suppose that @xmath11 is @xmath1-regular and not @xmath1-restricted .
if @xmath611 then we are done .
assume @xmath612 .
note that @xmath554 by proposition [ lem : help ] .
so choose @xmath234 such that @xmath499 and @xmath547 ( see lemma [ lem : killhead ] ) .
then @xmath590 is irreducible by [ lem : spechtlength2help](3 ) .
so @xmath613 by proposition [ prop : spechtsimpleinduce ] . finally , if @xmath11 is @xmath1-singular and @xmath1-restricted then @xmath614 since @xmath11 is not a hook .
hence @xmath615 .
we conclude that @xmath596 .
[ cor : length3 ] let @xmath11 be a partition in @xmath5 .
the specht module @xmath12 has loewy length 3 if and only if @xmath11 is not a hook and either @xmath11 is @xmath1-regular , not @xmath1-restricted , and @xmath616 or @xmath11 is @xmath1-singular , @xmath1-restricted , and @xmath617 .
now that the upper bound on the loewy lengths has been established , we turn our attention to describing some of the radical layers of the specht modules in @xmath5 .
the discussion begins by first considering specht modules in the block corresponding to partitions that are both @xmath1-regular and @xmath1-restricted .
consider the following lemmas .
[ lem : regristreg ] assume @xmath618 is a partition in @xmath5 that is @xmath1-regular and @xmath1-restricted . 1 . if @xmath619 , @xmath620 then there is @xmath234 such that @xmath305 and @xmath290 is @xmath1-regular .
2 . if @xmath621 then @xmath622 and @xmath623 has simple head isomorphic to @xmath624 . part ( 1 ) follows immediately from the proof of lemma [ lem : remove ] .
next assume @xmath625 . obviously , @xmath626 .
now @xmath623 has simple head isomorphic to @xmath624 by lemma [ lem : help ] since @xmath627 and @xmath628 .
[ lem : mullineux421 ] we have the following .
we begin by finding the mullineux symbol for @xmath630 .
start with the @xmath211 display @xmath631 .
let @xmath632 be the last bead of @xmath258 , i.e. , @xmath632 is the bead at position @xmath633 .
now position @xmath125 of @xmath258 is unoccupied , but there is a proper bead at a position before position @xmath125 .
so let @xmath634 be the bead of @xmath258 at position @xmath635 .
position @xmath636 of @xmath258 is occupied by an improper bead , so we stop .
we move @xmath632 to position @xmath125 and @xmath634 to position @xmath637 ( the first unoccupied position of @xmath258 ) .
this process removes one rim @xmath1-hook with 3 parts and then one box .
so @xmath638 and @xmath639 , and we obtain another abacus display @xmath640 .
we repeat the process to the abacus display @xmath640 .
let @xmath632 be the bead at position @xmath641 ( i.e. , the last bead of the display ) .
position @xmath288 of @xmath640 is unoccupied and @xmath640 has no proper bead before position @xmath288 , so we stop .
move @xmath632 to position @xmath288 of @xmath640 .
this removes one rim @xmath1-hook with 3 parts .
hence @xmath642 and @xmath643 , and we produce a new abacus display @xmath644 .
let @xmath632 be the bead of @xmath644 at position @xmath645 .
now position @xmath637 is occupied by an improper bead .
so we stop the process and move @xmath632 to the first unoccupied position of @xmath644 which is position @xmath635 .
this removes a hook of length @xmath109 with @xmath19 parts .
so @xmath646 and @xmath647 , and we produce a abacus display @xmath648 for the empty diagram .
therefore , @xmath649 has mullineux symbol @xmath650 so @xmath651 has mullineux symbol @xmath652 we first prove the result for @xmath223 .
it suffices to show that @xmath653 and @xmath654 have the same mullineux symbol .
let @xmath85 be the @xmath655 display @xmath654 .
take @xmath632 to be the last bead of @xmath85 , i.e. , the bead of @xmath85 at position @xmath656 .
position @xmath657 of @xmath85 is unoccupied ; however , @xmath85 has a proper bead at a position before position @xmath657 .
so take @xmath634 to be the bead of @xmath85 at position @xmath658 .
now position @xmath659 of @xmath85 is occupied by an improper bead , so we stop .
move @xmath632 and @xmath634 to positions @xmath657 and @xmath660 of @xmath85 , respectively .
this removes one rim @xmath198-hook with 2 parts and then one box .
hence @xmath661 and @xmath662 , and we produce a new display @xmath663 .
let @xmath632 be the bead at position @xmath664 of @xmath663 .
now position @xmath665 of @xmath663 is unoccupied and @xmath663 has no proper bead before position @xmath665 , so we stop .
move @xmath632 to position @xmath665 of @xmath663 .
this removes one rim @xmath198-hook with 2 parts .
so @xmath666 and @xmath667 , and we obtain a new display @xmath668 .
let @xmath632 be the bead of @xmath668 at position @xmath657 .
now position @xmath660 of @xmath668 is occupied by an improper bead , so we stop .
move @xmath632 to position @xmath658 ( the first unoccupied position ) of @xmath668 .
this removes a hook of length @xmath669 with @xmath68 parts .
thus @xmath670 and @xmath671 , and we obtain a display for the empty diagram .
so @xmath672 therefore , @xmath673 .
we now show @xmath674 for @xmath245 .
let @xmath675 be the @xmath676 display @xmath677 .
take @xmath632 to be the last bead of @xmath675 , i.e. , the bead of @xmath675 at position @xmath678 .
now position @xmath679 of @xmath675 is unoccupied ; however , @xmath675 has a proper bead at an earlier position than @xmath679 .
so take @xmath634 to be the bead of @xmath675 at position @xmath680 .
now position @xmath681 of @xmath675 is occupied by an improper bead , so we stop .
move @xmath632 to position @xmath679 and @xmath634 to position @xmath424 ( the first unoccupied position of @xmath675 ) .
this removes one rim @xmath1-hook with @xmath522 parts and then one box .
thus @xmath638 and @xmath682 , and we obtain a new display @xmath683 . now let @xmath632 be the bead of @xmath683 at position @xmath684 . position @xmath685 of @xmath683 is unoccupied and @xmath683 does not have a proper bead before position @xmath685 , so we stop .
move @xmath632 to position @xmath685 .
this removes one rim @xmath1-hook with @xmath522 parts .
so @xmath642 and @xmath686 , and we obtain a new display @xmath687 .
let @xmath632 be the bead at position @xmath688 of @xmath687 .
position @xmath424 is occupied by an improper bead , so we stop .
move @xmath632 to position @xmath680 ( the first unoccupied position of @xmath687 ) .
this removes a hook of length @xmath109 with @xmath522 parts .
so @xmath646 and @xmath689 .
the resulting display is for the empty diagram .
@xmath690 so @xmath691 for @xmath245 .
lemmas [ lem : regristreg ] and [ lem : mullineux421 ] help us prove the following property of the ext@xmath9-quiver of @xmath5 .
the author wonders whether this property generalizes to ext@xmath9-quivers of arbitrary blocks of weight 3 .
[ prop : mullextend ] if @xmath11 is a @xmath1-regular and @xmath1-restricted partition in @xmath5 then @xmath692 first assume @xmath693 .
let @xmath694 , and assume by way of a contradiction that @xmath695 .
choose @xmath234 such that @xmath305 .
further , if @xmath696 we choose @xmath25 so that @xmath290 is @xmath1-regular , as in lemma [ lem : regristreg](1 ) .
. now @xmath695 implies either : 1 .
@xmath698 and @xmath699 or 2 .
@xmath700 and @xmath701 \neq 0 $ ] .
we claim that ( 1 ) can not happen .
suppose ( 1 ) holds .
then @xmath702 where @xmath57 is the unique @xmath286-normal node for @xmath703 .
now the partition @xmath410 is both @xmath1-regular and @xmath1-restricted , so @xmath537 has simple socle @xmath704 . by proposition [ prop : mullk ]
, there exists a normal node @xmath705 for @xmath39 such that @xmath706 and @xmath707 .
but then @xmath705 corresponds to a removable node for @xmath11 of residue @xmath708 .
therefore @xmath709 .
observe that @xmath710 .
hence , @xmath711 a contradiction .
so ( 2 ) must hold , that is , @xmath700 and @xmath712 \neq 0 $ ] .
in particular , @xmath713 .
first consider @xmath696 .
since @xmath714 and @xmath290 is @xmath1-regular , we have @xmath715 .
we have a short exact sequence .
@xmath716 apply the functor @xmath717 to equation [ eq : ses11 ] .
now @xmath713 and the eckmann - schapiro lemma yield @xmath718 we conclude that @xmath719 .
now , by theorem [ thm : rr ] , @xmath720 \geq 2 $ ] , a contradiction .
therefore , @xmath721 if @xmath696 .
next suppose that @xmath625
. then @xmath722 and @xmath723 . by lemma [ lem : regristreg ]
, @xmath623 has simple head isomorphic to @xmath624 .
observe @xmath724 . by lemma [ lem : mullineux421 ] , @xmath725 .
we have the following short exact sequence .
@xmath726 apply the functor @xmath717 to equation [ eq : ses22 ] .
now @xmath713 and the eckmann - schapiro lemma yield @xmath727 we conclude that @xmath719 .
now , by theorem [ thm : rr ] , @xmath728 \geq 2 $ ] , a contradiction .
therefore , it must be that @xmath721 for @xmath729 . if @xmath730 then @xmath731 . so the theorem holds for @xmath39 . now , since the simple modules are self - dual , @xmath732 this same argument can be used to show the theorem holds for @xmath733 by considering @xmath734 . the next proposition together with proposition [ prop : mullextend ] will allow us to determine the composition factors of @xmath12 from the @xmath15-quiver for @xmath5 , provided @xmath11 is in @xmath5 and is both @xmath1-regular and @xmath1-restricted .
see @xcite for a construction of the @xmath15-quiver for @xmath5 .
[ prop : ext2 ] let @xmath735 and @xmath100 be @xmath1-regular weight 2 partitions of @xmath4 with @xmath736
. then @xmath737 if and only if @xmath738 \neq 0 $ ] and @xmath739 and @xmath740 have different parities . taken together
, propositions [ prop : w2 ] , [ prop : ext2 ] , and [ prop : james ] imply proposition [ prop : ks ] holds for all weight 2 blocks of the symmetric group ( over odd characteristic ) with no restriction on the number of non - zero parts .
the goal now is to prove the restriction on the number of non - zero parts in proposition [ prop : ks ] can be removed for @xmath1-regular partitions in @xmath5 that are also @xmath1-restricted .
first we introduce an assumption to better organize our work . * assumption . *
say a @xmath1-regular partition @xmath11 satisfies assumption @xmath741 if there exists @xmath234 such that 1 .
@xmath742 such that @xmath305 and 2 .
@xmath290 is @xmath1-regular .
let @xmath743 denote the set of all partitions in @xmath5 that are both @xmath1-regular and @xmath1-restricted .
define @xmath744 next we define the following sets : @xmath745 and @xmath746 . by inspection
, we have the following lemmas .
adopting the notation from above , @xmath747 [ lem : help2 ] let @xmath28 be a @xmath1-regular partition in @xmath5 .
the @xmath15-quiver for @xmath5 gives the following .
1 . if @xmath748 and @xmath749 then @xmath750 .
2 . if @xmath751 and @xmath752 then @xmath753 .
we now prove the second loewy layer of @xmath12 is determined by the ext@xmath9-quiver of @xmath5 whenever @xmath11 is @xmath1-regular and @xmath1-restricted .
[ thm:2nd ] let @xmath11 and @xmath28 be @xmath1-regular partitions in the principal block of @xmath2 with @xmath11 @xmath1-restricted .
if @xmath11 does not strictly dominate @xmath28 then @xmath754 we may assume @xmath11 has at least @xmath1 nonzero parts and @xmath755 .
assume @xmath618 .
choose @xmath234 such that @xmath305 , and let @xmath53 be the normal @xmath286-node of @xmath11 such that @xmath756 . since @xmath755 we have either 1 .
@xmath318 and @xmath757 or 2 .
@xmath321 and @xmath758 \neq 0 $ ] .
begin by assuming ( 1 ) holds .
then @xmath759 , where @xmath52 is the normal node for @xmath28 with @xmath760 . by proposition [ prop :
ext2 ] , we have @xmath761 \neq 0 $ ] . then , by lemma 2.13 of @xcite , @xmath105 \neq 0 $ ] .
now @xmath755 implies @xmath762 .
hence @xmath317 must lie in the second radical layer of @xmath12 by proposition [ prop : mullextend ] .
so @xmath763 .
the result now follows since @xmath764 is one - dimensional .
assume ( 2 ) holds .
inducing @xmath537 back up to @xmath5 gives the following short exact sequence .
@xmath765 applying the functor @xmath766 to the short exact sequence in equation [ eq : ses2 ] we obtain the following exact sequence .
@xmath767 now @xmath768 since @xmath321 .
we conclude that @xmath769 by the eckmann - schapiro lemma and frobenius reciprocity . we claim that @xmath770 whenever @xmath11 satisfies @xmath741 .
if @xmath11 satisfies @xmath741 then @xmath576 has simple head isomorphic to @xmath329 .
moreover , @xmath293 implies @xmath770 since @xmath11 does not strictly dominate @xmath28 .
therefore , @xmath771 .
so @xmath772 as desired .
assume @xmath11 does not satisfy assumption @xmath741 .
first assume @xmath773 . then @xmath774 .
if @xmath775 and @xmath776 then @xmath753 by lemma [ lem : help2 ] . it is easily verified that @xmath777 .
hence , for @xmath753 , @xmath105 \neq 0 $ ] with @xmath778 .
now assume @xmath779 .
by lemma [ lem : help2 ] , if @xmath775 and @xmath776 then @xmath750 .
observe that @xmath780 and @xmath781 for @xmath782 .
also @xmath783 .
hence @xmath784 .
take @xmath785 .
we have a short exact sequence .
@xmath786 by lemma [ lem : help ] and proposition [ prop : james ] , @xmath623 has the simple head isomorphic to @xmath787 .
hence @xmath788 for any partition @xmath28 in the set @xmath789 .
the result now follows from equation [ eq:332 ] and frobenius reciprocity .
now consider the case of @xmath790 .
using the @xmath15-quiver for @xmath5 we see : if @xmath791 does not strictly dominate a @xmath1-regular partition @xmath28 of @xmath20 and @xmath792 then @xmath793 or @xmath794 . observe that @xmath795 and @xmath796 for all @xmath234 .
assume @xmath797 .
then @xmath798 , so @xmath105 \neq 0 $ ] .
the result follows since @xmath799 .
now suppose @xmath800 .
we have a short exact sequence @xmath801 where @xmath802 has a specht filtration @xmath803 with @xmath804 at the top . hence , @xmath805 . by frobenius reciprocity and equation [ eq:21 ] , @xmath806 at this point
the reader may wonder whether proposition [ prop : ks ] holds for all @xmath1-regular partitions in @xmath5 .
one is led to believe that it does since @xmath5 is closely related to the blocks of @xmath14 . in corollary
[ cor : ks ] we see proposition [ prop : ks ] does indeed hold .
[ cor : ks ] let @xmath11 and @xmath28 be a @xmath1-regular partitions in @xmath5 .
if @xmath28 does not strictly dominate @xmath28 then @xmath807 it suffices to prove the theorem for a @xmath1-regular partition @xmath11 in @xmath5 that is @xmath1-regular , not @xmath1-restricted , a non - hook , and has at least @xmath1 non - zero parts .
refer to the list of abacus displays from the proof of lemma [ lem : nicerest ] .
of these displays the following have at least @xmath1 non - zero parts : 1 .
@xmath808 or 2 .
@xmath809 , @xmath234 .
assume @xmath810 .
then @xmath811 .
write @xmath812 and @xmath813 .
suppose @xmath28 is a partition in @xmath5 such that @xmath775 and @xmath814
. then either 1 .
@xmath815 and @xmath816 where @xmath817 or 2 .
@xmath818 and @xmath819 \neq 0 $ ] . if ( 1 ) holds then @xmath820 \neq 0 $ ] by proposition [ prop : ext2 ] .
hence @xmath105 \neq 0 $ ] by lemma 2.13 of @xcite .
notice that @xmath12 has loewy length 2 or 3 and @xmath762 .
hence @xmath821 .
the result now follows since @xmath822 is one - dimensional .
now assume ( 2 ) holds .
so @xmath823 . we have a short exact sequence . @xmath824 since @xmath825 , @xmath826 .
we conclude that @xmath827 .
a similar argument can be used to prove the theorem for @xmath828 for @xmath234 .
it is known that if @xmath11 is @xmath1-restricted then @xmath12 has simple socle isomorphic to @xmath273 .
so the mullineux map gives the socle of @xmath12 whenever @xmath11 is @xmath1-restricted .
we are concerned with determining the socle of a specht module in @xmath5 that has loewy length 3 .
we claim that the socle is indeed the third radical layer .
[ thm : socles ] let @xmath11 be a @xmath1-regular partition in @xmath5 .
if @xmath12 has loewy length 3 then @xmath829 . by corollary [ cor : length3 ]
, @xmath11 must have @xmath211 notation of the form : 1 .
@xmath376 , @xmath830 , 2 .
@xmath213 , @xmath377 and @xmath831 , or 3 .
@xmath378 . by lemma [ lem : help ] , we know @xmath832 .
so we assume @xmath554 .
suppose @xmath833 .
we prove that @xmath575 is not in the socle of @xmath12 . by lemma [ lem : killhead ] we may choose
is @xmath234 such that @xmath499 and @xmath547 .
furthermore , @xmath834 for the unique @xmath1-regular partition @xmath28 in @xmath5 .
write @xmath835 .
now either @xmath836 or @xmath837 is simple . if @xmath836 then @xmath838 . so @xmath575 is not in the socle of @xmath12 by lemma [ lem : spechtlength2help](1 ) .
now suppose @xmath839 is simple . since @xmath840 , @xmath833 , and @xmath841 has simple head @xmath320 , we have @xmath842 . next , by proposition [ prop : spechtsimpleinduce ] , @xmath841 is reducible since @xmath11 is not in the set @xmath843 .
so @xmath844 since @xmath845 = 1 $ ] .
therefore @xmath575 is not in the socle of @xmath12 by lemma [ lem : spechtlength2help](1 ) .
in the previous section we saw that the loewy length of any specht module in @xmath5 is at most 4 .
moreover , @xmath12 has loewy length 4 if and only if @xmath11 is both @xmath1-regular and @xmath1-restricted . in this section
we establish an upper bound of 14 on the composition lengths of the specht modules in @xmath5 .
furthermore , we produce a specht module in the block that has 14 composition factors and we demonstrate that the only specht modules that can have composition length 14 must correspond to a partition that is both @xmath1-regular and @xmath1-restricted .
theorem [ thm:2nd ] describes the loewy structure of the specht module @xmath12 in @xmath5 , provided @xmath11 is @xmath1-regular and @xmath1-restricted . for such @xmath11 , @xmath846 is given by the @xmath15-quiver of @xmath5 . to obtain @xmath847
we apply theorem [ thm:2nd ] to @xmath848 and tensor @xmath849 with the sign representation .
we now place an upper bound on the number of composition factors seen in the second loewy layer of @xmath12 .
let @xmath11 be a @xmath1-regular partition in @xmath5 , and let @xmath850 be the set of all @xmath1-regular partitions @xmath28 in @xmath5 such that @xmath775 and @xmath851 is non - zero .
we claim that @xmath852 .
assume @xmath853 .
if @xmath854 , the quiver of @xmath283 ( see figure 16 of @xcite ) tells us that the second loewy layer @xmath855 has at most one composition factor since @xmath856 is @xmath1-regular .
in other words , @xmath857 . after analyzing the quivers ( see figures 5 - 8 of @xcite ) for the blocks @xmath858 ( @xmath859 ) we may conclude the following : given a @xmath1-regular partition @xmath287 in @xmath858 there are at most 3 @xmath1-regular partitions @xmath860 in @xmath858 such that @xmath860 strictly dominates @xmath287 and @xmath320 extends @xmath319 .
so , by proposition [ prop : ext2 ] , if @xmath859 then the second loewy layer of @xmath861 has at most 3 composition factors since @xmath862 is @xmath1-regular . in other words , @xmath863 . now given @xmath864 either there exists @xmath865 so that @xmath866 or @xmath867 whenever @xmath853 . in the former case , the quiver for the set @xmath868 tells us that @xmath317 extends @xmath308 . in latter case , we see that restricting to any of the blocks @xmath858 of @xmath14 misses a non - split extension of @xmath308 by @xmath317 . the next proposition , which is lemmas 4.2 , 4.3 , and 4.4 of @xcite , gives all pairs of @xmath1-regular partitions @xmath869 where @xmath870 and a non - split extension of @xmath317 by @xmath308 is missed by restriction .
[ prop : completemiss ] the following is a complete list of pairs of @xmath1-regular partitions @xmath869 where @xmath870 , @xmath871 , and @xmath867 whenever @xmath853 . 1 . for @xmath872 , @xmath873 and @xmath874 1 .
@xmath875 , @xmath876 2 .
@xmath877 , @xmath878 3 .
@xmath879 , @xmath880 4 .
@xmath881 , @xmath882 2 . for @xmath883 @xmath884 3 .
for @xmath885 @xmath886 [ lem : farapart ] assume @xmath245 .
if @xmath887 where @xmath248 and @xmath241 and @xmath250 then @xmath852 .
observe that @xmath888 and @xmath11 is not of the form ( 1 ) , ( 2 ) , nor ( 3 ) of proposition [ prop : completemiss ] . the discussion above gives the estimate @xmath889 .
we lower this bound on @xmath890 to 6 by applying the following strategy : use the ext@xmath9-quiver of @xmath5 to find three distinct partitions @xmath891 so that @xmath892 , @xmath893 , and @xmath894 . if such partitions exist then @xmath895 , for we have over counted these partitions .
if the triple @xmath896 does not exist , we provide the set @xmath850 .
we consider four cases
. .1 in * case 1 . *
assume @xmath897 and @xmath898 .
first suppose that @xmath899 , and consider the partitions @xmath900 , @xmath901 , and @xmath902 .
then @xmath903 , @xmath904 , and @xmath905 and @xmath891 ( by corollary [ cor : cpext ] ) .
if @xmath906 then @xmath907 , @xmath908 , and @xmath909 .
therefore @xmath852 .
.1 in * case 2 .
* assume @xmath897 and @xmath910
. then @xmath911 where @xmath897 .
if @xmath906 then @xmath912 . if @xmath913 then @xmath914 .
hence , @xmath852 .
.1 in * case 3 . *
assume @xmath915 and @xmath916 .
then @xmath917 where @xmath918 .
first suppose that @xmath906 . then @xmath919 .
if @xmath899 then @xmath920 . therefore , @xmath852 .
.1 in * case 4 . *
assume @xmath915 and @xmath921 .
then @xmath922 where @xmath923 .
first assume @xmath490 .
take @xmath924 , @xmath925 , and @xmath926 .
now if @xmath927 we have @xmath928 .
therefore , @xmath852 .
if @xmath929 is a @xmath1-regular partition @xmath5 not of the form @xmath214 where @xmath248 and @xmath930 then @xmath852 . by proposition [ prop : nremovables](5 ) , we know that @xmath11 is in at most two of the sets @xmath868 .
we consider two cases .
.1 in * case 1 .
* whenever @xmath931 and @xmath870 there is @xmath865 so that @xmath11 and @xmath28 are both in @xmath868
. then we have @xmath862 and @xmath932 are both @xmath1-regular and @xmath933 .
first assume there is only one @xmath934 such that @xmath853 . if @xmath935 then @xmath855 lies in the defect 1 block @xmath283 of @xmath14 .
hence the second loewy layer of @xmath855 has at most one composition factor .
we conclude that @xmath936 . if @xmath937 then @xmath861 lies in the defect 2 block @xmath858 of @xmath14 .
the second loewy layer of @xmath861 has at most 3 composition factors .
we conclude that @xmath938 .
now suppose that @xmath939 for some @xmath940 .
if @xmath941 then , using the argument just given , @xmath942 . finally , if @xmath943 then @xmath944 .
.1 in * case 2 .
* there is a @xmath1-regular partition @xmath870 in @xmath5 so that @xmath931 and @xmath867 whenever @xmath853 . first assume @xmath11 is in @xmath868 for the unique @xmath865 .
then , by proposition [ prop : completemiss ] , @xmath945 if @xmath946 and @xmath947 .
now suppose that @xmath939 for some @xmath940 .
if @xmath941 then @xmath948 .
so assume that @xmath943 .
after analyzing the @xmath211 display for the partition @xmath11 in the pair @xmath869 of proposition [ prop : completemiss ] , we need only consider @xmath11 from the pairs @xmath949 of ( 1 ) or ( 2 ) of proposition [ prop : completemiss ] and @xmath950 .
for such @xmath11 , @xmath951 .
we use the ext@xmath9-quiver of @xmath5 to show that there is a partition @xmath864 such that @xmath952 .
if such a partition @xmath28 exists then @xmath953 .
first consider @xmath954 for @xmath955 .
now @xmath956 . using the ext@xmath9-quiver of @xmath5
we see that @xmath957 .
observe that @xmath958 .
next assume that @xmath959 where @xmath883 . then @xmath960 .
if @xmath961 then @xmath962 , and @xmath963 . if @xmath964 then @xmath965 .
by the ext@xmath9-quiver , @xmath966 . observe that @xmath967 .
finally , consider @xmath968 for @xmath883 .
then @xmath956 .
first suppose @xmath969 . if @xmath223 then @xmath970 and @xmath971 .
if @xmath245 then @xmath972 and @xmath971 .
finally , assume that @xmath973 .
the partition @xmath974 is in both @xmath850 and @xmath975 . if @xmath929 is a @xmath1-regular partition in @xmath5 then @xmath852 .
[ cor : compregrest ] if @xmath11 in @xmath5 is @xmath1-regular and @xmath1-restricted then the specht module @xmath12 has at most 14 composition factors .
corollary [ cor : compregrest ] gives an upper bound on the number of composition factors @xmath12 has when @xmath11 is @xmath1-regular and @xmath1-restricted .
for such a partition @xmath11 in @xmath5 , we think of the corresponding specht module @xmath12 as `` large '' compared to other spechts in @xmath5 ( in the sense of loewy length ) .
we will see that any specht module in @xmath5 has at most 14 composition factors . specifically , if @xmath11 in @xmath5 is either @xmath1-regular and not @xmath1-restricted , @xmath1-singular and @xmath1-restricted , or neither @xmath1-regular nor @xmath1-restricted then @xmath12 has strictly less than 14 composition factors .
we begin the discussion by providing a specht module that has exactly 14 composition factors .
[ prop : ex14 ] there is at least one specht module in @xmath5 with composition length 14 .
consider @xmath976 which is @xmath1-regular and @xmath1-restricted .
note that @xmath977 , so @xmath11 is self - conjugate only if @xmath59 . observe that @xmath11 has exactly @xmath1 parts .
define the sets .1 in * case 1 . *
assume @xmath59 , so then @xmath978 .
from the ext@xmath9-quiver , @xmath979 by theorem [ thm:2nd ] , @xmath317 is in the second loewy layer of @xmath12 for each @xmath980 .
now @xmath981 and theorem [ thm:2nd ] give @xmath982 is in the third loewy layer of @xmath12 for each @xmath980 .
hence @xmath12 has 14 composition factors .
.1 in * case 2 . *
assume @xmath245 .
from the ext@xmath9-quiver , @xmath983 and .1 in @xmath984 .
.1 in the result now follows from theorem [ thm:2nd ] and proposition [ prop : james ] . for @xmath985 , the specht module @xmath986 has composition length 14 .
the case @xmath906 is proposition [ prop : ex14 ] since @xmath987 .
now assume that @xmath899 .
we saw in the proof of lemma [ lem : farapart](case 4 ) that @xmath988 is @xmath989 now @xmath990 hence @xmath991 , so @xmath992 has composition length 14 by proposition [ prop : james ] and corollary [ cor : ks ] .
take @xmath130 in proposition [ prop : ex14 ] .
the partition @xmath993 has @xmath994 display @xmath995 .
so we see that the specht module @xmath996 has 14 composition factors .
we provide the explicit partitions which label the simple modules lying in the second radical layer of @xmath996 .
@xmath997 computing the image of @xmath675 under the mullineux map we obtain the labels for the third radical layer of @xmath996 .
@xmath998 finally , @xmath996 has simple head isomorphic to @xmath999 and simple socle isomorphic to @xmath1000 .
so we know the radical series of @xmath996 .
we now turn our attention to showing every specht module in @xmath5 corresponding to a partition that is @xmath1-regular and not @xmath1-restricted has composition length at most 14 .
furthermore , we show that any such specht module in fact has composition length at most 13 .
the decomposition numbers in weight 3 blocks of the symmetric group are not well understood .
it is known that they are at most one ; however , there is no nice closed form for the numbers .
the next proposition helps us relate the decomposition matrix for @xmath5 to the decomposition matrix for @xmath14 .
[ prop : jm ] suppose @xmath1001 .
then every decomposition number of the principal block of @xmath1002 is either zero or can be equated with an explicit decomposition number of @xmath1003 .
suppose @xmath12 and @xmath317 are in the principal block @xmath5 with @xmath105 \neq 0 $ ] .
let @xmath1004 be the removable nodes for @xmath11 . by proposition [ prop : jm ] , @xmath28 has at least one normal node of @xmath1-residue res@xmath64 for some @xmath1005 .
it is known that a specht module in defect 2 has at most 5 composition factors ( cf .
@xcite , theorem 5.6 ) and a specht module in defect 1 has at most 2 composition factors . in light of this we obtain an upper bound on the composition length of @xmath12 ( by lemma 2.13 of @xcite ) . in this case
@xmath1006 is an upper bound for the composition length of @xmath12 .
now suppose @xmath11 is @xmath1-regular with @xmath1007 normal nodes .
if @xmath1008 then we can improve upon our upper bound , for we have over counted the composition factor @xmath308 by @xmath1009 ( this counting argument works since @xmath1010 = 1 $ ] ) .
hence @xmath1011 is an upper bound on the composition length of @xmath12 .
we restrict our attention to non - hook partitions @xmath11 in @xmath5 that are @xmath1-regular but not @xmath1-restricted .
any such partition must have @xmath211 notation of the form : 1 .
@xmath1012 where @xmath501 , 2 .
@xmath213 where @xmath377 , or 3 .
@xmath378 . by lemma [ lem :
help ] , @xmath526 has at most 13 composition factors since the specht module @xmath1013 has at most 14 composition factors .
so we only consider the partitions from ( 1 ) and ( 2 ) above .
define @xmath1014 and define the collection @xmath1015 .
the discussion that follows proposition [ prop : jm ] together with propositions [ prop : jm ] , [ prop : nremovables ] gives : if @xmath11 is in @xmath1016 then the specht module @xmath12 has at most 14 composition factors .
now , consider @xmath1017 .
the display @xmath1018 has removable beads on runners @xmath1019 , @xmath68 , and @xmath1 , so @xmath1020 has a non - zero restriction to one defect 1 and two defect 2 blocks of @xmath14 . using our counting argument
we see that @xmath1020 has at most @xmath1021 composition factors .
hence we consider partitions that lie in @xmath1022 .
first fix @xmath349 . the removable beads for @xmath809 lie on runners @xmath1023 , and the normal bead lies on runner @xmath1024 .
so @xmath1025 . furthermore , the partition @xmath1026 is @xmath1-regular and not @xmath1-restricted ( @xmath1027 ) .
now fix @xmath1028 that satisfy @xmath353 and @xmath241 .
the removable beads for @xmath213 lie on runners @xmath1029 , and the normal beads for @xmath213 lie on runners @xmath1030 .
so @xmath1031 . finally , the partition @xmath1032 is @xmath1-regular and not @xmath1-restricted ( @xmath1033 ) . by propositions
[ prop : w2 ] , [ prop : spechtsimpleinduce ] we conclude the following . [ prop : lastcase ] the following statements hold . 1 . for each @xmath349 and @xmath1034 ,
the specht module @xmath1035 has loewy length 2 . 2 . for each @xmath353 with @xmath241 and @xmath1036 ,
the specht module @xmath1037 has loewy 2 .
proposition [ prop : lastcase ] allows one to use the ext@xmath9-quivers given by martin and russell ( see @xcite figures 5 - 8 ) to place the desired upper bound on the composition length of the specht module @xmath12 where @xmath1038 . henceforth , we assume @xmath346 is expressed in @xmath1039 notation ( see proposition [ prop : images ] )
. begin by fixing @xmath349 . using proposition [ prop : images](1 ) and
figures 5 - 8 of @xcite , deduce that @xmath1026 is connected to at most 3 vertices in the ext@xmath9-quiver of the block @xmath1040 for each @xmath1027 . in other words , @xmath1035 has at most 4 composition factors ( @xmath1041 ) .
hence , the specht module @xmath1042 has at most @xmath1043 composition factors .
next fix @xmath1028 that satisfy @xmath353 and @xmath241 . by the same argument given above the specht module
@xmath1044 has at most @xmath1045 composition factors .
we use the following argument to prove @xmath12 has at most 13 composition factors where @xmath1046 . find all blocks @xmath1040 ( @xmath1047 ) of @xmath14 such that @xmath1048 and @xmath1049 .
choose the unique partition @xmath1050 , which is necessarily @xmath1-regular , such that @xmath870 and @xmath1051 .
then @xmath1052 .
proposition [ prop : cp ] will tell us that we have over counted the composition factor @xmath317 of @xmath12 . from above , the removable beads for @xmath213
lie on runners @xmath1029 and @xmath1031 .
observe that @xmath1053 by propositions [ prop : decnumb ] and [ prop : cp ] , @xmath1054 = 1 $ ] and @xmath1055 = 1 $ ] .
now @xmath1056 and @xmath1057 implies we have over counted the composition factors @xmath1058 and @xmath1059 of @xmath1044 .
hence @xmath1044 has at most 13 composition factors . the argument used to lower the composition length of @xmath1044 from 15 to 13
may be used to show that for any @xmath1060 the corresponding specht module @xmath12 has at most 13 composition factors .
we summarize the main result of the section in the following theorem .
[ thm : complengthbound ] suppose the specht module @xmath12 lies in the principal block @xmath5 of @xmath2
. then @xmath12 has at most 14 composition factors .
furthermore , if @xmath12 has composition length 14 then @xmath11 is both @xmath1-regular and @xmath1-restricted .
in this section we summarize the results of this paper , indicate what is still not known about the specht modules in @xmath5 , and discuss specht modules in other weight 3 blocks of the symmetric group .
we summarize the results of this paper in the following proposition . 1
. the loewy length of @xmath12 is at most 4 . 1 .
@xmath12 has loewy length 1 if and only if @xmath1061 .
@xmath12 has loewy length 2 if and only if @xmath596 .
( see theorem [ thm : spechtlength2 ] ) 3 .
@xmath12 has loewy length 3 if and only if @xmath11 is not a hook and either @xmath11 is @xmath1-regular and not @xmath1-restricted or @xmath1-singular and @xmath1-restricted .
@xmath12 has loewy length 4 if and only if @xmath11 is @xmath1-regular and @xmath1-restricted .
2 . if @xmath11 is @xmath1-regular and @xmath1-restricted then @xmath1062 .
3 . the module @xmath12 has composition length at most 14 .
furthermore , if @xmath12 has composition length 14 then @xmath11 is @xmath1-regular and @xmath1-restricted .
4 . proposition [ prop : ks ] holds for all @xmath1-regular partitions in @xmath5 .
suppose @xmath123 lies in a defect 3 block of @xmath87
. if the partition @xmath28 is @xmath1-regular and @xmath1-restricted then @xmath123 has loewy length 4 . we have discovered a number of results related to the structure of specht modules in @xmath5 ; however , we still do not have a general description of the socle of many of the modules .
let @xmath11 be a @xmath1-regular partition in @xmath5
. if @xmath12 has loewy length 2 then the socle of @xmath12 is determined by the ext@xmath9-quiver of @xmath5 ( by corollary [ cor : ks ] ) .
if @xmath12 has loewy length 3 then @xmath1063 is known in theory .
indeed , one may pick a prime @xmath279 and list the composition factors of @xmath12 . in this case , by the bipartite nature of the ext@xmath9-quiver and theorem [ thm : socles ] , the socle of @xmath12 is the direct sum of all composition factors @xmath317 of @xmath12 such that @xmath1064 and @xmath1065 .
so in theory we know the socles of the modules for each fixed prime @xmath279 , but we do not have a general answer .
for example , in the case of @xmath59 we found that every specht module in the block has simple head and simple socle , except @xmath1066 and @xmath1067 .
explicitly , @xmath1068 and @xmath1069 . we are left with the question : is there only one @xmath1-regular partition in @xmath5 such that @xmath12 does not have a simple socle ? one may wonder whether the results ( 2 ) - ( 4 ) extend to all defect 3 blocks of the symmetric group ( over characteristic at least 5 ) .
the author knows of no example where @xmath308 extends @xmath278 in a defect 3 block of the symmetric group .
perhaps this can be resolved using the algorithm , which constructs the ext@xmath9-quiver of a weight 3 block from the ext@xmath9-quiver of another weight 3 block , given by martin and tan in proposition 4.25 of @xcite .
finally , we saw that @xmath12 in @xmath5 has loewy length 4 if and only if the corresponding partition is @xmath1-regular and @xmath1-restricted .
we proved the ` only if ' direction for any defect 3 block of @xmath83 .
it remains open whether the ` if ' direction holds .
.1 in * acknowledgements .
* i would like to thank my advisor , dr .
david hemmer , for introducing me to this problem and for all of his guidance throughout this research .
i would also like to thank the referee for his or her helpful comments and suggestions .
m. fayers , _ decomposition numbers for weight three blocks of symmetric groups and iwahori - hecke algebras _ , transactions of the american mathematical society , vol .
3 ( mar . , 2008 ) , pp .
1341 - 1376 .
|
let @xmath0 be a field of characteristic @xmath1 at least 5 .
we study the loewy structures of specht modules in the principal block of @xmath2 .
we show that a specht module in the block has loewy length at most 4 and composition length at most 14 .
furthermore , we classify which specht modules have loewy length 1 , 2 , 3 , or 4 , produce a specht module having 14 composition factors , describe the second radical layer and the socle of the reducible specht modules , and prove that if a specht module corresponds to a partition that is @xmath1-regular and @xmath1-restricted then the head of the specht module does not extend the socle .
|
if one assumes the existence of a mixing between an active neutrino flavor @xmath1 and a sterile neutrino flavor , then a lepton number can be generated in the early universe @xcite .
to this purpose a small vacuum mixing angle ( @xmath2 ) is necessary ( the mass eigenstates almost coincide with the interaction eigenstates ) .
moreover the sterile neutrino must be lighter than the active neutrino and the absolute value of @xmath3 , the difference of squared masses , must be in the range @xmath4 .
in fact the critical temperature at which the generation occurs is given , if the sterile neutrino production is until that time negligible , by the expression : @xmath5 it follows that if @xmath6 , then @xmath7 and the lepton number variation would be dominated by the msw effect . in this case a lepton number larger than @xmath8 can not be generated @xcite .
on the other hand if @xmath9 then @xmath10 and one should be able to describe the neutrino oscillations in a quark - gluon plasma , something beyond the present level of matter effects account ( moreover it is not interesting in this context , because it would imply active neutrino masses much higher than @xmath11 , cosmologically excluded if neutrinos are stable ) .
a rigorous description of neutrino oscillations in the early universe requires a quantum kinetic approach able to describe the evolution of the statistical density matrix for the two mixed neutrino flavors @xcite .
such a description must include the different effects of matter on the neutrino mixing : a coherent effect due to forward scattering @xcite , a loss of coherence due to the collisions that change the neutrino momentum @xcite and the repopulation of active neutrino states ( depleted by the oscillations into the sterile neutrino states ) through the collisions @xcite . for a better understanding of the generation of a lepton number
, it has been shown to be more convenient to turn to a simpler description that neglects the possibility of a msw effect at the resonance .
this approximation proves to be valid at temperatures @xmath12 because the collisions destroy the coherence at the resonance . in this way
the physics underlying the lepton number generation is isolated and can be clearly understood @xcite .
the role of collisions , from this point of view , is crucial : they can be usefully considered , in a heuristic sense , as a measurement process able to make the two - quantum state collapsing in one of the two interaction eigenstates . through this effect
sterile neutrinos are produced and the presence of a small baryon number induces a tiny asymmetry between the production rates of neutrinos and antineutrinos . if the sterile is lighter than the active neutrino and if the values of mixing parameters are in the intervals previously indicated , this initial asymmetry is amplified through the generation of a lepton number that starts at the critical temperature @xmath13 .
in this mechanism the momentum dependence plays an important role @xcite and a monochromatic approximation provides only a rough description .
the interesting feature is that , within this simplified physical picture , the quantum kinetic description through the density matrix collapses into an _ effective ( classic ) kinetic description _
( pauli - boltzmann approach ) .
this is possible because in the regime where collisions are rapid enough , the evolution of the off - diagonal terms can be disentangled by the diagonal terms evolution and explicit expressions can be derived for them . in this way , in the description of the evolution of the statistical properties of mixed neutrinos , only the diagonal terms are left , the usual statistical distributions : a simpler physical picture yields a simpler mathematical description .
the derivation of the equations for the distributions can be done either directly from the simplified physical model @xcite or also , with a more formal procedure , from the quantum kinetic equations themselves , via some approximations , indicated by the authors globally as _ static approximation _ ,
valid under appropriate conditions on the mixing parameters @xcite .
the differences that one expects from a full quantum kinetic description are at low temperatures ( @xmath14 ) , when the msw effect becomes important and must be taken into account , and at the critical temperature for large enough vacuum mixing angles , when the growth of lepton number is so rapid that an adiabatic condition to describe the process does not hold anymore .
this adiabatic condition is the possibility to neglect any change in the effective potentials , and thus in the mixing angle , between the collisions on average .
the collisions , if vacuum mixing angles is not too large , are able to average out the coherent effects of the oscillations in the macroscopic quantity ( like the neutrino asymmetry ) .
this analysis on the validity of the approach is confirmed by the numerical calculations performed using the qke . in @xcite
it has been shown how the msw effect at low temperature is able to amplify the growth of lepton number up to values slightly lower than the maximum absolute value obtainable of @xmath15 , corresponding to a situation when all active antineutrinos ( or neutrinos ) are converted into sterile neutrinos and to a value @xmath16 for the chemical potential .
on the other hand it has been shown how , for temperatures @xmath17 , an effective kinetic approach agrees almost perfectly with the quantum kinetic one for small mixing angles .
it is also possible to extend the numerical study of qke to large mixing angles @xcite and it has been found that for @xmath18 and @xmath19 , in the quantum kinetic approach , at the critical temperature @xmath20 , rapid oscillations in the lepton number take place . this behaviour was first studied in @xcite and later confirmed in @xcite ( even though in a smaller region of mixing parameters ) , but in these works the momentum dependence was not taken into account and the rapid oscillations are observed for a much larger region of mixing parameters . in @xcite is concluded that , outside the special region of mixing parameters where rapid oscillations are observed at the resonance , at temperatures @xmath17 , the effective kinetic approach provides a very good description not only for the evolution of the absolute value of lepton number but also in predicting its final sign . in this paper
we will extend this analysis to the case when some tiny inhomogeneities in the baryon number ( baryon isocurvature perturbations ) are present , dealing with the region of mixing parameters where the effective kinetic approach can be used .
this can be done including into the equation a term that describes the diffusion of neutrinos and we will show how this effect can induce the generation of lepton domains .
the plan of the paper is the following . in section 2
we will show how the problem of sign of lepton number is fully determined in the effective kinetic approach , re analyzing and extending a procedure presented in @xcite .
in section 3 we will extend the analysis to the case where small inhomogeneities in the baryon number are present , through the introduction of a diffusion term .
we will illustrate how this term can induce , in some regions , an inversion in the sign of lepton number growth , bringing to the formation of lepton domains . in section 4
we describe qualitatively the evolution of lepton domains once they have been generated . in section 5
we conclude and discuss the possible applications of the new proposed mechanism .
within the pauli - boltzmann approach developed in @xcite , the evolution of the lepton number carried by an active @xmath1-neutrino flavor that is mixed with a sterile neutrino flavor , is described by the following equation : @xmath21\ ] ] where we defined : const .
this term can give effects only when an initial @xmath1-lepton number much higher than @xmath22 is assumed .
here we do not consider this situation and in this case this third term can be neglected . ]
@xmath23 @xmath24 the variable @xmath25 in the integrals is the adimensional momentum @xmath26 . the quantity @xmath27 is rigorously defined as the lepton number of the @xmath1-active neutrino in the portion of comoving volume that contains a fixed number of photons @xmath28 at some initial temperature @xmath29 . considering that the evolution of lepton number freezes at temperatures around @xmath30 , one can safely consider the number of photons in the element of comoving volume as a constant ( neglecting muon annihilations ) and write : @xmath31 we indicated with @xmath32 the particle densities and with @xmath33 the numbers of particles in the comoving volume @xmath34 .
we also introduced the _ ( effective ) total lepton number _
@xmath35 defined as : @xmath36 where @xmath37 holds for @xmath38 and @xmath39 for @xmath40 .
the field @xmath22 is the total charge number of non oscillating neutrinos plus a contribution from the baryon number carried by neutrons : we will refer to it as the _ background charge_. it is constant while the oscillations occur , and must be considered as a parameter given by some earlier phase of baryo - leptogenesis . in this section
we will assume that it is also strictly homogeneous , while in the next section we will study the effect of the presence of inhomogeneities .
both in @xmath35 and in @xmath22 we dropped an index @xmath1 to simplify the notation .
the quantities @xmath41 , @xmath42 are the production rates for sterile neutrinos and antineutrinos , @xmath43 is the fermi - dirac distribution with zero chemical potential and @xmath44 are the sum and difference distributions of neutrinos , relative to the fermi - dirac one .
the active neutrinos distributions can be safely described by thermal equilibrium distributions because the process of generation of lepton number occurs for temperatures @xmath45 ( the total collision rate is therefore much higher than the expansion rate ) and for small mixing angles ( the collision rates that refill the quantum states of active neutrinos are much higher than the sterile neutrinos production rates that deplete them ) . on the other hand
the sterile neutrinos distributions must be described by two other rate equations that , together with the expressions for the production rates , can be found in @xcite ( we do not need them in the present context ) . for a qualitative understanding of the evolution of lepton number , with a specific attention to its sign
, it is useful to recast the equation ( [ eq : lnua ] ) in the form : @xmath46\;(l_{\nu_{\alpha}}-l_{eq})\ ] ] where we introduced the quantity : @xmath47 while the term @xmath48 is always positive , the term @xmath49 changes sign at a critical temperature @xmath20 , being negative for higher temperatures and positive for lower temperatures .
when the temperature is far from the critical value ( @xmath50 ) , the @xmath48 term acts as a correcting term ( @xmath51 ) and the fixed point @xmath52 .
moreover , until the lepton number is below a threshold value @xmath53 ( see @xcite ) , its non linear effect inside @xmath49 and @xmath48 can be neglected . in this situation
the equation ( [ eq : lnua ] ) becomes extremely simple : @xmath54 while for temperatures above @xmath20 the quantity @xmath49 is negative , the fixed point @xmath55 is stable and thus the total lepton number is destroyed , for temperatures below @xmath20 , @xmath49 is positive , the fixed point is unstable and the total lepton number starts to grow . to answer the question toward which direction it grows
, one has to take into account the term @xmath48 and study its action in the vicinity of the critical temperature when @xmath49 is small and @xmath48 becomes important in driving the evolution of lepton number .
let us see in different stages what happens when temperature approaches the critical value from the above .
when @xmath48 is still small compared to @xmath49 , expanding up to the first order in @xmath56 , one gets : @xmath57 during this period the term @xmath58 is negative and the solution tracks the fixed point @xmath55 . an approximate expression for the quantity @xmath59 can be derived neglecting a term @xmath60 in the eq .
( [ eq : dlnuab ] ) : @xmath61 until @xmath62 , the solution tracks the fixed point . from the expression ( [ eq : leq ] ) it is clear that the term @xmath48 drives the growth of the lepton number toward the same sign as the background charge @xmath22 . at the critical temperature the term @xmath49 vanishes and simply @xmath63 . in this moment
the fixed point is changing very rapidly while the term @xmath58 , that should force the solution @xmath27 to track @xmath55 , is small and the solution starts to diverge from the fixed point .
immediately below the critical temperature , when @xmath64 , the fixed point has a vertical asymptote and changes sign and at still lower temperatures it rapidly approaches zero again .
this time however @xmath49 is positive and the fixed point @xmath55 does not attract the solution any more and this continues to grow toward the same direction transmitted by the `` derailment '' action of the term @xmath48 at the critical temperature .
the behaviour of the solution around the critical temperature is shown in fig . 1 for a particular choice of the mixing parameters .
in this section we will generalize the process of active - sterile neutrino oscillations to the case when spatial inhomogeneities are present . from the observation of cmb we know that small inhomogeneities in the temperature field ( adiabatic perturbations ) were present in the early universe .
however taking into account the presence of these perturbations does not change the basic results of the homogeneous scenario .
the lepton number growth starts at slightly different times in different points , but the final result is unchanged : the presence in the end of a final lepton number with the same sign everywhere in the space and with the same absolute value as in the homogeneous case .
more interesting is to consider the possibility that small inhomogeneities are present in the background charge @xmath22 ( isocurvature perturbations ) .
this quantity is the sum of the lepton number carried by the non oscillating neutrinos and a term given by the presence of a baryon number . whatever is the mechanism that created the inhomogeneities , it is reasonable to think that the inhomogeneities in the lepton numbers carried by the non oscillating neutrinos become soon much smoother than those in the baryon number , due to the much higher diffusion of neutrinos .
more precisely we can consider the baryon number field expressed in the comoving coordinates as a constant , neglecting the neutron diffusion .
this will make much simpler our following discussion , without altering the main results . a first trivial effect on the equation ( [ eq : lnua ] ) is simply that now @xmath65 , where @xmath66 are the comoving coordinates .
at temperatures above @xmath20 this effect is enough to understand how the process is modified by the presence of the inhomogeneities
. the @xmath67-lepton number will evolve in a way to destroy in any point the total lepton number @xmath35 and a zero order solution , considering only the @xmath49-term , is given by @xmath68 .
when temperature drops down , approaching the critical value , again the action of the correcting term @xmath48 must be considered in order to understand toward which direction , positive or negative values , the total lepton number will grow .
this time however a new correcting term must be considered in the equation , a term that arises due to the neutrino diffusion , and the full equation becomes now : @xmath69
-b(t , l)\;l_{\nu_{\alpha } } + \frac{\cal d}{r^{2}}\nabla^{2}l_{\nu_{\alpha}}\ ] ] where clearly @xmath70 , @xmath71 is the _ diffusion coefficient _ and @xmath72 is the scale factor that appears because we are dealing with comoving coordinates ( we normalize @xmath72 in a way that @xmath73 ) , while the diffusion term must be calculated in physical lengths .
the appearance of a diffusion term is something intuitively clear . for a more rigorous derivation one
has simply to write the liouville operator in the left hand side of the boltzmann equation in the inhomogeneous case : @xmath74 where @xmath75 is the expansion rate @xmath76 . in this way , integrating on the momenta the boltzmann equation for the distribution difference of active neutrinos @xmath77 and dividing for the photon number density , one obtains : @xmath78_{\rm osc}+\frac{1}{r^{2}}{\bf\nabla}_{\bf x } \cdot\left({\cal d}\,{\bf\nabla}_{\bf x}l_{\nu_{\alpha}}\right)\ ] ] where the first term in the r.h .
side is the contribution from the oscillations into the sterile neutrinos ( the r.h .
side in the eq .
( [ eq : lnua ] ) ) , while the second term is the contribution due to the diffusion .
the diffusion coefficient is defined by the expression : @xmath79 as we are considering the case of small inhomogeneities in the baryon number , the diffusion coefficient can be safely considered homogeneous and one gets the diffusion term in the form written in the equation ( [ eq : lnuad ] ) .
the order of magnitude of @xmath71 is given by @xmath80 , where @xmath81 is the interaction length and @xmath82 is the total collision rate , with @xmath83 @xcite for @xmath84 . if the diffusion term is negligible compared to the @xmath48 term , then the homogeneous scenario is practically unchanged . in the equation ( [ eq : leq ] ) one
has simply to consider that now @xmath65 and , as we are considering small inhomogeneities so that the sign of @xmath22 is spatially constant , the @xmath48 term pushes the solution toward the same direction in all the points . on the other hand
if we assume that the diffusion term is dominant , the situation can be much different .
let us assume for definiteness that @xmath85 . in the regions where the background charge is lower
, the @xmath1-neutrino lepton number , before the critical temperature , would be higher ( because @xmath86 ) . in these regions
the diffusion term pushes the lepton number to be depleted with the result that at the critical temperature one can have @xmath87 and the growth starts toward a negative sign .
the vice versa would happen in the regions where the background charge is higher .
the final result is the creation of regions with different sign of lepton number with the same size as the scale of baryon number inhomogeneities .
the sign of lepton number is in fact determined at any point by the curvature of the background charge field and thus the global properties of the baryon number inhomogeneities are somehow transmitted to the lepton domains .
we have now to study the conditions for which this mechanism can work .
the involved parameters are the mixing parameters , the amplitude of the inhomogeneities and the size of the inhomogeneities .
we can assume , for simplicity , that inhomogeneities have only one size scale . let us introduce the field @xmath88 of the inhomogeneities , writing the _ background charge _
field as : @xmath89\ ] ] where @xmath90 is the mean value , and where we assume that the field @xmath91 is a perturbation ( @xmath92 at any point ) . as we already said , while also non oscillating neutrinos give a non negligible contribution to the mean value , the baryon number gives the dominant contribution to the inhomogeneities . at temperatures higher than @xmath20
the solution will approach again the fixed point that at the zero order will be simply @xmath93 .
the final sign of lepton number will be again determined by the correcting terms that this time include also the diffusion term .
this can be generically estimated through the following expression ( we drop the dependence on @xmath66 in @xmath91 and @xmath94 ) : @xmath95 where we indicated with @xmath96 the scale of the inhomogeneities and , given the generic physical length @xmath97 , we mean with @xmath98 , the comoving length normalized at the present .
we also introduced the convenient quantity @xmath99 . in this way the expression ( [ eq : leq ] ) , valid for the fixed point in the homogeneous case , becomes : @xmath100 or at the first order in @xmath101 : @xmath102 this expression is now describing what we previously said in words : if the term @xmath99 is negative ( it means @xmath103 , the regions where the background charge is lower than the mean value if we assume it is positive ) and its absolute value higher than @xmath48 , then the lepton number growth is addressed toward the opposite sign than in the regions where the opposite condition holds .
to determine the conditions for which an inversion of sign is possible , we have thus simply to compare the term @xmath48 with the absolute value of @xmath99 at temperatures around the critical temperature .
an estimation of the order of magnitude of @xmath48 can be easily done considering that the integral on the momenta receives a dominant contribution only around the resonant value .
this procedure has already been employed in @xcite ( in that paper it was used to estimate the @xmath49-term ) and here we only give the result ( valid for temperatures @xmath104 , when the lepton number is still small and can be neglected ) : @xmath105 where we defined @xmath106 .
a calculation of the comoving interaction length expressed in parsec gives the result : @xmath107 for @xmath108 . using this expression and imposing the condition @xmath109 for the generation of lepton domains , it is thus straightforward to derive the following condition on the ( orders of magnitude ) of the involved parameters : @xmath110 this analytical expression is confirmed by the numerical calculations . in figure 2
we show the evolution of the terms @xmath111 , @xmath48 , calculated numerically for a particular choice of the mixing parameters and compared with the term @xmath112 , calculated through the expression ( [ eq : d ] ) with different values of the parameter @xmath113 .
the result is in agreement with the one that could be derived using the eq .
( [ eq : rho ] ) .
this condition can also be expressed as a condition on the order of magnitude of the maximum size of lepton domains that can be generated : . below this scale neutrino are free streaming and they can destroy the lepton domains more rapidly than they can be generated , considering that @xmath114 . ]
@xmath115^{-\frac{2}{3 } } \left(\frac{\rho}{10^{-1}}\right)^{\frac{1}{2}}\ , \left(\frac{10^{-10}}{s^{2}}\right)^{\frac{1}{2 } } \stackrel{<}{\sim}10\,{\rm kpc}\ ] ] in this last expression we indicated the extreme possible values of the different quantities that yield the maximum value for @xmath116 .
in particular the condition @xmath117 must be imposed to have a not negligible final absolute value of lepton number . in next section
we will use this result to sketch a scenario of how lepton domains with different sizes evolve and affect the bbn .
this analysis will suggest a way to circumvent the upper limit found on the maximum size of a lepton domain .
in conclusion of this section we notice that because of the presence of the term @xmath48 there are always points where , even though @xmath103 , the condition @xmath118 is not verified .
therefore we can say that the @xmath48 term will always favour a dominance of regions where the generation of lepton number occurs with the same sign of @xmath22 : we will refer to it as the _ dominant sign _ , compared to regions with _
inverted sign_. there are however two much different extreme situations of dominance . in a first one @xmath119 everywhere , except in thin walls around the surfaces where @xmath120 ( _ weak dominance _ ) . in this case
the size of lepton domains coincides with the size of baryon number inhomogeneities and in a comoving sphere with a much larger radius than the size of inhomogeneities , the volume of regions with dominant sign lepton number is almost equal to that of regions with inverted sign lepton number .
the topology of lepton domains is that one of a cubic lattice with each cube surrounded by first closest neighbours cubes with opposite sign and second closest neighbours cubes with the same sign . in a second extreme situation
the condition @xmath118 is verified only in small regions around the points where @xmath121 is maximum , with a size much smaller than that of the inhomogeneities ( _ strong dominance _ ) . in this case one
has a structure of lepton domains with inverted sign ( 3-dim ) islands in a dominant sign background .
lepton domains start to be formed at @xmath122 . with the increase of lepton number the diffusion at the border of domains also increase .
there is some interplay between the rate of generation ( the @xmath49-term ) and the rate of diffusion .
let us assume that the dominant sign is positive . in this case
the diffusion can be described as a process that gradually fills at the border the regions with negative sign ( the `` holes '' ) . at the same time , in the positive sign regions , the lepton number that has been used to fill the holes , is restored by the oscillations through the growth term . in this situation
it is clear that if the diffusion is able to cover the whole hole , the generation of lepton domains is only a transient regime without any practical effect .
we can refer to this phase as _ cannibalization regime _ , in which the dominant sign regions enlarge at the expense of the inverted sign regions , until eventually their complete disappearance .
however this process can last only until the generation of a lepton number can occur at the border of domains , where there is a change of sign and the lepton number is kept small by the diffusion .
when the temperature drops down to @xmath123 , in the tail of the distribution . for small values of @xmath35 , neutrinos and antineutrinos
are both resonant but if @xmath124 and if @xmath125 , the number of resonant antineutrinos ( neutrinos ) is a little bit higher than the number of resonant neutrinos ( antineutrinos ) : this explains why the fixed point @xmath126 is unstable and lepton number starts to grow ( see @xcite ) . ] or in any case down to @xmath127 mev , the growth starts to be inhibited .
we have to calculate the diffusion length to know how large the lepton domains with inverted sign must be in order to survive to the cannibalization regime .
this is given by the following expression that takes into account the universe expansion : @xmath128 the _ diffusion velocity _ is approximately given by @xmath129 .
considering that @xmath130 and that @xmath131 , it is an easy task to get the following expression for the comoving diffusion length : @xmath132,\ ] ] for @xmath133 .
the generation of lepton number can only occur down to temperatures @xmath134 , therefore for @xmath135 , using the expression ( [ eq : tc ] ) , we get : @xmath136 for sizes below this value ( _ very small scales _ ) lepton domains with inverted sign are destroyed before they can produce any effect . for @xmath137
the diffusion length between the beginning of the generation of lepton number at @xmath20 and its end at @xmath138 tends to zero for @xmath139 , but in any case there is a lower limit on the size of lepton domains that can be generated @xcite and in the end the same expression ( [ eq : min ] ) can be approximately used . for larger scales the cannibalization can not destroy completely the lepton domains before that the generation of lepton number at the border stops .
when this happens , the lepton number that diffuses to fill the hole is not generated any more : the result is an effect of dilution of lepton number .
the generation of lepton domains leaves a mark that modifies the homogeneous scenario . in this case , to understand the fate of lepton domains , it is useful to compare the scale of lepton domains with the horizon scale , given by the following expression : @xmath140 it coincides with the diffusion length at @xmath141 when neutrinos start to free stream . all lepton domains with _ small scales _
@xmath142 , do not survive and one gets in the end a homogeneous lepton number field .
however , as already stated , this time the generation of lepton domains produced the effect to dilute the final value of lepton number .
if the condition for the formation of domains ( [ eq : rho ] ) is satisfied only in the peaks of the inhomogeneities ( _ strong dominance _ ) , then the contribution from the regions with inverted sign is negligible and the dilution effect too .
otherwise in the other extreme case , when the relation ( [ eq : rho ] ) is satisfied in almost all the space ( _ weak dominance _ ) , then there would be an almost total reciprocal cancelation of regions with opposite sign , with only a small relic value of lepton number due to the presence of the @xmath48-term .
all intermediate values of lepton number are possible .
scales of inhomogeneities @xmath143 can produce lepton domains able to survive until the freezing of neutron to proton ratio and therefore the presence of these scales would give rise to an inhomogeneous bbn scenario .
another important distinction is between scales able to produce primordial nuclear abundance inhomogeneities that survive until the present ( _ large scales _ ) or not ( _ medium scales _ ) .
in fact , even though nuclear abundances are produced in a inhomogeneous scenario of bbn , subsequent astrophysical mixing mechanisms , such as shock waves induced by supernovae explosions , would be able to homogenize the products of bbn @xcite .
these processes are effective on scales @xmath144 . from this point of view the mechanism we proposed , in its simplest version , is unable to produce visible primordial abundance inhomogeneities .
we can however circumvent this limit if we relax our initial assumption on the presence of only one characteristic scale in the field of the inhomogeneities @xmath88 .
we can in fact imagine a simple extension with two characteristic scales contemporarily present ( for examples two fourier modes ) : one small scale and one large scale .
the amplitude of the first one could be modulated by the presence of the second . in this case
one can have the simultaneous existence of regions where the small scales inhomogeneities are able to efficiently dilute the lepton number generated at the critical temperature down to negligible values , and regions where the amplitude of small inhomogeneities is depressed and there is no dilution effect . in this way one can produce islands where lepton number is present in a background with zero lepton number or vice versa .
this time the size of the island regions has no intrinsic limit , but it is a feature of the model that produced the baryon inhomogeneities . more realistically one has to think that the inhomogeneities are described by a spectrum of lengths .
the interesting property is that the spectral features of the isocurvature perturbations would be reflected in the lepton domains features and eventually in the bbn products .
another interesting aspect is that very large scale lepton domains would not be constrained by microwave background observations as models where large amplitude isocurvature perturbations are directly present in the baryon number @xcite .
this because the inhomogeneity in the active neutrino energy density would be compensated by an opposite inhomogeneity in the sterile neutrino energy density .
in fact during active - sterile neutrino oscillations one has that @xmath145const . even though sterile neutrinos free stream while active neutrinos are still diffusing , for very large scales ( surely larger than the diffusion length of active neutrinos ) this different behaviour can not change the conservation of the sum of the asymmetries of active neutrinos and sterile neutrinos at each point . in the end of this section
we want also to show that small amplitudes of inhomogeneities are sufficient for the mechanism to work . from the expression ( [ eq : rho ] ) , imposing the lower limit ( [ eq : min ] ) on @xmath116 ( in order to have lepton domains with inverted sign large enough not to be cannibalized ) , one gets the following condition on the mixing parameters : @xmath146 that is satisfied for large regions of possible values of the mixing parameters , even for very small values of @xmath91 .
these regions are represented in figure 3 for three different values of @xmath91 .
it can be noticed that even for @xmath147 , there is a significant region where lepton domains can be generated and produce interesting effects .
we proposed a mechanism according to which active - sterile neutrino oscillations would amplify small baryon number inhomogeneities generating a structure of lepton domains .
the assumption on the presence of baryon number inhomogeneities is a very reasonable one , because very small amplitudes are sufficient to produce lepton domains large enough to give some effect .
the structure of the lepton domains and its effect on the bbn would reflect the spectral features of the same perturbations that seeded the formation of the domains .
the main attraction of this mechanism is that the size of these domains is not limited by the horizon scale .
moreover large scale lepton domains can not be ruled out by current cmb observations ( but maybe could be constrained or discovered in future experiments ) .
the mechanism presents different applications .
it provides a scenario for a non standard bbn ( see @xcite for a review and see also the recent paper @xcite where the effects of inhomogeneous chemical potentials on bbn are considered ) .
if the active neutrino is an electron neutrino the effects on the nuclear abundances can be remarkable @xcite and some of them could help to explain the present observational picture , for example claimed inhomogeneities in the d / h abundances from measurements in quasar absorption systems @xcite . from this point of view
one possibility is that we live in a background of zero electron lepton number but some observed high redshift absorption systems , for example the one observed with an high abundance , could be included in an island of non zero lepton number .
we also mention that isocurvature perturbations can have interesting effects on large scale structure @xcite .
very generically we can say that the generation of lepton domains enlarge the possibility of the existence of some observational signature of active - sterile neutrino oscillations in the early universe . in the very fortunate case that active - sterile neutrino oscillations , with the right parameters for a lepton domain formation , do really occur in nature ,
then we would have a powerful probe for baryogenesis models that predict a spectrum of inhomogeneities .
the simple non observation of a signature of the generation of lepton domains would put strong constraints .
we conclude saying that the planned earth experiments will be able in next future to test the hypothesis of active - sterile neutrino oscillations and thus to rule out or support the proposed mechanism .
* acknowledgements * p. di
bari acknowledges many interesting and stimulating discussions with r. foot , p. lipari , m. lusignoli and r.r .
volkas and is grateful to them for many suggestions that made this manuscript clearer .
he also thanks k. jedamzik for a nice discussion during cosmo-99 .
10 r.foot , m.j.thomson and r.r.volkas , r5349 ( 1996 ) .
k.enqvist , k.kainulainen and j.maalampi , 754 ( 1991 ) .
r.barbieri and a.d .
dolgov , 743 ( 1991 ) .
dolgov . 700
( 1981 ) .
l. wolfenstein .
2369 ( 1978 ) .
r.a.harris and l. stodolsky , 464 ( 1982 ) .
l.stodolsky , 2273 ( 1987 ) .
m.j.thomson , 2243 ( 1992 ) .
b.h.j.mckellar and m.j.thomson , _
phys.rev._*d49 * 2710 ( 1994 ) .
r.foot and r.r.volkas , 5147 ( 1997 ) ; p.di bari , p.lipari , and m.lusignoli , hep - ph/9907548 .
nicole f.bell , raymond r.volkas , and yvonne y.y.wong , 113001 ( 1999 ) .
r.foot and r.r.volkas , _
phys.rev._*d56 * 6653 ( 1997 ) ; erratum _ phys.rev._*d59 * 029901 ( 1999 ) .
r. foot , 10 ( 1999 ) .
a.d.dolgov , s.h.hansen , s.pastor and d.v.semikoz , hep - ph/9910444 p.di bari and r.foot ( in preparation ) .
x.shi , 2753 ( 1996 ) .
k.enqvist , k.kainulainen and a.sorri , hep - ph/9906452 .
x. shi and g.m.fuller astro - ph/9904041 k.enqvist , k.kainulainen and m.thomson , 498 ( 1992 ) .
l.d landau , e.m .
lifsits , , ( pergamon , new york , 1981 ) .
r. foot , private communication .
k.jedamzik and g.m .
fuller , 33 ( 1995 ) .
c.j.copi , k.a.olive and d.n.schramm , _ ap.j . _
* 451 * ( 1995 ) , k.jedamzik , proceedings of neutrino astrophysics , ringberg , 1997 .
s.e.whitmire and r.j.scherrer , astro - ph/9910352 .
r.foot , hep - ph/9906311 .
x.shi , g.m .
fuller and k.abazajian , _ phys .
* d60 * 063002 ( 1999 ) .
j.k.webb et al .
, 250 ( 1997 ) .
a.d.dolgov and b.e.j.pagel , new astronomy , in press .
d.kirkman , d.tytler , s. burles , d. lubin and j.m .
omeara , astro - ph/9907128 .
peebles _ nature _ * 327 * 210 ( 1987 ) .
r.cen , p.ostriker , and p.j.e .
peebles , 423 ( 1993 ) .
the behaviour of @xmath148 ( solid line ) around the critical temperature when the sign is determined by the action of the correcting term @xmath48 .
the dotted line is the quantity @xmath149 , with @xmath55 defined by the expression ( [ eq : leq ] ) and @xmath150 .
we used @xmath151 , while the values of mixing parameters are those indicated . * fig .
2 * comparison of the quantities @xmath152 times @xmath153 ( for numerical calculations it is more convenient to replace time with temperature ) .
the term @xmath49 vanishes at the critical temperature @xmath154 .
the values of mixing parameters are those indicated .
the diffusion term is plotted for different values of the parameter @xmath155 .
it can be noticed that the diffusion term @xmath112 is dominant on @xmath48 , around the critical temperature , for @xmath156 , as it could be deduced by the expression ( [ eq : rho ] ) . *
3 * represantation of the regions where the condition ( [ eq : domains ] ) for the surviving of lepton domains to the canibalization regime is satisfied .
the regions are those below the three solid lines and correspond to the three indicated values for the amplitude @xmath91 of the baryon number inhomogeneities that seed the lepton domain formation . above the dotted line the contribution of sterile neutrinos to the number of effective neutrinos during the bbn is higher than @xmath157 for @xmath38 ( see @xcite ) . even for very small amplitudes ( @xmath147 ) there is a significant region of parameters where the condition ( [ eq : domains ] ) is satisfied .
|
we show how the generation of a lepton number in the early universe induced by @xmath0 , in presence of small baryon number inhomogeneities , gives rise to the formation of lepton domains .
the structure of these domains reflects the spectral features of the baryon number inhomogeneities that generated it and interestingly the size of the lepton domains can be super - horizon .
pacs numbers : 14.60.pq ; 14.60.st ; 98.80.bp;11.30.fs keywords : early universe , neutrino oscillations , sterile neutrino , baryon isocurvature perturbations , leptogenesis .
|
the period of low solar activity between 1645 to 1715 is commonly known as the maunder minimum and has attracted attention of researchers as it put forward many unexpected questions . during this period ,
the sunspots were infrequent and as a result it was assumed that the solar dynamo and the solar wind were either weak or switched off .
anomalies in the surface differential rotation during the maunder minimum @xcite also point out the global changes in the internal solar dynamo mechanism .
measurements of @xmath0be concentration in the dye 3 ice core @xcite recently showed that the magnetic cycles persisted throughout the maunder minimum , although the sun s over all activity was drastically reduced .
this period is also associated with the so called `` little ice age '' in europe and suggested that it might be due to a decrease in solar irradiance .
many authors have estimated the total solar irradiance during the maunder minimum and find a decrease in the range of 0.1% at a time of relatively high activity down to 1% at a time of no or low activity . in a different approach
, we estimate the change in the total solar irradiance and 10.7 cm radio flux during the maunder minimum .
this is achieved by calculating the _ p_-mode frequencies using the historic record of sunspots and the empirical relations derived by . since these relations between mode frequencies and solar activity indices are shown to be independent of solar cycles , these can be reliably used to estimate variations in solar irradiance and radio flux .
we estimate the change in _ p_-mode frequencies during the maunder minimum using the following relation derived by :
@xmath1 where @xmath2 is given in nhz and @xmath3 is the smoothed sunspot number .
the estimated @xmath2 for the epoch 1610 1750 and 1965 1998 using the mean annual sunspot numbers from and equation ( 1 ) is plotted in figure 1 .
it is observed that during the deep maunder minimum ( 1655 - 1685 ) , the frequency shifts are negligible corresponding to the low or no solar activity period . the maximum change in the _ p_-mode frequencies during the entire period
is found to be 43 @xmath4 3 nhz , which corresponds to 11% of the maximum change during the solar cycle 22 .
this small change in frequency may be interpreted as due to the internal structural stability in the sun during that period . from the historic photoheliograpic records , and have shown that the solar diameter during the maunder minimum increased by 7 arc seconds , as compared to the present value .
have studied the correspondence of _ p_-mode frequency shifts with solar radius and neutrino flux and found that @xmath2 is anticorrelated with both . from these findings
, we infer that during the maunder minimum , the neutrino flux would have increased from the present day value .
this indicates that there was a real expansion of the solar envelope during this period and a decrease in solar surface temperature and irradiance .
these observed correlations between the solar activity , _
p_-mode frequencies , solar radius , neutrino flux and solar irradiance , strongly suggest the possibility that the dynamical processes occuring in the upper layers are closely related to the perturbations inside the sun , perhaps even down to the core .
since the sun s radiation is the principal driving force for global climate , the variation in total solar irradiance ( tsi ) may lead to the climatic change .
the coincidence of maunder s `` prolonged solar minimum '' with the coldest period in europe has been reported earlier .
it was suggested by that this cold period might be due to the reduction in solar irradiance .
the estimates for solar irradiance during the maunder minimum have been provided by many workers and they have found a decrease with respect to the present value . to obtain an independent estimate of the variation of the solar irradiance , we follow and derive the following relation between @xmath5 and @xmath6 : @xmath7 using this relation and the value of @xmath5 for the maunder minimum obtained earlier
, we estimate that the total solar irradiance during maunder minimum decreased by 2.68 w / m@xmath8 ( 0.19% ) from the average value of 1366.06 w / m@xmath8 for cycle 22 .
had found that the tsi during maunder minimum was lower by 1% than the value for 1980 . using empirical correlation between tsi and the integrated ca ii emission , suggested a decrease between 0.15% and 0.35% . obtained a 0.2% decrease in the irradiance by comparing the variation in the differential helio - latitudinal rotation during the maunder minimum and present time . calculated an average decrease of 0.25%0.5% in luminosity using the solar radius measurements .
they had assumed an anticorrelation between the sun s diameter and its luminosity . modified these values to 0.11%0.43% using the maunder minimum sunspot numbers and near - equatorial rotation rates .
further , from the contemporary and past rotation rates and solar radii , reported a decrease of 1.23% in tsi for the year 1683 , when there were almost no sunspots and 0.37% for the year 1715 , when solar activity begun to rise .
thus , our estimate of tsi fairly agrees with the earlier results using different acitivity indicators or physical parameters . from the estimated change in frequency during the maunder minimum and the following equation from @xmath9 we calculate the change in 10.7 cm radio flux during maunder minimum .
as compared to 128 sfu - the average value for cycle 22 , the radio flux is found to decrease by 66.2 sfu _
i.e. _ by 52% during the maunder minimum .
using the historical sunspot data and the empirical relations between the _ p_-mode frequencies and activity indices derived by , it is estimated that during the maunder minimum from 1645 to 1715 ad , the mode frequencies showed a maximum change of 43 @xmath4 3 nhz . using the derived mode frequencies as a proxy of the solar activity indicators , we have further estimated the total solar irradiance and 10.7 cm radio flux and find that these are lowered by 0.19% and 52% respectively than the average values of the solar cycle 22 .
the decrease in _ p_-mode frequency and irradiance during maunder minimum probably indicates stability in the dynamical processes occuring in the solar convection zone and even deep down up to the core .
beer , j. , tobias , s. and weiss , n. : 1998 , _ solar phys . _ * 181 * , 237 .
delache , ph . ,
gavryusev , v. , gavryuseva , e. , laclare , f. rgulo , c. and roca corts , t. : 1993 , , 801 .
eddy , j. a. : 1976 , _ science _ * 192 * , 1189 .
jain , k. , tripathy , s. c. , bhatnagar , a. and kumar , b. : 2000 , _ solar phys .
_ , 192 , 487 .
lean , j. , skumanich , a. and white , o.r . :
1992 : _ geophys .
* 19 * , 1951 .
mendoza , b. : 1996 , _ geofis .
int . _ * 35 * , 161 .
mendoza , b. : 1997 , , 523 .
nesme - ribes , e. and mangeney , a. : 1992 , _ radiocarbon _ * 34 * , 263 .
nesme - ribes , e. , ferreira , e. n. , sadourny , r. , le trent , h. and li , z. x. : 1993 , _ geophys .
res . _ * 98 * , 18923 .
reid , g. : 1991 , _ j. geophys .
res . _ * 96 * , 2835 .
ribes , j.c . and nesme - ribes , e. : 1993 , , 549 .
ribes , e. , ribes , j. c. and barthalot r. : 1988 , in j. christensen - dalsgaard and s. frandsen ( eds . ) _ advance in helio- and asteroseismology _ , d. reidel publishing company , dodrecht , holland , p. 227 .
|
using the sunspot numbers reported during the maunder minimum and the empirical relations between the mode frequencies and solar activity indices , the variations in the total solar irradiance and 10.7 cm radio flux for the period 1645 to 1715 is estimated .
we find that the total solar irradiance and radio flux during the maunder minimum decreased by 0.19% and 52% respectively , as compared to the values for solar cycle 22 .
|
a wide variety of real - life problems involve the simultaneous optimization of several objectives or criteria .
objectives that contradict each are common , thus resulting in situations where one objective can not be improved without worsening at least one other objective .
for example , in cancer radiotherapy treatment , one goal is to maximize the amount of cancerous tissue exposed to radiation treatment while another ( contradicting ) goal is to minimize the radiation exposure of healthy organs . here
, there is typically no trivial compromise .
problems of this kind are called _ multi - criteria optimization _ ( mco ) problems .
the number of competing goals in the treatment planning for _ intensity modulated radiation therapy _ ( imrt ) is quite large ( about 10 - 25 ) . exploring the solutions to the mco problem
involves a high - dimensional pareto boundary @xcite . approximating
this set is computationally expensive , and experience shows that there are many tradeoffs that are either clinically unacceptable or irrelevant .
thus a significant part of the computational effort is wasted on identifying useless solutions .
the complexity of the approximated high - dimensional pareto boundary ( see e.g. @xcite ) makes it difficult for the decision maker to navigate .
an alternative approach would be to first identify the most important aspects of the imrt plan and make sure that these are met by the solution ; compromises , if necessary , would only be acceptable for subsidiary goals . toward this end , we want to incorporate a priori knowledge about priorities into the treatment plan optimization process .
this information translates naturally into _ lexicographic optimization _ ( lo ) problems , which constitute a special class of mco problems .
classical lo methods minimize the objective functions sequentially , starting with the most important one and proceeding according to the lexicographic order of the objectives . here , the optimal value found for each objective is added as a constraint for subsequent optimizations
. these constraints can be quite rigorous , however , depending on the correlation of the corresponding functions and the decision maker might not always want to sacrifice so much optimizing freedom for the sake of a single goal .
we therefore propose the more intuitive approach of gathering goals in groups of descending importance , as in @xcite , and formulating objective functions as weighted sums of the functions contained in these groups .
this reduces the number of objective functions to be minimized , which leads to fewer optimization problems to be solved and thus lowers the computational effort .
typically , the subsequent objectives are in conflict with each other , which implies that the minimum of one objective function is a relatively bad starting point for subsequent optimization levels . to address this issue
, we propose to combine the superiorization methodology with an mco algorithm .
superiorization can speed up convergence by steering the mco algorithm towards solutions that are better suited to minimizing subsequent objective functions . in the best case
, one could even find the minimizer of all subsequent objective functions before reaching the last optimization level .
here , subsequent minimization becomes unnecessary , and one need only verify that the solution is indeed the minimizer of all subsequent objective functions under the given constraints .
the superiorization methodology @xcite was recently developed as a framework for algorithms that lie conceptually between feasibility - seeking and optimization algorithms .
it is a heuristic tool that does not guarantee to find the optimum value of a given functional , rather it obtains a solution that is superior ( with respect to a given objective function ) to the solution achieved by a classical feasibility seeking algorithmic operator .
the main advantage of an algorithm that uses superiorization as the driving tool ( as opposed to optimization ) is that it requires less computational resources - given a good choice of parameters - while providing comparable solutions , from the point of view of real - world applications , to those that one would get with algorithms that use optimization . within this framework
, we wish to include the class of projection methods , which are applied successfully in many real - world feasibility problems , see e.g. , @xcite .
this class has witnessed great progress in recent years and its member algorithms have been applied with success to problems in sensor networks , image reconstruction , image processing , imrt planning , resolution enhancement and in many others ; see @xcite . apart from theoretical interest , the main advantage of projection methods is their computational efficiency .
they commonly have the ability to handle huge - size problems of dimensions in which other , more sophisticated methods cease to be efficient . in this paper
we propose to apply the superiorization methodology with projection methods to improve the speed of convergence of lexicographic optimization methods .
we demonstrate the plausibility of using superiorization with a simple example with two optimization variables .
we also present results from calculations on four real head neck cases in imrt for two different choices of superiorization parameter sets suited to yield fast convergence for each case individually or robust behavior for all four cases .
the paper is organized as follows . in section [ sec : methods ] we present some preliminaries , definitions and algorithms that will be needed in the sequel . later , in section [ sec : numerical illustration ] our general scheme is illustrated on a 2d example and we present the results of our calculations on the aforementioned imrt head neck cases .
finally we discuss our findings and conclusions in sections [ sec : discussion ] and [ sec : conclusion ] .
let us start by giving a definition of what is a multicriteria optimization problem
. given a mapping @xmath2 with @xmath3 for all @xmath4 and let @xmath5 for @xmath6 .
the ` multicriteria optimization ` ( mco ) problem is the following .
@xmath7 where for all @xmath8 , @xmath9 . in this work we focus on a particular mco method which is called _ lexicographic optimization _
( lo ) . for lo
we define an order for the objectives according to their importance and minimize them in that order under the given constraints . after having minimized one objective ,
we impose an additional constraint which ensures that the objective function value of subsequent solutions do not deviate from the previously found minimum by more than a user - defined value .
in the context of imrt we consider a more general setting for the formulation of the optimization problems . here , we gather the objective functions @xmath10 in @xmath11 priority groups which are ranked from most to least important ( compare e.g. , @xcite ) . denote by @xmath12 an ordered subset of objective function indices such that @xmath13 is a _ partition _ of @xmath14 , meaning that @xmath15 it is clear that if for all @xmath16 the set @xmath17 is a singleton then this reformulation reduces to the classical lexicographic optimization problem .
now we define@xmath18 where for each @xmath19 , @xmath20 is the weight of the @xmath21-th objective function @xmath22 .
so @xmath23 gathers the objective functions whose indices are elements of @xmath17 . at optimization level @xmath24 @xmath25
we thus solve the following problem .
@xmath26 where @xmath27 are the minimum values found for each @xmath28 during the previous optimization levels , @xmath29 are small user - chosen constants for @xmath30 and @xmath31 are as in ( [ problem : mco ] ) .
let us illustrate the above for a two - stage lo also known as _ bi - level optimization_. assume that @xmath32 @xmath33 , and @xmath34 ; thus @xmath35 and @xmath36 .
at the first optimization level we solve the following single - objective optimization problem.@xmath37after we have solved ( [ eq : p1 ] ) and obtained @xmath38 ( @xmath39 ) , we continue with the second optimization level , where our problem formulates as @xmath40 where @xmath41 for some small user - chosen constant @xmath42 . in the above , the choice of @xmath43 and in general @xmath44 for @xmath45 ,
defines in a fixed way how far from previously found optima of objective functions of higher priority the algorithm is allowed to deviate in the current optimization level .
a different approach has been taken by @xcite which allows the decision maker to choose the tradeoff between two subsequent ( in terms of their lexicographic order ) objective functions in an interactive way .
observe that if the constraint set of ( [ eq : p1 ] ) @xmath46 ( or of ( [ eq : p2 ] ) with @xmath47 ) contains only one element , then this is the optimal solution of the bi - level optimization problem and it is found already at the first optimization level . otherwise , in case of multiple feasible points , a feasibility seeking algorithm for finding an element of ( [ eq : p1 ] ) might end up with a point which is relatively far from the optimal solution of the second optimization level , this is due to the fact that the second objective function has not been taken into account yet .
this effect is likely to occur when the objective functions of the subsequent optimization levels represent conflicting goals as commonly happens in cancer therapy . motivated by these difficulties we propose to steer the algorithm towards solutions that provide good starting points for subsequent optimization levels by applying the superiorization methodology together with projection methods .
before we can do so we need to transform the optimization problem such that we can use projection methods to solve it .
for the aforementioned purpose we use the _ level set scheme _
@xcite which is a new projection - based scheme for solving convex optimization problems .
it transforms a convex optimization problem into a sequence of auxiliary feasibility problems by iteratively constraining the objective function from above until the feasibility problem is inconsistent .
to illustrate the idea let @xmath48 be convex and continuously differentiable function and @xmath49 nonempty , closed and convex set .
following the level set scheme we can reformulate the problem @xmath50 in the following form @xmath51 now let @xmath52 be some user chosen positive sequence , for example @xmath53or @xmath54 . in the initial step of the level set scheme
we choose any projection method to find a feasible point @xmath55 , set @xmath56 and @xmath57 .
now at the iterative step @xmath58 , when we are given the current point @xmath59 , we try to solve the following _ convex feasibility problem _ ( cfp ) @xmath60 if there exists a feasible solution , set @xmath61 and continue , else there exists no feasible solution , and @xmath62 is an @xmath63-optimal solution . in this work ,
we use _ projection methods _ to solve the auxiliary convex feasibility problems generated by the level set scheme .
projection methods are iterative algorithms that use projections onto sets .
they rely on the general principle that when a family of ( usually closed and convex ) sets is present , projections onto the given individual sets are easier to perform than projections onto other sets ( intersections , image sets under some transformation , etc . ) that are derived from the given individual sets .
projection methods come in various algorithmic structures , some of which are particularly suitable for parallel computing , and they demonstrate nice convergence properties , for example bounded perturbations resilience .
this fact allows us to use them to solve the cfps resulting from the level set scheme and to incorporate superiorization , which is presented in the following paragraph , while retaining convergence to the optimal point .
consider the cfp : @xmath64 with @xmath31 defined as in .
the particular method we use in this paper is the _ simultaneous subgradient projections method _ which is defined as follows .
let @xmath65 be an arbitrary starting point .
given the current iterate @xmath62 , calculate the next iterate @xmath66 via @xmath67 where @xmath68 ( subgradient of @xmath69 at @xmath62 ) , @xmath70 are _ weights _ and @xmath71 $ ] ( relaxation parameters ) for arbitrary @xmath72 .
simultaneous projection methods are also referred to as parallel methods . in this case , in order to evaluate the next iterate , all ( or a block of more than one ) constraints are taken into account . in case of a system of linear equalities , this method is known as cimmino s method @xcite .
see @xcite for more details
. given a problem @xmath73 , an algorithmic operator @xmath74 is said to be ` bounded perturbations resilient ` if the following is true .
if the sequence @xmath75 generated by @xmath76 , for all @xmath77 , converges to a solution of @xmath73 , then any sequence @xmath78 of points in @xmath79 generated by @xmath80 for all @xmath77 , also converges to a solution of @xmath73 provided that , for all @xmath77 , @xmath81 are bounded perturbations , meaning that @xmath82 for all @xmath77 such that @xmath83 and the sequence @xmath84 is bounded .
the _ superiorization methodology _ of @xcite was recently developed as a framework for algorithms that lie conceptually between feasibility - seeking and optimization algorithms .
it is designed to find a solution to a cfp which is superior with respect to a given objective function @xmath85 , meaning with a value of @xmath85 at least as low , but possibly lower , compared to the solution obtained by a classical feasibility seeking algorithmic operator .
the state of current research on superiorization can best be appreciated from the superiorization and perturbation resilience of algorithms : a bibliography compiled and continuously updated by yair censor which is at : http://math.haifa.ac.il/yair/bib-superiorization-censor.html
. in particular , @xcite and @xcite are recent reviews of interest . given a function @xmath86 and a point @xmath87
, we say that a vector @xmath88 is ` nonascending ` for @xmath85 at @xmath89 if and only if @xmath90 and there is a @xmath91 such that for all @xmath92 $ ] we have @xmath93 consider again the convex feasibility problem : @xmath94 where @xmath95 is a closed and convex set and let @xmath96 be any feasibility - seeking algorithmic operator , that defines the basic algorithm @xmath76 to solve the convex feasibility problem .
let @xmath86 be a given convex and continuously differentiable function .
the _ superiorized _
( with respect to @xmath85 ) version of @xmath96 is @xmath97 where @xmath98 is nonascending for @xmath85 at @xmath99 .
one option for @xmath98 , as it is used in @xcite , is @xmath100 if the solution set of the feasibility problem contains several elements , the superiorized version of @xmath96 obtains a solution that is superior with respect to @xmath85 to the solution achieved by the basic algorithmic operator @xmath96 .
we now combine all of the mentioned concepts to derive our approach .
we consider the multicriteria optimization problem ( [ problem : mco ] ) and define the priority groups @xmath17 and weights @xmath101 , @xmath102 .
we then transform it into a lo problem like ( [ eq : mclo ] ) .
now we apply the level set scheme and consequently solve the following optimization problem at every optimization level @xmath103 : @xmath104 where @xmath105 to do that , we transform ( [ cfplexopt ] ) into a sequence of auxiliary cfps of the form : @xmath106 for decreasing sequences of @xmath107 .
we solve the auxiliary cfps by using the simultaneous subgradient projections method as the basic algorithm @xmath96 .
let @xmath108 be a subset of indices contained in the subsequent priority groups @xmath109 and let @xmath110 be weights to the corresponding objective functions . then
, after having successfully solved @xmath111 auxiliary cfps we include superiorization with respect to @xmath112 by defining @xmath98 as in ( [ eq : nabla ] ) with @xmath113 and setting @xmath114 where @xmath115 for all @xmath116 such that @xmath117 and @xmath118 , meaning that @xmath98 is a direction of nonascend for @xmath85 at @xmath62 . in the following we refer to this method as _ superiorized lexicographic
( slo ) .
the imrt planning problem is to find a treatment plan , i.e. energy fluence intensities , which results in a dose distribution in the patient s body that irradiates the tumor as homogeneously as possible while sparing critical healthy organs .
the dose ( measured in gy ) delivered to the patient s body is evaluated according to clinical goals which are translated into the functions given below .
the clinical goals are ordered according to their importance and can also be grouped together which reduces the number of optimization levels and therefore accelerates the optimization process . for our calculations on four head neck cases we chose one group of goals as constraints and three subsequent priority groups .
the volume of the patient s body is discretized into a three - dimensional grid of voxels .
the voxels are assigned to different planning structures , e.g. organs or tumor tissue .
let @xmath119 denote the so - called dose matrix that maps a vector @xmath120 of fluence intensities to the dose @xmath121 received by the individual voxels .
thus a planning structure @xmath122 is essentially a subset of indices of the vector @xmath123 . to evaluate the dose received by different planning structures we used lower and upper tail penalty functions as well as mean upper tail penalty functions .
the lower tail penalty function of the dose @xmath123 for a planning structure @xmath122 penalizes dose values below a given threshold @xmath124 .
it is given by @xmath125 the upper tail penalty function is @xmath126 and is used to penalize dose values exceeding a given threshold @xmath127 .
the mean upper tail function is defined as @xmath128 this function is used to keep the mean dose values below a given threshold @xmath11 , which is a less strict way of avoiding overdosage than using the upper tail penalty function .
table [ tab : problemformulation ] shows our problem formulation for the first head neck cases according to @xcite .
the other cases were treated similarly , depending on the individual geometry of the tumor tissue .
to gain an impression of the geometry of some of the planning structures involved see figure [ fig : imrtrsloct ] .
note that the function values of the evaluation functions listed as constraints were required to be 0 which translates to the dose prescription of the corresponding planning structure being met by the dose distribution .
the weights for the evaluation functions were chosen empirically for each case individually such that the optimization would yield a clinically acceptable treatment plan .
non - tumor tissue 1 denotes the three - dimensional margin of 1 cm around the lymphatic drainage pathways ( which provide the largest of the target volumes and contain both ptv 60 and ptv 70 ) .
non - tumor tissue 2 is the tissue that consists of everything but the lymphatic drainage pathways and non - tumor tissue 1 .
these structures are introduced to reduce the radiation to non - tumor tissue , especially to areas which are not covered by any of the organs at risk .
additionally , it offers an incentive to the optimization algorithm to reduce the dose values of the organs at risk that lie outside of the lymphatic drainage pathways even below the prescribed maximum or maximum mean dose values .
in optimization level @xmath103 we minimize the objective function @xmath129 which is a weighted sum of the functions belonging to priority group @xmath17 .
.evaluation functions of the structures under consideration gathered in different priority groups . [ cols="<,^,<,^",options="header " , ] in this section we present pseudo codes of the algorithms we implemented to solve the imrt optimization problems .
algorithm [ alg:1 ] implements the lexicographic optimization problem turned into a sequence of convex feasibility problems and algorithm [ alg:2 ] implements the superioriation methodology . in the following paragraph
we give some details about the submethods mentioned in the codes . `
findfeasiblesolution(t^{(\mu ) } ) ` tries to solve the cfp with @xmath130 within @xmath131 simultaneous projections .
if this is successful , it returns [ @xmath120 , true ] with @xmath132 and @xmath120 feasible .
if this is not successful , the method returns [ @xmath120 , false ] with @xmath133 , where @xmath134 and @xmath120 feasible .
@xmath135 is the smallest upper bound for @xmath129 that the algorithm was able to find a feasible solution for . `
reduce(\phi_{\mu}(\mathbf{p}x ) ) ` returns a new upper bound @xmath136 for @xmath129 that is smaller than @xmath137 .
` addconstraint(\phi_{\mu}^{\ast } ) ` adds @xmath138 to the set of constraints . `
supdirection(x ) ` returns the direction of superiorization at @xmath120 . in our calculations we used superiorization with respect to the objective function @xmath139 of the following optimization level .
@xmath24 = 1 ; @xmath140 @xmath141 reduce(@xmath142 ) levelissolved = false counter = 0 = findfeasiblesolution(@xmath136 ) counter++ levelissolved = true @xmath143 addconstraint(@xmath144 ) @xmath24++ x = superiorize ( ) @xmath136 = reduce(@xmath142 ) levelissolved = true @xmath143 addconstraint(@xmath144 ) @xmath24++ coefficientbigenough = true base @xmath145 @xmath146 = 0 direction = supdirection(@xmath120 ) loop = true exponent = 1 coefficient = base@xmath147 @xmath148 = @xmath120 + coefficient @xmath149 direction @xmath150 loop = false @xmath146++ exponent++ coefficientbigenough = false
in the first part of this section we demonstrate the plausibility of using the superiorized lexicographic optimization methodology on a simple example with two optimization variables which is taken from stanimirovi stanimirovic12 . in the second part we present the performance of our method on four real imrt head neck cases . to demonstrate the mechanism of superiorization and how it can be useful for lexicographic optimization we consider the following example .
solve the following multicriteria optimization problem @xmath151 where the functions @xmath152 @xmath153 and @xmath154 are given in lexicographic order .
the optimal solution @xmath155 is @xmath156 with @xmath157 .
we stopped the optimization process at the iterate @xmath62 if @xmath158 . in the following ,
points are given with precision @xmath159 .
in this example calculating the gradients and projections is computationally very cheap , but for imrt cases gradient- and dose - evaluations consume most of the computation time , so we compare the number of projections and gradient evaluations .
figure [ fig : toyextrajectoryboth ] shows the feasible region and the trajectories of the iterates of the classical and the superiorized lo for the starting point @xmath160 .
both methods first seek feasibility and find the point @xmath161 . in optimization level 1
, the classical level set scheme minimizes with respect to @xmath162 only and ends up at the solution @xmath163 of the first optimization level with @xmath164 using 327 projections and 327 gradient calculations .
additionally , we need another 260 projections and 518 gradient evaluations to find out that we can not improve @xmath162 any further while staying feasible . in optimization level 2
, we want to preserve @xmath165 for all subsequent iterates @xmath62 and minimize @xmath166 under this additional condition .
we arrive at the solution @xmath167 of optimization level 2 after a total of 4743 projections and 9107 gradient calculations .
we can not improve any further with respect to @xmath168 and have found the optimal solution .
the superiorized level set scheme minimizes with respect to @xmath162 in optimization level 1 but uses superiorization with respect to @xmath166 .
it arrives at the solution @xmath169 of optimization level 1 after 108 projections and 144 gradient calculations , 34 of which were needed for the superiorization . using the superiorized lo we get already sufficiently close to the optimum in optimization level 1 . from the formulae of the utilized functions it is clear that the minimal value we can hope to achieve for any of the @xmath170 is 0 .
thus we considered an optimization level @xmath103 to be solved , if @xmath171 given that @xmath120 is feasible .
what takes most of the computational time with imrt cases are multiplications of the dose matrix @xmath119 with the fluence vector @xmath120 which occur when we evaluate the dose or calculate the objective function gradient for a given @xmath120 .
the number of these multiplications is what we will use as units to measure progress , i.e. , reduction of objective function values , of the classical and the superiorized lo . in algorithm
[ alg:1 ] and [ alg:2 ] in subsection [ subsec : implementation ] we find parameters @xmath111 and @xmath172 which determine the behavior of the method .
@xmath111 is the number of cfps that have to be successfully solved before superiorization is applied and @xmath172 is the maximum number of superiorization steps that are taken .
if the superiorization step size is smaller than a predefined threshold or the maximum number of steps is reached , the superiorization stops and we solve the next cfp . in tables [ tab
: imrtfastresults1 ] and [ tab : imrtfastresults2 ] we first present results we achieved by trying different parameter sets @xmath173 and then picking for each case individually the set that resulted in the fastest convergence of the algorithm measured in the number of multiplications .
we refer to the sets of parameters picked according to this strategy as @xmath174 .
all values presented in the tables in this section are rounded to @xmath175 .
the results presented in table [ tab : imrtfastresults1 ] show that fast slo results in equal or lower objective values than unsuperiorized lo within less multiplications . in table
[ tab : imrtfastresults2 ] we see that fast slo produces solutions of the single optimization levels with potentially lower objective values of the subsequent objective functions than unsuperiorized lo .
fast slo thus offers a better starting point to the single optimization levels than unsuperiorized lo which reduces the overall number of multiplications .
@xmath176 denotes the value of @xmath170 at the solution @xmath177 of optimization level @xmath103 obtained using the classical lo , @xmath178 denotes the value of @xmath170 at the solution @xmath179 of optimization level @xmath103 obtained using the superiorized lo with @xmath174 .
@xmath180 is the number of multiplications needed by the classical lo to solve optimization level @xmath181 to level @xmath182 .
@xmath183 is the total number of multiplications .
analogously , @xmath184 denotes the number of multiplications needed by the superiorized lo with @xmath174 to solve optimization level @xmath181 to level @xmath182 and @xmath185 . for our calculations we chose @xmath186 for the classical lo and @xmath187 for the superiorized lo .
lcccc ' '' '' & @xmath188 & @xmath189 & @xmath190 & @xmath191/@xmath192 + case 1 & @xmath193 & 0.9598 & 0.9923 & 0.8998 + case 2 & @xmath193 & 0.9743 & 0.9960 & 0.6612 + case 3 & @xmath193 & 0.9996 & 1.0000 & 0.9210 + case 4 & @xmath193 & 0.9708 & 0.9941 & 0.8959 + lccccc ' '' '' & @xmath194 & @xmath195 & @xmath196 & @xmath197 & @xmath198 + case 1 & 1.6521 & 0.5669 & 0.9895 & 0.3741 & 0.9856 + case 2 & 1.4066 & 0.3588 & 1.0065 & 0.3682 & 1.0008 + case 3 & 1.0403 & 0.8594 & 1.0000 & 0.5625 & 1.0075 + case 4 & 0.3818 & 1.0536 & 0.8591 & 0.8196 & 0.9939 + it is not clear a priori how to choose values for @xmath111 and @xmath172 for a given problem .
therefore we aimed to find a robust set of parameters that yields good results for all imrt cases and could possibly be applied for other imrt head neck cases as well .
we refer to this set of parameters as @xmath199 .
the results for the robust parameter set are presented in table [ tab : imrtrobustresults1 ] and tab : imrtrobustresults2 .
the results in table [ tab : imrtrobustresults1 ] show that robust slo achieves comparable objective values to unsuperiorized lo within fewer multiplications . in table
[ tab : imrtrobustresults2 ] we see that robust slo , too , produces solutions of the single optimization levels with potentially lower objective values of the subsequent objective functions than unsuperiorized lo . like fast slo
, robust slo also offers a better starting point to the single optimization levels than unsuperiorized lo which again reduces the overall number of multiplications .
following the previous notation @xmath200 is the value of @xmath170 at the solution @xmath201 of optimization level @xmath103 obtained using the superiorized lo with @xmath199 .
@xmath202 denotes the number of multiplications needed by the superiorized lo with @xmath199 to solve optimization level @xmath181 to level @xmath182 and @xmath203 .
lcccc ' '' '' & @xmath204 & @xmath205 & @xmath206 & @xmath207/@xmath208 + case 1 & @xmath193 & 1.0046 & 1.0002 & 0.9177 + case 2 & @xmath193 & 1.0020 & 1.0043 & 0.8877 + case 3 & @xmath193 & 0.9829 & 0.9969 & 0.9541 + case 4 & @xmath193 & 1.0245 & 1.0071 & 0.8859 + lccccc ' '' '' & @xmath209 & @xmath210 & @xmath211 & @xmath212 & @xmath213 + case 1 & 1.1825 & 0.6972 & 0.9849 & 0.3916 & 1.0008 + case 2 & 0.7447 & 0.8409 & 0.9999 & 0.5194 & 1.0058 + case 3 & 1.1320 & 0.9029 & 0.9999 & 0.5530 & 1.0007 + case 4 & 1.2453 & 0.8575 & 0.8664 & 0.4959 & 1.0103 + in figure [ fig : imrtrsloct ] we present one slice of the ct of case 1 with the dose distribution resulting from the solution of superiorized lo with the robust parameter set @xmath214 .
the corresponding dvh is shown in figure [ fig : imrtrslodvh ] .
both the images and the dose matrix were created using cerr @xcite .
note that the dose calculation via the dose matrix , which is used in the optimization algorithm , provides only an approximation to the dose values calculated by cerr which are depicted in figure [ fig : imrtrsloct ] and [ fig : imrtrslodvh ] .
thus the dvh curves show e.g. that dose values below 66.5 gy occur in the ptv70 even though in the optimization the mathematical constraint ensuring that this does not happen is not violated .
+ + + + in figure [ fig : imrtobjfunctions ] we present the objective values of @xmath215 and @xmath154 plotted against the number of dosematrix - vector - multiplications for case 1 . to show the fundamental behavior of the superiorization method we intentionally did not give any actual numbers in the plots in figure [ fig : imrtobjfunctions ] . in figure
[ fig : imrtobjfunctionsa ] we can see how @xmath153 increases while we minimize @xmath216 in level 1 using classical lo .
we see how the superiorization with respect to @xmath153 disturbs the minimization of @xmath217 and leads to @xmath218 being smaller than @xmath219 .
figure [ fig : imrtobjfunctionsb ] shows that @xmath220 is a better starting point for the minimization of @xmath153 in level 2 than @xmath221 because its objective value of @xmath222 is significantly lower .
this results in @xmath223 being notably smaller than @xmath224 . due to the correlation of the clinical goals there are relatively few auxiliary cfps to be solved in optimization level 2 for all cases . when superiorization is applied ( both with optimal and robust @xmath173 ) we even get so close to the minimum of @xmath166 at the end of level 1 that for these parameter pairs the cfp in level 2 turns inconsistent within less than @xmath225 or @xmath226 reductions of @xmath227 .
therefore no superiorization is applied in level 2 and the objective value of @xmath228 is about the same for fast slo , robust slo and unsuperiorized lo . as we can see from figure
[ fig : imrtobjfunctionsc ] superiorization with respect to @xmath153 during level 1 can in some cases depending on the correlation of the objective functions reduce the value of @xmath168 and @xmath229 .
similar behavior can be observed for the superiorized lo that uses the robust parameter set @xmath199 .
we used lo as an intuitive approach to lower the complexity of decision making in imrt and to include a priori knowledge about the priorities of the decision maker .
we combined this mco technique with the superiorization methodology to speed up the optimization in the next optimization level .
this is not the only application for superiorization .
one could for example choose as superiorization function @xmath230 a weighted sum of different evaluation functions than we did , or one could choose @xmath230 as a function that is not included in the lexicographic optimization model at all.in this work we identified the superiorization parameters manually .
this process takes of course too much time to be of any practical use . to be able to find robust parameters that work for many optimization problems in an automated way
, one should make sure to identify problems with similar tradeoffs among the objective functions and the superiorization function(s ) . an issue with using projection methods to solve cfps is that in general we do not have a criterion that tells us whether a cfp is inconsistent .
therefore we defined a maximum number @xmath131 of attempts , i.e. simultaneous projections , to solve it . if the cfp can not be solved within @xmath131 simultaneous projections , we assume it to be inconsistent .
this means that finding out that a cfp actually is inconsistent takes @xmath131 projections , which drastically increases the total number of projections . on the other hand , low values of @xmath131 might not suffice to reach the actual minimum of an objective function under the given constraints , but only an approximation thereof .
we decided to stop solving an optimization level @xmath24 after encountering the first inconsistent auxilary cfp and continue with level @xmath231 , accepting that we might miss the minimal value of @xmath129 accessible with the chosen value of @xmath131 by a certain gap ( which is bounded by the reduction strategy of @xmath136 , in our case by 10 @xmath232 ) . with this strategy , the computationally expensive process of finding out that a cfp is inconsistent occurs ( at most ) once per optimization level .
for this reason it is a big advantage to know the objective values of the optimal solution a priori as in example [ toyex ] .
in general we do not have this information but as in our calculations for the imrt cases , sometimes it is possible to exploit a priori information about the objective functions to define a stopping criterion that is computationally cheap to evaluate .
a question arising in the context of optimization problems is how algorithm 1 performs compared to other solvers . in the following , we will refer to algorithm 1 as ( superiorized ) projection solver .
we implemented the projection solver in matlab . to keep computation time for algebraic operations comparable we chose to compare our solutions to the solutions of matlab s fmincon function using the interior point solver .
we did our calculations in matlab 2015b on an intel core i7 - 5600u processor with 2.6 ghz and 8 gb ram .
matlab s interior point solver took 16904 to 30690 seconds to solve the lexicographic optimization problem for our four cases . within a maximum of 1000 iterations per optimization level it yielded solutions which were slightly infeasible ( maximal constraint violation of @xmath233 - @xmath234 ) . due to the nature of the evaluation functions this infeasibility had no major effect on the dose values .
these solutions were still acceptable from a medicinal point of view . for a detailed comparison of computation time see table [ tab : matlabsolvercomparisoncomptime ] .
lcccc ' '' '' & ip & proj & robustsproj & fastsproj + case 1 & 30689.8 & 55.3 & 50.8 & 49.8 + case 2 & 17649.8 & 14.4 & 12.8 & 9.5 + case 3 & 26596.2 & 13.9 & 13.6 & 12.9 + case 4 & 16903.9 & 14.8 & 13.0 & 13.5 + all iterates produced by the projection solver are feasible by construction of the algorithm .
this enables us to decide how much computational effort we want to put into the calculations - of course at the expense of mathematical solution quality . for our calcuations
we chose the maximum number @xmath235 of simultaneous projections , to solve a given cfp for th projection solver .
when comparing the resulting solutions to the ones produced by matlab s solver ( see tables [ tab : matlabsolvercomparisonobjvalues ] and [ tab : projsolvercomparisonobjvalues ] ) it becomes apparent that they are suboptimal .
solutions with equal or smaller objective function values can be found by the projection solver , too , if @xmath131 is raised .
lcccc ' '' '' & @xmath236 & @xmath237 & @xmath238 & maximal constraint + & & & & violation + ' '' '' case 1 & 9.1529@xmath239 & 425 & 145 & 9.1419@xmath240 + case 2 & 3.6419@xmath241 & 363 & 795 & 7.2779@xmath240 + case 3 & 0.3834 & 172 & 371 & 9.8338@xmath241 + case 4 & 0.4778 & 66 & 382 & 9.6511@xmath242 + lcccc ' '' '' & @xmath236 & @xmath237 & @xmath238 & maximal constraint + & & & & violation + ' '' '' case 1 & @xmath243 & 720 & 97 & @xmath244 + case 2 & @xmath243 & 502 & 1725 & @xmath244 + case 3 & @xmath243 & 1477 & 1918 & @xmath244 + case 4 & @xmath243 & 1318 & 851 & @xmath244 + the dvh curves of case 1 in figure [ fig : imrtsolvercomparison ] show that the additional 30639 seconds of computation time the matlab solver takes compared to robustsproj result in an improvement of 3 gy in the average dose value of the ldp ( whose dose evaluation function is contained in @xmath245 ) at the cost of 2 gy increase in the average dose value of non - tumor tissue 1 ( whose dose evaluation function is contained in @xmath246 and thus considered less important ) . for case 1
, the matlab solver yields a solution with has a lower evaluation function value corresponding to the ldp than the projection solver .
because in case 1 the ldp has a relatively big weight @xmath247 in , this difference is amplified .
we observed similar behavior with the other imrt cases . concerning the question of performance compared to other solvers
, we observe that the projection solver with @xmath235 offers a fast way to reach a solution that is in many aspects already as good as the solution produced by matlab s solver .
if one is willing to invest additional hours of computation time , however , matlab s solver offers solutions with lower objective function values .
we demonstrated the concept of superiorization in combination with an mco method on a simple example as well as on four imrt head neck cases .
we showed that for lo we can speed up the optimization process by providing a better starting point for the next level of optimization .
we found a common set of superiorization parameters which yielded robust results for all four cases .
we thank the department of radiation oncology ( head : prof . dr . c. belka ) of the ludwig - maximilians university ( lmu ) , munich , germany , for providing the clinical data for our calculations and visualizations .
d. butnariu , r. davidi , g. t. herman and i. g. kazantsev , stable convergence behavior under summable perturbations of a class of projection methods for convex feasibility and optimization problems , _ ieee journal of selected topics in signal processing _ * 1 * ( 2007 ) , 540547 . y. censor , w. chen , p. l. combettes , r. davidi and g. t. herman , on the effectiveness of projection methods for convex feasibility problems with linear inequality constraints , _ computational optimization and applications _ * 51 * ( 2012 ) , 10651088 . y. censor , r. davidi , g. t. herman , r. w. schulte and l. tetruashvili , projected subgradient minimization versus superiorization , _ journal of optimization theory and applications _ * 160 * ( 2014 ) , 730747 .
r. davidi , y. censor , r. w. schulte , s. geneser , and l. xing , feasibility - seeking and superiorization algorithms applied to inverse treatment planning in radiation therapy , _ contemporary mathematics _ * 636 * ( 2015 ) , 8392 . w. jin , y. censor and m. jiang , a heuristic superiorization - like approach to bioluminescence , _ international federation for medical and biological engineering ( ifmbe ) proceedings _ * 39 * ( 2013 ) , 10261029 . k. izui , t. yamada and s. nishiwaki , a gradient - based multiobjective optimization technique using an adaptive weighting method , _ 10th world congress on structural and multidisciplinary optimization _
may 1924 , 2013 , orlando , florida , usa . k .- h .
kfer , m. monz , a. scherrer , p. sss , f. alonso , a. azizi sultan , t. bortfeld and c. thieke , multicriteria optimization in intensity modulated radiotherapy planning , _ springer optimization and its applications , ( handbook of optimization in medicine ) _ * 30 * ( 2009 ) , 123168 .
t. long , m. matuszak , m. feng , b. a. fraass , r. k. ten haken and h. e. romeijn , sensitivity analysis for lexicographic ordering in radiation therapy treatment planning , _ medical physics _
* 39 * ( 2012 ) , 34453455 .
|
multicriteria optimization problems occur in many real life applications , for example in cancer radiotherapy treatment and in particular in intensity modulated radiation therapy ( imrt ) . in this work we focus on optimization problems with multiple objectives that are ranked according to their importance .
we solve these problems numerically by combining lexicographic optimization with our recently proposed level set scheme , which yields a sequence of auxiliary convex feasibility problems ; solved here via projection methods .
the projection enables us to combine the newly introduced superiorization methodology with multicriteria optimization methods to speed up computation while guaranteeing convergence of the optimization .
we demonstrate our scheme with a simple 2d academic example ( used in the literature ) and also present results from calculations on four real head neck cases in imrt ( radiation oncology of the ludwig - maximilians university , munich , germany ) for two different choices of superiorization parameter sets suited to yield fast convergence for each case individually or robust behavior for all four cases .
@xmath0 optimization department , fraunhofer - itwm , kaiserslautern 67663 , germany + @xmath1 mathematics department , ort braude college , karmiel 2161002 , israel e - mail : [email protected] and [email protected] _ keywords _ : superiorization , projection methods , feasibility problems , multicriteria optimization , iterative methods + _ msc _ : 65k10 , 65k15 , 90c25 , 90c90 +
|
stellar streams in the milky way halo produced by the accretion of smaller galaxies are a standard prediction of hierarchical merging cosmogonies ( e.g. , * ? ? ?
* and references therein ) .
the most spectacular example is the disrupting sagittarius dwarf spheroidal ( sgr dsph ) , originally discovered by ibata , gilmore & irwin ( 1995 ) .
it has a heliocentric distance of @xmath2 kpc and is centered at galactic coordinates of @xmath3 and @xmath4 @xcite .
it is dominated by an intermediate age population ( between 6 and 9 gyrs , bellazzini et al .
2006 ) , but there is evidence for a much older population ( @xmath5 10 gyrs ) as well @xcite .
the metallicity [ fe / h ] ranges from very metal - poor ( as low as -2 based on the globular clusters ) up to approximately solar , with probably a mean of @xmath6 ( * ? ? ?
* and references therein ) .
it was realised early on that there was some tidal debris in the neighbourhood of the sgr dsph @xcite and that the distribution of this material traced the sgr dsph orbit .
subsequently , @xcite used sloan digital sky survey ( sdss ) first - year commissioning data to identify an over - density of blue a - type stars in two strips located at ( @xmath7 ) = ( @xmath8 , @xmath9 , 46 kpc ) and ( @xmath10 , @xmath11 , 33 kpc ) , which were then matched with the sgr stream @xcite .
likewise , @xcite noticed that clumps of rr lyrae stars in sdss commissioning data lay along the sgr stream s orbit .
the best panorama of the sgr stream to date was obtained by @xcite using m giants selected from the two micron all - sky survey ( 2mass ) .
they saw the trailing tidal tail very clearly in the southern galactic hemisphere , as well as part of the leading arm reaching towards the north galactic cap . here , we use sdss data release 5 ( dr5 ) to provide a picture of the leading arm of the sgr stream in the vicinity of the north galactic cap with remarkable clarity , together with a number of other notable stellar structures in the field .
[ cols="<,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ]
sdss @xcite is an imaging and spectroscopic survey that has mapped @xmath12 of the sky .
imaging data are produced simultaneously in five photometric bands , namely @xmath13 , @xmath14 , @xmath15 , @xmath16 , and @xmath17 @xcite .
the data are processed through pipelines to measure photometric and astrometric properties @xcite and to select targets for spectroscopic follow - up . for de - reddening
, we use the maps of @xcite .
dr5 covers @xmath18 square degrees around the galactic north pole , together with 3 strips in the galactic southern hemisphere .
we use the catalogue of objects classified as stars with artifacts removed , together with magnitude limits @xmath19 and @xmath20 . at low right ascension and declination ,
we are limited by the boundary of dr5 .
we choose to cut our sample at @xmath21 and @xmath22 .
this gives us a total of @xmath23 stars .
figure [ fig : sag ] shows the density of stars satisfying the color cut @xmath1 in sdss dr5 . as we show shortly , these are upper main sequence and turn - off stars belonging to the sgr stream .
this is a rgb - composite image assembled from three grayscale images colored red , green and blue corresponding to the density of the selected stars in the following three magnitude bins : red is @xmath24 , green is @xmath25 , and blue is @xmath26 . for a given stellar population ,
these magnitude bins are distance bins , with red being the most distant and blue the nearest .
there is a clear distance gradient along the stream , from the nearer parts at right ascensions @xmath27 to the more distant parts at @xmath28 .
there is also a bifurcation visible in the stream starting at about the north galactic pole ; in what follows , we will refer to the lower declination branch as branch a , and the higher declination as branch b. majewski et al .
( 2003 ) traced the northern stream of the sgr for right ascensions @xmath29 between @xmath30 and @xmath31 .
for @xmath32 , majewski et al .
( 2003 ) did not see a clear continuation of the stream .
the combination of majewski et al.s ( 2003 ) m giants together with the sdss stars in figure [ fig : sagnew ] shows for the first time the entirety of the stream , including its continuation through the galactic cap and into the galactic plane .
figure [ fig : sagnew ] also shows the locations of a number of globular clusters , some of which are known to be associated with the sgr stream .
for example , bellazzini et al .
( 2003 ) used 2mass data to conclude that ngc 4147 was physically immersed in the stream .
figure [ fig : sag ] displays such a remarkable wealth of galactic substructure that it might appropriately be called the `` field of streams '' . among the most visible of these is the whitish - blue colored , and hence relatively nearby , stellar overdensity centered at ( @xmath33 ) , found by juri et al .
( 2005 ) and named the virgo overdensity ; this is perhaps the same structure as the nearby 2mass northern fluff ( majewski et al .
parts of the monoceros ring ( newberg et al .
2002 ) are visible as the blue - colored structure at @xmath27 .
figure [ fig : saglb ] , an rgb - composite image of the sdss stars in galactic coordinates ( @xmath34 ) , also shows the arc - like structures of the monoceros ring , as predicted by the simulations of pearrubia et al .
two of the globular clusters with tidal tails previously identified in sdss data namely pal 5 ( odenkirchen et al . 2001 ) and ngc 5466 ( belokurov et al . 2006 ) can be discerned in the figures , together with their streams .
finally , a new stream is shown clearly , running from @xmath35 to @xmath36 ( @xmath37 and @xmath38 in figure [ fig : saglb ] ) .
it is distinct from the sgr stream , which it crosses ; we discuss its progenitor in a future contribution .
to analyze the three - dimensional structure of the stream , we set up a series of @xmath39 fields along branches a and b , shown as red ( for a ) and black ( for b ) squares in figure [ fig : sagnew ] .
the first three a fields actually probe both the a and b branches , which are merging at these locations .
the coordinates of the field centers are listed in table 1 . for each on - stream field
, there is a companion off - stream field of size @xmath40 , which has the same right ascension but is offset in declination as noted in table 1 .
the off - stream fields are larger , so that the background is as smooth as possible
. color - magnitude diagrams ( cmds , @xmath41 versus @xmath16 ) are constructed for each of the fields and then normalised by the number of stars .
the difference between each on - stream field and its companion off - stream field reveals the population of stars belonging to the sgr stream .
an example for the field a7 is shown in figure [ fig : cmdex](a ) , together with a one - dimensional slice at @xmath42 in figure [ fig : cmdex](b ) .
the subgiant branch is clearly visible and its location can be found to good accuracy .
in fact , we fit a gaussian to the one - dimensional slice to obtain its @xmath16 band magnitude and uncertainty . for the example in figure [ fig : cmdex](a ) , there are two subgiant branches visible in the cmd , and two distinct peaks in the luminosity functions in figure [ fig : cmdex](b ) .
these correspond to two distinct structures at different distances , possibly different wraps of the sgr stream .
although they could correspond to different populations at the same distance , this seems unlikely as the magnitude difference between the two subgiant branches changes on moving along the stream .
the sgr dsph is known to contain a variety of stellar populations with different ages and metallicities .
the most recent study of this is by @xcite , who presented a comprehensive cmd of the body of sgr . rather than attempting to fit multiple theoretical isochrones to our cmds , we show a direct comparison between @xcite s cmd and a composite cmd for the whole of the stream . the composite cmd is created by using the subgiant branch location to measure the magnitude offset .
we use fields a4 to a16 , with a16 as the reference .
the result , transformed to @xmath43 versus @xmath44 @xcite , is shown as the black contours in figure [ fig : cmdex](c ) .
this is overlaid upon the colored contours of bellazzini et al.s ( 2006 ) cmd , corrected for extinction in @xmath43 and @xmath44 by -0.12 and -0.4 magnitudes , respectively . the upper main sequence , turn - off , and subgiant branches are all well - matched .
the only significant difference occurs in the sparsely populated upper red giant branch ( the dark blue contours ) .
hence , this stream is entirely consistent with being composed of the same mix of populations as the sgr dsph .
the 0.6 magnitude offset between our composite cmd and bellazzini et al.s is a direct measure of distance .
figure [ fig : cmdex](d ) shows the @xmath16 band magnitude of the subgiant branch versus right ascension of the fields .
the triangles and stars correspond to the two structures in the a branch , the diamonds to the b branch .
only one sequence is detected in the b fields .
although its distance offset is significant only at about the 1 @xmath45 level , the b branch is systematically brighter and hence probably slightly closer . assuming the distance to the sgr dsph is @xmath2 kpc , then figure [ fig : cmdex](d ) can be converted directly into heliocentric distance versus right ascension , as shown on the right - hand axis .
the two distinct structures in the a fields are separated by distances up to @xmath46 kpc .
note that there is no evidence for any part of the sgr stream passing close to the solar neighbourhood , as has sometimes been conjectured @xcite .
this structure of three branches two close together and one more distant is understandable on comparison with the numerical simulations .
for example , the effect is clearly visible in the lower panel of figure 3 of helmi ( 2004 ) , where there are two close branches representing material stripped between 3 gyr and 6.5 gyr ago , and less than 3 gyr ago .
branches a and b are therefore probably tidal debris , torn off at different times .
the older material , stripped off more than 6.5 gyr ago , lies well behind the two close branches in helmi s simulations .
this earlier wrap of the orbit may correspond to the distant structure seen behind the a branch .
we used a simple color cut @xmath1 to map out the distribution of stars in sdss dr5 .
the `` field of streams '' is an rgb - composite image composed of magnitude slices of the stellar density of these stars .
it reveals a super - abundance of galactic substructure , including the leading arm of the sgr stream , as well as a number of sibling streams ( some hitherto unknown ) .
part of the sgr stream has previously been seen in the northern hemisphere by majewski et al .
( 2003 ) . here ,
we have mapped out the continuation of the stream , as it passes by the galactic cap and returns to the galactic plane .
at least two branches of the stream labelled a and b have been identified , corresponding to material torn off at different epochs .
there is also evidence for a still more distant structure behind the a branch .
the sgr stream provides a probe of the shape of the galactic halo .
previous work has been hampered by the absence of data in the most important regions of the sky .
for example , @xcite provides simulations of the sgr stream for a range of halo shapes from extreme oblate to prolate , all of which broadly agree with the data available at that time .
our data covers the critical region where the models differ substantially .
examination of the simulations displayed in figure 2 of @xcite enables some preliminary conclusions to be drawn . in oblate haloes ,
the sgr stream is much fatter and the stars much more scattered than shown in our figure 1 . only the spherical and very mildly prolate haloes ( @xmath47 , where @xmath48 is the axis ratio in the potential ) seem reasonable matches to the data . in both these simulations
, the sgr stream shows the same bifurcation and overall morphology as in the sdss data .
we suggest that this may be a powerful discriminant of halo shape ( fellhauer et al .
2006 ) .
we thank m. bellazzini and the anonymous referee for helpful discussions .
funding for the sdss and sdss - ii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s .
department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium for the participating institutions .
the participating institutions are the american museum of natural history , astrophysical institute potsdam , university of basel , cambridge university , case western reserve university , university of chicago , drexel university , fermilab , the institute for advanced study , the japan participation group , johns hopkins university , the joint institute for nuclear astrophysics , the kavli institute for particle astrophysics and cosmology , the korean scientist group , the chinese academy of sciences ( lamost ) , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , ohio state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington .
|
we use sloan digital sky survey ( sdss ) data release 5 ( dr5 ) @xmath0 photometry to study milky way halo substructure in the area around the north galactic cap . a simple color cut ( @xmath1 )
reveals the tidal stream of the sagittarius dwarf spheroidal , as well as a number of other stellar structures in the field .
two branches ( a and b ) of the sagittarius stream are clearly visible in an rgb - composite image created from 3 magnitude slices , and there is also evidence for a still more distant wrap behind the a branch .
a comparison of these data with numerical models suggests that the shape of the galactic dark halo is close to spherical .
|
one of the most important discoveries from extragalactic observations at mid- and far - infrared wavelengths has been the identification of a class of `` luminous infrared galaxies '' ( ligs ) , objects that emit more energy in the infrared ( @xmath15500@xmath2 ) than at all other wavelengths combined ( see @xcite for a comprhensive review ) .
the first all - sky survey at far - infrared wavelengths carried out in 1983 by the _ infrared astronomical satellite _ ( _ iras _ ) resulted in the detection of tens of thousands of galaxies , the vast majority of which were too faint to have been included in previous optical catalogs .
it is now clear that part of the reason for the large number of detections is the fact that the majority of the most luminous galaxies in the universe are extremely dusty .
previous assumptions , based primarily on optical observations , about the relative distributions of different types of luminous galaxies e.g .
starbursts , seyferts , and quasi - stellar objects ( qsos)need to be revised .
galaxies bolometrically more luminous than @xmath14@xmath3 ( i.e. _ l_@xmath4 ) appear to be heavily obscured by dust .
although luminous infrared galaxies ( hearafter ligs : _
l_@xmath5 ) are relatively rare objects , reasonable assumptions about the lifetime of the infrared phase suggest that a substantial fraction of all galaxies with _
@xmath7___l_@xmath8 pass through such a stage of intense infrared emission @xcite . _
_ l@xmath8 ) , infrared galaxies selected from the iras survey outnumber optically selected seyferts and quasars .
for references see @xcite . ]
a comparison of the luminosity function of infrared bright galaxies with other classes of extragalactic objects in the local universe is shown in figure [ fig1 ] . at luminosities below @xmath9
_ l_@xmath8 , _ iras _ observations confirm that the majority of optically selected objects are relatively weak far - infrared emitters .
surveys of markarian galaxies confirm that both markarian starbursts and seyferts have properties ( e.g. @xmath10 and @xmath11 ratios ) closer to infrared selected samples as does the subclass of optically selected interacting galaxies .
however because the most luminous galaxies are enshrouded in dust , relatively few objects in optically selected samples are found with @xmath12__l@xmath8 . for a 60@xmath13 m sample of infrared galaxies .
( insert ) examples of the subset ( @xmath115% ) of uligs with `` warm '' infrared color ( f@xmath14/ f@xmath15 @xmath160.3 ) .
three objects ( 1the powerful wolf - rayet galaxy iras010022238 , 2the `` infrared qso '' iras07598 + 6508 , 3the optically selected qso izw1 ) are shown in the inset .
for references see @xcite . ]
the high luminosity tail of the infrared galaxy luminosity function is clearly in excess of what is expected from the schechter function . for @xmath17_l_@xmath8 , ligs are as numerous as markarian seyferts and @xmath13 times more numerous than markarian starbursts .
ultraluminous infrared galaxies ( hereafter uligs : @xmath18_l_@xmath8 ) appear to be @xmath12 times more numerous than optically selected qsos , the only other previously known population of objects with comparable bolometric luminosities .
m ) of the nuclear regions of ngc4038/39 , ngc7252 , and iras192547245 , and in the _ r_-band ( 0.65@xmath13 m ) of arp220 .
white contours represent the co(1@xmath190 ) line integrated intensity as measured by the ovro millimeter - wave interferometer .
no hi or co interferometer data are available for the southern hemisphere object iras192547245 .
the scale bar represents 10kpc . ]
although ligs comprise the dominant population of extragalactic objects at @xmath20_l_@xmath8 , they are still relatively rare . for example , figure [ fig1 ] suggests that only one object with @xmath21_l_@xmath8 will be found out to a redshift of @xmath10.033 , and indeed , arp220 ( @xmath22 ) is the only ulig within this volume . the total infrared luminosity from ligs in the _ iras _ bright galaxy survey ( bgs ) is only @xmath16% of the infrared emission in the local universe @xcite .
there are preliminary indications that uligs have been more numerous in the past .
comparison of the space density of nearby uligs with the more distant population provides evidence for possible strong evolution in the luminosity function at the highest infrared luminosities .
assuming pure density evolution of the form @xmath23 , @xcite found @xmath243 for a complete flux - limited sample of uligs . , countours )
image of the antennae galaxies obtained with the infrared space observatory , on the composite optical image with v ( 5252 ) and i ( 8269 ) filters recovered from the hubble space telescope archive .
about half of the mid - infrared emission from the gas and dust that is being heated by recently formed massive stars comes from an off - nuclear region that is clearly displaced from the most prominent dark lanes seen in the optical .
the brightest mid - infrared emission comes from a region that is relativelly inconspicuous at optical wavelengths .
the isocam image was made with a 1.5@xmath25 pixel field of view .
contours are 0.4 , 1 , 3 , 5 , 10 , and 15 mjy .
the lower figure shows the spectrum of the brightest mid - infrared knot and of the nuclei of ngc 4038 and ngc 4039 .
the rise of the continuum above 10 @xmath13 m and strong neiii line emission observed in the brightest mid - infrared knot indicate that the most massive stars in this system of interacting galaxies are being formed in that optically obscured region , still enshrouded in large quantities of gas and dust . ] the infrared properties for the complete _ iras _ bright galaxy sample have been summarized and combined with optical data to determine the relative luminosity output from galaxies in the local universe at wavelengths @xmath10.11000@xmath2 @xcite .
figure [ fig2 ] illustrates how the shape of the mean spectral energy distribution ( sed ) varies for galaxies with increasing total infrared luminosity .
systematic variations are observed in the mean infrared colors ; the ratio @xmath26 increases while @xmath27 decreases with increasing infrared luminosity .
figure [ fig2 ] also illustrates that the observed range of over 3 orders of magnitude in @xmath28 for infrared - selected galaxies is accompanied by less than a factor of 34 change in the optical luminosity .
@xcite showed that a small but significant fraction of uligs , those with `` warm '' ( _ f_@xmath14/_f_@xmath15 @xmath160.3 ) infrared colors , have seds with mid - infrared emission ( @xmath1540@xmath13 m ) over an order of magnitude stronger than the larger fraction of `` cooler '' uligs .
these warm galaxies ( figure [ fig2 ] insert ) , which appear to span a wide range of classes of extragalactic objects including powerful radio galaxies ( prgs : @xmath29 ) and optically selected qsos , have been used as evidence for an evolutionary connection between uligs and qsos ( e.g. @xcite ) .
there is a strong correlation between the broad band colors ( from optical to far - infrared ) and morphological type @xcite . in particular
, the fraction of objects that are interacting / merger systems appears to increase systematically with increasing infrared luminosity .
the imaging surveys of objects in the local universe @xcite have shown that the fraction of strongly interacting / merger systems increases from @xmath110% at @xmath30 ) = 10.511 to @xmath1100% at @xmath31 ) @xmath1612 . in pannel ( c ) of figure [ fig3 ]
is shown the `` super - antennae '' , which is the prototype of ulig @xcite .
iso observations @xcite have shown that more than 98% of the mid - infrared flux from this object comes from the southern component which hosts a seyfert 2 nucleus . from the detailed studies of nearby ultraluminous infrared galaxies
the following conclusions were reached .
1 ) they are mergers of evolved gas - rich giant spiral galaxies ( e.g. milky way with andromeda ) , and not primival " galaxies .
2 ) to boost the luminosity above 10@xmath32 l@xmath33 the nuclei must have approached at least 10 kpc , namely , they are advanced mergers .
3 ) due to the gravitational impact the interstellar gas decouples from the stars and large amounts of interstellar matter fall at high rates to the central region .
this is the condition to produce a nuclear starburst , and/or feed a supermassive black hole at super - eddington accretion rates . to produce such large accretion rates , the gravitational potential wheels of massive buldges
are needed . a workshop on the question concerning the ultimate source of energy ( starbursts versus agn s ) took place in ringberg on october 1998 .
below 2 10@xmath32 l@xmath33 starbursts dominate the energy budget , but above 3 10@xmath32 l@xmath33 agn s seem to be always present and become an important source of energy . in this respect it is interesting to note that it is found with iso that in the prototype seyfert 2 galaxy ngc 1068 , about 80% of the mid - infrared flux between 4 and 18 @xmath13 m comes from the agn @xcite .
a caveat for the subject of this conference is that the pre - encounter objects that merged at high redshifts must have been different from the metal - rich evolved galaxies merging at present .
another caveat is that ultraluminous ir galaxies at high redshifts may be very difficult to detect using the lyman break technique . due to the large amounts of dust in ultraluminous objects ,
very little or none continuum leaks out at ultraviolet wavelengths .
therefore , surveys with submillimeter arrays as alma will be needed to detect ultraluminous galaxies at high redshifts .
the starbursts in ultraluminous galaxies take place in the nuclear region .
one of the new findings with iso is a class of very luminous dust - enshrouded extranuclear starbursts in nearby spiral - spiral mergers .
when the pre - encounter galaxies do not have prominent buldges , namely , when the mergers are -for instance- two sc galaxies , the most luminous starbursts may take place in extranuclear regions that are inconspicuous at optical wavelengths .
these extranuclear starbursts have sizes @xmath0 100 pc in radius and can produce up to 50% of the overall mid - infrared output from these systems .
furthermore , the analyses of the mid - infrared spectra indicate that the most massive stars in these systems are formed inside these optically invisible knots . in figure [ fig4 ]
is shown in contours the mid - infrared ( 12 - 17 @xmath13 m ) image of the antennae galaxies obtained with iso @xcite , superimposed on the optical image from hst .
below are shown representative spectra of the two nuclei and the brightest mid - infrared knot .
it shows that the most massive stars are formed in an obscured knot of 50 pc radius , which produces about 15% of the total luminosity from the antennae galaxies between 12.5 and 17 @xmath13 m .
a more extreme case is found in ngc 3690 @xcite , where it is observed an extranuclear region @xmath0100 pc in radius that radates @xmath145% of the overall mid - infrared output from this system .
if the fraction of far - infrared fluxes is the same as in the mid - infrared , such compact region produces a luminosity of 2 10@xmath34 l@xmath33 . therefore , the luminosity of a few compact starburst knots of this type would be comparable to the total bolometric luminosity of a ulig such as arp 220 ( figure [ fig3]d ) .
the multiwavelength view of this nearby sample of prototype merging systems suggests caution in deriving scenarios of early evolution of galaxies at high redshift using only observations in the narrow rest - frame ultraviolet wavelength range @xcite .
although the actual numbers of this type of systems may not be large , we must keep in mind that the most intense starbursts are enshrouded in dust and no ultraviolet light leaks out from these regions .
m emission ( dark structure ; @xcite ) and vla 20 cm continuum in contours @xcite , overlaid on an optical image from the palomar digital sky survey . the emission from dust with a bisymmetric morphology at
the centre is about 10 times smaller than the overall size of the shell structure in the elliptical and lies on a plane that is almost parallel to the minor axis of its giant host .
whereas the gas associated to the spiral rotates with a maximum radial velocity of 250 km s@xmath35 , the ellipsoidal stellar component rotates slowly approximately perpendicular to the dust lane @xcite .
the synchrotron radio jets shown in this figure correspond to the inner structure of a double lobe radio source that extends up to 5@xmath36 ( @xmath1 300 kpc ) on the sky .
the jets are believed to be powered by a massive black hole located at the common dynamic center of the elliptical and spiral structures . ]
giant radio galaxies are thought to be massive ellipticals powered by accretion of interstellar matter onto a supermassive black hole .
interactions with gas rich galaxies may provide the interstellar matter to feed the active galactic nucleus ( agn ) . to power radio lobes that extend up to distances of hundreds of kiloparsecs , gas has to be funneled from kiloparsec size scales down to the agn at rates of @xmath11 m@xmath33 yr@xmath35 during @xmath3710@xmath38 years .
therefore , large and massive quasi - stable structures of gas and dust should exist in the deep interior of the giant elliptical hosts of double lobe radio galaxies .
recent mid - infrared observations with iso revealed for the first time a bisymmetric spiral structure with the dimensions of a small galaxy at the centre of centaurus a @xcite .
the spiral was formed out of the tidal debris of accreted gas - rich object(s ) and has a dust morphology that is remarkably similar to that found in barred spiral galaxies ( see figure [ fig5 ] ) .
the observations of the closest agn to earth suggest that the dusty hosts of giant radio galaxies like cena , are symbiotic " galaxies composed of a barred spiral inside an elliptical , where the bar serves to funnel gas toward the agn .
the barred spiral at the centre of cena has dimensions comparable to that of the small local group galaxy messier 33 .
it lies on a plane that is almost parallel to the minor axis of the giant elliptical .
whereas the spiral rotates with maximum radial velocities of @xmath1250 km s@xmath35 , the ellipsoidal stellar component seems to rotate slowly ( maximum line - of - sight velocity is @xmath140 km s@xmath35 ) approximately perpendicular to the dust lane .
the genesis , morphology , and dynamics of the spiral formed at the centre of cena are determined by the gravitational potential of the elliptical , much as a usual spiral with its dark matter halo . on the other hand ,
the agn that powers the radio jets is fed by gas funneled to the center via the bar structure of the spiral .
the spatial co - existence and intimate association between these two distinct and dissimilar systems suggest that cen a is the result from a cosmic symbiosis .
in disk - disk collisions of galaxies , dynamical friction and subsequent relaxation may produce a mass distribution similar to that in classic elliptical galaxies . from the relative numbers of mergers and ellipticals in the new general catalogue @xcite estimated that a large fraction of ellipticals could be formed via merging .
the first direct observational evidence for the transition from a disk - disk merger toward an elliptical was presented in the optical study of ngc7252 by @xcite .
the brightness distribution over most of the main body of this galaxy which is shown in figure [ fig3]c is closely approximated by a de vacouleurs ( @xmath39 ) profile .
however , ngc7252 still contains large amounts of interstellar gas and exhibits a pair of prominent tidal tails ( see figure [ fig3]b ) ; neither property is typical of ellipticals .
near - infrared images are less affected by dust extinction and also provide a better probe of the older stellar population , which contains most of the disk mass and therefore determines the gravitational potential .
_ k_-band images of six mergers by @xcite showed that the infrared radial brightness profiles for two ligs arp220 and ngc2623follow an _
r_@xmath40 law over most of the observable disks . among eight merger remnants
, @xcite found _
k_-band brightness profiles for four objects that were well fitted by an _
r_@xmath40 law over most of the observable disks .
@xcite finds a similar proportion ( @xmath150% ) of uligs whose _
k_-band profiles are well fit by a _ r_@xmath40 law .
more recently , @xcite have proposed that uligs are elliptical galaxies forming by merger - induced dissipative collapse .
the extremely large central gas densities ( @xmath1@xmath41@xmath42 ) observed in many nearby uligs , and the large stellar velocity dispersions found in the nuclei of arp220 and ngc6240 are comparable to the stellar densities and velocity dispersions respectively , in the central compact cores of ellipticals . despite the _ k_-band and co evidence that ligs may be forming ellipticals
, we still need to account for two important additional properties of ellipticals : 1 . the large population of globular clusters in the extended halos of elliptical galaxies , which can not be accounted for by the sum of globulars in two preexisting spirals , and 2 .
the need to remove the large amounts of cold gas and dust present in infrared - luminous mergers in order to approximate the relative gas - poor properties of ellipticals .
these two issues have been discussed by @xcite .
0 ) line integrated intensity as measured by the iram millimeter - wave interferometer .
co emission has not been detected in ngc5291a / b .
the scale bar represents 20kpc .
collisions between giant disk galaxies may trigger the formation of dwarf galaxies .
this idea , which was first proposed by @xcite and later by @xcite , has received recent observational support @xcite . renewed interest in this phenomenon arose from the inspection of the optical images of uligs , which frequently exhibit patches of optically emitting material along the tidal tails ( see figure [ fig3]a c ) .
these objects appear to become bluer near the tips of the tails at the position of massive clouds of hi .
these condensations have a wide range of absolute magnitudes , _ m_@xmath43 @xmath1@xmath44 to @xmath45 , and hi masses , @xmath46 to @xmath47_m_@xmath8 . @xcite
have shown that objects resembling irregular dwarfs , blue compacts , and irregulars of magellanic type are formed in the tails .
these small galaxies of tidal origin are likely to become detached systems , namely , isolated dwarf galaxies . because the matter out of which they are formed has been removed from the outer parts of giant disk galaxies , the tidal dwarfs we observe forming today have a metallicity of about one third solar @xcite .
it is interesting that in these recycled galaxies of tidal origin there is as in globular clusters no compelling evidence for dark matter @xcite . to find the true fraction of dwarf galaxies that may have been formed by processes similar to the tidal interactions we observe today between giant spiral galaxies , more extensive observations of interacting systems
will be needed .
a recent step forward is the statistical finding that perhaps as much as one half of the dwarf population in groups is the product of interactions among the parent galaxies @xcite .
tidal dwarfs are formed not only during spiral - spiral mergers , but also in encounters of spirals with massive ellipticals in clusters of galaxies . in figure
[ fig6 ] are shown the results from the multiwavelength study of arp 105 and ngc 5291a / b which are in clusters of galaxies . in arp 105 ,
@xcite find tidal dwarfs that resemble magellanic irregulars and a blue compact . in ngc 5291 ,
about 10 tidal dwarfs of irregular morphology are found associated to the 200 kpc hi ring shown in figure [ fig6 ] @xcite .
\1 ) scenarios on the history of star formation that use only observations in the uv and optical rest - frames result in luminosity functions that are strongly biased in the high luminosity end .
\2 ) the most luminous nuclear and off - nuclear starbursts are enshrouded in dust . in merging
galaxies iso revealed off - nuclear starburst knots with sizes @xmath0100 pc that produce bolometric luminosities of up to 2 10@xmath34 l@xmath33 ( e.g. ngc 3690 ) .
a few of these starburst knots can produce the overall bolometric luminosity of an ultraluminous galaxy such as arp 220 .
\3 ) the observation with iso of the nearest agn to earth ( centaurus a ) opens the general question on whether the hosts of giant radio galaxies are symbiotic galaxies composed of spirals at the centre of giant ellipticals .
\5 ) collisions between giant disk galaxies trigger the formation of dwarf galaxies out tidal debris .
a fraction of these re - cycled galaxies become detached systems with diverse morphologies : blue compact dwarfs , dwarf irregulars , and irregulars of magellanic type .
|
ultraluminous galaxies in the local universe ( [email protected] ) emit the bulk of their energy in the mid and far - infrared .
the multiwavelength approach to these objects has shown that they are advanced mergers of gas - rich spiral galaxies .
galaxy - galaxy collisions took place on all cosmological time - scales , and nearby mergers serve as local analogs to gain insight into the physical processes that lead to the formation and trans - formation of galaxies in the more distant universe . here
i review multiwavelength observations with particular emphasis on recent results obtained with iso of mergers of massive galaxies driving the formation of : 1 ) luminous infrared galaxies , 2 ) elliptical galaxy cores , 3 ) luminous dust - enshrouded extranuclear starbursts , 4 ) symbiotic galaxies that host agns , and 5 ) tidal dwarf galaxies .
the most important implication for studies on the formation of galaxies at early cosmological timescales is that the distant analogs to the local ultraluminous infrared galaxies are invisible in the ultraviolet and optical wavelength rest - frames and should be detected as sub - millimeter sources with no optical counterparts .
|
since the first achievement of bose - einstein condensation ( bec ) in atomic gases , techniques for the production of quantum degenerate gases ( qdgs ) from laser cooled atoms have been continually improved , refined and simplified . because the phase space density achievable with laser cooling is limited by light - induced , density - dependent collisional losses and by the temperature limits of laser cooling , qdgs are typically made in a two step process .
the only exception to this is sr which has been directly laser cooled to degeneracy @xcite .
after laser cooling , the ensemble is transferred into a conservative potential formed by an off - resonant magnetic or optical field and forced evaporative cooling is performed to achieve the requisite temperatures and densities .
efficient evaporative cooling relies on rapid thermalization of the atoms in the trap and requires that the evaporation time be shorter than the lifetime of a particle in the trap .
since the speed of evaporation is limited by the elastic collision rate , @xmath0 ( where @xmath1 is the atomic density , @xmath2 is the elastic collision cross section , and @xmath3 is the average collision velocity ) , this requirement places simultaneous constraints on the minimum initial atom number loaded into the conservative trap and on the quality of the vacuum system because the density of room - temperature , background - gas particles sets an upper bound on the trap lifetime .
these two constraints are in conflict since the atom number available for the initial stage of laser cooling is proportional to the partial pressure of those atoms , and it is the presence of this atomic vapor that also ultimately limits the trap lifetime . in this work ,
we demonstrate that these requirements can be met for the production of qdgs of lithium by a magneto - optic trap loaded from a simple effusive atomic source .
this apparatus does not require a separated cold atom source such as a zeeman slower @xcite or a 2d mot @xcite or an additional 3d mot from which the atoms are transferred @xcite .
nor does it require any separation or differential pumping between the effusive atom source and the science chamber .
the result is a small and inexpensive vacuum system ideal for miniaturization .
we fully characterize our system and demonstrate with it the production of a mbec , and we use it to observe evidence of the pairing gap in a strongly interacting two - component dfg in the bec - bcs crossover regime .
.reported experimental parameters for qdg production using odts loaded directly from a mot .
@xmath4 is the total atom number including all spin states and corresponds to twice the number of feshbach molecules for those experiments that report on a molecular bec . [
cols="^,^,^,^,^,^,^,^,^ " , ] fig .
[ fig : ovencurrentdata ] shows the behavior of this oven loaded lithium mot at a field gradient of 35 g@xmath5 .
the loading rate ( @xmath6 ) , the lifetime or inverse loss rate ( @xmath7 ) of both the mot and odt , and the steady state atom number ( @xmath8 ) in the mot are shown as a function of the oven temperature .
these parameters were determined by fitting the loading curve of the mot to the solution , @xmath9 of the differential equation @xmath10 .
the steady state number is then the product of the loading and inverse loss rates , @xmath11 .
this model does not include a two - body loss term to model particle loss due to light assisted collisions between cold atoms within the mot , and therefore the loss term we report is an overestimate of the mot losses due only to collisions with the residual background vapor or atoms from the hot atomic beam .
in addition , because the li li@xmath12 collision cross section ( with one atom in the excited state ) is much larger than for ground state collisions , the loss rate for the excited state atoms in the mot will be significantly larger due to collisions with fast moving li atoms in the atomic beam .
these two effects make the mot lifetime only an estimate of the expected lifetime for the ensemble in the @xmath13 . for comparison , the inverse loss rate ( i.e. lifetime ) of a 50 @xmath14k deep @xmath13
is also provided for a few of the oven current settings .
since the mot is significantly deeper than the @xmath13 , the mot trap loss rate due to background vapor collisions is expected to be smaller @xcite .
in addition , residual losses from the @xmath13 will occur due to spontaneous emission and evaporation losses .
thus the lifetime in a shallow @xmath13 will always be an overestimate of the total background collision rate . at low oven current settings where the loss rate is dominated by collisions with atoms or molecules other than li atoms ,
the mot lifetime is , as expected , much larger than the lifetime in the @xmath13 . whereas at high oven current settings where the loss rate is primarily determined by the collisions with fast moving atoms in the li atomic beam , the mot lifetime is similar to the @xmath13 lifetime .
the longest inverse loss rate measured for the @xmath13 was on the order of 40 s ( at an oven current of 8 a and a temperature of 550 k ) corresponding to a background vapor pressure of approximately @xmath15 torr @xcite .
as the temperature of the oven increases so does the captured flux and the loss rate . at a current of @xmath16 ( corresponding to an oven temperature of 680 k ) ,
the steady state atom number is maximized .
there exists a clear tradeoff between the inverse loss rate of the lithium mot and the steady state atom number , and by running the oven at 625 k ( with @xmath17 ) instead of 680 k , the inverse loss rate can be increased by more than a factor of four with only a factor of two reduction in the trapped number .
we note that the mot performance is similar to that previously reported in ref .
the primary difference is that here the oven aperture is 30% larger and a portion of the mot light is sent counter propagating to the oven flux .
this additional slowing beam enhances the loading rate by a factor of two . in this experiment , a high power but broad linewidth laser creates an optical dipole trap ( 200 w total power ) that is loaded from the mot .
a large power and small beam waist is necessary to achieve a 3.5 mk odt depth required to capture the atoms from the mot .
the atoms are then transferred to a shallow , narrow linewidth , low power @xmath13 ( 18 w total power ) .
this is done because an eventual goal is to confine the atoms in a 1d optical lattice formed by the narrow linewidth odt .
however , we note that this procedure is not optimal in terms of the final atom number at degeneracy since the transfer between the odts is only 25% efficient .
the geometry of the high and low power odts relative to the glass cell is shown in fig .
[ fig : geometry ] . versus atom number in the mot . plotted
are the initial atom number in the high power @xmath13 at 200 w total power ( circles ) , the atom number in the high power @xmath13 after evaporation to 100 w total with 50 w per beam ( squares ) , and the atom number transferred into the low power @xmath13 ( diamonds ) for various initial mot atom numbers.,scaledwidth=49.0% ] in order to transfer the atoms into the @xmath13 , we compress and cool the atoms in the mot by increasing the axial magnetic field gradient from @xmath18 to @xmath19 g @xmath5 , lowering the intensity and shifting the frequency of both the pump and repump light to 10 mhz below resonance . during this compression and cooling phase , a @xmath13 of 200 w
total power is turned on and , in less than 10 ms , up to 10% of the @xmath20li atoms are transferred into the @xmath13 .
fine adjustments of the mot position to optimize its overlap with the @xmath13 are essential to achieving efficient transfer and are made by adding a small ( 5 to 10 g ) homogeneous magnetic field that shifts the magnetic quadrupole center during the compression phase .
this field is generated by three pairs of large ( 30 cm diameter ) compensation " coils for independent control of the field along each of the three orthogonal spatial axes .
we observe extremely rapid trap losses ( with trap lifetimes on the order of a few ms ) due to light assisted collisions and hyperfine relaxation , and we therefore optically pump to the lower hyperfine state ( @xmath21 ) during the transfer by extinguishing the repump light 400 @xmath14s before the pump light .
this procedure produces an almost equal population of the two sub - levels of the lower hyperfine state : @xmath22 and @xmath23 .
the light for the @xmath13 is derived from a multi - longitudinal mode , 100 w fiber laser ( spi lasers , sp-100c-0013 ) operating at 1090 nm with a spectral width exceeding 1 nm . the @xmath13 is comprised of two nearly co - propagating beams crossing at an angle of 14@xmath24 .
each beam has a maximum power of 100 w ( for a total power of 200 w achieved by recycling the first beam ) and is focused to a waist ( @xmath25 intensity radius ) of @xmath26 @xmath14 m and @xmath27 @xmath14 m . after the mot light is extinguished ,
the @xmath13 beam power is ramped down linearly in time to 100 w total ( 50 w per beam ) in 100 ms while applying a homogenous magnetic field of 300 g produced by the same coils used to generate the quadrupole magnetic field for the mot .
the output power of this spi laser is varied by an analog input controlling the pump diode current of the post amplifier stage .
rapid thermalization occurs because of the large collision rate between the @xmath28 and @xmath29 states at 300 g @xcite . at the end of this forced evaporation stage , there are approximately @xmath30 atoms remaining at a temperature of @xmath31 @xmath14k ( verified by a time - of - flight expansion measurement ) .
we observe thermal lensing effects which produce beam aberrations of our high power @xmath13 .
in particular , we observe the position of the beam waists of the @xmath13 arms moving at approximately 780 @xmath14 m s@xmath32 when the high power @xmath13 is initially turned on with 100 w per beam.m in the first 500 ms and 280 @xmath14 m in the next 500 ms due to thermal lensing ] these thermal effects were worse with a vacuum cell made from borosilicate glass .
for this reason , we are now using a quartz cell . not only does quartz have a higher transmission in the near ir ( e.g. at 1090 nm ) , but the thermal expansion coefficient of quartz is ten times smaller than that of borosilicate glass .
at the end of the pre - evaporation stage the atoms are transferred into a lower - power @xmath13 whose intersection is aligned to the crossing of the high - power @xmath13 .
the light for this second @xmath13 is generated by a single frequency , narrow - linewidth ( @xmath33 khz ) , 20 w fiber laser operating at 1064 nm ( ipg photonics ylr-20 - 1064-lp - sf ) .
this transfer is done to avoid ensemble heating observed in the spi laser @xmath13 and allows further forced evaporative cooling to much lower powers and thus ensemble temperatures .
in addition , as mentioned already , this narrow line - width laser trap can be used to realize an optical lattice .
the disadvantage of this step is that the transfer is not yet well optimized and we only retain 25% of the atoms .
further optimization of this transfer stage by aligning the trapping beams to be co - linear ( in order to increase the spatial overlap of the trapping potentials due to each laser ) is expected to produce significant improvement in the final atom number at the transition to quantum degeneracy .
the ipg @xmath13 is comprised of two beams in the horizontal plane and crossing at an angle of 60@xmath24 and with a total power of 18 w ( 9 w per beam ) .
the beams are focused to a waist ( @xmath25 intensity radius ) of @xmath34 @xmath14 m and @xmath35 @xmath14 m .
the polarization of the two beams ( which have the same frequency ) is chosen to lie in the horizontal plane to minimize reflections at the cell ( the cell is uncoated but the beams enter the cell close to brewster s angle ) .
unexpected heating and loss was observed when the polarization of the second beam was fixed perpendicular to the first ( i.e. perpendicular to the horizontal plane and parallel to the magnetic field axis ) .
the ipg @xmath13 beam power is controlled by a water cooled aom ( intraaction atd-1153da6 m ) .
the diffraction efficiency of the aom is 60% . due to the experimental constraints
the polarization of the ipg @xmath13 is perpendicular to the polarization that , according to the manufacturer of the aom , would lead to the diffraction efficiency of 75% .
after the transfer of atoms into the ipg @xmath13 further evaporation is done at a field specific to a given experiment ( below @xmath36 to create molecules and above @xmath36 to create a strongly interacting fermi gas ) .
it is worth noting here that the power of the ipg laser is not actively stabilized .
over the past decade , quantum degenerate fermi gases of either @xmath37li or @xmath38k atoms have enabled the experimental study of the so - called bardeen - cooper - schrieffer bose - einstein condensation crossover physics ( see @xcite and references therein ) . in the following , we present data obtained with this apparatus showing evidence of both bose einstein condensation of molecules and pairing in a strongly - interacting , degenerate fermi gas .
the goal here is not to quantitatively interpret these data , but rather to present representative data obtained with this apparatus showing signatures of these phenomena studied in detail previously .
laser power ( i.e. trap depth ) in the final forced evaporation stage at ( a ) b=800 g ( below the fr ) and ( b ) b=900 g ( above the fr )
. the number of spin - up " atoms ( atoms in the @xmath39 state ) is detected at these high magnetic fields by absorption imaging .
for the evaporation below the fr , when the temperature is well below the binding energy all of the atoms form fr molecules and the number detected in the @xmath39 state corresponds to the total fr molecule number .
the solid line indicates the expected particle number for an efficient evaporative cooling .
the dashed line indicates the fermi temperature for the @xmath39 component at each power computed from the particle number and the trap frequencies . in (
a ) the dotted line shows the critical temperature for bose einstein condensation for the molecules .
, scaledwidth=49.0% ] the trap depth is lowered using a series of linear ramps of different duration , for a total evaporation time of 4 s. we use an additional gradient " coil , concentric with the top main magnetic coil , to add a magnetic field gradient to both compensate for the residual magnetic field gradient of our main magnetic coil pair and to provide a magnetic force on the atoms equal and opposite to the gravitational force .
this additional coil is essential for the production of quantum degenerate samples as in this case the evaporation must be continued to the very lowest optical trap depths where gravitational sagging significantly compromises the confinement .
figure [ fig : evap ] shows the result of evaporation both below and above the feshbach resonance center at @xmath36 .
both the ensemble temperature and particle number are shown as a function of the trap depth during the final forced evaporation stage .
the solid line indicates the expected particle number for an efficient evaporative cooling where the ensemble temperature remains a factor of ten below the trap depth ( @xmath40 ) . for a discussion of the scaling laws for evaporative cooling in time - dependent optical traps ,
see ref .
@xcite .
li@xmath41 feshbach molecules after a 3 ms time of flight expansion at a magnetic field of @xmath42 for different ensemble temperatures , ( a ) @xmath43 nk ( @xmath44 ) , ( b ) @xmath45 nk ( @xmath46 ) , ( c ) @xmath47 nk ( @xmath48 ) , ( d ) @xmath49 nk ( @xmath50 )
. the wings of the profiles are fit to a gaussian function and the bimodal character of the density profile is evident for temperatures well below the critical temperature . for the coldest molecular cloud ( d )
we show in ( e ) the absorption images after different free expansion times .
the in - situ shape of the molecular cloud ( image taken at 0 ms ) has the anisotropic shape of the @xmath13 , and the cloud anisotropy reverses during the time - of - flight expansion .
, scaledwidth=49.0% ] for evaporation at @xmath51 g , we observe the formation of fr molecules when the temperature nears the molecule binding energy ( 250 nk ) , and at the end of the evaporation ramp , we observe @xmath52 molecules at a temperature of 70 nk . to image the molecular cloud , we reduce the magnetic field from 800 to @xmath53 and release the ensemble from the dipole trap . after some fixed time - of - flight ( tof ) expansion , we then take an absorption image of the molecular cloud . as mentioned before , the spin - up ( or spin - down ) component of this very weakly bound molecule can be imaged as if it were a free atom since the binding energy is far below the excited state width . at the end of the evaporation ramp at @xmath51
g the molecular gas is strongly interacting and deep in the hydrodynamic regime .
in particular , the molecule - molecule @xmath54-wave scattering length is predicted to be on the same order as the atom - atom scattering length ( @xmath55 @xcite ) , and the atom - atom scattering length at b=800 g is @xmath56 @xcite . in order to reduce the interaction strength during the tof expansion and
thus increase the visibility of the characteristic bimodal distribution of the mbec , the tof is performed at @xmath53 @xcite .
figure [ fig : mbec ] shows a series of absorption images and resulting horizontal density profiles of these feshbach molecules after a 3 ms tof expansion at a magnetic field of @xmath42 for different final ensemble temperatures .
the wings of the profiles are fit to a gaussian function and the bimodal character of the density profile is evident for temperatures below the critical temperature for bose einstein condensation .
for the coldest molecular cloud , we show in fig .
[ fig : mbec ] the absorption images after different free expansion times .
the in - situ shape of the molecular cloud ( image taken at 0 ms ) has the anisotropic shape of the @xmath13 , and the cloud anisotropy reverses during the time - of - flight expansion .
this anisotropy reversal is expected for a bec of non - interacting particles due to the larger confinement frequency along the vertical direction in these images .
however , even at the magnetic field of @xmath42 , the molecules are still strongly interacting , and the inversion of the aspect ratio of the molecular cloud can also be the result of hydrodynamics @xcite .
g. the number of atoms lost from the @xmath57 state is plotted after the application of an rf field as a function of its frequency offset from the unpaired atom resonance . in ( a ) and ( b ) where the temperature is above the pair dissociation temperature , @xmath58
, the rf - spectrum shows a single resonance .
this corresponds to the energy required to flip the spin of an unpaired atom . in ( c ) and
( d ) , the ensemble temperature is @xmath59 , and a second maximum appears at higher energy due to pair formation . here
the temperature is slightly above the critical temperature for pair condensation , and these so - called pre - formed pairs exist in what is known as the `` pseudogap '' region @xcite . in
( d ) where the temperature is lower than in ( c ) , the ratio of pairs to free atoms is larger as expected ; however , the gap energy is less because the fermi temperature in ( d ) is smaller than in ( c ) .
the lines show a fit of the data to two lorentzians and are meant only as a guide to the eye .
, scaledwidth=49.0% ] one of the first experimental signatures of fermionic pairing above the fr where the atom - atom interactions are attractive was the observation of the pairing gap in a strongly interacting fermi gas of @xmath37li atoms by radio frequency spectroscopy @xcite .
while the correct interpretation of these spectra requires a full accounting of both the final - state and trap effects to understand the contributions from the pairing - gap , pseudogap , and no - gap phases , nevertheless , it is a simple measurement that can be intuitively understood and that provides a clear signature of pair formation ( both so - called pre - formed " and condensed pairs ) @xcite .
here we perform the final evaporation stage at @xmath60 g , just above the @xmath61 to different final trap depths producing different final ensemble temperatures . at the end of this evaporation
, we apply radio frequency ( rf ) radiation for 1 second that transfers atoms in state @xmath57 to the unoccupied @xmath62 state .
we monitor the loss of atoms from state @xmath57 by state - selective absorption imaging , and in fig .
[ fig : pairing_gap ] we plot the loss as a function of the rf radiation frequency detuning @xmath63 from the free atom resonance at 81.34 mhz ( corresponding to the energy splitting between the @xmath57 and @xmath62 states at this magnetic field including the difference in the mean field interaction energies ) .
we note that when the loss of atoms from state @xmath39 is monitored instead of the loss from state @xmath57 the spectrum is the same .
this is likely due to the loss of atoms in state @xmath39 due to the collisional instability of these mixtures when @xmath62 state atoms are present @xcite . in fig .
[ fig : pairing_gap](a ) and ( b ) where the temperature is above the pair dissociation temperature , @xmath58 , the rf - spectrum shows a single resonance peak at a frequency offset of zero .
this corresponds to the energy required to flip the spin of an unpaired atom . in ( c ) and
( d ) , the ensemble temperature is @xmath59 , and a second maximum appears at higher energy
@xmath57 atoms undergo the transition at a zero offset whereas an additional energy due , in part to the pairing gap , must be added to flip those bound to a @xmath39 state atom as a pair .
here the temperature is slightly above the critical temperature for pair condensation , and these so - called pre - formed pairs exist in what is known as the `` pseudogap '' region @xcite . in fig .
[ fig : pairing_gap](d ) where the temperature is lower than in ( c ) , the ratio of pairs to free atoms is larger as expected ; however , the gap energy is less because the fermi temperature in ( d ) is smaller than in ( c ) .
in addition , the unpaired spin flip energy is shifted to a negative offset in ( c ) and ( d ) .
this is because we produce a colder ensemble by evaporating to a lower trap depth and this changes the atomic density producing a different differential mean field interaction energy for the unpaired atom transition @xcite .
in summary , we demonstrate that the conditions for achieving a quantum degenerate gas can be met for @xmath37li with a simple oven - loaded mot . our apparatus does not require a zeeman slower or a 2d mot nor does it require any separation or differential pumping between the effusive atom source and the science chamber .
the result is a simple , inexpensive , and compact vacuum system ideal for miniaturization .
we transfer @xmath64 @xmath37li atoms into a @xmath13 from a mot of @xmath65 atoms accumulated from the atomic beam flux in 20 s. we demonstrate with this system the production of a mbec , and we use it to observe evidence of the pairing gap in a strongly interacting two - component dfg in the bec - bcs crossover regime .
the key results of this work are that ( 1 ) a sufficient number of @xmath37li atoms for evaporative cooling to degeneracy in a @xmath13 can be collected in a mot directly from an effusive atomic source and ( 2 ) the vacuum limited lifetime imposed by this atom source can be exceedingly long when properly degassed .
the authors gratefully acknowledge florian schreck for helpful advice and abraham olson for useful feedback on this manuscript .
the authors also acknowledge financial support from the canadian institute for advanced research ( cifar ) , the natural sciences and engineering research council of canada ( nserc / crsng ) , and the canadian foundation for innovation ( cfi ) .
this work was done under the auspices of the center for research on ultra - cold systems ( crucs ) .
|
we report on a simple oven - loaded magneto - optical trap ( mot ) apparatus for the creation of both molecular bose - einstein condensates ( mbec ) and degenerate fermi gases ( dfgs ) of lithium .
the apparatus does not require a zeeman slower or a 2d mot nor does it require any separation or differential pumping between the effusive atom source and the science chamber .
the result is an exceedingly simple , inexpensive , and compact vacuum system ideal for miniaturization .
we discuss our work in the context of other realizations of quantum degenerate gases by evaporation in optical dipole traps and illustrate that our apparatus meets the key requirements of atom number and trap lifetime . with this system ,
we also demonstrate the production of a mbec , and we use it to observe the pairing gap of a strongly interacting two - component dfg in the bec - bcs crossover regime .
|
financial markets can be regarded as a non - equilibrium open system .
understanding how they work remains a great challenge to researchers in finance , economics , and statistical physics .
fluctuations in financial market prices are sometimes driven by endogenous forces and sometimes by exogenous forces .
business news is a typical example of exogenous forces .
casual observation indicates that stock prices respond to news articles reporting on new developments concerning companies circumstances .
market reactions to news have been extensively studied by researchers in several different fields @xcite , with some researchers attempting to construct models that capture static and/or dynamic responses to endogenous and exogenous shocks @xcite .
the starting point for neoclassical financial economists typically is what they refer to as the `` efficient market hypothesis , '' which implies that stock prices respond at the very moment that news is delivered to market participants .
a number of empirical studies have attempted to identify such an immediate price response to news but have found little evidence supporting the efficient market hypothesis @xcite .
investors seek to forecast what will happen in the near future , and buy and sell securities based on such forecasts . through this process , some newsworthy developments are factored into market prices before they occur , so that stock prices do not respond at all when they are reported @xcite .
this means that it is important for researchers to distinguish between anticipated and unanticipated news and focus only on unanticipated news in detecting the immediate response to news . to do this
, we need to measure the extent to which a news article is novel to market participants , which is the first issue we will discuss in this paper . on the other hand ,
even if a particular piece of news is unanticipated , market responses differ depending on the importance of that piece of news to market participants .
specifically , it has been shown that market reaction to news differs depending on how it is interpreted by market participants @xcite , on how it is reported by the media ( i.e. , whether it is reported in a pessimistic or an optimistic context ) @xcite , and on how many times the same news item is reported @xcite .
it has also been shown that transaction volumes tend to be greater for stocks with a larger number of searches on the internet @xcite .
all of these pieces of evidence suggest that we need to distinguish news that attract a lot of attention from market participants and news that receive little attention , and focus on news attracting a lot of attention in assessing the market response to such news .
this means that we need to measure the extent to which a news item attracts attention from market participants , which is the second issue we will discuss in this paper .
our approach to measure the novelty and topicality of news is closely related to recent studies on the application of text mining techniques to the analysis of financial market activities .
specifically , it has been shown that linguistic and statistical characteristics of news articles extracted using text mining techniques contain useful information to predict future stock prices and trading volumes @xcite .
also , in the context of information filtering , several new methods for detecting and eliminating redundant text in blogs and on twitter have been developed and applied to the identification of the novelty content of social networking service ( sns ) texts @xcite .
our paper is most closely related to studies by the thomson reuters corporation , which propose to measure the novelty of news by counting the number of linguistically similar news articles that are found in a particular time period in their news products @xcite .
based on this method , it was shown that financial market activities respond more strongly to follow - up news than to initial news @xcite .
another study closely related to ours is ref @xcite , which attempts to measure the importance of a news article by counting the number of retweets of a tweet mentioning the article @xcite . in this paper
, we measure the novelty of a news article by comparing it with other news articles reported before that article in terms of linguistic similarity : the article is regarded as novel if there was no linguistically similar news article before it .
this approach is almost the same as that taken in previous studies . on the other hand
, we say a news item attracts a lot of attention and thus is highly topical if it is simultaneously reported by multiple news agencies and read by many investors who acquire news from those agencies .
the topicality measure for a news article is obtained by counting the number of news articles which have a similar content to the original news article but are delivered by other news agencies .
this measure is similar to the measure proposed by ref @xcite but differs from it in that our measure is able to capture the extent to which a news article is topical immediately after it is delivered , while the measure proposed by ref @xcite does not work that quickly because the number of retweets of a tweet mentioning the article increases only gradually . to check the performance of the indicators ,
we empirically examine how they are correlated with intraday financial market indicators such as the number of transactions and price volatility .
specifically , we use a dataset consisting of over 90 million business news articles reported in english and a dataset consisting of minute - by - minute stock prices on the new york stock exchange ( nyse ) and the nasdaq stock market from 2003 to 2014 , and show that stock prices and transaction volumes exhibited a significant response to a news article when it is novel and topical .
the rest of the paper is organized as follows .
we first provide a detailed description of our dataset containing over 90 million english - language business news articles , and show that breaking news have much more impact on stock prices and transaction volume than other news .
next , we examine the statistical laws regarding linguistic similarity among news articles , and propose a measure for the novelty of a news article as well as a measure for the topicality of an article .
we then examine how these indicators are correlated with intraday financial market indicators .
the thomson reuters corporation ( rtrs ) and the dow jones & company inc .
( dj ) deliver news to market participants around the world within fractions of a second through electronic systems @xcite .
news items published by over 300 third parties are displayed on rtrs s electronic trading platform . in this paper
, we use only english - language news articles published by rtrs , the business wire news service ( bsw ) , the canada newswire news service ( cnw ) , marketwire ( mkw ) , the pr newswire news service ( prn ) , and market news publishing inc .
( vmn ) on rtrs s platform as well as all of the english - language news articles by dj from 2003 to 2014 .
the total number of news articles exceeds 90 million .
journalists include keywords in their articles on rtrs s platform .
for example , news articles for general motors company , llc have a keyword , gm.n , where .n means the new york stock exchange ( nyse ) .
there are three types of news events on rtrs s platform .
alert articles , which provide a one - line summary of breaking news , are displayed in red .
headline articles provide a one - line summary of non - breaking news .
an alert and a headline are up to 80 - 100 characters long .
a story shows a complete news article .
the percentages of alerts and headlines in our dataset are about 12% and 42% , respectively . on the other hand ,
dj s news also has keywords like gm . in this paper
, we use the alerts , the headlines , and the titles of dj s news . , news articles with keyword `` gm.n '' are displayed on rtrs s electronic trading platform .
upper right side figure shows semilogarithmic graph .
solid and dashed lines show market response to alert and headline , respectively .
black lines follow exponential function ( @xmath0).,width=240 ] [ cols="<,<,<,<,<,<,<,<,<,<,<,<",options="header " , ] [ tabl_2 ]
using novelty @xmath1 and topicality @xmath2 of news , we investigate the intraday market reactions to both novel and topical news .
[ fig_5 ] shows the market activities ( i.e. , volatility @xmath3 , number of transactions @xmath4 , and transaction volume @xmath5 defined by eqs .
( 2)-(4 ) ) of aapl stock before and after alert with `` aapl.o '' was reported . when @xmath6 , market activities sharply increased just after the alert was reported at time lag @xmath7 . when @xmath8 , the market has already responded to the previous alerts and headlines before additional current alert occurs .
we investigate the relationship between market reaction and topicality @xmath2 of news ( i.e. , alert and headline ) . as shown in fig .
[ fig_6 ] , when @xmath9 , the market responds greatly ; when @xmath10 , the market tends to avoid responding to the news article .
the size of the response three minutes just after the news article was reported ( i.e. , @xmath11 , @xmath12 , @xmath13 ) is proportional to its news topicality @xmath2 .
is above average .
dashed lines show market activities when novelty of alert @xmath1 is below average.,width=240 ] is above average .
dashed lines show market activities when topicality of news @xmath2 is below average .
( b ) relationship between topicality and mean of market activity for three minutes just after news was reported for gm , ibm , pfe , aapl , yhoo . for ibm , pfe , aapl , yhoo ,
market activity expresses mean of number of transaction , volume , and volatility.,title="fig:",width=240 ] is above average .
dashed lines show market activities when topicality of news @xmath2 is below average .
( b ) relationship between topicality and mean of market activity for three minutes just after news was reported for gm , ibm , pfe , aapl , yhoo . for ibm , pfe , aapl , yhoo ,
market activity expresses mean of number of transaction , volume , and volatility.,title="fig:",width=240 ]
we observed that the stock market strongly responds to the alerts that were displayed on rtrs s electronic trading platform . on the other hand
, none of the stocks greatly responded to the headlines through which most news articles are reported .
these results suggest that we need to measure the importance of news to predict market responses to it . in this paper
, we focused on an indicator to measure the degree to which a particular news article is novel , as well as an indicator to measure the degree to which a particular news article acquires attention from investors .
the novelty measure is obtained by comparing a news article with other news articles reported before that article in terms of linguistic similarity . on the other hand ,
we say a news article attracts a lot of attention and thus is highly topical if it is simultaneously reported by other news agencies and read by many investors who acquire news from those agencies .
the topicality measure for a news article is obtained by counting the number of news articles which have similar content to the original news article but are delivered by other news agencies . in order to check whether novel or topical news articles are caught by these indicators
, we observed that the novelty of follow - up news is lower than that of initial news and confirmed that the topicality of alert exceeds headline .
we found the characteristics of intraday market reactions to both novel and topical news .
for a news article with high novelty , market activities ( i.e. , number of transactions , volume , volatility ) sharply increased just after the news article was reported . on the other hand , for a news article with low novelty , market activities
have already increased based on past similar news before the news article was reported .
the increase of market activities based on news is proportional to its topicality . by these results
, we can empirically relate price movements to particular news to find convincing supportive evidence for efficient market hypothesis .
exogenous shocks often trigger or burst financial bubbles .
future work will investigate the characteristics of novel and topical news that cause bubbles to burst .
this work was partially supported by jsps kakenhi grant number 24710156 .
99 t. ito and v. v. roley , `` news from the us and japan : which moves the yen / dollar exchange rate ? , '' _ journal of monetary economics _ , vol . 19 , pp . 255 - 277 , 1987 . w. s. chan , `` stock price reaction to news and no - news : drift and reversal after headlines , '' _ journal of financial economics _ , vol . 70 , pp . 223 - 260 , 2003 . c. vega , `` stock price reaction to public and private information , '' _ journal of financial economics _ , vol . 82 , pp . 103 - 133 , 2006 . s. dellavigna and j. m. pollet , `` investor inattention and friday earnings announcements , '' _ the journal of finance _ , vol . 64 , pp . 709 - 749 , 2009 .
a. m. petersen , f. wang , s. havlin , and h. e. stanley , `` market dynamics immediately before and after financial shocks : quantifying the omori , productivity , and bath laws , '' _ phys .
82 , 036114 , 2010 .
a. m. petersen , f. wang , s. havlin , and h. e. stanley , `` quantitative law describing market dynamics before and after interest - rate change , '' _ phys .
81 , 066121 , 2010 . g. mitra and l. mitra , eds .
, _ the handbook of news analytics in finance_. wiley , 2011
. j. g. rangel , `` macroeconomic news , announcements , and stock market jump intensity dynamics , '' _ journal of banking & finance _ , vol .
35 , pp . 1263 - 1276 , 2011 .
j. e. engelberg and c. a. parsons , `` the causal impact of media in financial markets , '' _ the journal of finance _ , vol .
66 , pp .
67 - 97 , 2011 .
t. mizuno , k. takei , t. ohnishi , and t. watanabe , `` temporal and cross correlations in business news , '' _ progress of theoretical physics supplement _
194 , pp .
181 - 192 , 2012 .
j. elder , h. miao , and s. ramchander , `` jumps in oil prices : the role of economic news , '' _ energy journal _ , vol .
34 , pp . 217 - 237 , 2013 . l. a. smales , `` non - scheduled news arrival and high - frequency stock market dynamics : evidence from the australian securities exchange , '' _ research in international business and finance_. vol . 32 , pp . 122 - 138 , 2014 .
a. storkenmaier , _ financial markets and public information_. kit scientific publishing , 2014
. v. filimonov and d. sornette , `` quantifying reflexivity in financial markets : toward a prediction of flash crashes , '' _ phys .
e _ , vol .
85 , 056108 , 2012 .
v. filimonov , s. wheatley , and d. sornette , `` effective measure of endogeneity for the autoregressive conditional duration point processes via mapping to the self - excited hawkes process , '' _ communications in nonlinear science and numerical simulation _ , vol .
22 , pp .
23 - 37 , 2015 .
d. cutler , j. poterba , and l. summers , `` what moves stock prices ? '' _ journal of portfolio management _ , vol .
15 , pp . 4 - 12 , 1989 . g. mcqueen and v. v. roley , `` stock prices , news , and business conditions , '' _ review of fin studies _ , vol . 6 , pp . 683 - 707 , 1993
. m. j. fleming and e. m. remolona , `` what moves the bond market , '' _ economic policy review _ ,
vol . 3 , pp . 31 - 50 , 1997 .
r. fair , `` events that shook the market , '' _ journal of business _ , vol . 75 , pp .
713 - 731 , 2002 .
a. joulin , a. lefevre , d. grunberg , and j. p. bouchaud , `` stock price jumps : news and volume play a minor role , '' _ e - print arxiv_:0803.1769 , 2008 . o. erdogan and a. yezege , `` the news of no news in stock markets , '' _ quantitative finance _ , vol .
897 - 909 , 2009 .
a. n. bomfim , `` pre - announcement effects , news effects , and volatility : monetary policy and the stock market , '' _ journal of banking & finance _ , vol . 27 , pp . 133 - 151 , 2003 .
g. birz and j. r. lott jr .
, `` the effect of macroeconomic news on stock returns : new evidence from newspaper coverage , '' _ journal of banking & finance _ , vol .
35 , pp . 2791 - 2800 , 2011 .
p. c. tetlock ,
`` giving content to investor sentiment : the role of media in the stock market , '' _ the journal of finance _ , vol .
62 , pp . 1139 - 1168 , 2007 .
f. lillo , s. micciche , m. tumminello , j. piilo , and r. n. mantegna , `` how news affect the trading behavior of different categories of investors in a financial market , '' _ quantitative finance _ , vol .
15 , pp .
213 - 229 , 2014 .
i. bordino , n. kourtellis , n. laptev , and y. billawala , `` stock trade volume prediction with yahoo finance user browsing behavior , '' _ in data engineering ( icde ) , 2014 ieee 30th international conference on _ , pp . 1168 - 1173 , 2014 .
r. p. schumaker and h. chen , `` textual analysis of stock market prediction using breaking financial news : the azfin text system , '' _ acm transactions on information systems ( tois ) 27 _ , article no . 12 , 2009 .
j. bollen , h. mao , and x. zeng , `` twitter mood predicts the stock market , '' _ journal of computational science _ , vol . 2 , pp . 1 - 8 , 2011 . g. m. mian and s. sankaraguruswamy , `` investor sentiment and stock market response to earnings news , '' _ the accounting review _ ,
87 , pp .
1357 - 1384 , 2012 .
r. hisano , d. sornette , t. mizuno , t. ohnishi , and t. watanabe , `` high quality topic extraction from business news explains abnormal financial market volatility , '' _ plos one _ , vol . 8 , e64846 , 2013 .
g. ranco , i. bordino , g. bormetti , g. caldarelli , f. lillo , and m. treccani , `` coupling news sentiment with web browsing data predicts intra - day stock prices , '' _ _ e - print arxiv:__1412.3948 , 2014 .
r. luss and a. daspremont , `` predicting abnormal returns from news using text classification , '' _ quantitative finance _ , vol .
15 , pp .
999 - 1012 , 2015 .
y. zhang , j. callan , and t. minka , `` novelty and redundancy detection in adaptive filtering , '' _ proceedings of the 25th annual international acm sigir conference on research and development in information retrieval _ , pp .
81 - 88 , august , 2002 .
e. gabrilovich , s. dumais , and e. horvitz , `` newsjunkie : providing personalized newsfeeds via analysis of information novelty , '' _ proceedings of the 13th international conference on world wide web _ , pp . 482 - 490 , 2004 .
l. zhao , m. zhang , and s. ma , `` the nature of novelty detection , '' _ information retrieval _ , vol . 9 , pp .
521 - 541 , 2006 .
k. w. ng , f. s. tsai , l. chen , and k. c. goh , `` novelty detection for text documents using named entity recognition , '' _ in information , communications & signal processing , 2007 6th international conference on _ , pp . 1 - 5 , 2007 . h. liang , f. s. tsai , and a. t. kwee , `` detecting novel business blogs , '' _ in information , communications and signal processing , 2009 .
icics 2009 .
7th international conference on _ , pp . 1 - 5 , 2009 .
thomson reuters , `` reuters newsscope sentiment engine : guide to sample data and system overview , '' _ thomson reuters white paper _
thomson reuters , `` reuters newsscope sentiment engine : output image and file format , '' _ thomson reuters white paper _ , 2008 .
a. gross - klussmann and n. hautsch , `` when machines read the news : using automated text analytics to quantify high frequency news - implied market reactions , '' _ journal of empirical finance _ , vol .
18 , pp .
321 - 340 , 2011 .
s. phuvipadawat and t. murata , `` breaking news detection and tracking in twitter , '' _ in web intelligence and intelligent agent technology ( wi - iat ) , 2010 ieee / wic / acm international conference on _ , vol . 3 , pp . 120 - 123 , 2010 .
thomson reuters eikon .
[ online ] .
available : http://financial.thomsonreuters.com/en/products/tools-applications/trading-investment-tools/eikon-trading-software.html .
( accessed : 1st june 2015 ) dow jones factiva .
[ online ] .
available : http://new.dowjones.com / products / factiva/. ( accessed : 1st june 2015 )
|
we propose an indicator to measure the degree to which a particular news article is novel , as well as an indicator to measure the degree to which a particular news item attracts attention from investors .
the novelty measure is obtained by comparing the extent to which a particular news article is similar to earlier news articles , and an article is regarded as novel if there was no similar article before it . on the other hand ,
we say a news item receives a lot of attention and thus is highly topical if it is simultaneously reported by many news agencies and read by many investors who receive news from those agencies .
the topicality measure for a news item is obtained by counting the number of news articles whose content is similar to an original news article but which are delivered by other news agencies . to check the performance of the indicators ,
we empirically examine how these indicators are correlated with intraday financial market indicators such as the number of transactions and price volatility .
specifically , we use a dataset consisting of over 90 million business news articles reported in english and a dataset consisting of minute - by - minute stock prices on the new york stock exchange and the nasdaq stock market from 2003 to 2014 , and show that stock prices and transaction volumes exhibited a significant response to a news article when it is novel and topical .
novelty ; topicality ; exogenous shocks ; financial markets ; business news ;
|
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.